ATTD 2017 Invited Speaker Abstracts A-1

ATTD 2017 Oral Abstracts A-10

ATTD 2017 E-Poster Discussion Abstracts A-36

ATTD 2017 E-Poster Viewing Abstracts A-45

ATTD 2017 Read By Title A-127

ATTD 2017 Abstract Author Index A-134

DIABETES TECHNOLOGY & THERAPEUTICS
Volume 19, Supplement 1, 2017
Mary Ann Liebert, Inc.
DOI: 10.1089/dia.2017.2525.abstracts

ATTD 2017 ABSTRACTS

ATTD 2017 Invited Speaker Abstracts

001 The importance of inpatient glycemic control on outcomes not

only applies to patients managed with IV insulin in critical care
MOVING TO A DYNAMIC VIEW OF DIABETES: areas but also patients on general surgical and medical wards. In
A CORNERSTONE ON THE ROAD TOWARD such patients, hyperglycemia is associated with increased length
GLUCOSE CONTROL of stay, infection, mortality, and readmissions. Time to target
E. Renard1 glucose is important for success of a subcutaneous (SubQ) insulin
1 regimen in any inpatient setting. This study evaluated the effec-
Montpellier University Hospital, Department of tiveness of an Electronic Glycemic Management System (eGMS)
Endocrinology, Diabetes, Nutrition, Montpellier cedex 5, to achieve prescribed glucose target ranges on SubQ insulin
France therapy while measuring incidence and severity of hypoglycemia.
Methods: This retrospective study evaluated 5,718 hypergly-
Blood glucose is a biological parameter which is influenced by cemic patients who required SubQ insulin. They were admitted to
almost all physiological body regulators and unlimited envi- one of 7 hospital systems over 45 months from 02/2013 to 11/
ronmental impacts. In contrast to this eminently moving element, 2016. Insulin regimens were targeted to glucose ranges of 100-
from the discovery of insulin and its therapeutic use in diabetes, a 140, 120-160 or 140-180 mg/dL using the eGMS SubQ program
paradoxical mode of insulin management has been followed. GM. Primary outcome measure was mean Time-To-Target (TTT).
Punctual measurements of glucose initially in urine and since the Secondary measures included: average initial BG (SD), daily BG
1970s in blood have been used to decide which iterative doses of average, time on eGMS, hypoglycemia <40 & <70 mg/dL, and
insulin had to be delivered. Hence most of time blood glucose percent of BG readings and patient days in target once target was
levels fluctuate above or below the physiological glucose range reached.
and it can be considered as a miracle that patients may obtain an Results: Patients reached their prescribed target in 0.8 days.
acceptable life in these conditions. Average initial BG was 261.7 mg/dL (SD+/-129.2), average time
During the two last decades, the technology has made available on eGMS was 4.56 days with 67.9% of BG readings and 68.5% of
tools that fully change this paradigm. Insulin infusion from minia- patient day values remaining 70-180 mg/dL once prescribed tar-
turized pumps has first allowed dynamic, prospective or corrective get was reached. Hypoglycemia was <40 = 0.0011% & <70 mg/
adjustments of insulin delivery. Using this dynamic mode of insulin dL = 0.013%.
therapy based upon static iterative measurements of blood glucose is Conclusion: These results suggest that eGMS can achieve a
however just like driving a car with only distant occasions to see the prescribed target range quickly with limited mild or severe hy-
road. The availability of continuous glucose monitoring has pro- poglycemia.
vided the instrument to see the road. Moreover, the combination of
continuous adjustable insulin delivery and permanent monitoring of
glucose level has opened the way to efficient prediction and antici-
pation. Glucose control can then be switched from human brain to 003
control closed-loop algorithms. Hence targeted diabetes control is no
more having a good HbA1c but keeping blood glucose as much time CHALLENGES AND PROBLEMS IN INTRODUCING
as possible in a close-to-normal range. Because blood glucose is a NEW TECHNOLOGIES INTO CLINICAL PRACTICE
dynamic parameter, a dynamic view of glucose status and a dynamic
G. Grunberger1
management of insulin therapy are needed. Today, the request of
1
patients with diabetes to the health insurances can copy Winston Grunberger Diabetes Institute, N/A, Bloomfield Hills, USA
Churchills request to the US allies during WW2: Give us the tools,
and we will do the job! Diabetes technology has come a long way. From the initial use of
continuous subcutaneous insulin infusion via insulin pumps in late
1970s and capillary glucose monitoring introduced in early 1980s
002 we are now contemplating use of automated insulin delivery devices
COMPUTERIZED SOFTWARE USING (also known as artificial pancreas to bionic pancreas systems)
POC GLUCOSE IN THE HOSPITAL in clinical practice within the next year. Tremendous progress has
AND OUTPATIENT SETTING been made by investigators, clinicians and patients on all fronts to
enable this discussion. Insulin delivery devices have shrunk in size
B. Bode1 and grew in complexity and flexibility while blood glucose monitors
1
Atlanta Diabetes Associates, Endocrinology & Metabolism, have allowed faster, more accurate and convenient acquisition of
Atlanta, USA glucose concentration by patients. Introduction of continuous glu-
cose monitoring systems measuring glucose concentration in the

A-1
A-2 ATTD 2017 INVITED SPEAKER ABSTRACTS

interstitial fluid brought in the next stage necessary for automation dicted for prior studies. As each new study becomes available,
of insulin delivery and eventual freeing up of patients from constant the process becomes increasing difficult as the controller needs
need for finger sticks, carbohydrate counting, manual adjustments to effect acceptable control over all prior studies. The BCH-
of insulin pump delivery and fear of nocturnal hypoglycemia. While approach is compared to alternate approaches that have achieved
these exciting developments had been taking place, discussions had broad use within the AP community.
started about eventual deployment of these technological advances
into clinical practice beyond the research setting. Relevant stake-
holders have been asking essential questions: Is the technology
ready for implementation in the real world? If so, how will it be 005
done? Will any clinician be allowed to introduce it in his/her
practice setting? If not, who, when and how will train and certify the THE MOLECULAR BASIS FOR THE REMISSION
practitioner in his/her proficiency? How will their performance be OF TYPE 2 DIABETES AND OPPORTUNITIES
monitored and acted on? Who will select the appropriate patients for FOR DRUG DEVELOPMENT
this technology to optimize its promise of better glycemic control W. Pories1
while minimizing risks of intensive insulin therapy? How, when and 1
by whom will the patients be trained? How will their progress be Metabolic Surgery Research Group, Surgery, Brody School of
monitored and judged? What is and will be the role of experts and Medicine, East Carolina University, Greenbille, USA
professional societies? How will their recommendations influence Bariatric surgery has forced a re-examination of type 2 dia-
the marketplace? And, most importantly, how will be the new betes (T2D). It is no longer an incurable disease that ends, twenty
technology paid for? Who and by what parameters will decide on years after diagnosis, with renal failure, blindness and amputa-
reimbursement levels for the professionals, practice settings and tions. Operations that d.ecrease contact between food and the gut,
patients? It is not too early to start getting correct answers so the if done early enough, can produce long-term cures with decreases
advances in diabetes technology can be enjoyed by our patients. in mortality by 80%. Further, the procedures do not only reverse
T2D but also produce full and durable remission of other ex-
pressions of the metabolic syndrome such as dyslipidemias,
004 hypertension, severe obesity and polycystic ovary syndrome.
Current therapies for advancing T2D are directed at over-
HARVARDS PID APPROACH coming insulin resistance by decreasing insulin resistance, in-
G. Steil1 creasing insulin secretion or the administration of insulin. Our
1 studies suggest that these approaches are misdirected, that insulin
Boston Childrens Hospital, Division of Medicine Critical resistance may not be an appropriate target and that interfering
Care, Boston, USA with gut signaling offers additional, so far overlooked, oppor-
tunities for drug development.
Different Proportional Integral Derivative (PID) control de-
signs have produced substantially different results. In many
cases, the differences can be attributed to improvements in the
PID algorithm that are effected as new data become available, or 006
to improvements in the continuous glucose monitor (CGM) used
by the controller. In other cases, changes in the PID design have INSULIN PUMP THERAPY IN TYPE 2 DIABETES:
resulted in a worsening of control performance. In these latter AN INDIVIDUAL PATIENT DATA META-ANALYSIS
cases, it is often difficult to assess whether the decrease in per- J. Pickup1
formance is due to a change in the PID design per se, a change in
1
CGM performance, or a change in any number of indeterminate Kings College London, Diabetes & Nutritional Sciences,
factors inherent in all cross-study comparisons e.g., differ- London, United Kingdom
ences in the subjects studied, differences in the types or size of
meals consumed, presence or absence of manual meal insulin Insulin pump therapy (CSII) has an established place in the
dosing, or difference in the allowed activity level all of which treatment of selected people with type 1 diabetes but its role in
can be reasonably expected to affect Artificial Pancreas (AP) the management of type 2 diabetes is less certain. There has been
control performance. One approach to assess the impact of these equivocal evidence from some randomized controlled trials
factors is through the use of model identification and simulation. (RCTs) for the effectiveness of CSII vs. MDI in type 2 diabetes,
In this approach - used extensively at Boston Childrens Hos- but a recent large-scale, multicenter RCT indicates that and some
pital (BCH) to effect improvements in PID control design and patients who remain poorly controlled after a period of optimized
open-loop pump therapy - CGM and insulin delivery data are MDI may benefit from a switch to insulin pump therapy.
combined with meal and exercise data to identify time varying In order to test the hypothesis that HbA1c is improved on CSII
changes in insulin sensitivity and other metabolic parameters vs. MDI in type 2 diabetes, and to identify those patients with
characterizing insulins effect to lower blood glucose. Once type 2 diabetes who benefit the most from CSII, we performed an
these parameters have been obtained, and the underlying met- individual patient data meta-analysis of published RCTs in this
abolic model shown to predict the glucose profile observed in group. We searched for trials that met the inclusion criteria in
the study in question without error (statistical runs test), the PID- several data bases including Cochrane, Medline and Google. We
controller used in the study showing poor performance is re- used Bayesian meta-regression models to identify determinants
placed with a previously studied PID controller to assess if the of final HbA1c and insulin dose.
degradation in performance is related to a difference in the Five RCTs met the selection criteria, with 287 patients ran-
control design. Once this step is completed, the controller is domized to MDI and 303 to CSII. Data from all individual patients
optimized to address the problems in observed in the most recent were available. Meta-regression models and recommendations for
study while at the same time not degrade the performance pre- patients with type 2 diabetes best treated by CSII will be presented.
ATTD 2017 INVITED SPEAKER ABSTRACTS A-3

007 In several reported studies, age at surgery is *16 years (range

14-20 years), mean body mass index is >50, and females pre-
CGM IN TYPE 2 DIABETES: IS IT A REALISTIC
dominate (75%).Weight loss ranges from 20%-30%, remission of
TECHNOLOGY?
T2DM occurs in *95% in those who have this condition at
I.B. Hirsch1 baseline, and remits in *75% in those with pre-diabetes; remission
1 of elevated BP also occurs in *75%, dyslipidemia in *66%, and
University of Washington, Seattle, USA significant improvement in weight-related quality of life is reported
by most. I discuss possible mechanisms for the rapid improve-
Continuous glucose monitoring (CGM) is not nearly as mature ments seen before significant weight loss occurs. Complications
in its development in type 2 diabetes (T2D) as it is in type 1 include deficiency of essential micronutrients which occurs in
diabetes (T1D). In fact, after a decade of CGM there are only a *50% and 10%-20% require one or more additional procedures.
few studies addressing its utility in T2D. Reasons for this in- Contra-indications for surgery include an unstable home environ-
cludes a wide heterogeneity of patients, particularly in terms of ment, history of recidivism, drug or alcohol abuse, established
medication use. While it could be argued one of the mechanisms psychiatric disorders and emotional immaturity.
of CGM success is behavioral modification no matter the type of To date, follow up studies have been of short duration so
diabetes, for those not receiving insulin these alterations in longer follow up is essential before widespread application of
lifestyle are often the primary mechanism glucose control can these surgical techniques for a medical problem.
improve. Besides classifying patients on insulin-requiring or not,
CGM can also be divided as masked (professional) or real-
time (personal). Furthermore, the newer Flash Glucose Mon-
itoring (FGM) needs to be viewed separately from CGM since 009
alarms do not notify the patient about extremes in glycemia.
Studies to date, particularly randomized controlled studies for OPPORTUNITIES TO ENHANCE TECHNOLOGY USE
CGM or FGM in T2D are sparse. Nine studies of professional IN PEDIATRIC PATIENTS WITH T1D
CGM observed for 12 to 24 weeks have shown HbA1c reductions L. Laffel1
generally between 0.4 to 0.6%. However, only 3 of these trials were 1
RCTs. There also appears to be a CGM-behavioral memory in Joslin Diabetes Center, Pediatrics, Boston, USA
some with T2D. For example, in a separate 40-week study (after 12
weeks of intermittent CGM) of patients with T2D not receiving Intensive insulin therapy and use of advanced diabetes tech-
prandial insulin there was a 0.8% reduction in HbA1c compared to nologies for type 1 diabetes (T1D) management offer opportu-
0.2% in those receiving conventional finger stick blood glucose nities to optimize glycemic control in children, adolescents, and
testing (Diabetes Care 2012; 35:32-38). Other shorter studies using young adults. Despite these treatment tools, the majority of
CGM for 12 weeks have shown similar improvements in HbA1c. young patients do not achieve the recommended HbA1c targets
For FGM, a 26-week RCT reported in 2016 no improvement of of <7.5% (58 mmol/mol) for patients <18 years old and <7% (53
HbA1c but significant reductions in hypoglycemia exposure. mmol/mol) for those 18 and older; only *1/4 youth <18 reach
In conclusion, experience in CGM and FGM in T2D is target and only *1/5 aged 18-25 achieve target. To enhance
growing but not nearly what it is in T1D. The heterogeneity of the uptake and sustain use of diabetes technologies, there is need to
populations studied makes it difficult to compare studies. More understand potential barriers and burdens associated with use as
research is required to better define the utility of this technology well as a need to demonstrate and promote potential benefits of
in T2D and in particular which T2D patients would most benefit. use. Understanding and articulating the balance between barriers
Finally, robust cost effectiveness studies will be required to show and benefits will become increasingly important in the era of
benefit before widespread use can be adopted. automated glucose control with the impending availability of the
first commercial hybrid closed loop system in 2017.
There is likely a bidirectional interaction between psychoso-
cial factors and technology. Devices such as CGM can add
008 burdens to care due to the need for repeated insertions, pain/
discomfort with insertions, site care, potential skin reactions,
PROS AND CONS OF BARIATRIC SURGERY daily calibrations, concerns with accuracy, need to carry multiple
IN ADOLESCENTS WITH OBESITY AND T2DM devices, and alert/alarm fatigue, which should be counter-
M. Sperling1 balanced by the benefits of real-time glucose data to guide
1
management and provide timely alerts to patients (and family
New York, USA members connected remotely) regarding out-of-range glucose
levels.
The world-wide epidemic of obesity involves children and Pediatric patients are particularly burdened by technologies as
adolescents who manifest several endocrine (PCOS, early or past experience has shown regarding CGM; young persons may
delayed puberty, abnormal thyroid function) and biochemical initiate CGM, but many fail to sustain consistent and durable
disturbances including metabolic syndrome and T2DM, both of CGM use. Understanding burdens as well as benefits will be
which are increasing in prevalence and are associated with particularly important with automated glucose control systems in
morbidity, earlier mortality and decreased quality of life mea- order to promote use and to yield glycemic benefits for youth. In
sures. Medical treatment options, including lifestyle interven- order to enhance technology use in pediatric patients with T1D, it
tions, are limited in effectiveness. will be important to ensure provision of realistic expectations,
The seriousness of the problem has spawned surgical inter- including review of patient/family efforts required to use such
vention(s) via bariatric surgery procedures, the 2 most common systems and the limitations of device performance as well as the
being gastric sleeve and Roux en Y gastric bypass. need for ongoing support and education regarding system use.
A-4 ATTD 2017 INVITED SPEAKER ABSTRACTS

010 of dopaminergic pathways associated with motivation and re-

ward, as well as of brain regions involved in stress responses,
MAKING THE MOST OF DIABETES TECHNOLOGY
arousal and vigilance, in adults with type 1 diabetes and IAH and
K. Barnard1 people with IAH are less likely to follow treatment plans re-
1 commended by their health care professionals. They describe
Bournemouth University, Faculty of Health & Social Science, unhelpful health beliefs that may act as barriers to their en-
Bournemouth, United Kingdom gagement with hypoglycaemia avoidance. Anecdotally, these
may be associated with use of technology further to lower
Diabetes technologies represent some of the most advanced glucose or avoidance of technological solutions. Nearly 1 in 10
pieces of engineering ever created. Devices including insulin adults with type 1 diabetes may be at high risk of severe hy-
pumps (CSII), continuous glucose monitors (CGM) and automated poglycaemia but express relatively low concern about it. A pilot
insulin delivery or closed loop (CL) systems, as well as self- study of a group intervention, delivered by trained diabetes
monitoring of blood glucose (SMBG) devices with varying levels educators supported by clinical psychology, and using psy-
of functionality, are associated with improved glycaemic control chotherapeutic techniques of motivational interviewing and
and psychosocial benefits. It is also often reported that such devices cognitive behavioural theory specifically targeting the de-
contribute to relief from diabetes self-management in terms of scribed health beliefs of IAH, has shown great promise in
flexibility, freedom, greater awareness of glycaemic information helping people with treatment-resistant IAH regain awareness
for both individuals with T1D and their families / significant others and avoid severe hypoglycaemia. The programme is now the
for example with shared CGM data now possible. subject of a formal randomized controlled trial.
Both disease and disease management contribute to subopti-
mal quality of life however, with the burden of managing tech-
nologies also reported to have a negative impact on quality of 012
life. SMBG, insulin pump therapy and CGMs have all been as- 18 MONTHS USE OF CLOSED-LOOP
sociated with increased burden of disease management in the AT HOME STUDY
literature, with mixed reports of the psychosocial impact of CL
systems. Commonly believed to reduce time spent thinking about S. Garg1
diabetes, the STAR3 trial reported that CSII was associated with 1
Barbara Davis Center for Diabetes, University of Colorado
more time spent on diabetes care than multiple daily injection Denver, Department of Medicine and Pediatrics, Aurora, USA
therapy. The amount of data provided by CGMs has been re-
ported to be overwhelming, with mixed impact on psychosocial Several new technologies have emerged in the past 15 years to
functioning of users. improve diabetes outcomes as it relates to hypoglycemia (safety)
Technology in itself is insufficient to effect long-term im- and improving glucose control (A1c values). Improvements in
provements in diabetes outcomes, as can be seen in successive continuous glucose monitoring (CGM) devices have consistently
national diabetes audits. Optimal diabetes outcomes require timely shown reduction in HbA1c without increasing hypoglycemia in
education and ongoing support of personalised use of devices. insulin requiring patients with diabetes. These improvements
Ensuring robust, consistent and effective assessment of psycho- have been shown to be consistent in patients using insulin pump
social factors associated with diabetes technologies is crucial to or multiple daily injections a day (MDI). However, when CGM
understanding the barriers and facilitators of uptake and continued is used alone without automatic feedback to the insulin delivery
successful use for each individual in the context of their daily lives system, the subjects have to intervene to take the necessary steps
both bio-medically and psychologically. of either have a snack to avoid hypoglycemia or take additional
insulin to avoid impending hyperglycemia.
011 With continued improvements in the CGM technology
(MARD now consistently less than 10%) and in the new insulin
ENHANCING AWARENESS OF HYPOGLYCAEMIA - delivery systems (pumps), several studies have now incorporated
WHERE TO GO WHEN TECHNOLOGY FAILS! the two technologies together to reduce diabetes burden in pa-
S. Amiel1 tients with type 1 diabetes. Several studies on closed-loop systems
have shown significant reduction in nocturnal hypoglycemia,
1
Kings College Hospital, London, United Kingdom significant increase in time in range with a reduction of hyper-
glycemic glucose excursions. Most of these studies have been
Impaired awareness of hypoglycaemia (IAH), associated with short term lasting from a few days to three months. We report the
extensive defects in the physiological counterregulatory re- first large real-life unsupervised at-home use of the hybrid closed-
sponses to hypoglycaemia, increases risk of severe hypogly- loop system (HCL) in subjects with type 1 diabetes.
caemia 6- (in type 1 diabetes) to 17- (in insulin-treated type 2) The original cohort of 124 adolescent and adult subjects with
fold. Awareness of hypoglycaemia can be restored by strict type 1 diabetes (single-arm) after two weeks of run-in phase
avoidance of exposure to plasma glucose less than 3 mmol/l, clearly showed the HCL system to be safe in reducing hypo-
54 mg/dl, but this can be difficult to achieve and sustain, espe- glycemia and improving glucose control. The study results were
cially while maintaining target concentrations of glycated hae- recently reported in JAMA 2016 (Bergenstal, Garg, et al., JAMA
moglobin (HbA1c). For adults with type 1 diabetes, structured 2016). More than 85% of the subjects chose to continue with the
education in flexible intensive insulin therapy can lower HbA1c, HCL system which was allowed by the FDA and the sponsor. To
reduce rates of severe hypoglycaemia and, in at least one pro- everyones surprise in September 2016 the FDA approved the
gramme, restore awareness of hypoglycaemia to about 40% of first artificial pancreas (HCL) and making it an historic landmark
people entering the programme with impaired awareness. Insulin in use of technology for subjects with type 1 diabetes. It is likely
pump therapy, and sensor-augmented pump therapy also reduce to be available in the US early spring of 2017.
severe hypoglycaemia rates but have not, to date, been able to In this study, we report results of 18 subjects out of the 24 who
restore awareness. Neuroimaging has shown abnormal activation participated in the original study reported above from the Barbara
ATTD 2017 INVITED SPEAKER ABSTRACTS A-5

Davis Center for Diabetes. The data clearly indicates continued nighttime CGM values of 152 and 142 mg/dl for the hotel study.
benefits to both adolescents and adults in improving glucose con- The tuning change reduced hypoglycemia from 1.6 to 0.85
trol and reducing hypoglycemia. All subjects have been satisfied events/patient/day.
with the use of HCL and are hoping not to give up the system next Conclusions: The MMPPC system achieved a mean glucose
spring. The detailed data will be presented at the ATTD 2017. equivalent to an estimated HbA1c of 7% and may be particularly
helpful for people who have an elevated HbA1c as a result of
frequent missed meal boluses. The current MMPPC algorithm
may have a higher risk for hypoglycemia when compared to
013 algorithms using meal announcement. More research is required
IMPLANTABLE AP: BENEFITS AND CHALLENGES to develop a fully closed loop system.
OF IP-IP AP
E. Dassau1
1 015
Harvard University, Harvard John A. Paulson School
of Engineering and Applied Sciences, Cambridge, USA MULTI-WEEK OUTPATIENT STUDIES
WITH ADAPTATION
Automated insulin delivery (AID) systems have shown to be F.J. Doyle1
superior to sensor augmented pump (SAP) and continuous sub-
1
cutaneous insulin infusion (CSII) in multiple clinical studies. Harvard, School of Engineering & Applied Sciences,
However, the first generation of subcutaneous automated insulin Cambridge, USA
delivery systems such as the Medtronic 670G will most likely
have limited capability and will be highly dependent on user In this talk, I will report on a multi-site clinical trial that
involvement. One of the main limitations of the first generation studied the performance of an automated glucose control system
of AIDs is current insulins and the delivery methods. in home settings for subjects with type 1 diabetes. We investi-
In this presentation, the benefits and challenges for an im- gated an enhanced version of zone model predictive control,
plantable AID will be discussed, presenting the case for the next which included run-to-run optimizations of basal rates (BR) and
generation of AID. Designed to be user centric and with superior insulin to carbohydrate ratios (CR). Specifically, the protocol
ability to reject disturbances such as meals and exercises. included algorithmic adjustment of CRs prior to closed-loop
initiation, and continued BR and CR algorithmic optimization
during closed-loop use for a longer duration. The system was
evaluated for 15 weeks at 3 different sites (William Sansum
014 Diabetes Center, University of Virginia, and Mayo Clinic,
MULTIPLE MODEL PROBABILISTIC PREDICTIVE Rochester, MN). Thirty-two subjects (17 F) were enrolled and 29
CONTROL: IS A PRE-MEAL BOLUS NECESSARY? subjects completed the protocol. The primary outcome measure
was the change in Hemoglobin A1c (HbA1c) (%) from baseline
D. Maahs1, F. Cameron2, L. Trang3, B. Buckingham3, C. Levy3, (week 2) to the end of week 15 (end of study), with A1c measured
G. Forlenza4, D. Lam3, P. Clinton1, L. Messer4, E. Westfall4, at weeks 7 and 11 as well for repeated measures.
C. Levister3, Y.Y. Xie3, N. Baysal2, D. Howsmon2, S.D. Patek5,
W. Bequette2
1
Stanford University, Pediatric Endocrinology, Stanford, USA 016
2
Rensselaer Polytechnic Institute, Engineering, New York, USA
3
Icahn School of Medicine at Mount Sinai, Endocrinology, ADD-ON THERAPIES IN OBESE PRE-DIABETIC
New York, USA WOMEN WITH POLYCYSTIC OVARY SYNDROME
4
Barbara Davis Center, Pediatrics, Aurora, USA A. Janez1, M. Jensterle2
5
University of Virginia, USA
1
Univerzitetni klinicni center Ljubljana, KO za endokrinologijo-
Background: Due to the action time of current insulin for- diabetes in bolezni presnove, Ljubljana, Slovenia
2
mulations, closed loop systems require a pre-meal bolus to re- University Medical center Ljubljana, Dept of Endocrinology-
duce post-prandial hyperglycemia. However, systems that Diabetes and Metabolic Diseases, Ljubljana, Slovenia
eliminate patients need to bolus insulin for meals may be de-
sirable. We report preliminary data assessing initial safety and Obesity is frequently present in women with polycystic
efficacy from a fully closed-loop insulin-only system that does ovary syndrome (PCOS). It aggravates the adverse features of
not require announcement of meals or activity. the syndrome and increases the metabolic risk in this popula-
Research Design and Methods: The multiple model proba- tion. Weight management by lifestyle intervention often re-
bilistic predictive controller (MMPPC) anticipates meals when mains unsatisfactory and Glucagon like peptide 1 receptor
the patient is awake. The system (UVa DiAs, Roche Spirit agonists (GLP1RA) are a class of glucose-lowering drugs
Combo insulin pump, Dexcom G4 CGM) was tested at two sites which act through the GLP-1 receptors. Through approved for
on 10 participants in a 30-hour inpatient study followed by 15 use in type 2 diabetes, the ubiquitous distribution of GLP-1
participants at three sites in a 54 hour supervised hotel study with receptors suggests that these molecules can be utilized to
exercise and unannounced meals challenges. benefit persons with other conditions, including endocrine and
Results: The mean overall (24-hour) and nighttime (11 pm7 metabolic dysfunctionnon-sustainable. GLP1RA liraglutide is
am) CGM values were 142 and 125 mg/dl for the inpatient study. currently approved as anti-obesity agents, yet the experience
Tuning adjustments to reduce daytime aggressiveness made with their use in PCOS-related obesity and insulin resistance is
before the hotel study resulted in mean overall (24-hour) and still limited Short-term 12 week randomized combined
A-6 ATTD 2017 INVITED SPEAKER ABSTRACTS

treatment with liraglutide 1.2 mg QD s.c. alone or in combi- these analogues do not solve the major issue of insulin therapy:
nation with metformin 1000 mg BID was associated with sig- weight gain and risk of hypoglycaemia. Next generation insulins,
nificantly greater weight loss in obese women with PCOS who like smart insulins or hepato-preferred insulins, might bring a
had been previously poor responders regarding weight reduc- solution.
tion on lifestyle intervention and metformin when compared to
metformin monotherapy. The reported mean weight losses
were 6.5 kg with liraglutide plus metformin, 3.8 kg with lir- 018
aglutide alone and 1.2 kg with metformin alone. Furthermore,
treatment with liraglutide was associated with significantly CLOSED LOOP AND PHYSICAL ACTIVITY - A DREAM
greater weight loss in a subset of obese patients with newly OR ALREADY REALITY?
diagnosed PCOS and higher metabolic risk profile when com- N. Bratina1, K. Dovc2, T. Battelino2
pared to metformin and lifestyle intervention. Recognizing that
1
the weight reducing effects of GLP-1 RAs are mediated through Lubljana University Medical Centre, Department of
GLP-1 receptor its genetic variability could be hypothetically Endocrinology- Diabetes and Metabolism, Lubljana, Slovenia
2
associated with the inter-individually different response to University Childrens Hospital Ljubljana, Department of
weight lowering potential of liraglutide in metabolically bal- Endocrinology, Diabetes and Metabolic Diseases, Ljubljana,
anced and BMI matched obese population. We have demon- Slovenia
strated that some GLP1-R polymorphisms were associated with
inter-individual differences in response to liraglutide regarding Physical activity is important for children and adults and es-
weight reduction in phenotypically and metabolically homo- pecially for people with chronic diseases. It helps to reduce stress
geneous cohort of obese women with PCOS.The novel phar- in everyday life, keeps body in good physical condition, it can
macological treatment concept in obesity and obesity related help to slow down atherosclerotic diseases, improves HDL
conditions should focus on distinct regulatory mechanisms of cholesterol level, it can delay osteoporosis.
energy homeostasis and eating behaviour. Agents mediating Athletes with diabetes can participate in recreational sports or
through GLP-1 effects in combination with lifestyle interven- be a part of the competitive teams. Physical activity is always a
tion and metformin could potentially improve treatment out- challenge to the athlete with diabetes, as well as for members of
comes in obese PCOS via co-targeting multifactorial origin of the medical teams who care for them on long term basis. We must
obesity and concomitant abnormalities intrinsically related to be aware that different sports and exercise intensity have their
PCOS. Larger and longer randomized studies are needed to own effects on diabetes. Many factors can significantly affect
establish the metabolic, reproductive, and cardiovascular risk blood glucose levels such as the level of hydration, the secretion
reduction and to assess the sustainability and safety profile of of counter regulatory hormones, timing of meals and the level of
weight reduction achieved by this potential new treatment active insulin. If talking physical activity in childhood next to
strategies. diabetes, growth, injuries, sudden heart death syndrome, and
doping problems, protein overload and unhealthy eating patterns
017 must be discussed.
Diabetic athlete must exactly plan how much aerobic and
NEW INSULINS IN TYPE 1 DIABETES: GETTING anaerobic exercise will be a part of the training, the intensity,
BETTER ALL THE TIME? duration and timing of the exercise, but also the risk of hypo-
C. Mathieu1 glycemia and injuries are a concern. Most frequently athletes
learn to manage their diabetes during exercise by trial and error
1
Laboratory and Clinic of Experimental Medicine and or with sharing personal experiences with other athletes. Fear of
Endocrinology, University Hospital Leuven- Catholic hypoglycaemia can be a big burden for many active people.
University of Leuven, Leuven, Belgium Nowadays technological development gives us new possibil-
ities for diabetes treatment modern insulins, insulin pumps and
Therapy of people with type 1 diabetes consists of external sensors. Closed-loop systems combines glucose monitoring
replacement of all functions of the beta-cell aimed at achieving system and a modern insulin pump with a computer algorithm
glucose levels as close to normal as possible. This demands re- that regulates insulin delivery according to glucose oscillations.
placing glucose sensing and achieving insulin levels that mimic the Closed loop would be an important tool improving metabolic
physiological insulin profiles, with basal coverage and meal time control during physical activity. By the time data from studies on
excursions. Patient education and training is crucial to achieve closed loop and physical activity are modest.
good glycaemic control, but having insulin preparations at hand We performed an open label, in-hospital, randomized, cross-
that have profiles that provide stable basal insulin coverage and over designed trial with two different, 40 minutes protocols on a
appropriate mealtime insulin peaks helps patients with type 1 di- cycle ergometer: moderate intensity (55% VO2max) physical
abetes live active lives without sacrificing tight glucose control. activity, and moderate activity with incorporated high intensity
The availability of insulin analogues, with profiles that allow (80% VO2max) sprints either with close or open loop use. The
better coverage of mealtime glucose excursions and give a results are promising, showing that closed loop system increased the
more stable basal insulin coverage than the human insulins have proportion of time within the target glucose range 70 180 mg/dl.
allowed many people with type 1 diabetes to achieve tighter The proportion of time in hypoglycemia below 60 mg/dl was 0.00 %
glucose control, with less hypoglycaemia risk. Still, the first for both groups.
generation rapid-acting and long-acting insulin analogues do not Further studies are needed, including competitive and duration
provide the perfect fit for insulin demands and novel analogues sports.
(long-acting: insulin degludec and U300 insulin glargine) as well References
as rapid-acting insulin analogues (faster insulin aspart) are be- Robertson K, Riddell MC, Guinhouya BC, Adolfsson P, Ha-
coming available and will allow yet another step towards good nas R. Exercise in children and adolescents with diabetes: Ex-
glycemic control in patients with type 1 diabetes. However, even ercise. Pediatr Diabetes 2014;15(S20):20323.
ATTD 2017 INVITED SPEAKER ABSTRACTS A-7

Adolfsson P, Nilsson S, Albertsson-Wikland K, Lindblad B. Nevertheless, recently it was recognized that with proper and
Hormonal response during physical exercise of different inten- careful management, children and adolescents with Diabetes can
sities in adolescents with type 1 diabetes and healthy controls. dive safely.
Pediatr Diabetes. 2012;13(8):58796. Practical guidelines are needed for diabetic patients and their
Wilson DM, Calhoun PM, Maahs DM, Chase HP, Messer L, caregivers, in order to decrease the extra-risk during diving and
Buckingham BA, et al. Factors associated with nocturnal hypo- consequently prevent hypoglycemia and hyperglycemic peaks.
glycemia in at-risk adolescents and young adults with type 1 Continuous Subcutaneous Insulin Infusion (CSII), Continuous
diabetes. Diabetes Technol Ther 2015;17(6):38591. Glucose Monitoring (CGM) and remote monitoring in case of
Phillip M, Battelino T, Atlas E, Kordonouri O, Bratina N, Multi Daily Injections (MDI) seem to be valid options to manage
Miller S, et al. Nocturnal glucose control with an artifi- children with diabetes approaching to diving. During the last ten
cial pancreas at a diabetes camp. N Engl J Med 2013;368(9): years some papers demonstrated that pump technology and ac-
82433. curacy of glucose sensors are still valid even in a such extreme
condition as scuba diving.
Recently a CGM System (CGMS) has been developed to real-
019
time monitor glucose profiles during diving and allow a specific
CONTINUOUS AND FLASH GLUCOSE MONITORING control of glycemia and prevention of hypos.
IN YOUTH WITH TYPE 1 DIABETES HIKING GORGES Ten adolescents, 12 to 18 years old, with Type 1 Diabetes,
IN GREECE AND VOLCANOES IN ICLAND intensively insulin treated with both, CSII or MDI schemes, were
involved in a Project named Sweet Abyss, in order to evaluate
O. Kordonouri1 the accuracy and utility of a diving suited CGMS. A detailed
1 management protocol was purposely built for a safe dive and
Kinder- and Jugandkrankenhaus AUF DER BULT, Diabetes
Center for Children and Adolescents, Hannover, Germany favorable results in terms of hypoglycemia prevention and glu-
cose control have been obtained. Further studies are needed to
Aim: To demonstrate that well-educated young patients using fully evaluate the accuracy of glucose monitoring during pres-
modern insulin treatments are able to perform successfully even sure changes.
extraordinary physical challenges without limitations because of
their diabetes.
Patients and Methods: Two challenges, a 4-days trekking in
Crete, Greece (2015) and a 5-days trekking in Iceland (2016) have 021
taken place so far. In total, 21 adolescents and young adults with THE INTERNET OF THINGS, DIABETES
T1D [13 male, 8 female; age 18.0 years (16-25), BMI 22.4 kg/m2 MANAGEMENT AND AUTOMATED INSULIN
(19.5-26.2), diabetes duration 7.3 years (1-17), HbA1c 7.2% (5.8- MANAGEMENT
8.5); median (range)] from 17 countries worldwide have suc-
cessfully participated. Ten youngsters (47.6%) were on CSII, E. Dassau1
eleven on MDI. During the first trekking in Crete, participants 1
Harvard University, Harvard John A. Paulson School
were wearing a continuous glucose monitoring system (CSII:
of Engineering and Applied Sciences, Cambridge, USA
Paradigm Veo or 640G, Medtronic; MDI: DexCom G4 Platinum,
DexCom), while during the second challenge in Iceland they were
Automated insulin delivery (AID) and decision support (DS)
using a Flash glucose system (Freestyle libre). The glycemic
systems can be considered as the first implementation of the
targets during the challenge were defined as 80-180 mg/dl (4.4-
internet of things in health. With smart-electronics that can es-
10 mmol/l).
tablish local body network and remote communication to cloud
Results: All participants completed the challenge. In total,
services and other stakeholders. These smart insulin pumps and
the groups walked 54.5 km and 88.5 km under varying cli-
glucose sensors are just the first step toward wearable medical
mate conditions, respectively. Insulin requirements decreased
devices. The introduction of other on-body sensors provides
significantly compared to baseline: total daily insulin by
endless opportunities to improve diabetes management for pa-
31.5 17.6% (p
tients with type 1 diabetes mellitus. The Benefit to glycemic
Conclusion: Despite of very challenging physical, climate
control has been proven, however this amazing technology has
and cultural conditions, the youth succeeded reaching the sum-
opened the door to the integration of IOT as well as the need for
mits while maintaining very good glycemic control and without
better cybersecurity, privacy of data, and human interaction. In
any acute complication. Intensive glucose management using
this talk I will discuss the integration of wearable IOT with AID
Continuous and Flash Glucose Monitoring systems, flexible
and DS systems.
insulin treatment and food management guaranteed the excel-
lent performance of the young people.

020 022
CSII, CGM, AND SCUBA DIVING IN PEDIATRICS NOVEL APPROACHES TO CYBER SECURITY
1
R. Schiaffini H. Goldman1
1 1
Bambino Gesu, Hospital, Pediatric Diabetology, Rome, Italy The MITRE Corporation, Center for National Security,
Bedford, USA
Scuba diving is usually prohibited in children and adolescents
with Type 1 Diabetes, especially due to the particular risk of Medical devices are increasingly exposed through network-
hypoglycemia. ing. Consequently, their security posture is degrading as the
A-8 ATTD 2017 INVITED SPEAKER ABSTRACTS

attack surface and cyber threats increase. They were originally secretion, delaying gastric emptying, and inducing satiety.
built for functionality, not security, at a time when cyber threats Pramlintide, a synthetic amylin analog, is used as an adjunct to
were not what they are today. Practitioners and many device insulin in the treatment of T1D and has been shown to effectively
manufacturers do not understand the number of, or extent of lower A1c, primarily through blunting of post-prandial glycemic
commodity and proprietary embedded components. They are excursions. Because most closed-loop (CL) insulin delivery
also typically unaware of the amount of hardware and software systems rely on manual meal bolusing to control post-prandial
reuse that, if vulnerable, could result in pervasive compromise glucose levels, we hypothesized that pramlintide may be an ef-
across technologies and devices and cause systemic failures and fective adjunctive therapy and eliminate the need for manual
cascading effects. More importantly, many cybersecurity coun- insulin dosing during CL. In two studies of adolescents and
termeasures designed for traditional information systems may young adults with T1D, pre-meal injections of 30-60mcg doses
not be useful or appropriate given embedded system constraints, of pramlintide mitigated post-prandial glycemic excursions and
environmental and user considerations, and compromise impact lowered overall mean glucose levels during CL control. These
to safety, medical efficacy, or loss of life. improvements were achieved with lower insulin requirements,
We will describe cybersecurity challenges unique to embed- due to the blunting of endogenous glucagon secretion after
ded systems including system constraints and operating trends meals. We conducted similar CL studies with adjunctive lir-
that affect cybersecurity risk. We will also identify inherent aglutide, a synthetic glucagon-like peptide-1 (GLP-1) receptor
embedded systems characteristics to leverage that help mitigate agonist that also acts to suppress endogenous glucagon secretion,
vulnerabilities and reduce consequences. Finally we will offer and found comparable reductions in prandial glycemic excur-
approaches to address embedded systems threats, risk manage- sions and reduction in insulin requirements, although glucagon
ment, secure resilient architectures, and countermeasure cost suppression was inferior. Other groups have demonstrated that
effectiveness. exenatide, another synthetic GLP-1 agonist, may have even
greater suppression of endogenous glucagon during CL control.
Amylin and GLP-1 agonists show potential to mitigate prandial
glucose excursions during CL control but their effectiveness,
023 safety, and tolerability during longer ambulatory CL use needs
further characterization.
AUTOMATED GLYCEMIC REGULATION WITH A
DUAL-HORMONE (INSULIN AND GLUCAGON)
ARTIFICIAL PANCREAS: THE MONTREAL
APPROACH 025
1
A. Haidar ISSUES RELATED TO INFUSION OF GLUCAGON
1
McGill University, Biomedical Engineering, Montreal, OR AMYLIN
Canada R. Rabasa-Lhoret1
1
The artificial pancreas is a long-awaited goal for the manage- Institut de recherches cliniques de Montreal, Metabolic
ment of type 1 diabetes and its development was recently trig- diseases, Montreal, Canada
gered by advances in continuous glucose sensors. Two
configurations of the artificial pancreas have been proposed: one Single-hormone closed-loop systems using insulin only have
that infuses insulin (single-hormone) and one that infuses insulin been shown to significantly improve overall glucose control. To
and glucagon (dual-hormone). We review two potential usages of better reproduce physiology and bypass limitations of current
glucagon in the artificial pancreas: to reduce hypoglycemia and to subcutaneous insulin pharmacokinetic and pharmacodynamics,
allow more aggressive insulin delivery, with discussion of the multi-hormonal closed-loop systems are under investigation.
potential benefits and drawbacks of each approach. We review Aims are to further improve mean glucose, reduce hyperglyce-
results from our inpatient and outpatient studies that assess the mic risk (by mainly targeting postprandial glucose excursions),
ability of the dual-hormone artificial pancreas to reduce the risk of further reduce hypoglycemic and/or simplify meal insulin dos-
hypoglycemia compared to the single-hormone artificial pancreas ing. Up to now, short-term data with two hormones have been
and conventional pump therapy. Advances in novel insulin ana- reported: 1) Glucagon has been shown to further reduce hypo-
logs and non-insulin adjunctive therapies are also discussed in the glycemic risk and/or to allow more aggressive insulin infusion
context of dual-hormone artificial pancreas systems. without increasing the risk of hypoglycemia risk; 2) Pramlintide,
an amylin analogue, has been shown to improve postpran-
dial glucose excursions. Along with these significant potential
benefits, moving to multi-hormonal closed-loop systems raise
024 important challenges. By essence, these systems will be more
complex necessitating technological improvements (e.g. multi-
AMYLIN PLUS INSULIN chamber pumps and specific algorithms), new indications (e.g.
S. Weinzimer1 glucagon for hypoglycemia prevention rather than to treat severe
hypoglycemia), and new pharmacological formulation (e.g.
1
Yale University School of Medicine, Pediatric Endocrinology, stable glucagon). In addition, compliance issues will need to be
New Haven, USA addressed if Pramlintide still needs to be injected subcutaneously
with a pen rather than with a pump. These hormones could by
Amylin is a naturally-occurring polypeptide co-secreted with themselves lead to additional costs, specific side effects (e.g.
insulin from beta cells in response to food ingestion, and like nausea and vomiting with Pramlintide), and issues for hypogly-
insulin, is deficient in people with type 1 diabetes (T1D). Amylin cemia treatment or specific situations (e.g. glycogen depletion
improves post-prandial glucose control by inhibiting glucagon with low carbohydrate diet). Though some data is available for
ATTD 2017 INVITED SPEAKER ABSTRACTS A-9

Pramlintide, no long-term safety data is available for chronic lymph nodes, pancreas, and peripheral blood of all treated mice,
glucagon usage. Available data are reassuring but in order to independent of metabolic outcome. Neutralization of CTLA4
determine if such addition is worth, longer trials will have to and TGF-b partially abrogated the suppressive function of
confirm benefits, investigate potential issues and finally assess therapy-induced Tregs. Ablation or functional impairment of
the risk-to-benefit ratio. Foxp3+ Tregs in vivo at start or stop of therapy impaired im-
mune tolerance, highlighting the dependence of the therapy-
induced tolerance in new-onset diabetic mice on the presence
and functionality of CD4+Foxp3+ T cells.
026 Other antigens have been introduced into the platform (GAD,
LACTOCOCCUS AS A DELIVERY SYSTEM IA2) with similar results. Other cytokines have been introduced
FOR IMMUNOTHERAPY as well, some of which allow the elimination of the anti-CD3
induction therapy.
C. Mathieu1, C. Gysemans2
1
Laboratory and Clinic of Experimental Medicine and
Endocrinology, University Hospital Leuven- Catholic 027
University of Leuven, Leuven, Belgium
2
KULeuven, Experimental and clinical endocrinology, Leuven, ANTIGEN-BASED VACCINE THERAPY FOR T1D
Belgium M. Peakman1
1
Prevention of type 1 diabetes or arrest of the autoimmune Kings College London, London, United Kingdom
destruction of the beta-cell after onset of hyperglycemia remain
an elusive goal. Interventions with immune modulators have The main focus of my research has been the role of T lym-
proven to be able to arrest the progression of beta-cell destruction phocytes in the aetiology of Type 1 diabetes, a chronic autoim-
for a short time, but despite doses being used at the edge of mune disease. In particular, we have defined the critical targets
acceptable side effects, like for anti-CD3 monoclonal antibodies, for T cells that appear to have a role in the destruction of insulin-
no long lasting protection of the beta-cell was observed. producing cells, and key immunological pathways through
Combining immune modulation with antigen-specific tolerance which this damage is mediated. These findings represent the
(re-)induction offers an attractive path for intervention. fundamental breakdown in immunological self-tolerance that
The Lactococcus lactis (L. lactis) platform offers an elegant underlies type 1 diabetes, and identify it as a major barrier to
tool to introduce beta cell auto-antigens in combination with preventive and curative therapies. Work in preclinical models
immune modulating agents like cytokines via the gut has been has highlighted the potential for restoration of tolerance through
demonstrated to be a promising approach for diabetes reversal in various manipulations, including administration of antigens - so
NOD mice, when combined with a short course (5days) of low called antigen-specific immunotherapy (ASI). ASI has given
doses of anti-CD3 antibodies. A combination of a 5 day course of mixed results in human trials to date, probably reflecting the
low-dose anti-CD3 with a 6 week treatment with clinical-grade importance of selection of route of administration, timing and the
self-containing L. lactis appropriate for human application se- nature of the antigen. Work in our laboratory has led to the
creting human pro-insulin and IL10 cured 66% of new-onset definition of target antigenic peptides enabling the design of a
diabetic mice, comparable to plasmid-driven L. lactis. Initial novel approach to ASI. This strategy, termed peptide immu-
blood glucose concentrations (<350 mg/dl) and insulin autoan- notherapy is the first of its kind in diabetes. We have reported
tibody positivity were predictors of stable reversal of hypergly- proof-of-concept studies in different disease stages and further
cemia and decline in IAA positivity was an immune biomarker of phases of this programme are now progressing. In the future, a
therapeutic outcome. Assessment of the immune changes in- better understanding of the role of the immune response in Type
duced by the L. lactis-based therapy revealed elevated fre- 1 diabetes will promote the further development of these novel
quencies of CD4+Foxp3+ T cells in the pancreatic draining therapeutics into the clinical setting.
ATTD 2017 Oral Abstracts

028 Conclusions: Use of CGM without regular use of confirma-

tory BGM is as safe and effective as using CGM with BGM in
REPLACE-BG: A RANDOMIZED TRIAL COMPARING adults with T1D.
CONTINUOUS GLUCOSE MONITORING WITH
AND WITHOUT ROUTINE BLOOD GLUCOSE
MONITORING IN ADULTS WITH TYPE 1
DIABETES 029
G. Aleppo1, T. Riddlesworth2, R. Beck3, K. Ruedy3, E. Eyth4,
DETECTION OF UNANNOUNCED MEAL AND
R. Bergenstal5
ANOMALIES OF CONTINUOUS GLUCOSE MONITORS
1
Northwestern University, Medicine Endocrinology, USING SENSOR REDUNDANCY AND THE
Chicago, USA UNSCENTED KALMAN FILTER
2
Jaeb Center for Health Research, REPLACE-BG, Z. Mahmoudi1, K. Nrgaard2, N. Kjlstad Poulsen1,
Tampa, USA H. Madsen1, J. Bagterp Jrgensen1
3
Jaeb Center for Health Research, REPLACE-BG Study,
1
Tampa, USA Technical university of Denmark, Applied Mathematics
4
University of South Florida, USF Diabetes Center, and Computer Science, Lyngby, Denmark
2
Tampa, USA Copenhagen University Hospital Hvidovre, Department
5
International Diabetes Center Park Nicollet, Endocrinology, of Endocrinology, Copenhagen, Denmark
Minneapolis, USA
Background and Aims: In an artificial pancreas (AP), it is
Background and Aims: To determine whether the use of critically important to detect the faults of the continuous glucose
continuous glucose monitoring (CGM) without confirmatory monitoring (CGM) sensor in order to e.g. avoid dosing insulin in
blood glucose measurements (BGM) is as safe and effective response to an erroneous high glucose signal. We developed a
as using CGM as an adjunct to BGM in adults with type 1 method for detecting CGM anomalies as well as detecting un-
diabetes (T1D). announced meals for the AP.
Methods: Multicenter, randomized non-inferiority clinical Methods: The module consists of two CGM sensors, two fault
trial conducted by the T1D Exchange Clinic Network. Parti- detectors, a fault isolator, and an adaptive unscented Kalman filter
cipants were 25 years (mean 44 14 years), had T1D for 1 (UKF). Each of the fault detectors has a local UKF that works as a
year (mean duration 24 12 years), were being treated with residual generator. Each of the fault detectors detects drift and
an insulin pump, and had HbA1c 9.0% (mean 7.0 0.7%); compression artifacts associated with one of the sensors. When a
47% were CGM users and 53% were not using CGM when fault is detected, the fault detectors tune the noise covariance of the
enrolled. Each participant was randomly assigned 2:1 to ei- adaptive UKF. By applying a set of rules on the filter-predicted
ther the CGM-only (N = 149) or CGM+BGM Group (N = 77). CGM measurements, the meal detector is a fault isolator differ-
The primary outcome was time 70-180 mg/dl over the 6 entiating between a sensor fault and an unannounced meal.
months of the trial, with a pre-specified non-inferiority limit Results: The drift detection sensitivity, specificity, and pre-
of 7.5%. cision are 80.9%, 100.0%, and 100.0%, respectively. The sensi-
Results: CGM use averaged 6.6 0.6 days/week in each tivity, specificity, and precision of detecting compression artifacts
group over the 26 weeks of the trial. BGM tests/day (including are 63.2%, 99.2%, and 95.2%, respectively. The fault detectors
the 2 required daily BGM tests) averaged 2.8 0.8 in the CGM- detect 100% of the simulated drifts and 97% of the compression
only Group and 5.4 1.4 in the CGM+BGM Group (P < 0.001). artifacts. The fault isolator detects 100% of the unannounced
Mean time 70-180 mg/dL was 63 13% at both baseline and 26 meals, without any false positive. The average change in glucose
weeks in the CGM-only Group and 65 12% and 65 11%, between the meals and detection points is 62.8 [mg/dL].
respectively in the CGM+BGM Group (adjusted difference = 0%; Conclusions: The adaptive UKF compensates the effect
one-sided 95% confidence interval 1%). Mean change in HbA1c of drift and compression artifact on the CGM data, and
was 0.0% in each group (P = 0.40). One severe hypoglycemic compared to a non-adaptive UKF, reduces the deviation of
event occurred in the CGM+BGM Group and none in the CGM- the CGM measurements from their paired blood glucose
only Group. concentrations.

A-10
ATTD 2017 ORAL ABSTRACTS A-11

030
PATIENTS WITH TYPE 2 DIABETES USING
MULTIPLE DAILY INSULIN INJECTIONS HAVE
HIGH ADHERENCE AND BENEFIT FROM CGM: A
PROSPECTIVE, RANDOMIZED CONTROLLED TRIAL
R. Bergenstal1, T. Riddlesworth2, K. Ruedy2, C. Kollman2,
D. Price3, R. Beck2
1
Park Nicollet Institute - International Diabetes Center, n/a,
St. Louis Park, USA
2
Jaeb Center for Health Research, n/a, Tampa, USA
3
Dexcom Inc., n/a, San Diego, USA

Background and Aims: To determine the effectiveness of

CGM in adults with type 2 diabetes (T2D) using multiple daily
injections (MDI)
Methods: In a randomized trial conducted at 25 clinics in the
US and Canada, 158 adults (mean age 60 + 10, range 35-79 years)
with T2D (mean duration 18 + 9 years) using MDI and with
HbA1c 7.5%-9.9% (mean 8.5% +0.6%) were randomly assigned Methods: FSLP-FGM was done in T2DM patients (n = 388,
to CGM (Dexcom G4 Platinum CGM System with software Age = 55.42 13.05 yrs., Diabetes duration- 14.17 8.37 yrs.,
505, N = 79) or Control Group, N = 79. The primary outcome was 70.62 % Male, 29.38 % Female). Our multi-disciplinary team
HbA1c at 24 weeks. analysed the Ambulatory Glucose Profiles (AGP) generated by
Results: Mean HbA1c decreased from 8.5 + 0.6% at baseline to FSLP and advised patients on therapeutic and lifestyle inter-
7.7 + 0.7% at 24 weeks in the CGM Group and from 8.5 + 0.7% to ventions. Improvements in health status in terms of HbA1C, BMI
8.0 + 0.9% in the Control Group (adjusted difference = -0.3%, 95% and FBS after 6 months of undergoing FSLP- FGM were eval-
confidence interval -0.5% to -0.0%, P = 0.02). The CGM Group uated by comparing them to previously reported values.
averaged 6.7 + 0.9 days/week of CGM use and reported a high degree Results: A mean wear time of 12.69 2.53 days was achieved
of satisfaction with CGM. Benefits were similar across age groups, with FSLP. Statistically significant reductions in HbA1C
educational level, and numeracy skills. There was minimal hypo- (p < 0.0001), FBS (p < 0.0001) and BMI (p = 0.0226) levels
glycemia at baseline and during the study. Mean CGM-measured (Figure 1) were noted among patients after 6 months of un-
time in the range (TIR) of 70-180 mg/dL increased by about an hour dergoing FSLP-FGM.
more per day in the CGM Group than the Control Group. Conclusions: FSLP generated collated AGP visualizes ac-
Conclusions: A high percentage of adults with T2D being tual glycemic exposure patterns beyond HbA1c, discloses
treated with MDI used CGM on a daily or near-daily basis for persistent and significant hypo- and/or hyperglycemia, as well
24 weeks with a high degree of satisfaction and a significant as considerable glycemic variability. It also increases the fea-
reduction in HbA1c and increased TIR compared with the con- sibility of a productive interactive session among patients and
trol group. Since few MDI-treated patients with T2D currently health management team which facilitates better treatment
are using CGM, the study results support an additional man- outcomes.Newer technologies with advantages such as pre-
agement modality that may be beneficial for these patients. calibrated mode which avoids frequent finger-pricks, increased
wear time, larger volume of data available for analysis, lower
MARD etc. are likely to eventually replace glucosemeters and
conventional CGMs.

031
BETTER OUTCOMES IN TYPE 2 DIABETES
032
MANAGEMENT WITH A USER FRIENDLY FLASH
GLUCOSE MONITORING SYSTEM: FREESTYLE NOVEL FLASH GLUCOSE-SENSING TECHNOLOGY
LIBRE PRO REDUCES HYPOGLYCAEMIA IN INDIVIDUALS
J. Kesavadev1, G. Krishnan1, B. Saboo2, A. Shankar1, WITH TYPE 1 DIABETES USING MULTIPLE DAILY
A.D. Ashok1, G. Sanal1, L. Ramachandran1, S. Jothydev1 INJECTIONS (MDI)
1 J. Bolinder1, R. Antuna2, P. Geelhoed-Duijvestijn33,
Jothydevs Diabetes Research Centre, Diabetes,
J. Kroeger4, R. Weitgasser5
Thiruvanathapuram, India
2 1
Dia Care & Hormone Clinic, Diabetes, Ahmedabad, India Karolinska Universitetssjukhuset, Department of Medicine,
Stockholm, Sweden
2
Background and Aims: Continuous Glucose Monitoring Clinica Diabetologica, Department of Medicine, Gijon, Spain
3
(CGM) provides more distinct glycaemic patterns thereby al- Medisch Centrum Haaglanden, Department of Diabetes and
lowing appropriate therapeutic interventions to pursue glycaemic Endocrinology, Den Haag, The Netherlands
4
targets safely. FreeStyle Libre Pro (FSLP) Flash Glucose Mon- Zentrum fur Diabetologie Hamburg Bergedorf, Department
itoring (FGM) was launched in India with several unique fea- of Diabetes, Hamburg, Germany
5
tures over CGM. With an intention to reach better glycemic goals, Wehrle-Diakonissen-Krankenhaus and Paracelsus
T2DM patients were deployed FSLP sensor for 14 days and im- Medizinische Privatuniversitat, Department of Medicine,
provements in HbA1C, FBS and BMI were assessed. Salzburg, Austria
A-12 ATTD 2017 ORAL ABSTRACTS

Background and Aims: The effectiveness of FreeStyle during explantations. Early performance measures indicate
LibreTM Flash Glucose Monitoring System in patients with type same-pocket re-implantations may be feasible for sensor re-
1 diabetes (T1DM) using MDI has not been documented. In this placement/renewal.
subgroup analysis of MDI users in the IMPACT trial, we as- Conclusions: Use of the fully-implanted ICGM Sensor re-
sessed its impact on hypoglycaemia compared to conventional quires an annual user decision (whether to implant/re-implant or
self-monitoring of blood glucose (SMBG). not) and an occasional decision to recalibrate; no body-worn
Methods: 161 patients with well-controlled T1DM, (HbA1c components or other regular user intervention is required to re-
50.3 6.3 mmol/mol (meanSD) [6.76 0.58%]), age 43 13.1 ceive glucose readings. This combination of features offers
years and duration of diabetes 21 10 years) using MDI were minimal barriers for adherence to treatment modalities requiring
randomised to the intervention group (FreeStyle Libre; n = 81) or continuous glucose monitoring.
to the control group (SMBG; n = 80). CAUTION - Investigational Device. Limited by United States
Results: After 6 months, those using FreeStyle Libre signifi- law to investigational use. Eclipse is a trademark of GlySens
cantly reduced time in hypoglycaemia (<3.9 mmol/L [70 mg/dL]) Incorporated. GlySens and ICGM are trademarks of GlySens
by 46.0% vs. control (meanSE: -1.65 0.283 hours per day; Incorporated registered in the U.S. Patent and Trademark Office.
p < 0.0001). Time <2.2 mmol/L (40 mg/dL) was reduced by
65.6%; (p = 0.0012). The proportion of patients who achieved
1 hour per day in hypoglycaemia (<3.9 mmol/L) was signifi-
cantly higher for those using FreeStyle Libre vs. control; 33.3% 034
vs. 10.0%, p = 0.0005.
EVIDENCE OF A STRONG ASSOCIATION BETWEEN
Time in range (3.9-10.0 mmol/L [70-180 mg/dL]) signifi-
FREQUENCY OF FLASH GLUCOSE MONITORING
cantly improved by 0.9 0.37 hours per day (meanSE);
AND GLUCOSE CONTROL MEASURES DURING
p = 0.0106 vs. control. There was no change in HbA1c.
REAL-WORLD USAGE
Using FreeStyle Libre, scanning frequency at 6 months av-
eraged 14.7 per day, whereas SMBG tests dropped from a me- T. Dunn1, Y. Xu1, G. Hayter2
dian of 5.4 (baseline) to 0.1 per day. In the controls, SMBG tests 1
were 5.1 per day at 6 months. Abbott Diabetes Care, Clinical Affairs, Alameda, USA
2
Treatment satisfaction (DTSQ/DQoL) and perception of Abbott Diabetes Care, Research and Development,
hypo- and hyperglycaemia (DTSQ) were significantly improved. Alameda, USA
Conclusions: Well-controlled T1DM patients using FreeStyle
Libre with MDI significantly reduced time in hypoglycaemia Background and Aims: The aim was to evaluate association
without deterioration of HbA1c, and reported improvements in of the real-world scanning with flash glucose monitoring (Free-
treatment satisfaction. Style Libre) and glucose control measures. FreeStyle Libre is a
sensor-based glucose monitor, and the reader scans the sensor to
collect the current glucose and glucose trend, along with up to
8 hours of glucose readings automatically stored every 15 min-
utes. When connected to the PC-based software with an active
033 internet connection, the readers 90-day memory is de-identified
CLINICAL UPDATE WITH A LONG TERM, and uploaded to a database.
UNOBTRUSIVE, FULLY-IMPLANTED CONTINUOUS Methods: For analysis, sensors were required to have at least
GLUCOSE MONITORING SYSTEM 120 hours of operation, and all sensors were grouped per reader,
resulting in 50,831 readers with 279,446 sensors (86.4 million
J. Lucisano1, L. Kurbanyan1, S. Martha1, T. Routh1 monitoring hours by 63.8 million scans). Twenty equally-sized
1
GlySens Incorporated, Technology Development, groups by scan rate were analyzed (n = 2,542 each).
San Diego, USA Results: Users performed an average of 16.3 scans per day
(median:14, interquartile range: 10-20). Estimated HbA1c re-
Background and Aims: Recent investigational human ex- duced (p < 0.001) as scan rate increased, from 8.0% to 6.7% from
perience with a second generation long term fully-implanted (no the lowest (mean 4.4 scans/day) to highest (mean 48.1 scans/day)
skin-attached elements) continuous glucose monitoring system
(the GlySens Eclipse ICGM System) includes sensor im-
plantations in five new adult human subjects, as well as same-
pocket re-implantations with new sensors in six adult human
subjects following completion of a 12-month initial implant
period.
Methods: The Eclipse ICGM system is implanted in a
minor outpatient surgical procedure utilizing local anesthesia.
Following sensor implantation, study subjects self-monitor blood
glucose four times per day via finger stick glucose meter and
undergo monthly clinic visits that include meal-based glucose
excursions with YSI plasma glucose comparison measurements.
Some subjects also utilize a Dexcom G4 CGM to provide ad-
ditional paired values. Monthly subject interviews including a
standardized survey questionnaire are conducted to assess tol-
erance of the device.
Results: All implantations were completed successfully and
no significant sensor-adherent capsular tissue was observed
ATTD 2017 ORAL ABSTRACTS A-13

groups, while simultaneously time below 70, 55 and 45 mg/dL Background and Aims: We aimed to evaluate the impact of
decreased by 15%, 40% and 49%, respectively (all p < 0.001). real-time continuous glucose monitoring (RT-CGM, Dexcom G5)
Time above 180 mg/dL decreased from 10.4 to 5.7 h/day (44%, and intermittent flash glucose monitoring (Abbott Freestyle Libre)
p < 0.001), and time in range 70-180 mg/dL increased from 12.0 on diabetes-related emotional distress and fear of hypoglycaemia,
to 16.8 h/day (40%, p < 0.001). using the Problem Area in Diabetes (PAID) and Hypoglycaemia
Conclusions: In real-world use, higher rates of scanning to self- Fear Survey-II (HFS-II) questionnaires respectively, in adults with
monitor glucose were found to strongly associate with improved type 1 diabetes (T1D) and impaired awareness of hypoglycaemia
glucose measures, including decreased mean glucose and time in (IAH).
hyper- and hypoglycemia as well as increased time in range. Methods: This is a prospective randomized parallel group
study. Participants were randomized to either Libre or G5 for 8
weeks after two weeks run-in with blinded CGM. They were
asked to complete the PAID (score range 0-100, higher score is
035 worse) and HFS-II (range 0-132, higher score is worse) at
EFFICACY OF CONTINUOUS GLUCOSE baseline and endpoint.
MONITORING IN DIABETIC PREGNANCY: Results: 32 adults with T1D on multiple daily injections of
THE GLUCOMOMS TRIAL insulin and IAH (Gold score 4) have completed the study (66%
male, mean (SD) age 50 (15) years, diabetes duration 29 (12)
D. Voormolen1 years, Gold score 4.7 (1), HbA1c 58 (11) mmol/mol). There was
1 significant reduction in mean (SD) HFS-II score from baseline to
UMC Utrecht, obstetrics, Utrecht, The Netherlands
endpoint with G5 (52.7(22.5) vs 47.3 (25.3), p = 0.03) compared
to Libre (52.1(24.5) vs 50.8 (28.1), p = 0.7). There was no sig-
Background and Aims: Hyperglycemia in pregnancy is as-
nificant change in PAID score from baseline to 8 weeks with
sociated with poor neonatal outcome. Glycemic control is tra-
either G5 (27.9 (19) vs 27.1 (16.5), p = 0.7) or Libre (29.9 (20.3)
ditionally monitored with self-measured glucoseprofiles and
vs 27.2 (20.2), p = 0.2). The HbA1c improved significantly in
periodical HbA1c measures. We investigated the efficacy of
both groups.
Continuous Glucose Monitoring (CGM) in diabetic pregnancies.
Conclusions: Our preliminary data suggest that real-time
Methods: In a multicenter open label randomized controlled
CGM has a significantly greater benefit compared to intermittent
trial, pregnant women >18 years with diabetes type 1, type 2 on
flash glucose monitoring in reducing fear of hypoglycaemia in
insulin therapy (gestational age (GA) <16 weeks) or insulin depen-
this high risk T1D population group.
dent gestational diabetes (GA <30 weeks) were randomly allocated
to intermittent use of retrospective CGM for 5-7 days every six
weeks Glycemic control was monitored by day-curves and HbA1c
checks. Macrosomia (birthweight >90th percentile), was the primary
037
outcome. Secondary outcomes were glycemic control, maternal and
neonatal complications. Primary analyses were according to inten- HYBRID CLOSED-LOOP (HCL) THERAPY IN
tion to treat, while a secondary per-protocol analysis was limited to ADOLESCENTS AND YOUNG ADULTS WITH TYPE 1
women using the CGM at least once every trimester. DIABETES (T1D) INCREASES TIME IN RANGE
Results: Between July 2011 and September 2015, we ran-
domized 304 women (109 type 1, 83 type 2, 112 gestational L. Messer1, G. Forlenza1, R.P. Wadwa1, E. Westfall1, E. Jost1,
diabetes), 150 to CGM and 154 to conventional treatment. The G.T. Alonso1, D. Maahs2, R. Slover1
incidence of macrosomia was 29% in both the intervention and 1
Barbara Davis Center for Childhood Diabetes, University
control group (RR .99, 95%CI .76-1.28). No difference was of Colorado, Denver, USA
observed in the per protocol analysis (66% of total population, 2
Stanford University School of Medicine, Department
RR 1.00, 95%CI .70-1.42). Glycemic control in terms of HbA1c of Pediatric Endocrinology, Stanford, USA
measures throughout pregnancy are presented in figure 1 and
other outcomes in the table. Preeclampsia was less common in Background and Aims: Changes in total daily insulin dose
the CGM group, while otherwise no differences were seen on (TDD) and use patterns of the Medtronic HCL system in ado-
maternal and neonatal outcomes. lescents and young adults, during a 3-month outpatient trial are
Conclusions: Intermittent CGM use in diabetic pregnancy described.
does not reduce the incidence of macrosomia. Methods: Twelve subjects with T1D, aged 14-21 years, wore a
Medtronic MiniMed 670G pump and Guardian Sensor 3 for
7 days in open loop mode, and 3 months in auto mode (HCL with
manual meal boluses). Remaining in auto mode required blood
036 glucose calibrations and avoidance of prolonged hypoglycemia
IMPACT OF REAL-TIME CONTINUOUS GLUCOSE
MONITORING AND INTERMITTENT GLUCOSE DATA
ON HYPOGLYCAEMIA FEAR IN ADULTS WITH
IMPAIRED AWARENESS OF HYPOGLYCAEMIA
AND TYPE 1 DIABETES
N. Jugnee1, M. Reddy1, S. Anantharaja1, N. Oliver1,
E. Spanudakis1
1
Imperial College London, Department of Medicine- Diabetes-
Endocrinology and Metabolism division, LONDON, United
Kingdom
A-14 ATTD 2017 ORAL ABSTRACTS

and hyperglycemia. Data were analyzed for patterns of use and Results: The proposed methods are able to estimate the PIC
changes in TDD from baseline to trial conclusion. in real time by using only CGM measurements and infused
Results: TDD did not significantly change from baseline to insulin data.
end of trial (p = 0.47), however insulin-to-carbohydrate (I:C) Conclusions: These methods will be beneficial for an AP
ratios were significantly more aggressive after 3 months of auto system in terms of real-time estimation of PIC for preventing
mode. Time in range (70-180 mg/dl) increased during auto excessive insulin infusions if plasma insulin levels indicate
mode (p = 0.01), however not when the system was operating in insulin-on-board.
open loop (p = 0.91). Mean time in auto mode decreased sig-
nificantly from the first two weeks of use (86 11%) to the final
2 weeks of use (72 21%) (p = 0.01), however there were no
differences in number of events triggering the system to exit 039
auto mode (p = 0.96), including hypoglycemia and calibration VARIABILITY OF AUTOMATED NIGHTTIME
alerts. INSULIN DELIVERY AND GLUCOSE LEVELS
Conclusions: Increased time in range was achieved with HCL
compared with open loop without changing TDD in adolescents/ A. Roy1, B. Grosman1, N. Parikh1, D. Wu1, N. Kurtz1, R. Brazg2,
young adults. More aggressive I:C ratios were required during R. Slover3, S. Garg4, J. Ilany5, B. Bode6, B. Buckingham7,
HCL. Decreased time spent in auto mode during the study re- S. Weinzimer8, S. Anderson9, T. Bailey10, R. Bergenstal11,
quires further evaluation to optimize patient outcomes. F. Kaufman12
1
Medtronic Diabetes, R&D, Northridge, USA
2
Rainer Clinical Research Center, Metabolism and
Endocrinology, Renton, USA
038 3
Barbara Davis Center for Childhood Diabetes, Pediatric
REAL-TIME PLASMA INSULIN ESTIMATION Diabetes, Aurora, USA
4
USING CONTINUOUS SUBCUTANEOUS GLUCOSE Barbara Davis Center for Childhood Diabetes, Adult Diabetes,
MEASUREMENTS FOR AN ARTIFICIAL PANCREAS Aurora, USA
5
SYSTEM Institute of Endocrinology- Sheba Medical Center,
Endocrinology, Tel Hashomer, Israel
A. Cinar1, I. Hajizadeh1, K. Turksoy2, S. Samadi1, J. Feng1, 6
Atlanta Diabetes Associates, Endocrinology & Metabolism,
M. Sevil2, N. Frantz2, C. Lazaro Martinez3, Z. Mahoney2, Atlanta, USA
E. Cengiz4 7
Stanford University, Pediatric Endocrinology, Stanford, USA
8
1 Yale University School of Medicine, Pediatric Endocrinology,
Illinois Institute of Technology, Chemical and Biological
New Haven, USA
Engineering, Chicago, USA 9
2 UVA Center for Diabetes Technology, Endocrinology and
Illinois Institute of Technology, Biomedical Engineering,
Metabolism, Charlottesville, USA
Chicago, USA 10
3 AMCR Institute, Advanced Metabolic Care Research,
Illinois Institute of Technology, Electrical and Computer
Escondido, USA
Engineering, Chicago, USA 11
4 Park Nicollet Clinic-International Diabetes Center,
Yale University, Division of Pediatric Endocrinology,
International Diabetes Center, Minneapolis, USA
New Haven, USA 12
Medtronic Diabetes, Global Medical- Clinical & Health
Affairs, Northridge, USA
Background and Aims: Artificial pancreas (AP) systems
use CGM data to calculate the optimum amount of insulin to be
Background and Aims: The MiniMed 670G hybrid closed-
infused with an insulin pump for regulating blood glucose
loop (HCL) system was shown to be safe and effective in a multi-
concentration. Real-time plasma insulin concentration (PIC)
center pivotal trial conducted with 124 type 1 diabetes patients.
estimations would be beneficial for increasing the accuracy of
The algorithm calculates a patient-specific maximum insulin
AP control algorithms to calculate more realistic insulin in-
delivery rate daily.
fusion rates and prevent hypoglycemia caused by overdosing
of insulin. Our objective is to fulfill a real-time PIC estimation
from CGM data using different mathematical models and es-
timation techniques.
Methods: Two different glucose-insulin compartmental mod-
els, Hovorkas model and extended Bergmans minimal model,
which were developed to describe glucose-insulin dynamic in
different parts of the human body, have been incorporated into
different estimation techniques namely, continuous-discrete ex-
tended Kalman filtering, unscented Kalman filtering, and moving
horizon estimation, to provide an estimate of PIC. Furthermore,
because of variability in system dynamics, some uncertain pa-
rameters have been considered as new states in the compartmental
models to be estimated by the estimators. Partial least squares
models are developed for the initial guess of the time-varying un-
known model parameters used in the estimators. Thirteen datasets
from nine different subjects with T1D are used. The study was
conducted in adolescents with T1D who attended the Yale Chil-
drens Diabetes Clinic (New Haven, CT).
ATTD 2017 ORAL ABSTRACTS A-15

4
Atlanta Diabetes Associates, Endocrinology & Metabolism,
Atlanta, USA
5
Stanford University, Pediatric Endocrinology, Stanford,
USA
6
Yale University School of Medicine, Pediatric Endocrinology,
New Haven, USA
7
UVA Center for Diabetes Technology, Endocrinology and
Metabolism, Charlottesville, USA
8
AMCR Institute, Advanced Metabolic Care Research,
Escondido, USA
9
Park Nicollet Clinic-International, International Diabetes
Center, Minneapolis, USA
10
Medtronic Diabetes, Clinical Affairs, Northridge, USA

Background and Aims: Medtronics hybrid closed-loop

(HCL) system was evaluated in a 10-center, at-home pivotal trial.
The system was designed to automatically adjust basal insulin
delivery every 5 minutes based on sensor glucose values. The
system included the MiniMed 670G pump, Guardian Sensor
3, and a control algorithm. Patients needed to calibrate the sen-
sor, enter mealtime carbohydrate estimates, and notify the sys-
tem of exercise.
Methods: A 2-week run-in phase was followed by a 3-month Methods: During the 2-week run-in phase, the system was
at-home study phase of HCL. Sensor glucose (SG) data and the used in manual mode (open-loop). During the 3-month at-home
amount of insulin delivered was analyzed during the overnight study phase, subjects enabled the auto-mode (closed-loop) fea-
period (10:00pm7:00am). ture. A total of 124 type 1 diabetes patients (age range 14 75yrs)
Results: The top panel of Figure-1. represents overnight with A1C <10% were studied.
SG tracings from the run-in (gray-patch) and the study phases Results: The HCL system was used for 12,389 patient-days
(red- patch). The inter-quartile range (IQR) reduced signifi- with no severe hypoglycemia or DKA. Mean glycated hemo-
cantly comparing the run-in and study phases (71 versus globin (A1C) was 7.4 0.9% at run-in and 6.9 0.6% at study
37mg/dL, at 7:00am). This was mostly achieved by trans- end. A linear fit of the delta A1C (end of study A1C minus
ferring the variability from glucose levels to the insulin in- baseline A1C), as shown in the figure (red dashed line), indicates
fusion rate (bottom panel of Figure-1). The maximum IQR of that subjects experienced a drop in A1C levels if the initial A1C
insulin delivered by the algorithm increased by *3 fold was >6.4%. A Pearsons correlation coefficient also shows a
compared to the pre-set basal rates (0.55 versus 1.58 units/hr). significant inverse correlation between initial and delta A1Cs
The nightly variation of insulin delivery by the HCL system is (R = -0.75), demonstrating a larger drop in A1C levels for pa-
further demonstrated in Figure-2. During the nighttime, the tients with higher initial A1C. 88% of subjects with initial A1C
system suspended insulin a mean of 100 minutes and deliv- >7.0%, experienced a drop in their A1C levels by the end of the
ered insulin at the maximum limit a mean of 175 minutes. For study.
the remainder of 265 minutes, it delivered insulin between Conclusions: The HCL system was safe for unsupervised at-
1 99% of the maximum limit. home use by adults and adolescents. The system resulted in a
Conclusions: The greater variability of automated insulin significant decrease in A1C levels, especially when the initial
delivery, such as that achieved with the HCL algorithm com- A1C was >7.0%.
pared to pre-set basal rates, improves glucose control during the
night. The value of the HCL system is its ability to transfer
variability from glucose to insulin.

040
DELTA A1C IN THE PIVOTAL TRIAL
EVALUATING MEDTRONICS HYBRID
CLOSED-LOOP SYSTEM
A. Roy1, B. Grosman1, N. Parikh1, D. Wu1, N. Kurtz1,
R. Slover2, S. Garg2, J. Ilany3, B. Bode4, B. Buckingham5,
S. Weinzimer6, S. Anderson7, T. Bailey8, R. Bergenstal9,
F. Kaufman10
1
Medtronic Diabetes, R&D, Northridge, USA
2
Barbara Davis Center for Childhood Diabetes, Pediatric
Diabetes, Aurora, USA
3
Institute of Endocrinology, Sheba Medical Center,
Endocrinology, Tel Hashomer, Israel
A-16 ATTD 2017 ORAL ABSTRACTS

1
041 University of Colorado Denver, Barbara Davis Center for
Childhood Diabetes, Aurora, USA
ONE WEEKEND ON HYBRID-CLOSED-LOOP (HCL): 2
Rainier Clinical Research Center, Rainier Clinical Research
THE DREAM5-STUDY Center, Renton, USA
3
T. Biester1, I. Muller2, K. Remus1, S. Blasig1, E. Atlas2, AMCR Institute, AMCR Institute, Escondido, USA
4
R. Nimri3, K. Dovc4, N. Bratina4, T. Battelino4, M. Philip3, Atlanta Diabetes Associates, Endocrinology & Metabolism,
O. Kordonouri1, T. Danne1 Atlanta, USA
5
1 Sheba Medical Center, Institute of Endocrinology, Tel
Kinder- und Jugendkrankenhaus AUF DER BULT, Diabetes Hashomer, Israel
Center for Children and Adolescents, Hannover, Germany 6
Stanford University, Department of Pediatric Endocrinology,
2
DreaMed Diabetes Ltd., DreaMed Diabetes Ltd., Petach Stanford, USA
Tikvah, Israel 7
UVA Center for Diabetes Technology, UVA Center for
3
Schneider Childrens Medical Center of Israel, The Jesse Z Diabetes Technology, Charlottesville, USA
and Sara Lea Shafer Institute for Endocrinology and Diabetes, 8
University of Colorado Denver, Barbara Davis Center for
Petah Tikvah, Israel Diabetes, Aurora, USA
4
Ljubljana University Medical Centre, Department of 9
Yale University School of Medicine, Pediatric Endocrinology,
Endocrinology- Diabetes and Metabolism, Ljubljana, Slovenia New Haven, USA
10
International Diabetes Center, Park Nicollet Clinic,
Background and Aims: Previous DREAM studies showed Minneapolis, USA
the safety of the CE-marked closed loop (DreaMed Substance 11
Medtronic Diabetes, Clinical Research, Northridge, USA
Administration System) in overnight use (1 night, adolescents) 12
Medtronic Diabetes, Medical Affairs, Northridge, USA
at a Camp and at home (4 nights, all age groups). The objective of 13
Medtronic Diabetes, Clinical and Medical Affairs,
the present study was to evaluate the system for a 60 hours con- Northridge, USA
tinuous use, weekend time at home, without remote monitoring.
Methods: Subjects were randomized to one weekend with Background and Aims: At-home use of the MiniMed
sensor-augmented pump therapy (SAP) or HCL: respectively in a hybrid closed-loop (HCL) system on A1c levels, glycemia
crossover study. In the intervention arm only the amount of and glucose variability in adolescent and adult subjects with
carbohydrate was entered into the bolus calculator, automated type 1 diabetes (T1D) has been previously reported. The effect
insulin dosing was guided wirelessly by a tablet computer. of this system on nighttime (10:00pm-7:00pm) insulin delivery
Primary endpoint was the percentage of glucose values be- and sensor glucose (SG) time-in-range of 70-180 mg/dL were
tween 70-180 mg/dl. evaluated.
Results: 5 adults, 5 adolescents, 5 children (10f, 5m) experi- Methods: After a 2-week run-in period (HCL not enabled),
enced in sensor use were included: (median, [IQR]): age 16.8y 30 adolescents (14-21 years) and 94 adults (22-75 years) with
[12.9-18.5], diabetes duration 10.66y [7.1-13.8], pump use 10.7y T1D used the system with HCL enabled, for a 3-month at-
[5.3-12.6], HbA1c 7.6% [7.2-8.2].HCL lead to improved mean home study phase. Relative to the pre-set basal insulin de-
glucose: 173 57 mg/dl vs. 150 47 mg/dl, SAP vs. HCL, livery rates during run-in, the enabled HCL system allowed a
p = 0.021 and increased. variable basal insulin delivery rate ranging from 0 units to a
percentage of time in 70-180 mg/dl : 50.2% vs. 71.2% SAP vs. daily individualized insulin limit that could be delivered every
HCL, p = 0.006. 5 minutes.
Overall 11 events <60 mg/dl in SAP- and 12 in HCL-use oc- Results: The table shows the mean nighttime insulin delivered,
curred. No events of ketosis or severe hypoglycemia were observed. two measures of variability of nighttime insulin delivery (stan-
dard deviation (SD) and coefficient of variation (CV)), and SG
percentage of time in-target range (70-180mg/dL) during the run-
in and study phase; for all subjects, adolescents and adults. While
mean nighttime insulin delivered between the run-in and study
phase increased, it was not significant for either group. However,
there was a significant increase in the variability of insulin de-
livery between the run-in and the study phase, in addition to an
increase in the percentage of time in-target range during the study
Conclusions: The results confirm the security of this HCL in phase.
an around-the-clock setting. The system is safe and efficacious Conclusions: The dynamic insulin delivery rate, available
both in the administration of automated correction doses and in with the MiniMed 670G system, may partly contribute to im-
prevention of hypoglycemia. proved nighttime SG, suggesting that patients with sub-optimally
The lack of a remote monitoring did not lead to a worsening of controlled glycemia should benefit from this hybrid closed-loop
results or emerging dangerous events. system.

042
DYNAMIC INSULIN DELIVERY RATE AT NIGHTTIME
R. Slover1, R. Brazg2, T. Bailey3, B. Bode4, I. Jacob5,
B. Buckingham6, S. Anderson7, S. Garg8, S. Weinzimer9,
R. Bergenstal10, J. Shin11, S. Huang11, T. Cordero12,
F. Kaufman13
ATTD 2017 ORAL ABSTRACTS A-17

1
043 CERITD, Research, EVRY, France
2
CHSF, Service dendocrinologie, CORBEIL ESSONNES,
ORCHESTRA REGISTRY: A MULTICENTER France
OBSERVATIONAL STUDY ON THE USE OF CSII/SAP 3
CHU BESANCON, Service dEndocrinologie-Metabolisme
THERAPY BEFORE AND DURING PREGNANCY IN et Diabetologie-Nutrition, BESANCON, France
WOMEN WITH TYPE 1 DIABETES MELLITUS IN 4
CHU GRENOBLE, Pole DigiDune - Endocrinologie
POLAND Diabetologie, GRENOBLE, France
5
K. Cyganek1, J. Sieradzki2, K. Cypryk3, E. Wender-O_zegowska4, CEA_Leti- technology research institute, Micro Technologies
B. Telejko5, S. Runzis6, J. Shin7, K. Skoczylas8 Department for Biology and Health, GRENOBLE, France
6
1 CHU NANCY, Service de Diabetologie, NANCY, France
The University Hospital in Cracow, Department of Metabolic 7
CHU TOULOUSE, Service de Diabetologie- Maladies
Diseases, Cracow, Poland metaboliques et Nutrition, TOULOUSE, France
2
Jagiellonian University of Cracow, Collegium Medicum, 8
DIABELOOP SAS, Industrial development, GRENOBLE,
Cracow, Poland France
3
Medical University of Lodz, Diabetology and Metabolic 9
CHU STRASBOURG, Service dEndocrinologie- Diabete
Diseases Department, Medical University of Lodz, Lodz, Poland et des maladies de la Nutrition, STRASBOURG, France
4
Poznan University of Medical Sciences-, Department of 10
CHU MONTPELLIER, Service des maladies endocriniennes,
Reproduction, Poznan, Poland MONTPELLIER, France
5
Medical University of Bialystok, Department of 11
CAEN, Service dEndocrinologie, CAEN, France
Endocrinology, Diabetology and Internal Medicine Diseases, 12
Hopital de la conception AP-HM, Service de Nutrition-
Bialystok, Poland Endocrinologie-Maladies Metaboliques, MARSEILLE, France
6
Medtronic, Tolochenaz, Switzerland
7
Medtronic, Clinical Research, Northridge, CA, USA Background and Aims: Controlling BG after intensive
8
Medtronic, Clinical Department, Warsaw, Poland physical exercises (PE) is an issue in type-1-diabetes because
of an increased risk of immediate and delayed nocturnal hy-
Background and Aims: The ORCHESTRA Registry study is poglycemia. Diabeloop is a Closed-loop (CL) system with a
a multi-center, prospective, observational, post-market study MPC algorithm reinforced by a decisional matrix, uploaded
investigating the use of insulin pump therapy (Continuous Sub- on an android smartphone linked to Dexcom CGM and a
cutaneous Insulin Infusion, CSII) alone or incorporating sensor- Cellnovo insulin patch-pump. The primary aim of the study
augmented therapy (SAP) before and during the pregnancy in was to compare BG control following PE with Diabeloop
women with type 1 diabetes mellitus. CL versus sensor-augmented insulin pump (SAP), in well-
Methods: The study is conducted in 24 centers in Poland and educated patients practicing moderate or intensive PE repeat-
included total of 499 women. The aim is to follow up 100 women edly during 3-days.
from preconception phase and second group is observing already Methods: Fourteen T1D patients on insulin pump (48.7
pregnant women. Each participant is contributing data for up to 22 11.8 years old, diabetes duration: 20.9 10.6 years, HbA1c: 8.0
months (up to 12 months preconception, throughout pregnancy, and 0.5%) participated in a randomized three-center crossover study,
6 weeks after delivery): pregnant women can be enrolled up to the including two 72-hour periods, either with SAP or CL, included
16th week of pregnancy, while patient in the planning group have up similar standardized meals and exercises. PE were performed on a
to 1 year to conceive. Data are collected during routine follow-up cycle ergometer (1st-day: 45mn-intense exercise VO2max = 75%;
visits. Maternal outcomes include HbA1c before and during preg- 2nd-day 30mn moderate exercise VO2max = 50%; 3rd-day: 30mn-
nancy, serious adverse events and neonatal outcomes will be eval- moderate exercise in the morning and 45mn-intense exercise in the
uated for both groups. The CSII devices used in the study are the afternoon). Meals and exercises were announced in CL period.
Medtronic Paradigm REAL-time (722) insulin pump, including Results: Time in range 80-140mg/dL during the 3-nights, was
the sensor augmented Paradigm Veo (754) system (insulin significantly higher in CL compared with SAP (66.6[57.8;76.9]
pump plus Enlite sensor plus MiniLink transmitter). vs. 30.9[17.2;55.3]% P < 0.0076) as was time in range 70-180 mg/dL
Results: The enrollment started in May 2013 and was com- during the whole 72hrs (79.8[74.2;85.7] vs. 61.8[52.7;72.5]%
pleted in December 2015, and to date 221 and 60 patients de- P = 0.0021). Three-day mean BG was significantly lower in CL
livered from pregnancy group and planning group respectively. than in SAP (138.0[129.4;147.1] vs. 155.8[142.8;170.0]mg/dL,
Last patient will complete the study in September 2017. P = 0.0037). Day and nocturnal times spent in hypoglycemia
Conclusions: The Orchestra Registry offers the opportunity to were significantly lower in CL.
describe pregnancy outcomes according to the use of CSII with Conclusions: In case of repeated PE, Diabeloop was able to
or without CGM either before or during pregnancy. double the time spent in the 80-140mg/dL range and to reduce
time in hypoglycemia during the night following PE, compared
to SAP.

044
045
DIABELOOP CLOSED-LOOP DOES BETTER THAN
SENSOR-AUGMENTED-PUMP ON BLOOD GLUCOSE DIABELOOP CLOSED-LOOP DOES BETTER THAN
DURING 3 DAYS WITH INTENSIVE PHYSICAL SENSOR-AUGMENTED PUMP ON BLOOD GLUCOSE
EXERCISES: A RANDOMIZED CROSSOVER TRIAL CONTROL DURING 3 DAYS WITH GASTRONOMIC
DINNERS: A RANDOMIZED CROSSOVER TRIAL IN
S. Franc1,2, S. Borot3, P.Y. Benhamou4, M. Doron5, B. Guerci6,
PATIENTS WITH T1D
H. Hanaire7, E. Huneker8, N. Jeandidier9, E. Renard10,
Y. Reznik11, I. Xhaard1, A. Penfornis2, G. Charpentier1,2, H. Hanaire1, P.Y. Benhamou2, S. Borot3, M. Doron4,
P. Schaepelynck12 S. Franc5,6, B. Guerci7, E. Huneker8, N. Jeandidier9,
A-18 ATTD 2017 ORAL ABSTRACTS

Y. Reznik10, P. Schaepelynck11, I. Xhaard5, A. Penfornis6, 046

G. Charpentier5,6, E. Renard12
GLUCOSE CONTROL WITH MD-LOGIC
1
University hospital of Toulouse, Service de Diabetologie- CLOSED-LOOP SYSTEM USE IN YOUTH WITH
Maladies metaboliques et Nutrition, TOULOUSE, France TYPE 1 DIABETES DURING PHYSICAL ACTIVITY
2
University hospital of Grenoble, Pole DigiDune, AND THE NIGHT AFTER
Endocrinologie Diabetologie, GRENOBLE, France
3 K. Dovc1, M. Macedoni2, N. Bratina1, D. Lepej3, R. Nimri4,
University hospital of Besancon, Service dEndocrinologie-
M. Phillip4, E. Atlas5, I. Muller5, T. Danne6, O. Kordonouri6,
Metabolisme et Diabetologie-Nutrition, BESANCON,
T. Biester6, T. Battelino1
France
4 1
CEA_Leti, technology research institute, Micro University Childrens Hospital, Department of Endocrinology-
Technologies Department for Biology and Health, Diabetes and Metabolism, Ljubljana, Slovenia
2
GRENOBLE, France Ospedale dei Bambini Vittore Buzzi, Department of Pediatrics-
5
CERITD, Research, EVRY, France University of Milano- Milan- Italy, Milano, Italy
6 3
Hospital center of Sud Francilien, Service de diabetologie, University Childrens Hospital, Department of Pulmonology-,
EVRY, France Ljubljana, Slovenia
7 4
University hosiptal of Nancy, Service de Diabetologie, Schneider Childrens Medical Center of Israel, Jesse Z. and
NANCY, France Sara Lea Shafer Institute for Endocrinology and Diabetes-
8
DIABELOOP SAS, Industrial development, GRENOBLE, National Center for Childhood Diabetes, Petah-Tikva, Israel
5
France DreaMed Diabetes Ltd, DreaMed Diabetes Ltd, Petah-Tikva,
9
University hospital of Strasbourg, Service dEndocrinologie- Israel
6
Diabete et des maladies de la Nutrition, STRASBOURG, Kinder- und Jugendkrankenhaus AUF DER BULT, Diabetes
France Centre for Children and Adolescents, Hannover, Germany
10
University hospital of CAEN, Service dEndocrinologie,
CAEN, France Background and Aims: The objective of this clinical study
11
University hospital of Marseille, Service de Nutrition- was to evaluate the safety and the efficacy of closed-loop (CL)
Endocrinologie-Maladies Metaboliques, Marseille, insulin delivery system MD-Logic and the occurrence of hypo-
France glycemia during physical activity and the night after in children
12
University hospital of Montpellier, Service des maladies and adolescents with type 1 diabetes (T1D).
endocriniennes, MONTPELLIER, France Methods: In a two-arm, randomized, open-label, in-hospital,
crossover clinical trial 20 children on insulin pump therapy (9
Background and Aims: Blood glucose (BG) control after females, mean age 14.24 2.02 years, HbA1c 7.67 0.62%)
gastronomic dinners is an issue in type 1 diabetes (T1D) as performed two exercise protocols: moderate (55% of VO2max)
CHO can be very difficult to evaluate and increased amount or physical activity (PA) and moderate PA with integrated sprints
fat or protein can induce delayed long duration hyperglycaemia (55/80% of VO2max) in 4 visits with CL use and in a control
during night-time due to transient impaired insulin sensitivity open-loop (OL) group in random order. Subjects in control group
and/or hyperglucagonemia. Diabeloop is a Closed-loop (CL) disconnected the pump during the exercise and reduced the basal
system with a MPC algorithm reinforced by a decisional ma- for 20% for 4 hours after exercise.
trix, uploaded on a dedicated android smartphone linked to Results: Closed-loop system use increased the proportion of time
Dexcom CGM and a Cellnovo insulin patch-pump. The pri- within the target glucose range 70 180 mg/dl (3.9 10 mmol/l) for
mary aim of the study was to compare BG control with Dia- the whole study period (84.08 (IQR: 70.03 85.48) % for CL and
beloop CL versus sensor-augmented insulin pump therapy 68.67 (IQR: 56.00 77.70) % for OL, p = 0.006) This was achieved
(SAP), in well-educated patients enjoying gastronomic dinners with significantly less basal insulin delivered (p = 0.007) in CL. The
during 3 days. proportion of time in hypoglycemia below 60 mg/dl (3.3 mmol/l)
Methods: Ten T1D patients on insulin pump therapy (46.4 was 0.00% for both groups (IQR: 0.00 0.76% for CL and 0.00
16.1 years old, diabetes duration: 23.6 13.1 years, HbA1c: 1.06% for OL, p = 0.79).
7.7 0.8%) participated in a randomized three-center crossover Conclusions: CL insulin delivery was safe and effective in
study, including two 72-hour periods, either with SAP or CL. maintaining glucose values with significantly more time spend
Each session included three different dinners (1st-day: Japanese within target range without increased risk of hypoglycemia in the
meal, 2nd-day: French gastronomy, 3rd-day: pizza-tiramisu meal) in-hospital environment.
and three usual breakfasts and cafeteria lunches. Meals were
announced for bolus computing. 047
Results: Percent time in 80-140mg/dL range during the 3 nights,
was significantly higher in CL compared with SAP (58.4[49.7;68.7] FEASIBILITY OF OMNIPOD HYBRID CLOSED-LOOP
vs. 14.7[7.6;28.6], P < 0.0001) as was %time in 70-180mg/dL target CONTROL IN ADULTS WITH TYPE 1 DIABETES
range during the whole 72hrs: 80.2[73.9;87.1] vs. 51.7[43.1;62.0], USING A PERSONALIZED MODEL PREDICTIVE
P < 0.0001. Three-day mean BG(mg/dL) was significantly lower CONTROL ALGORITHM
in CL than in SAP (141.7[131.7;152.4] vs. 169.7[153.7;187.5],
B. Buckingham1, J. Pinsker2, M. Christiansen3, J. Schneider4,
P = 0.0003).
T. Peyser4, E. Dassau5, J. Bok Lee6, J. OConnor6, J. Layne7,
Conclusions: In difficult situations such as repeated heavy
T. Ly7
dinners, Diabeloop CL system was able to increase by nearly
1
four-fold the %time spent in 80-140mg/dL BG range during Stanford University, Department of Pediatrics, Stanford, USA
2
the night, and to decrease the average BG level compared to William Sansum Diabetes Center, Pediatric Endocrinology,
SAP. Santa Barbara, USA
ATTD 2017 ORAL ABSTRACTS A-19

3
Diablo Clinical Research, Endocrinology, Walnut Creek, USA cadaveric donor and chronic immunosuppression. Islet encap-
4
ModeAGC LLC, Research and Development, Palo Alto, USA sulation has been potential for decades, but has been limited
5
Harvard University, Harvard John A. Paulson School of especially in its application in autoimmune diabetes. We have
Engineering and Applied Sciences, Cambridge, USA improved on material selection, purification, and capsule design
6
Insulet Corporation, Research and Development, Billerica, to produce consistent 400 lm diameter purified islet capsules.
USA Methods: Rat islets were encapsulated using extra-purified
7
Insulet Corporation, Clinical Affairs, Billerica, USA UPLVM alginate (Gp1) or in unpurified alginate (Gp2). Diabetic
NOD mice (>3 days of non-fasting hyperglycemia >350 mg/dL), 9-
Background and Aims: To investigate the safety and per- 10 weeks old, received encapsulated islet transplant into the peri-
formance of an automated glucose control algorithm using the toneal cavity (Gp1 and 2, n = 10 per group), or non-transplanted as
Omnipod Insulin Management System. The system included a control (Gp3). Blood glucose and body weight was measured, and
modified version of Omnipod, Dexcom G4 continuous glucose glucose tolerance was tested at 1 month post-transplant. Mice were
sensor and a personalized model predictive control algorithm. euthanized after 2 months, and explants were analyzed by histology
Insulin dosing was implemented based upon a 60 minute glucose for CD68, Masson Trichrome, and insulin.
prediction horizon. Results: Using viable and functional encapsulated islet
Methods: The study consisted of a 36-hour inpatient hybrid transplant in Gp1 and Gp2, in Gp1, 8/10 mice returns to eu-
closed-loop assessment with announced meals ranging from 30- glycemia (bg <200 mg/dL) within 3 days of transplant. Gp2 mice
90g of carbohydrates and limited physical activity. Subjects aged also reversed hyperglycemia, but within 5 days retuned back to
18-65 y with type 1 diabetes, A1C between 6.0-10.0% and total the diabetic state. Control mice remained hyperglycemic for
daily dose (TDD) of insulin >0.4 U/kg/day were eligible. End- >30 days after diabetes induction Glucose tolerance test in Gp1
points included sensor glucose percentage of time 70-180 mg/dL, mice showed equivalent function to non-diabetic mice. Histol-
<70 mg/dL, >180 mg/dL and mean blood glucose. ogy showed intact islet capsules (Gp1), compared to free and
Results: Baseline characteristics for 24 subjects included damaged islets, immune cells, and capsule debris (Gp2).
(meanSD): age 36.3 15.0 y, diabetes duration 24.9 14.7 y, Conclusions: Our results indicated that improved purification
A1C 7.6 0.9% and TDD 0.6 0.2U/kg. Glycemic outcomes steps is necessary for a successful encapsulated islet transplant,
are reported in the table. Mean percentage of time in range and maybe indicative of the problems encountered in other en-
70-180 mg/dL was 69.3% overall and 90.0% during the over- capsulated islet and xenotransplant studies.
night period. There were 0.3 hypoglycemic events/day, with
meter glucose <70 mg/dL, during 787 hours of system use.

049
PATIENT PERSPECTIVES, TECHNOLOGY
ATTITUDES AND GLUCOSE CONTROL: A MIXED-
METHOD STUDY OF AN ARTIFICIAL PANCREAS
TRIAL IN PREGNANCY
C. Farrington1
1
University of Cambridge, Cambridge Centre for Health
Services Research, Cambridge, United Kingdom
Conclusions: The Omnipod prototype automated glucose
control algorithm performed well and was safe during day and Background and Aims: Recent publications (e.g. Stewart
night use in patients with type 1 diabetes. Additional studies are et al., N Engl J Med 2016:375) demonstrate the efficacy, safety
underway to further validate the performance in the pediatric and feasibility of artificial pancreas (AP) systems in pregnancy.
population. Less is known about users perceptions of AP systems - e.g. their
views about their level of control, or how their wider attitudes to
technology may influence their experiences.
Methods: A mixed-methods study embedded within the trial
048 reported by Stewart et al. The study incorporated semi-structured
qualitative interviews with 14 women alongside biomedical data
SUCCESSFUL ISLET XENOTRANSPLANTATION TO on glucose control response to AP therapy (relative to response to
RESTORE GLYCEMIC CONTROL IN DIABETIC NOD sensor-augmented pump therapy).
MICE WITH NOVEL ALGINATE PURIFICATION Results: Overall, patients control with AP therapy was
TECHNOLOGY slightly worse than patients believed. However, there was sig-
nificant inter-participant variation, with some patients reporting
M. Alexander1, A. Flores1, S. Lee1, S. Zhang1, T. Tucker1, S. Li1,
markedly better control than they in fact achieved (n = 3) and
P. de Vos2, J. Lakey1,3
others reporting control that was moderately better or worse than
1
University of California Irvine, Surgery, Orange, USA they achieved (n = 5). Six patients reported levels of control that
2
University Medical Center Groningen, Pathology and Medical matched biomedical data. With regard to technology attitudes,
Biology, Groningen, The Netherlands patients ending the study with more positive attitudes towards
3
University of California Irvine, Biomedical Engineering, technology not only demonstrated better opinions of their control
Irvine, USA but also worse control than those with more neutral attitudes
towards technology. Once again, participants differed markedly.
Background and Aims: Islet transplantation is an attractive Conclusions: These findings highlight the complexity of the
option for treatment of type 1 diabetes, but are limited by lack of relationship between patient attitudes and biomedical outcomes.
A-20 ATTD 2017 ORAL ABSTRACTS

3
It cannot be assumed that AP users have accurate perceptions of the LifeScan- Johnson & Johnson Medical GmbH, Strategic
level of control they achieve; nor can it be assumed that positive Affairs, Neuss, Germany
4
attitudes to technology will align with better control using AP Sciarc GmbH, Sciences, Baierbrunn, Germany
5
systems. To allow a range of users to achieve good control, future Diabetologische Schwerpunktpraxis, Institut fur
interventions should take account of this complexity. gesundheitliche Fortbildung, Rosenheim, Germany

Background and Aims: Self-monitoring of blood glucose

(SMBG) is a key component to personal diabetes management.
050 OneTouch Verio is a blood glucose meter system allowing for
an instant, color-coded display of glucose levels, which helps
DIGITALLY DELIVERED DIABETES SELF-
identify out of range glucose levels. The objective of the ob-
MANAGEMENT SUPPORT: KEYS TO SUSTAINING
servational study was to analyze diabetes self-management be-
AND SCALING PROGRAMS THAT WORK
fore and during system implementation in real-life settings and
N. Kaufman1 demonstrate its effects on glycemic control.
1 Methods: 193 insulin-treated diabetic patients (11% T1DM;
Canary Health, Medical, Los Angeles, USA 45% female, age 60 14 yr, diabetes duration 14 9 yr, HbA1c
8.7 1.9%) were provided a OneTouch Verio. Both the Diabetes
Background and Aims: Individuals with type 2 diabetes must Self-Management Questionnaire (DSMQ) and glycemic control
manage their everyday lives and their behaviors if they are going were analyzed at baseline, and 3 and 6 months after study initiation.
to remain healthy and slow the rate of adding additional condi- Results: OneTouch Verio resulted in improved glycemic
tions or complications. When trying to impact outcomes for a control, as shown by decreased mean HbA1c, from 8.7 1.9% at
population of patients, key challenges for success are to (1) en- baseline to 8.1 1.0% after 3 months (p = 0.0009) and to
gage enough of the population to take positive action, (2) provide 7.9 1.1% at 6 months (p < 0.0001).
an effective approach to those who engage which can promote Compared to baseline, patients attitudes with respect to nu-
sustainable behavior changes and improved outcomes, (3) pro- trition, lifestyle, and adherence showed a significant improve-
vide the intervention at price and logistical burden that is ac- ment, as observed from the results from the DMSQ (p < 0.001).
ceptable to whomever is paying for the services and supports, and Both patients and diabetes advisor personnel reported similar
(4) impact outcomes that matter to those who can scale and advantages of the color-coded indicator of glucose levels pro-
sustain a successful program. vided by OneTouch Verio and helpfulness in assisting pa-
Methods: 1,010 patients with Type 2 diabetes were recruited tients with diabetes management, as drawn from the results of the
from a large US Health Plan and provided a six week intervention self-reported satisfaction questionnaire.
to increase self-efficacy especially related to 3 key issues in self- Conclusions: This real-world study demonstrates that SMBG
management of a chronic disease- role management, chronic implemented via OneTouch Verio leads not only to an im-
condition management, and emotional management. Individuals provement in metabolic control, but is also associated with a
were part of a 20-30 person online group, facilitated by 2 peers. significant improvement in diabetes management.
In addition to specific content, individuals created action plans
and posted concerns and celebration which were commented on
by other participants in the workshops.
Results: N = 1,010 enrolled and demonstrated at 6 month post 052
the end of the program:
THE HYPO-TRIAD: A NEW COMPOSITE METRIC
1. HbA1c decreased 0.93% for those >9% FOR THE COMPREHENSIVE ASSESSMENT OF
2. Depression improved HYPOGLYCEMIA
3. Frequent hypoglycemic symptoms decreased
4. Medication-taking adherence improved R. Vigersky1, A. Thomas2, J. Shin3, B. Jiang4, C. McMahon4,
5. Exercise increased 43 minutes per week R. Kolassa5
1
Medication adherence, doctor-patient communication and Medtronic Diabetes, Medical Affairs, Northridge, USA
2
self-efficacy also improved. Medtronic GmbH, Market Development, Meerbusch, Germany
3
Conclusions: Digital delivery of a self-management support Medtronic Diabetes, Clinical Research, Northridge, USA
4
intervention proven effective in person can positively impact Medtronic Diabetes, Engineering, Northridge, USA
5
outcomes that matter for patients, providers and payers. Private Practice, Private Practice, Bergheim/Erft, Germany

Background and Aims: Hypoglycemia is usually expressed

uni-dimensionally (# of daily episodes; area-under-the-curve
051 (AUC) below a threshold, e.g., 50 or 70 mg/dL). This approach
describes neither the intensity nor overall impact of hypoglyce-
ONETOUCH VERIO: IMPACT ON DIABETES mia on a patient. The Hypo-Triad is a composite metric
SELF-MANAGEMENT AND GLYCEMIC CONTROL combining AUC, time, and frequency of events to more com-
pletely reflect a patients daily exposure to hypoglycemia.
O. Schnell1, G. Klausmann2, B. Gutschek3, R. Garcia-Verdugo4,
Methods: We first defined a 2-component metric - intensity of
M. Hummel5
hypoglycemia (IntHypo) - as the combination of AUC and time/
1
Forschergruppe Diabetes e.V., Diabetology, Munich day of sensor glucose less than 50 or 70 mg/dL expressed as the
Neuherberg, Germany vector length (Figure). The HypoSpace or Hypo-Triad is a
2
Gemeinschaftspraxis Drs. Klausmann, Studienzentrum 3-component metric adding frequency/day of hypoglycemia. We
Aschaffenburg, Aschaffenburg, Germany calculated these composite metrics using data from the ASPIRE
ATTD 2017 ORAL ABSTRACTS A-21

Conclusions: The novel smartphone-based blood glucose

monitoring system showed a high level of accuracy fulfilling ISO
15197:2013 (EN ISO 15197:2015) system accuracy criteria. The
possibility to discretely and conveniently measure blood glucose
levels might support regular self-monitoring of blood glucose
which is important for an efficient diabetes therapy.

054
IMPROVEMENT OF GLYCEMIC CONTROL
WITH REMOTE SUPPORT IN CHILDREN
ADOLESCENTS WITH TYPE 1 DIABETES
MELLITUS ON INSULIN-PUMP THERAPY
In-Home trial which compared low-glucose suspend (LGS)
D.N. Laptev1, A.O. Emelyanov1, S.V. Pereverzeva1,
(N = 121) to sensor-augmented pump (SAP) (N = 126) therapy (1).
A.A. Kukushkina1, A.V. Karpushkina1, V.A. Peterkova1,
(1) Bergenstal RM, et al. N Engl J Med 2013;369:224-232.
I.I. Dedov1
Results: LGS reduced IntHypo under 50 mg/dL by 56.1%
(181.6 277.6 to 79.2 85.4; p = 0.001) compared to a 2.2% 1
Endocrinology Research Centre, Department of Pediatric
increase for SAP (176.4 2173 to 180.4 181.4; p = 0.075) Endocrinology, Moscow, Russia
(p = 0.003 between groups). The HypoSpace under 50 mg/dL
rose 68% in the SAP group (p = 0.001) while the LGS group had a Background and Aims: Healthcare access plays a significant
9.3% decrease (p = 0.352) (p = 0.007 between groups). There was role in the maintaining glycemic control and quality of life for
a similar reduction in the IntHypo under 70/mg/dL by 31.4% diabetic patients on continuous subcutaneous insulin infusion
compared to a 14.3% decrease for SAP. The HypoSpace under (CSII). The aim of our study was to evaluate the feasibility of
70 mg/dL rose in the SAP group (+44.8%; p = 0.03) while there remote support in children and adolescents with type 1 diabetes
was no change in the LGS group (+2.2%; p = 0.308). mellitus (T1DM) and its effect on glycemic control.
Conclusions: The Hypo-Triad is a composite metric that Methods: In 40 children and adolescents (13 2.8 years, 18/
provides a comprehensive and clinically relevant way to assess 22 m/f) on CSII with inadequately controlled T1DM (HbA1c7.5%)
hypoglycemia numerically and graphically. we evaluated the effectiveness of telemedical support (TS), as
compared with conventional support (CS). Patients and their parents
in TM group twice a month sent their insulin pump data using
053 CareLink software to diabetes center and diabetologists sent back
ACCURACY EVALUATION OF A NOVEL SYSTEM their advice via e-mail or Skype. The primary end point was the
FOR SMARTPHONE-COUPLED BLOOD GLUCOSE change from the baseline HbA1c level.
MONITORING Results: At 6 month, the baseline mean HbA1c (8.7% in the
two study groups) had decreased to 7.7% in the TS group, as
N. Jendrike1, A. Baumstark1, S. Pleus1, D. Rittmeyer1, compared with 8.4% in the CS group (P < 0.05). The proportion of
U. Kamecke1, C. Haug1, G. Freckmann1 patients who reached the HbA1c target (<7.5%) was greater in the
1
Institut fur Diabetes-Technolgie Forschungs- und TS group (50%) than in the CS group (20%, p < 0.05). The pro-
Entwicklungsgesellschaft mbH, an der Universitat Ulm, Ulm, portion of readings above target (10 mmol/l) was 48.0% in CS
Germany group, as compared with 32.2% in TS group (P < 0.05).The pro-
portion of readings below target (4 mmol/l) was non-significantly
Background and Aims: While conventional blood glucose lower in TS group (P > 0.05).
monitoring systems consist of meters and corresponding test Conclusions: In children with inadequately controlled type 1
strips, the novel PixoTestTM (iXensor Co. Ltd.) does not require a diabetes, telemedical support proved to be feasible and resulted
separate meter but glucose values are measured by a smartphone in significant improvement in HbA1c levels, as compared with
camera through color change of test fields inside a small device conventional support.
integrated with a lancet and attached to the smartphone. In this
study, accuracy of this system was evaluated. 055
Methods: Capillary blood samples with defined glucose con-
centration distribution from 100 subjects were measured with each PROJECTION OF HEALTH-ECONOMIC BENEFITS
of three different reagent system lots of the system as well as with a OF SENSOR-AUGMENTED PUMP (640G AND PLGM
glucose oxidase-based comparison method. The applicable stan- SYSTEM) VERSUS INSULIN PUMP THERAPY ALONE
dard ISO 15197:2013 (EN ISO 15197:2015) specifies that 95% of (CSII) IN TYPE 1 DIABETES PATIENTS IN SWEDEN
measured values have to be within 15 mg/dl of a comparison J. Jendle1, A. Delbaere2, S. de Portu2, S. Roze3
method at glucose concentrations <100 mg/dl and within 15% at
1
higher concentrations. In addition, 99% of the results shall fall in Orebro University, Faculty of Medical Sciences, Orebro,
zones A and B of the consensus error grid. Percentages of the Sweden
2
results within the above-mentioned limits were calculated. Medtronic International Trading SARL, HEOR, Tolochenaz,
Results: All three lots fulfilled the ISO 15197:2013 (EN ISO Switzerland
3
15197:2015) criteria with 99%, 97% and 98.5% of values, re- HEVA HEOR, HEOR, Lyon, France
spectively, within the accuracy limits and 100% of values within
consensus error grid zone A. The relative bias ranged from Background and Aims: To assess the cost-effectiveness and
-0.6% to -1.0%. to project the clinical benefits of the Sensor-Augmented-Pump
A-22 ATTD 2017 ORAL ABSTRACTS

640G and PLGM system (SAP) versus Continuous Subcutaneous Background and Aims: Clinical decision support systems
Insulin Infusion alone (CSII) in type 1 diabetes (T1D) patients in (CDSS) can be specifically designed to empower people with
Sweden. type 2 diabetes (T2D), not on insulin, to interpret structured
Methods: The Core Diabetes Model was used to project the glucose data and make informed health decisions. Our CDSS
incidence of diabetes-related complications for a T1D population used best-in-class decision theory, control theory, artificial in-
with uncontrolled HbA1c and for a population with hypo- telligence, data mining, data science, and machine learning in a
unawareness. The Pickup meta-analysis showed, for the uncon- cloud based environment. Participants conducted paired glucose
trolled cohort based on the Swedish National Diabetes Registry tests using in-home glucose monitors with results transmitted,
NDR-2014, (mean HbA1c: 63.14 mmol/mol; age 46 years), a via remote monitoring platform, to a clinician portal. Feedback
HbA1c reduction of -0.58% versus -0.14% for SAP and CSII and recommendations were compiled and then delivered to pa-
respectively. In the hypo-unaware population (Hba1c 58.5 tients in an intuitive, supportive, and graphical manner with
mmol/mol, age 18.6 years) severe hypoglycemic events were dashboards for ease of understanding enabling a complete
observed at 2.2 and 0 (based on a randomized trial from Ly et al.) feedback loop. The core software facilitated customization by
per 100 patients months for CSII and SAP respectively. participant using up-to-date real-time processing.
Results: The incremental cost effectiveness ratio (ICER) was Methods: Data were analyzed using pattern analysis and
251896 SEK per Quality Adjusted Life Year gained (QALY) for summarized by premeal, postmeal, and change values following
the uncontrolled T1D-population. In the hypo-unaware popula- ADA guidelines. Individualized feedback was augmented and
tion the ICER was 139795 SEK/QALY. The improvement in personalized using secure messaging via the EHR. Participants
QALYs were 1.07 and 1.88 years in favor of SAP in the un- were encouraged to make one or two small changes. The com-
controlled and hypo-unaware population respectively. In the plete feedback loop repeated weekly.
uncontrolled T1D-population, as compared to CSII, SAP delayed Results: Participants (n = 40) averaged 38/90 (42%) paired
the onset of complications such as neuropathy (1.04 years), ulcer glucose tests over 12 weeks. 32% (n = 193/606) CDSS secure
(0.92 years) and stroke (0.64 years). Extensive sensitivity anal- messages were read by participants, of those 67% (n = 130/193)
ysis on key drivers confirmed the robustness of results. responding to the clinician engaging in discussion. 96% of par-
Conclusions: These results indicate that the pump 640G with ticipants agreed or strongly agreed they were more involved in their
Predictive Low Glucose Management can be considered good care and 93% reported their health was better following CDSS.
value for money compared to CSII alone in both uncontrolled Conclusions: CDSS designed for T2D, incorporating a com-
hyperglycemia and hypo-unaware T1D-populations. plete feedback loop, engages people in self-management increas-
ing the likelihood of improved health outcomes. Opportunities to
simplify participant-clinician communication are required to in-
crease productive interactions.

057
GLUCOMMANDER OUTPATIENT, A CLOUD-BASED
INSULIN MANAGEMENT SOLUTION, TITRATED
PATIENTS TO GOAL IN 12.5 DAYS AND SUSTAINED
A 2.5% DROP IN HBA1C OVER 6 MONTHS
J. Clarke1, B. Bode2
1
Atlanta Diabetes Associates, Education, Atlanta, USA
2
Atlanta Diabetes Associates, Research, Atlanta, USA

Background and Aims: Treatment of diabetes with multiple

daily injections, using a regimen of basal and rapid-acting insulins,
is widely accepted as a standard of care in the treatment of diabetes.
This study evaluated Glucommander Outpatient, a commercially
available, cloud-based software solution that analyzed patient
glucose data and calculated individualized insulin doses.
Methods: The study was comprised of 38 patients who were
previously treated with standard of care methods. The patient
mix included both type 1 and type 2 diabetes. All patients had
056 HbA1c measurements at start and 6-months; a 3-month HbA1c
was also collected. All insulin dose adjustments were calculated
ENGAGING PEOPLE WITH TYPE 2 DIABETES: by Glucommander Outpatient.
CLINICAL DECISION SUPPORT ANALYZES Glucose data was collected using the Telcare glucose meter,
STRUCTURED GLUCOSE DATA TO SUPPORT which was able to provide near real-time glucose results into the
SELF-MANAGEMENT cloud. Adjusted insulin doses from the software were provided to
a healthcare professional for review then sent to the patient,
D. Greenwood1, T. Forsman2
either on the display of the meter or via a text message.
1
Sutter Health, Research Development and Dissemination, Results: During treatment with Glucommander Outpatient,
Granite Bay, USA HbA1c decreased from a baseline average of 10.1% to 7.7% at 3
2
Lumina Decisions Systems- Inc, Senior Consulting Analyst, months and 7.6% at 6 months (P < 0.00001 by paired T-test).
Campbell, USA Hypoglycemia was infrequent throughout (see Table). Out of a
ATTD 2017 ORAL ABSTRACTS A-23

self-management and potentially improve post-prandial glucose

control in people with T1D.

059
MULTI-LAYERED DECISION SUPPORT SYSTEM
FOR INSULIN TREATED PATIENTS WITH DIABETES
SHOWS IMPROVEMENT IN SHORT TERM
GLUCOSE CONTROL

total of 17,058 BGs recorded, the average number of BG tests per C. Rogers1, M. Breton2, S. Patek2, P. Keith-Hynes1
day was 2.9 during the first 3 months and 2.0 during the final 1
TypeZero Technologies LLC, CEO, Charlottesville, USA
3 months. The median time for patients to achieve three con- 2
UVA, Center for Diabetes Technology, Charlottesville, USA
secutive days of BGs <10 mmol/l (180 mg/dl) was 12.5 days.
Conclusions: Glucommander Outpatient was found to be easy Background and Aims: Insulin dosing to control glucose
and effective to use by both client and provider. levels in Diabetes is a complex, multi-factorial process where
small errors can lead to significant discomfort and potential harm
via exposure to hypo and hyperglycemia. With the University of
Virginia, TypeZero Technologies has developed unique, user
058 centric, decision support technologies (DSTs) to facilitate a safe
A WEB-BASED CLINICAL PLATFORM FOR and stress free life with Diabetes.
AUTOMATED MEAL BOLUS ADJUSTMENTS Methods: TypeZero DSTs are informed by the concept of
actionable risk; they currently consist of 6 modules im-
P. Pesl1, M. Reddy2, N. Oliver2, D. Johnston2, C. Toumazou1, plemented on mobile devices: (i) a biweekly treatment para-
P. Herrero1, P. Georgiou1 digm review, proposing modifications to the insulin therapy
1
Imperial College London, Institute of Biomedical Engineering, to improve glucose control, (ii) an estimate of glycemic
Centre for Bio-Inspired Technologies, London, United Kingdom control trends (estimated HbA1c), (iii) a CGM based bolus
2
Imperial College London, Division of Diabetes, Endocrinology calculator, accounting for the metabolic state of the user, (iv)
and Metabolic Medicine, London, United Kingdom a CGM-based hypoglycemia alert: predictions up to 3h ahead,
(v) an exercise adviser: optimal insulin/carbohydrate modu-
lation, and (vi) a bedtime adviser: appropriate mitigation for a
Background and Aims: Insulin bolus calculators use pa-
safe night of sleep. DSTs are currently being tested in a ran-
rameters insulin:carbohydrate ratio (ICR) and insulin-sensitivity
domized crossover clinical trial, recruiting 35 subjects with
factor (ISF) to calculate the amount of insulin needed to cover a
type 1 diabetes (NCT02558491), consisting of two 48h ad-
meal. Because of significant variability in insulin require-
missions (usual control vs. adviser), with standardized meals
ments, these parameters often need re-adjustments by a clinical
and exercise.
expert in order to maintain or improve glucose control. How-
Results: 15 subjects have completed the trial, with enrollment
ever, adjustments are often based on retrospective glucose
continuing. Partial results demonstrate a strong trend to better
data and can be challenging if clinical visits are scheduled in-
glucose control: time between 70-180mg/dl rose from 56% to
frequently. Current diabetes management systems are able to
68%, <70mg/dl decreased to 1.8% from 2.9%, and glucose
visualise and summarise glucose related data but do not auto-
variability (ADRR) decreased to 27 from 41.
matically provide insulin therapy adjustments for people with
Conclusions: The TypeZero decision support technologies
type 1 diabetes (T1D).
are currently being validated in clinical trials and show encour-
Methods: We have developed a secure web-based clinical
aging results towards improved and safer glucose control. A
platform with the capability of automated insulin therapy ad-
3-months home study is planned for 2017.
justments, that works in combination with the previously pre-
sented Advanced Bolus Calculator for Diabetes (ABC4D).
ABC4D recommends meal boluses via the users smartphone
and periodically sends glucose related data (e.g. meal and in-
sulin information, exercise, continuous glucose monitoring
data) via WIFI or 3G to the clinical web-platform. The system
automatically analyses the data and optimises parameters ICR
and ISF, which are then synchronised with the ABC4D plat-
form.
Results: A prototype system has been developed and tested
with real-life glucose data obtained from previous clinical trials.
The web-based clinical platform is planned to be included in a
large-scale study evaluating the efficacy of ABC4D.
Conclusions: Automated insulin therapy adjustments, as im-
plemented within the proposed platform, may aid clinical experts
in optimising meal dosing parameters (ICR and ISF), encourage
A-24 ATTD 2017 ORAL ABSTRACTS

060 Background and Aims: To evaluate the long-term accept-

ability and usability of the Advanced Bolus Calculator for Dia-
CLINICAL DECISION SUPPORT FOR THE
betes (ABC4D) system. The ABC4D is an insulin bolus dose
MANAGEMENT OF TYPE 2 DIABETES IN HOSPITAL
decision support system based on case-based reasoning (CBR)
WARDS - DEVELOPMENT AND EVALUATION OF A
and works with real-time continuous glucose monitoring (RT-
MEDICAL DEVICE SOFTWARE SYSTEM
CGM). The system is implemented in a smartphone application
P. Beck1,2, K.M. Lichtenegger3, J.K. Mader3, K. Donsa2, to provide personalized and adaptive insulin bolus advice for
B. Holl2, S. Spat2, J. Plank3, T.R. Pieber2,3 people with type 1 diabetes(T1D).
1 Methods: The ABC4D system is currently being evaluated in
decide Clinical Software GmbH, SME, Graz, Austria a prospective 6-months blinded randomised controlled trial. At
2
Joanneum Research, HEALTH - Institute for Biomedicine the end of the study participants are asked to complete a ac-
and Health Sciences, Graz, Austria ceptability questionnaire. The ABC4D application looks exactly
3
Medical University of Graz, Department of Internal Medicine, the same for the intervention (ABC4D algorithm) and control
Graz, Austria (standard bolus calculator) group.
Results: 22 participants with T1D on either pump therapy or
Background and Aims: Despite affecting up to 1/3 of ad- multiple daily injections of insulin have completed the study to
missions, hospital management of type 2 diabetes is still date (59% female, mean (SD) age 39(12) years, diabetes duration
challenging. Clinical guidelines recommend basal-bolus in- 17(9) years and HbA1c 61(10) mmol/mol). 94% agreed that
sulin therapy to maintain blood glucose levels in range and entering information on the app was straightforward, 67% trusted
avoid further complications. However, the efficacy of this the insulin dose recommendation generated, 94% agreed that the
therapy depends on individualised patient-centric approaches use of RT-CGM was acceptable, 100% considered the ABC4D
with constant follow-ups during the day. Thus, this therapy app user-friendly and 78% would be happy to continue using the
becomes a complex task, increasing the workload for the al- ABC4D for bolus calculation. 11% were reluctant to use the
ready saturated personnel and enhancing the overall risk of ABC4D while at work, 6% while travelling abroad, 6% before/
medical errors. after exercise, 11% when eating out, 22% when consuming al-
Methods: In response, we have created a novel software cohol and 17% when feeling unwell.
based decision support system (GlucoTab) conceived as a Conclusions: Our preliminary data suggests that the ABC4D
medical device. The system provides guidance and support for smartphone application used with RT-CGM is user-friendly and
nursing staff and physicians involved with insulin therapy of remains well-accepted after 6 months of continuous use.
patients with type 2 diabetes at the hospital ward. It supports
the glycaemic management workflow and automatically sug-
gests the correct insulin dosage. GlucoTab was developed in
062
an interdisciplinary and research driven approach, early in-
volving end users in the design and utilizing the findings from QUANTIFYING THE EFFECT OF ANTECEDENT
clinical trials in a stepwise approach. PHYSICAL ACTIVITY ON PRANDIAL GLUCOSE
Results: A clinical trial performed in four different general CONTROL IN TYPE 1 DIABETES: DEFINING
wards showed that standardized glycaemic management could EXERCISE ON BOARD
be implemented efficiently and safely. 73% of all blood glucose
measurements were in the target range 70-180 mg/dl without B. Ozaslan1, S. Patek1, M. Breton2
causing additional hypoglycemic events. Compared to another 1
University of Virginia, Center for Diabetes Technology,
trial performed previously with standardized paper-based dia- Charlottesville, USA
betes management, the computerized system led to a consider- 2
UVA, Center for Diabetes Technology, Charlottesville, USA
able reduction of insulin dosing errors which were shown to be
associated with hypoglycaemic events. Background and Aims: Exercise is a major hurdle to insulin
Conclusions: GlucoTab improves quality in hospital care by dosing in Type 1 Diabetes (T1D). While physical activity (PA) is
standardising diabetes management according to current guide- encouraged in patients with T1D, metabolic adjustments to ac-
line recommendations and increases patient safety by reducing tivity (e.g. increased insulin sensitivity, increased glucose dis-
treatment errors. posal, and catecholamine secretion) creates a significant
glycemic imbalance with increased risk for hypo/hyperglycemia.
Quantifying the impact of antecedent PA on prandial glucose
061 (PG) could allow for better mealtime insulin dosing and reduced
glycemic risks.
ACCEPTABILITY AND USABILITY OF THE Methods: 12 subjects with T1D on insulin pump participated
ADVANCED BOLUS CALCULATOR FOR DIABETES in a clinical trial (NCT02558491), where they continuously wore
(ABC4D) AFTER 6 MONTHS OF USE a blinded continuous glucose monitor (Dexcom G4/505), a PA
tracker (Fitbit Charge HR), and followed regular pump therapy
M. Reddy1, P. Pesl2, N. Jugnee1, C. Toumazou2, D. Johnston1,
(integrated bolus calculator) for a month.
P. Georgiou2, P. Herrero2, N. Oliver3
To assess PG, dinner meals were identified from the dataset
1
Imperial College London, Division of Diabetes- Endocrinology and glucose area under the curve (AUC) was computed from 2h
and Metabolism, London, United Kingdom to 6h after dinner. Additionally, we derived a new index from
2
Imperial College London, Centre for Bio-inspired Technology- PA data: exercise on board (EOB): an exponentially weighted
Department of Electrical and Electronic Engineering, London, sum of steps over time (1h,3h,6h,12h). We also computed in-
United Kingdom sulin on board relative to total daily insulin (IOB/TDI) and
3
Imperial College London, Division of Diabetes- Endocrinology carbohydrate on board relative to bodyweight (COB/BW) from
and Metabolism, London, United Kingdom the pump records.
ATTD 2017 ORAL ABSTRACTS A-25

app, accessible on their smartphone or online, to implement on

their insulin pump.
Results: Initial trials indicate patients achieve better glucose
EOBs, IOB/TDI and COB/BW were associated with AUC control by implementing the algorithms insulin pump settings
using linear regression. adjustments.
Results: We collected complete datasets from 8 subjects Conclusions: This clinical decision support tool is well-
(13.4 5.1 dinners/subj), with variable conditions before dinner: positioned to revolutionize the way T1D data is used for insulin
BG = 161 63mg/dl, IOB/TDI = 0.14 0.06, COB/BW = 0.59 pump patients by:
0.23.  Offering a safe, effective self-management tool to op-
Linear regression coefficients for EOBs are shown below.
timize insulin pump settings during and between clinic
Conclusions: We have shown that an easily available quan-
tification of PA has a significant impact on glycemic exposure
visits
 Providing insights and recommendations for physicians
after a meal. By relating this effect to the effect of insulin, one
may build an exercise-informed bolus calculator. to individualize and optimize management of patients
on insulin pumps
 Offering a population health management tool that
highlights patients needing insulin pump adjustments to
063 their healthcare providers
DREAMED ADVISOR AUTOMATES INSULIN PUMP
SETTINGS ADJUSTMENTS WITH DATA FROM
GLOOKOS DIABETES MANAGEMENT PLATFORM
M. Phillips1, J. Chang2
1
Institute for Endocrinology and Diabetes, National Center for
Childhood Diabetes, Schneider Childrens Medical Center of
Israel, Israel
2
Glooko, Product, Mountain View, USA

Background and Aims: Adjusting insulin pump settings is

often a tedious and unsuccessful task, due to the complexity of
factors impacting blood glucose trends and the risk for hypo-
glycemia.
DreaMed Advisor features an insulin pump adjustments al-
gorithm, targeting Type 1 patients on insulin pumps, aims to
streamline the adjustment process. Leveraged in Glookos FDA
cleared, HIPAA compliant platform, which seamlessly aggre-
gates data from insulin pumps, continuous glucose monitors,
blood glucose meters, fitness apps and devices, biometric devices
and a built-in food and medication database, the DreaMed al-
gorithm analyzes blood glucose and contextual data to identify
trends and make insulin pump setting adjustment recommenda-
tions.
Methods: Healthcare providers are notified of adjustment
recommendations and review/modify the adjustments. Once
approved, the adjustments are delivered to the patients Glooko
A-26 ATTD 2017 ORAL ABSTRACTS

064 Background and Aims: Identification of factors influenc-

ing the outcome of insulin treatment intensification during the
IMPACT OF NON-SEVERE HYPOGLYCEMIC
8-week run-in period of the OpT2mise study could be useful in
EPISODES ON QUALITY OF LIFE, SLEEP AND WORK
informing management decisions.
PRODUCTIVITY IN PATIENTS WITH TYPE 1
Methods: In OpT2mise, patients with HbA1c 8% despite
DIABETES (T1D) ON CSII AND MDI THERAPY
optimized multiple daily dose (MDI) insulin therapy were ran-
E. Patrakeeva1, M. Dunicheva1, L. Chernilova1, A. Mosikian1 domized to continued MDI or pump therapy. Multivariate logistic
1 regression analysis (alpha level = 0.05), including factors that
Saint-Petersburg Medical University, Endocrinology, Saint- were significant (alpha level = 0.20) in univariate analysis, was
Petersburg, Russia performed to assess associations between baseline demographic
or clinical characteristics and the likelihood of randomization.
Background and Aims: To evaluate the impact of non-severe Sensitivity and specificity were calculated for several values of
hypoglycemic events on the quality of life (QoL), quality of sleep initial HbA1c to assess its predictive value in identifying partici-
and work productivity; to reveal the main reasons for the fear of pants in whom MDI treatment optimization was not beneficial.
hypoglycemia (FoH). Results: 134 of 495 participants were not randomized because
Methods: Anonymous validated questionnaire on minor hy- their HbA1c had decreased to <8.0% following insulin dose op-
poglycemic episodes was given to 1340 individuals registered at timization. On multivariate analysis, the only factor associated
diabetes-related on-line resource. To analyze the results we used with randomization was HbA1c at the beginning of dose opti-
hi-squared test, two-way ANOVA on ranks and linear regression mization. Two HbA1c cut-off values were identified: a baseline
model. HbA1c of 8.4% was found to give optimal sensitivity (90%) and
Results: 93 people (82% women) with T1D (60% on MDI, specificity (37%) for the prediction of dose optimization success
40% used CSII) agreed to fill out the questionnaire. The partic- or failure, while a value >9.0% (sensitivity = 62% and specific-
ipants are (meanSD) 30 9 years old and have (meanSD) ity = 76%) was considered on clinical grounds to indicate a low
12 8 years diabetes duration. Our analysis revealed a positive likelihood of achieving glycaemic goals with MDI alone.
correlation between the hypoglycemia duration and its negative Conclusions: This finding supports the early consideration of
influence on patients work quality (p = 0.048). The more time is alternative interventions, including insulin pump therapy, in pa-
needed for achieving normal range of blood glucose concentra- tients with HbA1c 9% despite optimized MDI therapy. Limita-
tion, the greater is negative influence on patients daily activity tions of this analysis include its post hoc nature and the lack of use
(p = 0.04) and sleep quality (p = 0.009). The main reasons for the of newer therapies introduced after the trial ended.
fear of hypoglycemia are being alone at the moment of hypo-
glycemia (p = 0.008), not being able to recognize the symp-
toms of hypoglycemia (p = 0.001) and causing disturbance to
other people (p = 0.013), especially in women (meanSD 066
1.8 0.2 and 0.8 0.3 points out of 5, respectively). ROUTINE USE OF BOLUS CALCULATOR IMPROVE
Conclusions: Non-severe hypoglycemic events have negative THE METABOLIC CONTROL IN CHILDREN WITH
impact on quality of life and quality of sleep (regardless of pa- TYPE 1 DIABETES IN THE FIRST YEAR OF INSULIN
tients gender). Physicians should pay special attention on pa- PUMP TREATMENT
tients knowledge of hypoglycemia symptoms. All the revealed
factors for fear of hypoglycemia are modifiable and the decrease B. Obermannova1, L. Petruzelkova1, S. Kolouskova1,
of its impact might improve the QoL. S. Pruhova1, Z. Sumnik1
1
2nd Faculty of Medicine- Charles University in Prague,
065 Department of Paediatrics, Prague, Czech Republic

FACTORS AFFECTING THE SUCCESS OF INSULIN Background and Aims: The treatment with CSII using bolus
DOSE INTENSIFICATION IN PATIENTS WITH TYPE 2 calculator (BC) may improve metabolic control in type 1 dia-
DIABETES: AN ANALYSIS FROM THE OPT2MISE betes (T1D). The aim of our study was to assess the impact of
RANDOMIZED TRIAL routine bolus calculator use in comparison with fixed boluses in
O. Cohen1, I. Schutz-Fuhrmann2, J. Castaneda3, Y. Reznik4, the first year of CSII treatment in children with T1D.
R. Aronson5, I. Conget6, S. Liabat7, S. Runzis7, S. de Portu7 Methods: In this single-centre, retrospective cohort study, 66
children (34 boys, 50%; age 10.0 3.1 years; diabetes duration
1
Sheba Medical Center, Institute of Endocrinology, Ramat Gan, 3.6 2.1 years; HbA1c 65.3 mmol/mol) in first year of insulin pump
Israel treatment, were observed. We compared children with routine use
2
City Hospital Vienna-Hietzing, Department of Endocrinology, of bolus calculator (BC+ group; 31 children, 47%) with children
Vienna, Austria without using this technology (BC- group; 35 children, 53%) from
3
Medtronic, Bakken Research Center, Maastricht, The CSII start. We have retrospectively assessed the HbA1c, total daily
Netherlands dose of insulin (TDD), basal/bolus ratio, number of boluses/day and
4
University of Caen Cote de Nacre Regional Hospital Center, BMI; all in 0, 3, 6, 9, and 12 months after the CSII initiation. The
Department of Endocrinology, Caen, France differences between groups were analyzed using non-parametric
5
LMC Diabetes & Endocrinology, LMC Diabetes & tests (Kruskal-Wallis, ANOVArepeated measurement).
Endocrinology, Toronto, Canada Results: In both groups the most significant drop in HbA1c
6
Endocrinology and Nutrition Department, Diabetes Unit- was in first three months of CSII treatment (p < 0.001). In the last
University Hospital Clinic, Barcelona, Spain quarter of year the HbA1c shifts differently: in BC- arises
7
Medtronic International Trading Sarl, Medtronic International (65.00 / 68.00 mmol/mol), in BC+ declines (62.00 / 61.00
Trading Sarl, Tolochenaz, Switzerland mmol/mol); p = 0.007, see graph. In BC+ group was significantly
ATTD 2017 ORAL ABSTRACTS A-27

higher boluses percentage of TDD (BC +58% vs. BC- 47% of insulin needs, but with a superior and no significant weight gain,
TDD, p = 0.01). The insulin TDD, number of boluses per day and probably linked to persisting high insulin needs.
BMI did not differ between the groups.
Conclusions: In children with T1D routine use of bolus cal-
culator precise the bolus size which improves the metabolic
control in the first year of CSII treatment. 068
EFFECT OF INSULIN DELIVERY ON GLYCEMIC
CONTROL IN PEOPLE WITH TYPE 2 DIABETES
067 D. Kruger1, R. Morlock2, R. Wood3, A. Davis4, C. Kuerschner5,
J. Warner5
BENEFIT OF AN AMBULATORY INSULIN PUMP (AIP)
1
IN INSULIN REQUIRING TYPE 2 DIABETIC (T2D) Henry Ford Health System, Endocrinology, Detroit, USA
2
PATIENTS WITH HIGH INSULIN NEEDS (1.3 U/KG/D) YourCareChoice, Managing Director, Ann Arbor, MI, USA
3
dQ&A, Research, San Francisco, USA
F. Travert1, D. Huet2, S. Clavel3, O. Dupuy4, E. Ghanassia5, 4
Metabolic Markets, Research, San Clemente, USA
K. Mohammedi1, J.F. Thuan6, K. Krompa1, J.P. Courreges7 5
CeQur, Clinical Development, Marlborough, USA
1
CHU Bichat, Diabetology, Paris, France
2
CH Saint Joseph, Diabetology, Paris, France Background and Aims: Insulin is critical therapy for people
3
CH Hotel Dieu, Diabetology, Le Creusot, France with T2 diabetes. However, insulin injections interfere with
4
CH Saint Jospeh, Diabetology, Paris, France daily life and can cause embarrassment. Choice of insulin delivery
5
Clinique Sainte Therese, Diabetology, Sete, France may impact glycemic control. This study examines HbA1c control
6
CH Narbonne, Diabetology, Narbonne, France in people withT2 diabetes (T2DM) using pumps, pens, and syringes.
7
Hopital General, Service de Medecine interne et Diabetologie, Methods: An online survey for people with diabetes was fielded
Narbonne, France 1Q2016 using the dQ&A Patient Panel. Type of insulin delivery
device, years using insulin, age at diagnosis, comorbidities, use of
Background and Aims: T2D patients may show very high glucose monitoring, self-reported adherence, insurance and patient
insulin needs with failure in glycemic control. How do these sociodemographics were collected. Descriptive statistics were used
patients evolve with an AIP on a day-to-day basis? to describe patients achieving HbA1c < 7 vs. HbA1c > 9 by insulin
Methods: 110 T2D patients who show insulin optimization delivery.
failure (HbA1c8% in spite of 3 injections/d - 0.7 U/kg/d) Results: Adults with T2DM using an insulin device (n = 1,579)
benefit from an AIP in an observational, multicentric trial. Among had mean age of 61.15(9.91) years, 60.27% were female and
them, 31 (HD) need 1.3 U/kg/d (21 1.5 U/kg/d) and are com- mean duration of insulin was 9.24 years. Less than 40% reported
pared with 32 patients (C) need 1.1 U/kg/d. achieving HbA1c<7 and 10% had an elevated HbA1c>9. Dif-
Results: ference in HbA1c by delivery device indicate patients using
HbA1c, decreased identically from T0 (identical initial level): syringe+pen were most likely to have HbA1c>9 followed by
patients using syringe alone, pen only and pump (30% vs.
26.5% vs. 20.3% vs. 12.5%; p = 0.03, respectively). Pump us-
ers reported using insulin for longest time and pen only users
for shortest (13.48 vs. 7.52 years; p < 0.001). Pump users were
most likely to be married followed by pen only, syringe+pen,
and syringe only (75.33% vs. 60.40% vs. 59.24% vs. 53.92%;
p < 0.001, respectively).
Conclusions: This study identifies differences in levels of
HbA1c control by insulin delivery. People using more discreet
(pens/pumps), easier to adhere and less inconvenient (pumps)
delivery report greater levels of control.
Insulin needs (significantly different at T0 p < 0.0001)
dropped very significantly in the HD Group vs. C Group.
069
LONG-TERM GLYCAEMIC CONTROL OF CHILDREN
WITH TYPE 1 DIABETES ON INSULIN PUMP
THERAPY IN A POPULATION BASED CASE-
CONTROL STUDY
M.A. Burckhardt1,2, G. Smith3, M.N. Cooper3, T.W. Jones1,2,3,
E.A. Davis1,2,3
1
No difference was noticed concerning hypoglycemia between Department of Endocrinology and Diabetes, Princess
both groups (0 severe hypoglycemia). Margaret Hospital for Children, Perth, Australia
2
Conclusions: In case of multi-injections failure associated School of Paediatrics and Child Health, The University of
with high insulin needs, the use of insulin infusion through AIP Western Australia, Perth, Australia
3
leads to a good glycemic control, with a very noticeable decrease Telethon Kids Institute, The University of Western Australia,
in insulin needs (-30%), better than in patients with moderate Perth, Australia
A-28 ATTD 2017 ORAL ABSTRACTS

Background and Aims: Long-term glycaemic outcomes in

patients with type 1 diabetes (T1D) on insulin pump therapy
(CSII) in real world are still debated. The aim of this study is to
investigate the long-term outcome of CSII compared to injection
therapy in children with T1D in a population-based paediatric
diabetes service in Western Australia.
Methods: Patients using CSII between January 1999 and July
2016 were matched to patients on injection therapy on the basis
of age, date of diagnosis and HbA1c at pump start.
Results: Of 760 eligible patients on CSII, 535 pump-injection
matches were identified. The mean age, duration of diabetes and
HbA1c at pump start was 11.1 years (SD 3.79), 3.3 years (SD
2.74) and 7.8% (SD 1.0). The mean duration of follow-up was
3.9 years (SD 2.5), the maximum length was 9.2 years.
CSII improved HbA1c significantly at all measured time points
except at 1 year compared to injection therapy throughout 6 years
of follow-up. The greatest improvements were observed early at 3 Background and Aims: Predictive low glucose management
months and as late as 6 years (N = 109 matched pairs) with a (PLGM) system suspends insulin infusion when patient glucose
decrease in HbA1c of 0.4% (SEM 0.07,p <0.001) and 0.5% (SEM levels are still in normal range to avoid reaching a pre-set low
0.22,p = 0.04) respectively. While the HbA1c beyond 7-years limit. The aim of this study was to determine the efficacy of this
duration was lower in the CSII group, statistical significance was algorithm in the prevention of exercise-induced hypoglycemia
not reached but the number of matched pairs was much smaller. under free-living conditions.
Conclusions: Patients using CSII had a sustained improve- Methods: This was an observational, controlled study in
ment in their glycaemic control compared to a matched cohort which 34 participants exercised (same amount for 2 consecutive
over nine years. To our knowledge, this is the longest and largest days with various sessions of high to moderate-intensity exercise
population-based cohort study in children and adolescents with during free-living conditions during a camp) using two preset
T1D on CSII. PLGM threshold (90 mg/dL vs 70 mg/dL). The primary outcome
was the requirement for hypoglycemia treatment (symptomatic
hypoglycemia with plasma glucose <70 mg/dL or plasma glu-
070 cose <50 mg/dL) and was compared in both groups.
Results: We analyzed 31/34 patients because of missing sen-
EXERCISE MANAGEMENT IN INSULIN PUMP sor data (mean age 15.0 1.5 years, 58.1% males, diabetes dura-
TREATED ADOLESCENTS WITH TYPE 1 DIABETES: tion 7.0 3.6 years, HbA1c at baseline 7.2 0.7%). No significant
PREDICTIVE LOW GLUCOSE MANAGEMENT IS difference has been observed about hypoglycemic events between
SAFE AND EFFECTIVE IRRESPECTIVE OF groups using threshold 70 or 90, with a mean blood glucose levels
THRESHOLDS USED in hypoglycemia 56.8 17.0 mg/dL vs 63.6 6.5 mg/dL (p = 0.07).
V. Cherubini1, I. Rabbone2, C. Arnaldi3, R. Bonfanti4, Time spent in target was similar in both groups, as well as time
G. dAnnunzio5, A.P. Frongia6, F. Lombardo7, E. Piccinno8, spent in hypo or hyperglycemia (Table), with only 3.1% and 3.2%
R. Schiaffini9, S. Toni10, S. Tumini11, P. Cipriano11, N. Minuto5, time respectively spent <70 mg/dL.
L. Lenzi12, S. Giorda2, L. Ferritto1, C. Ventrici7, F. Ortolani8, Conclusions: PLGM system in adolescents with type 1 dia-
O. Cohen13, A. Scaramuzza14 betes was effective to prevent hypoglycemia during and after
1 exercise, irrespective of the PLGM thresholds used.
Department of Womens and Childrens Health- AOU Salesi
Hospital, Division of Pediatric and Adolescent Diabetes,
Ancona, Italy 071
2
University of Turin, Paediatrics, Torino, Italy
3 CAN SMARTGUARD TECHNOLOGY RELIEVE
ASL Viterbo, Diabetologia Pediatrica, Viterbo, Italy
4 HYPOGLYCEMIA FEAR AND IMPROVE QUALITY
Scientific Institute San Raffaele, Paediatrics, Milano, Italy
5 OF LIFE? A MULTICENTER STUDY
Istituto Giannina Gaslini, Diabetologia Pediatrica, Genova,
Italy W.B. Gawel1, A. Tabor1, O. Goik1, G. Deja1, T. Klupa2,
6
Ospedale Brotzu, Paediatrics, Cagliari, Italy V. Szostek-Gawel3, M. Zalinska4, M. Mysliwiec4, A. Gawrecki5,
7
University of Messina, Paediatrics, Messina, Italy D. Zozulinska-Ziolkiewicz5, P. Jarosz-Chobot1
8
Hospital Giovanni XXIII, Division of Pediatric and Adolescent
1
Endocrinology and Diabetes, Bari, Italy Medical University of Silesia, Upper Silesia Centre for Childs
9
Bambino Gesu Hospital, Pediatric Diabetology, Rome, Italy Health, Paediatric Diabetology, Katowice, Poland
10 2
Anna Meyer Childrens Hospital, Juvenile Diabetes Center, Jagiellonian University Medical College, Metabolic Diseases,
Florence, Italy Krakow, Poland
11
University of Chieti, Center of Pediatric Diabetology, Chieti,
3 _
NZOZ Med-Art, Diabetology, Zory, Poland
4
Italy Medical University of Gdansk, Department of Paediatrics,
12
Anna Meyer Childrens Hospital, Juvenile Diabetes Center, Diabetology and Endocrinology, Gdansk, Poland
5
Florence, Italy Poznan University of Medical Sciences, Internal Medicine and
13
Sheba Medical Center, Institute of Endocrinology, Ramat Diabetology, Poznan, Poland
Gan, Israel
14
ASST Cremona, Ospedale Maggiore, Paediatrics, Cremona, Background and Aims: A multicenter study was performed
Italy to assess the quality of life (QoL) of the patients who suffer from
ATTD 2017 ORAL ABSTRACTS A-29

type 1 diabetes (DM1) after introducing personal insulin pump in hypoglycemia group. We also report three cases of KA, proved
(IP) 640G with SmartGuard technology. by elevated capillary b-hydroxybutyrate concentrations. Patho-
Methods: 23 women and 22 men with well-controlled DM1 physiological link between frequent hypoglycemia rate (17.3%)
(mean HbA1C was 7.26%) were examined, mean age 15.7 years. and KA development (2.2%) was suggested.
The mean time from diagnosis was 5.2 years. Patient were previ- Conclusions: To our knowledge, this is the first study inves-
ously treated with IP with or without hypoblocade. 2-11 months tigating the real-life rates of hypoglycemia and KA in SGLT2-i
after introducing 640G pump therapy two surveys were conducted: treated persons with the help of such precise tools as CGM and
PedsQLTM 3.0 Diabetes which measured the QoL in diabetic pa- capillary blood b-hydroxybutyrate measurement. Rates of hy-
tients (Survey I) and the authorial questionnaire (Survey II) which poglycemia and KA were discovered to be more frequent than
measured satisfaction of 640G therapy (11questions). previously reported. Pathophysiological link between two con-
Results: The mean scores of QoL in Survey I regarding ditions is assumed.
communication (76%), concerns (59%), treatment (74%) and
diabetes (68%) which according to our scale ( 0-19% very low
QoL, 20-39% low, 40-59% moderate, 60-79% high, 80-100% 073
very high) means QoL was intact. The results of Survey II showed
gladness and assurance of the patients with 640G pump therapy. EFFECTS ON GLYCEMIC VARIABILITY AND GLYCO-
Over a half of participants (35 people) certified a serious reduction METABOLIC CONTROL OF METFORMIN,
of both hypo/hyperglycemia episodes. 19 respondents highlighted PIOGLITAZONE AND SITAGLIPTIN IN TYPE 2
a better coherence between blood glucose(BG) measured by sensor DIABETIC PATIENTS
and glucose meter(GM) which enabled them to decrease the fre- G. Derosa1, D. Romano1, A. DAngelo1, P. Maffioli1
quency of pricking fingers with GM to measure BG and improve 1
QoL. 27 patients noticed easier management of DM1 and better University of Pavia, Internal Medicine and Therapeutics,
cooperation with 640G. Pavia, Italy
Conclusions: Patients with DM 1 using 640G pump are sat-
isfied with the effects of the therapy, feel safer and their QoL Background and Aims: The aim of the study was to evaluate
measured by PedsQL is relatively high. the effects of a triple therapy with metformin, pioglitazone and
sitagliptin on glycemic control and glycemic variability com-
pared to metformin monotherapy, and compared to a combina-
tion of metformin and pioglitazone.
Methods: We enrolled 66 not well controlled, type 2 diabetic
072
patients. Patients were instructed to take metformin 500 mg three
CGM-PROVED HYPOGLYCEMIA IN SGLT2-I times a day for three months, then pioglitazone 15 mg twice a day
TREATED TYPE 2 DIABETES PATIENTS: A LINK TO was added for further three months, and finally sitagliptin 100 mg
KETOACIDOSIS? once a day was added for the latest three months. At the baseline,
and every three months, a continuous glucose monitoring system
S. Levit1, J. Wainstein3, I. Abadi-Korek4, R. Barnea4 was performed. We assessed: glycemic control, high sensitivity C-
1 reactive protein (hs-CRP), lipid profile, metalloproteinase-2 (MMP-
Assuta Medical Center, Institute of Endocrinology,
Diabetology and Metabolism, Tel Aviv, Israel 2), metalloproteinase-9 (MMP-9), soluble adhesion molecules
2 (sICAM-1, sVCAM-1), sE-selectin, adiponectin (ADN).
Hebrew University Medical School, Jerusalem, Diabetes Unit-
Department of Internal Medicine, Hadassah Hebrew University Results: We recorded a significant decrease of glycated hemo-
Hospital, Jerusalem, Israel globin with metformin + pioglitazone, and a further decrease with
3 metformin + pioglitazone + sitagliptin. There was an improvement
Institute of Diabetes, Edith Wolfson Medical Center, Sackler
School of Medicine Tel-Aviv University, Israel, Holon, Israel of lipid profile with metformin + pioglitazone + sitagliptin. We
4 recorded a decrease of Hs-CRP, sICAM-1, sVCAM-1, and an in-
Assuta Medical Center, Tel-Aviv, Israel, Department of
Academy and Research, Tel Aviv, Israel crease of ADN with metformin + pioglitazone, and with metformin
+ pioglitazone + sitagliptin. A decrease of sE-selectin was recorded
Background and Aims: SGLT2- Inhibitors (SGLT2-i) are a only with metformin + pioglitazone + sitagliptin. No variations of
relatively novel class of anti-diabetic medications. Latest studies MMP-2, and 9 were recorded. Regarding glycemic variability, there
have shown their ability to reduce body weight, cardiovascular was a reduction of AUC>180 and of mean amplitude of glycemic
morbidity and mortality. Real-life data however is contradictory and excursion with metformin + pioglitazone + sitagliptin.
points out some serious side effects: renal impairment, ketoacidosis Conclusions: Combination of metformin + pioglitazone +
(KA), elevated fracture risk, urinary tract infections and even toe sitagliptin proved to be effective in improving glycemic control,
amputations. It was suggested that SGLT2-i per se do not cause and in decreasing glycemic variability.
hypoglycemia, until combined with some well-known hypoglycemic
agents like Insulin, Sulfonylurea or Glinides. The combination of
SGLT2-i with anti-hyperglycemic agents (GLP-1 analogues, Met- 074
formin, DPP4 inhibitors) was proposed as hypoglycemia-safe.
IMPROVED GLYCAEMIC CONTROL AND SUPERIOR
Methods: We report the results of a two-year retrospective
REDUCTION OF PPG INCREMENTS WITH DOUBLE-
analysis of 136 Type 2 Diabetes (T2DM) patients treated with a
BLIND MEALTIME FASTER-ACTING INSULIN
combination of SGLT2-i with Incretin-Based therapy (Mt-IBT). In
ASPART VS CONVENTIONAL INSULIN ASPART IN
52 patients, professional CGM recordings were performed. In nine
T1D: THE ONSET 1 TRIAL
patients, CGM-proved hypoglycemic episodes were discovered.
Results: SGM data analysis showed significantly lower av- D. Russell-Jones1, B.W. Bode2, C. De Block3, E. Franek4,
erage Sensor Glucose (129.0 22.5 vs. 148.4 30.7 mg%, S. Heller5, C. Mathieu6, A. Philis-Tsimikas7, L. Rose8, V. Woo9,
p = 0.04) and estimated HBA1C (6.1 0.7 vs. 6.8 1.1, p = 0.02) A.B. sterskov10, T. Graungaard10, R. Bergenstal11
A-30 ATTD 2017 ORAL ABSTRACTS

1
Diabetes and Endocrinology, Royal Surrey County Hospital, 075
Guildford, United Kingdom
2
Department of Clinical Research, Atlanta Diabetes Associates, FASTER-ACTING INSULIN ASPART VS INSULIN
Atlanta, USA ASPART AS PART OF BASAL-BOLUS THERAPY
3
Department of Endocrinology- Diabetology, and Metabolism, IMPROVES POSTPRANDIAL GLYCAEMIC CONTROL
Antwerp University Hospital, Antwerp, Belgium IN UNCONTROLLED TYPE 2 DIABETES: THE ONSET
4
Department of Human Epigenetics, Mossakowski Clinical 2 TRIAL
Research Center- Polish Academy of Sciences, Warsaw, Poland K. Bowering1, C. Case2, J. Harvey3, M. Reeves4, M. Sampson5,
5
Department of Oncology and Metabolism, University of R. Strzinek6, D.M. Bretler7, R.B. Bang7, B.W. Bode8
Sheffield, Sheffield, United Kingdom 1
6
Laboratory and Clinic of Experimental Medicine and Division of Endocrinology and Metabolism, University of
Endocrinology, University Hospital Leuven, Catholic Alberta, Edmonton, Canada
2
University of Leuven, Leuven, Belgium Medical Building, Jefferson City Medical Group, Jefferson
7
Scripps Whittier Diabetes Institute, Scripps Health, San Diego, City, USA
3
USA Wrexham Academic Unit, Bangor University, Bangor, United
8
Department of Internal Medicine, Institute of Diabetes Kingdom
4
Research, Munster, Germany Department of Endocrinology, Diabetes Clinical Trials,
9
Section of Endocrinology and Metabolism, University of Chattanooga, USA
5
Manitoba, Winnipeg, Canada Department of Diabetes, Endocrinology and General
10
Novo Nordisk A/S, Sborg, Denmark Medicine, Norfolk and Norwich University Hospitals NHS
11
International Diabetes Center, Park Nicollet, Minneapolis, USA Foundation Trust, Norwich, United Kingdom
6
CEO & Principal Investigator, Protenium Clinical Research,
Background and Aims: Limiting excursions of postprandial Hurst, USA
7
glucose (PPG) is desirable in people with diabetes. This multi- Novo Nordisk A/S, Sborg, Denmark
8
centre, treat-to-target, phase 3 trial evaluated the efficacy of Department of Clinical Research, Atlanta Diabetes Associates,
faster-acting insulin aspart (faster aspart) in adults with type 1 Atlanta, USA
diabetes (T1D). The primary endpoint was change from baseline
in HbA1c after 26 weeks treatment. Background and Aims: This multicentre, double-blind, treat-
Methods: Post 8-week run-in, subjects were randomised to to-target trial evaluated the efficacy of faster-acting insulin aspart
double-blind mealtime (administered 0-2 min before meal) faster (faster aspart) vs conventional insulin aspart (IAsp) in adults with
aspart (n = 381), or conventional insulin aspart (IAsp; n = 380), or inadequately controlled type 2 diabetes (T2D) on basal insulin
open-label postmeal (administered 20 min after the start of meal) and oral antidiabetic drugs.
faster aspart (n = 382); each with basal insulin detemir. Methods: Following optimisation of basal insulin glargine
Results: After 26 weeks of treatment, HbA1c was reduced for U100 during an 8-week run-in (mean HbA1c 7.9%), subjects
faster aspart and IAsp (Figure), confirming non-inferiority to IAsp were randomised 1:1 to mealtime faster aspart (n = 345) or IAsp
for both mealtime and postmeal dosing (estimated treatment dif- (n = 344), each with insulin glargine U100 and metformin, using
ference [ETD], % [95% confidence interval]: mealtime, 0.15 a simple daily patient-driven titration algorithm.
[0.23; 0.07]; postmeal, 0.04 [0.04; 0.12]; P < 0.0001 for non- Results: The primary endpoint, mean HbA1c change from
inferiority); the reduction in HbA1c was statistically significantly baseline to Week 26, was 1.38% and 1.36% for faster aspart
greater for mealtime faster aspart vs IAsp (P = 0.0003). Superiority and IAsp, respectively; mean HbA1c was reduced to 6.6% for
to IAsp for the 2-h PPG increment during a standardised meal test both arms. Faster aspart demonstrated non-inferiority vs IAsp
was confirmed for mealtime faster aspart (ETD: 0.67 [1.29; in reducing HbA1c; the estimated treatment difference (ETD;
0.04] mmol/L; 12.01 [23.33; 0.70] mg/dL; P = 0.0375). The 1-h [95% confidence interval]) was 0.02% [0.15; 0.10]. Both
PPG increment was also statistically significantly reduced (ETD: basalbolus regimens improved postprandial plasma glucose
1.18 [1.65; 0.71] mmol/L; 21.21 [29.65; 12.77] mg/dL; (PPG) control. The 1-hour PPG increment (meal test) was
P < 0.0001). There were no statistically significant differences in statistically significant in favour of faster aspart (P = 0.0198;
the overall rate of severe or blood glucose (BG)-confirmed (BG
<3.1 mmol/L [56 mg/dL]) hypoglycaemic episodes.
Conclusions: In summary, faster aspart effectively improved
glycaemic control with superior PPG control for mealtime faster
aspart vs IAsp, representing a clinical advance in treating T1D.
ATTD 2017 ORAL ABSTRACTS A-31

Figure). Rates of overall severe or blood glucose (BG) con- 077

firmed hypoglycaemia (BG <3.1 mmol/L [56 mg/dL]) were
USE OF THE GLUCAL APPLICATION FOR
comparable. In T2D, mealtime faster aspart and IAsp in a
CARBOHYDRATE COUNTING RESULTS IN
basalbolus regimen achieved glycaemic control and reduced
IMPROVED POST-MEAL GLUCOSE CONTROL
HbA1c from baseline to 6.6%, confirming non-inferiority of
IN PATIENTS WITH TYPE 1 DIABETES IN A
faster aspart to IAsp, using a simple daily patient-driven ti-
RANDOMISED CONTROL STUDY
tration algorithm.
Conclusions: Faster aspart effectively improved 1-hour PPG O. Diouri1, M. Traverso2, J. Place1, V. Georgescu3,
control vs IAsp without increasing overall hypoglycaemia. M.C. Picot3, A. Farret1,2, E. Renard1,2
1
Institute of Functional Genomics, CNRS UMR 5203- INSERM
U1191- University of Montpellier, Montpellier, France
2
Montpellier University Hospital, Department of
076
Endocrinology, Diabetes, Nutrition, Montpellier, France
3
INSULIN GLARGINE 300 U/ML (GLA-300) PROVIDES Montpellier University Hospital, Department of Medical
MORE STABLE AND MORE EVENLY DISTRIBUTED Information, Unit of Clinical and Epidemiological Research,
STEADY-STATE PHARMACODYNAMIC/ Montpellier, France
PHARMACOKINETIC PROFILES COMPARED WITH
INSULIN DEGLUDEC IN TYPE 1 DIABETES (T1DM) Background and Aims: The accurate estimation of carbo-
1 2 3 4 2 hydrates (CHO) is essential for computing meal insulin doses and
T. Bailey , R. Dahmen , J. Pettus , R. Roussel , K. Bergmann , achieving targeted post-meal glucose level in patients with type 1
M. Maroccia5, N. Nassr2, O. Klein6, G. Bolli7, T. Heise6 diabetes (T1D). We developed the Android application GLU-
1
AMCR Institute, Escondido, CA, USA CAL that automatically computes CHO content of declared
2 food intakes and needed insulin dose according to patients
Sanofi-Aventis Deutschland GmbH, Diabetes Division,
Frankfurt am Main, Germany CHO/insulin ratio. We assessed the outcomes of its use in free-
3 life on post-meal glucose control in a randomised control study.
University of California, San Diego, CA, USA
4 Methods: Thirty adult T1D patients, under basal-bolus insulin
Assistance Publique Hopitaux de Paris, Bichat Hospital, Paris,
France regimen, were randomised in a two-month cross-over study.
5 GLUCAL or patient (CONTROL) estimations of meal CHO
Umanis, Levallois-Perret, Paris, France
6 were used for computing meal insulin doses during two 1-month
Profil, Institut fur Stoffwechselforschung GmbH, Neuss,
Germany periods in random order. Computed CHO amounts, injected in-
7 sulin doses and 2-hour post-prandial blood glucose (PPG) levels
University of Perugia, School of Medicine, Perugia, Italy
were recorded for each meal.
Background and Aims: To compare steady-state pharma- Results: Twenty-for patients (age 43 15, HbA1c 7.5 0.6%)
codynamic (PD) and pharmacokinetic (PK) profiles of insulin completed the study. Number of GLUCAL vs. CONTROL study
glargine 300 U/mL (Gla-300) with insulin degludec 100 U/mL days and meals were respectively 30.6 2.6 vs. 31.3 4.3 and
(IDeg1) in two parallel cohorts with fixed once-daily dose regi- 80.8 17.4 vs. 86.8 19.8. Percentage of PPG values in 80-180mg/
mens in T1DM, in a multiple-dosing, crossover, euglycemic dl range (primary study endpoint) was higher with GLUCAL:
glucose clamp study. 53.8 13.1 vs. 48.1 12.5 (p = 0.0002), thanks to reduced %PPG
Methods: For both insulins participants received 0.4 U/kg/day >180mg/dl: 32.3 11.6 vs. 38.3 11.3 (p = 0.0010), while %PPG
(Cohort 1; n = 24) or 0.6 U/kg/day (Cohort 2; n = 24), before <80mg/dl were similar: 13.9 6.9 vs. 13.6 5.7 (p = 0.7875). Mean
breakfast, for 8 days. Metabolic activity was measured by glu- of PPG was also reduced significantly with GLUCAL: 158 23 vs.
cose infusion rate (GIR) over 30 hours. Main endpoint: within- 167 25 (p = 0.0094) while SD of PPG increased slightly: 72 19
day variability (fluctuation) of smoothed GIR over the dosing vs. 65 20 (p = 0.0156).
interval (GIR-smFL024; treatment ratios were obtained using a Conclusions: Our data supports the benefits of using GLU-
linear mixed-effects model). Insulin concentrations were mea- CAL to improve post-meal glucose control in T1D patients
sured using validated radioimmunoassays. through better adjusted insulin doses which reduce PPG values
Results: GIR-smFL024 was significantly lower with Gla-300 >180mg/dl without increasing hypoglycaemia. Longer term
than IDeg1 at 0.4 U/kg/day (p = 0.047; treatment ratio 0.7978 studies are expected to confirm sustainability with time.
[90% CI: 0.6637 to 0.9591]) (LOESS smoothing 0.15), but was
comparable for Gla-300 and IDeg1 at 0.6 U/kg/day. Both doses 078
of Gla-300 provided plateau-like insulin exposure from 2 to
16 hours post-injection, with a slight decline afterwards, whereas ACHIEVING NEAR EUGLYCAEMIA USING LONG-
IDeg1 concentrations (total insulin) after both doses increased TERM INSULIN PUMP THERAPY DIFFERENTIALLY
from *1 hour to a Tmax at *10 hours after dosing, followed by a IMPROVES C-PEPTIDE LEVEL OR DISPOSITION
steady decline with no plateauing. Both insulins provided ex- INDEX IN TYPE 2 DIABETES
posure and activity until 30 hours and were generally well tol-
S. Choi1, Y. Noh2
erated.
1
Conclusions: This PK/PD analysis supports a superior glu- School of Medicine- Konkuk University, Internal Medicine,
codynamic profile of Gla-300 versus IDeg1 at a dose clinically Seoul, Republic of Korea
2
relevant for T1DM (0.4 U/kg/day), in terms of within-day vari- School of Medicine- Konkuk University, Biochemistry, Seoul,
ability. An overall more stable and more evenly distributed in- Republic of Korea
sulin exposure over the dosing interval was observed at both dose
levels under Gla-300. Background and Aims: Through insulin pump (CSII) ther-
Study supported by Sanofi. apy, hyperglycemia can be controlled to near normal level in
A-32 ATTD 2017 ORAL ABSTRACTS

patients with type 2 diabetes (T2D). We wanted to find out the CGM event recorder to report DAD alerts, and completed
whether the changes in beta cell function, insulin sensitivity, and paper diaries detailing further information (e.g., alerts not re-
disposition function would be observed in T2D patients during corded electronically). CGM readings 3.9 mmol/L for 15 con-
long-term CSII therapy. secutive minutes were considered hypoglycemic events. To
Methods: We applied CSII therapy to T2D patients (number, account for anecdotal reports that DAD alerts may precede hy-
163 with 56.4% of male; age, 59.7 9.7 years; duration, poglycemia, alerts 30 minutes before to 30 minutes after the
11.1 6.9 years; HbA1c 8.9 1.9%; BMI 24.4 3.1 kg/m2). onset of an event were categorized as hits, with failures to alert
Blood samplings were performed yearly for 4 years at overnight in this time frame categorized as misses. Data was included
fasting and 120 minutes after ingestion of a standard mixed meal only if participants were awake and with their DADs.
(500 kcal). Serum C-peptide, glucose, and HbA1c were mea- Results: Table 1 displays an overview of each participants
sured and C-peptidogenic Index (CI), Matsuda Index (MI), and results. DAD sensitivity varied substantially between partici-
disposition Index (DI) were calculated. Patients were grouped pants, ranging from 0.0 100.0% (42.1% across all participants).
into high MI [insulin sensitive (IS) group; age 58.0 10.2 years] Five DADs demonstrated >50.0% sensitivity, or alerted to more
and low MI [insulin resistant (IR) group; age 61.1 9.1 years] by than half of hypoglycemic episodes. Eleven DADs alerted to
the mean value of baseline MIs. fewer than half of events.
Results: HbA1c decreased significantly from 8.9% to 6.6% in Conclusions: Sensitivity to hypoglycemia appears to be
both groups. In IS group, serum C-peptide increased significantly highly variable across DADs. More research is needed to assess
but DI did not change. Whereas, in IR group, serum C-peptide DAD accuracy and factors that contribute to DAD variability.
did not increase but the DI increased significantly. Glycemic
control to near normal level was associated with increase in C-
peptide level only in IS group but not in IR group, while it was
associated with increased DI level in IR group but not in IS 080
group.
Conclusions: Achieving near euglycaemia using long-term CORRELATION OF TRANSIENT ELASTOGRAPHY
insulin pump therapy seems to restore the original defect in each WITH OTHER NON-INVASIVE METHODS
diabetic group. OF NON-ALCOHOLIC LIVER FIBROSIS
ASSESSMENT IN DIABETIC PATIENTS
J. Jirkovska1, L. Vedralova1, M. Koula1, P. Hribek1,
P. Urbanek1, S. Solar1, M. Zavoral1
079
1
Medical Department of the First Faculty of Medicine and
USING CONTINUOUS GLUCOSE MONITORING Military University Hospital Charles University, Internal
(CGM) TO EVALUATE THE SENSITIVITY OF Medicine Dpt, Prague, Czech Republic
HYPOGLYCEMIA DETECTION IN DIABETES ALERT
DOGS (DADS) Background and Aims: Influence of non-alcoholic fatty liver
L. Gonder-Frederick1, J. Grabman1, J. Poler, Jr.1, J. Shepard1 disease (NAFLD) and steatohepatitis (NASH) on diabetes is
1 widely discussed. Liver biopsy as diagnostics is invasive and
University of Virginia, Psychiatry & NB Sciences, expensive. Therefore, non-invasive methods are recommended.
Charlottesville, USA Aim was to compare grade of fibrosis assessed by transient
elastography (Fibroscan) with NAFLD Fibrosis Score (NAFLD
Background and Aims: Despite more patients with type 1 Score) and FIB-4 Score.
diabetes (T1DM) using DADs to assist in their diabetes man- Methods: Study included 21 diabetics with NAFLD
agement, few studies have systematically investigated DAD (62% men, mean BMI 29.1). Liver stiffness examined with
accuracy. This study compared CGM data to participant reported transient elastography was assessed as a median of at least 10
DAD alerts to test real-world DAD sensitivity to hypoglycemia. valuable measurements (E_med) in kiloPascales (kPa). Corre-
Methods: All participants had DADs placed in their homes lation analysis was applied to compare E_med with NAFLD
for 5 months by the same training organization. Eight adults (22 Score and FIB-4 Score. F0-F4 grade od fibrosis was assessed for
43 yrs) and eight youth (8 17 yrs) with T1DM wore a Dexcom each method.
G4 Platinum for 13 50 days (Median = 29). Participants used Results: Mean E_med was 8.17 2.81kPa, mean NAFLD
Score -0.73 1.15, mean FIB-4 Score 1.45 0.64. F0-1 (absence
of significant fibrosis) was assessed in 47% of patients with Fi-
broscan, in 44% with NAFLD Score and in 62% with FIB-4
Score, respectively. F2 (grey zone) was assessed in 29% of
subjects with Fibroscan, in 39% with NAFLD Score and in 38%
with FIB-4 Score. F3-4 (significant fibrosis) was assessed in 24%
of patients with Fibroscan, in 17% with NAFLD Score and
in nobody with FIB-4 Score. Significant correlation was found
in comparison of Fibroscan with NAFLD score (r = 0.481,
p < 0.05), whereas correlation with FIB-4 Score was not signifi-
cant (r = 0.155 NS).
Conclusions: Significant correlation was found between
grade of liver fibrosis assessed by Fibroscan and NAFLD Score.
Combination of these methods in detection of liver fibrosis is
helpful. Physicians are then better able to select patients in higher
risk of NASH where biopsy should be considered.
ATTD 2017 ORAL ABSTRACTS A-33

081 (converted to estimated median glucose mmol/L) within 15-

months pre-admission. The association between AGN and CBG
IN VITRO PRODUCTION OF INSULIN FROM
variability (interquartile range), length of stay (LOS), and number
MONOCYTES WITH AND WITHOUT THE USE
of hypoglycaemic (<4mmol/L) episodes/day were investigated.
OF LIRAGLUTIDE AND GROWTH FACTORS
Hypoglycaemic episodes defined as sequences of contiguous
A. Kyventidis1, T. Didangelos1, G. Tzimagiorgis2, CBG measures below 4mmol/L where the time interval between
A. Hatzitolios1 each measure was <60 minutes.
1 Results: 10598 admissions with data identified. A U-shaped
AHEPA University Hospital, 1st Propeudetic Department association between AGN and CBG variability was seen with
of Internal Medicine, Thessaloniki, Greece lowest variability at AGN of 0-2 (IQR 3.7mmol/L), with peak
2
Aristotle University of Thessaloniki, Department of Biological variability AGN of negative 16-18 (6.9mmol/L) and 12-14
Chemistry, Thessaloniki, Greece (5.9mmol/L). Median LOS was negatively associated with AGN
from AGN negative 18-20 (6.1 days) to AGN 8-10 (1.9 days), with
Background and Aims: To investigate the abilities of AGN >10 associated with an increase to AGN 16-18 (2.9 days). A
monocytes to differentiate into insulin-producing cells in cell U-shaped association between AGN and hypoglycaemic episodes/
cultures in vitro using growth factors and liraglutide. day was seen with minimum 0.13 episodes/day at AGN (-2-0) and
Methods: Monocytes have been cultured in vitro for 7 and peak frequency at AGN negative 14-12 (0.34 episodes/day) and
14 days with and without growth factors and liraglutide in three AGN 8-10 (0.51 episodes/day) (p < 0.05 for all values compared
experiments. We used the following 4 growth factors: M-CSF with minimum).
Interleukin-3, EGF, and HGF. Afterwards, they were tested for in- Conclusions: AGN is a simple metric which combines 2 mea-
sulin production at 100 and at 250 mg/dl glucose concentrations. sures readily available very early during admission. AGN may be a
Results: In two experiments, monocytes that have been cul- useful tool to stratify patients for risk of glycaemic variability, hy-
tured for 7 days (1st) and 14 days (2nd) with growth factors and poglycaemia and prolonged hospital stay at the point of admission.
liraglutide produced 10.32 6.78 lU/ml insulin (n = 10) and
17.03 12.76 lU/ml insulin (n = 18), respectively. In the third
experiment, monocytes that have been cultured for 14 days
without growth factors and liraglutide produced 15.05 6.20 lU/ 083
ml insulin (n = 10). After the rise of glucose concentration from IMPAIRED AWARENESS OF HYPOGLYCAEMIA
100 to 250 mg/dl for one hour, they increased significantly the IS ASSOCIATED WITH BLUNTED THALAMIC,
amount of insulin to 22.81 6.02 lU/ml (p = 0.001). The differ- BUT INCREASED FRONTAL ACTIVATION
ence in the amount of insulin between the monocytes cultivated IN RESPONSE TO HYPOGLYCAEMIA
with and those cultivated without growth factors for 14 days was
not significant (p = 0.811). M. Nwokolo1, S. Amiel1, F. Zelaya2, M. Byrne1, B. Wilson1,
Conclusions: Monocytes can produce sufficient amounts of L. Green1, A. Pernet1, P. Choudhary1
insulin after 7 and 14 days of culture. Growth factors and liraglutide 1
Kings College Hospital, London, UK, Division of Diabetes
are not necessary to achieve efficient insulin production, but when and Nutritional Sciences, Faculty of Life Sciences and
added there was an increase in insulin production. In addition, cells Medicine, London, United Kingdom
grown without growth factors were able to increase insulin pro- 2
Kings College London, London, UK, Department of
duction according to the glucose increase of the supernatant. Neuroimaging, Institute of Psychiatry, Psychology and
Neuroscience, London, United Kingdom
082
Background and Aims: To investigate the impact of type 1
ADMISSION GLUCOSE NUMBER (AGN): A NOVEL diabetes and impaired awareness of hypoglycaemia (IAH) on the
POINT-OF-ADMISSION SCORE ASSOCIATED WITH brains response to hypoglycaemia, using arterial spin labelling
PROLONGED ADMISSION DURATION AND WITH functional MRI (ASL-fMRI).
GLYCAEMIC CHARACTERISTICS IN PATIENTS Methods: 15 hypoglycaemia-aware (HA) type 1 subjects (age
WITH TYPE 1 DIABETES 39 13.5years, diabetes duration 24 12.8years, Gold score 1.5
R. Maitland1, R. Srivastava2, S. Cunningham3, G. Jones1, 0.5, 6 male) and 15 matched subjects with IAH ( age 38 10.6years,
C. Sainsbury1 diabetes duration 22 7.8years, Gold score 5.7 1.4, 7 male) un-
derwent ASL-fMRI scans at 5mmol/L and 2.6mmol/L. Symptoms
1
Gartnavel General Hospital, Diabetes, Glasgow, United were scored using a visual analogue scale. Global and regional ce-
Kingdom rebral blood flow (rCBF) were analysed.
2
Queen Elizabeth University Hospital, Clinical Biochemistry, Results: Total symptom score increased significantly in HA,
Glasgow, United Kingdom euglycaemia vs hypoglycaemia; 18.9 6.9 vs 37.4 10.8,
3
Ninewells Hospital, Clinical Technology Centre, Dundee, p < 0.0001, but not in IAH, 17.8 5.4 vs 18.7 5.7, p = 0.4).
United Kingdom Global CBF increased significantly in both HA and IAH (HA
6.3%, p = 0.01; IAH 8.0%, p = 0.03). A within group, paired t-test
Background and Aims: To assess associations between ad- (euglycaemia vs hypoglycaemia) in HA showed an increase in
mission glucose number (AGN) and admission outcomes for rCBF in the thalamus and a decrease in the hippocampus and
inpatients with T1D. temporal lobes in response to hypoglycaemia. The same analysis
Methods: Inpatient capillary blood glucose (CBG) data for in IAH showed no changes in the thalamus, but significant in-
patients with T1D in our health board were identified for 6-year creases in the dorsolateral prefrontal cortex and orbitofrontal
period and associated with most recent pre-admission HbA1c. cortex, areas involved in executive function. A significant de-
AGN was calculated as first CBG measured during admission crease in rCBF was noted in the temporal cortex, involved in
(mmol/L), subtracted from most recent pre-admission HbA1c memory and processing of sensory input.
A-34 ATTD 2017 ORAL ABSTRACTS

Conclusions: IAH is associated with loss of the thalamic re- toring, reducing possible complications of subcutaneous sensors/
sponse to hypoglycaemia, a region involved in the modulation of lancing, cutting healthcare costs, and increasing better long-term
stress pathways. Differences in frontal responses between HA and outcomes due to more frequent monitoring.
IAH may be maladaptive, explaining dissimilarities in behavioural
responses to hypoglycaemia between IAH and HA. 085
SHORT-TERM EFFECTS OF LOW CARBOHYDRATE
084 DIET ON GLYCAEMIC CONTROL AND METABOLIC
NON-CONTACT, IMAGE ANALYSIS OF THE RISK MARKERS IN PATIENTS WITH TYPE 1
PUPILLARY PROCESS AS A MEANS OF DETERMINING DIABETES A RANDOMISED OPEN-LABEL
AND MONITORING HYPER/HYPOGLYCEMIA CROSS-OVER STUDY
K. Steinberg1 A. Ranjan1, S. Schmidt1, C. Damm-Frydenberg1, I. Steineck1,
1
J.J. Holst2, S. Madsbad1, K. Nrgaard1
GlucoSight, Engineering, Nashua, USA 1
Copenhagen University Hospital Hvidovre, Department
Background and Aims: To develop a completely non- of Endocrinology, Research Unit, Hvidovre, Denmark
2
invasive, no-contact method of determining blood glucose level. University of Copenhagen, Department of Biomedical
The intent is to allow a person with diabetes to measure his or her Sciences, Faculty of Health and Medical Sciences,
glucose level and only employ traditional blood-based analysis Copenhagen, Denmark
when out-of-range indication is provided.
Background and Aims: To compare effects of high versus
low carbohydrate diet on glycaemic control and on metabolic
risk markers in patients with type 1 diabetes.
Methods: A randomised cross-over study was performed on ten
patients with insulin pump-treated type 1 diabetes (4 females, me-
dianSD age 48 10 years, diabetes duration 23 7 years, HbA1c
53 6 mmol/mol (7.0 0.6%)). Patients followed one week of high
carbohydrate diet (HCD >250 g daily) and one week of isocaloric
low carbohydrate diet (LCD <50 g daily) in random order. At the
end of each week, we downloaded data from sensor-augmented
insulin pumps and collected fasting blood and urine samples.
Results: Diet compliance was high (meanSD 225 30
vs 47 10 g carbohydrates, p < 0.0001) and change in body weight
did not differ between diets (-0.8 1.5 vs -1.6 1.5, p = 0.50).
Mean sensor glucose values did not differ between diets (7.3 1.1
Methods: The controlled stimulation of the pupillary process vs 7.4 0.6 mmol/l, p < 0.05). However compared with HCD,
is correlated with the blood sugar level of a person with diabetes. patients on LCD had lower glucose variability (SD 2.6 0.4 vs
Infrared video is recorded while the patients eye is placed in a 1.9 0.4 mmol/l, p = 0.009) and lower total daily insulin dose
high-isolation environment and stimulated with a brief flash of (38.8 8.1 vs 21.6 3.5 U/day, p < 0.0001). LCD further increased
white light. The pupil diameter is analyzed over the duration of fasting levels of glucagon (5.0 2.6 vs 6.9 4.1 pmol/l, p = 0.02),
the stimulation (8 seconds), and correlated against previously ketones (214.9 100.3 vs 944.6 387.7 lmol/l, p < 0.001) and free
derived, patient specific, baselines. fatty acids (0.6 0.3 vs 0.9 0.4 mmol/l, p < 0.001). Other lipid
Results: Various data points along the pupillary reaction and renal biomarkers were unaffected by the diets.
curve have shown to be indicators of the patients blood sugar Conclusions: One week of LCD reduced glycaemic vari-
level, allowing the method to provide feedback as to whether the ability and daily insulin dose without altering mean glucose
patients glucose is normal, hyper- or hypoglycemic. levels but elevated concentration of ketones, free fatty acids and
Conclusions: The method will allow persons with diabetes to glucagon. Long-term studies are needed to determine whether
check their glucose range without the need for blood-based anal- these study effects are persistent.
ysis. This will increase the frequency with which a person will and ClinicalTrials.gov: NCT02578498
can check their glucose, increasing their ability to control their
glucose level, decreasing the pain associated with glucose moni- 086
CONTINUOUS GLUCOSE MONITORING IN VERY
PRETERM INFANTS: A RANDOMIZED
CONTROLLED TRIAL
A. Galderisi1, A. Facchinetti2, G. Steil3, P. Ortiz-Rubio3,
F. Cavallin4, E. Baraldi5, D. Trevisanuto5, C. Cobelli2
1
Yale University, Pediatric Endocrinology, New Haven, USA
2
University of Padova, Department of Information Engineering,
Padova, Italy
3
Harvard University, Boston Childrens Hospital, Boston, USA
4
Independent Statistician, Padova, Padova, Italy
5
University of Padova, Department of Womans and Childs
Health, Padova, Italy
ATTD 2017 ORAL ABSTRACTS A-35

Background and Aims: Impaired glucose control in very

preterm infants is associated with increased morbidity, mortality
and poor neurological outcome. Strategies based on insulin ti-
tration have been unsuccessful in achieving euglycemia in the
absence of severe complications. We assessed whether use of an
algorithm based on continuous glucose monitoring(CGM) data
for adjustment of glucose infusion is more effective than stan-
dard care in maintaining very preterm infants in euglycemia
during the first week of life.
Methods: We conducted a randomized controlled trial at the
Neonatal Intensive Care Unit of University of Padova (Italy).
Fifty newborns, born 32 weeks gestational and/or with birth-
weight 1500 g, were randomized (1:1) to two groups within
48 hrs of life: unblinded-CGM (UB-CGM), with glucose infu-
sion adjusted according to an algorithm driven by CGM (Dex-
comG4Platinum) continuously and hypoglycemic alarms active;
blinded-CGM (B-CGM), with display darkened and no alarms.
The primary outcome was defined as time spent in euglycemic
range (4-8mmol/L). Secondary outcomes were time spent in mild
(<4mmol/L) and severe (<2.6mmol/L) hypoglycemia; mild
(>8mmol/L) and severe (>10mmol/L) hyperglycemia; and glu-
cose variability. Clinicaltrials.govNCT02583776.
Results: UB-CGM group showed an increase of time spent
in euglycemic range vs. B-CGM (median, 84% vs. 68%, p <0.001),
with decreased time spent in severe hypoglycemia (1.5% vs. 0.2%,
p = 0.007), mild hypoglycemia (16.9% vs. 12.1%, p = 0.04) and
severe hypoglycemia (0.3% vs. 0.0%, p = 0.04), respectively. UB-
CGM showed decreased glycemic variability vs. B-CGM (glucose
SD 1.1 0.3mmol/L vs. 1.5 0.4mmol/L, p = 0.0106; coefficient of
variation 22.8 4.2% vs. 27.9 5.0%; p < 0.001). cemia and minimizes glycemic variability in preterm infants in
Conclusions: CGM coupled with a glucose infusion algorithm neonatal intensive care unit, during the first week of life.
increases the time spent in euglycemic range, reduces hypogly-

087
EFFECTS OF HAWTHORN ON HBA1C
AND LIPIDS LEVELS IN JORDANIAN
DIABETES PATIENTS (TYPE2)
A. Aljamal1
1
Al al-Bayt University, Biological Sciences, Mafraq, Jordan

Background and Aims: The present study was designed to

investigate the effects of supplementation of Hawthorn on
HbA1C and lipids levels among type2 diabetes patients.
Methods: The samples consisted of 55 subjects with type2
diabetes and the doses of Hawthorn were equally administered
orally in the form of capsules each capsules contain (500 mg),
with breakfast, lunch and dinner. The doses were given for 12
weeks. Blood samples were taken on the starting day of the
experiment and at the end of 12 weeks.
Results: The fasting blood glucose, HbA1C and lipid levels of
type2 patients were determined. The results at starting day were
found to be 223.6 mg/dl, the mean value of HbA1C was 8.5%, and
lipid levels were: triglycerides (235.5 mg/dl), total cholesterol
(310 mg/dl), low-density lipoprotein (LDL) (155.2 mg/dl), and
high density lipoprotein (HDL) (52.4 mg/dl). When the diabetic
individuals used the doses of Hawthorn for 12 weeks, their mean
fasting blood glucose level dropped to 186.34 mg/dl, HbA1C 7.2,
triglycerides (160 mg/dl), total cholesterol (187.6 mg/dl), LDL
(115.5 mg/dl), and HDL increased (69.2. mg/dl).
Conclusions: The administration of Hawthorn caused re-
duction in the blood glucose; HbA1C and lipid levels were sig-
nificant at P < 0.05, P < 0.001 respectively.
ATTD 2017 E-Poster Discussion Abstracts

088 M. Vidrio-Velazquez8, R.S. Nino-Vargas14,

E. Sainz-De La Maza-Viadero15, C. Magis-Rodriguez2
IS CANCER SCREENING IN DIABETES JUSTIFIED?
1
INSIGHTS FROM ELECTRONIC MEDICAL RECORDS Instituto Tecnologico de Monterrey / Centro Medico ABC,
1 1 1 1 Endocrinologia, Mexico, Mexico
J. Kesavadev , A. Shankar , L. Ramachandran , G. Krishnan , 2
Centro Nacional para la Prevencion y el Control del VIH y el
S. Jothydev1
Sida- Secretaria de Salud, Epidemiologia, Mexico City, Mexico
1 3
Jothydevs Diabetes Research Centre, Diabetes, Hospital General de Tepic, Departamento de Medicina
Thiruvanathapuram, India Interna, Tepic, Mexico
4
CEMEDIN, Endocrinologia, Monterrey, Mexico
5
Background and Aims: Evidences are on the rise which in- Instituto Tecnologico de Monterrey, Ciencias Basicas y
dicates a strong and consistently increased risk of incident cancer Clinicas, Mexico City, Mexico
6
associated with diabetes. Biomarkers constitute major tests for Hospital General Dr. Miguel Silva, Secretaria de Salud,
the early detection of cancer. To explore this association, we Endocrinologia, Morelia, Mexico
7
performed a retrospective analysis of our Electronic Medical Hospital DIF de la Ninez Hidalguense, Endocrinologia
Records to assess the utility of certain cancer markers and early Pediatrica, Pachuca, Mexico
8
detection of malignancy in asymptomatic subjects with diabetes. HGR 110 IMSS Guadalajara, Jal, Endocrinologia,
Methods: EMR of diabetes patients enrolled at our centre Guadalajara, Mexico
9
(n = 1300) from January 2014 to 2016, who were diagnosed of Hospital Espanol, Endocrinologia Pediatrica, Mexico City,
cancer were de-identified. Mexico
10
Results: 1324 patients were screened. 2.95% (n = 39) were Hospital Angeles Lomas, Endocrinologia, Mexico City,
diagnosed of cancer after the onset of diabetes. Of these, 17 cases Mexico
11
were screened and diagnosed during routine visits. Mean age Hospital Rovirosa, Secretaria de Salud, Diabetes,
63.77 9.77 yrs; males (64.10%). Type of cancers detected were- Villahermosa, Mexico
12
21.51% breast; 12.82% pancreas; 10.26% each of liver and Centro Medico ABC Santa Fe / Hospital Angeles Lomas,
prostrate; 7.69% each of multiple myeloma, CA colon/intestine, Endocrinologia Pediatrica, Mexico City, Mexico
13
stomach; and 2.56% each of gall bladder, lung, oral, kidney, Hospital De Gineco, Pediatra No. 31 IMSS, Endocrinologia
bladder, uterus, endometrium and non-Hodgkins lymphoma. Pediatrica, Mexicali, Mexico
14
CA of prostrate, liver and pancreas (16.00% each) were equally Centro para la Prevencion y Atencion Integral del VIH/SIDA
predominant among males whereas CA breast (42.86%) was del Distrito Federal, Secretaria de Salud, Jefatura de Sistemas
predominant among females. de Informacion, Mexico City, Mexico
15
Conclusions: Cancer markers such as CA 125, CA 19.9, AFP, UNIVERSIDAD IBEROAMERICANA, EDUCACION, Mexico
PSA etc. are non-specific but sensitive. Higher levels may not City, Mexico
always be significant. But early detection at a curable stage of
malignancy makes it an economically viable step. Our study Background and Aims: The information regarding treatment
highlight the significance of cancer screening in diabetes due to a practices and follow-up of type 1 diabetes (T1D) patients in
higher likelihood of incident cancers among this population. De- Mexico is limited. An online system, RENACED Diabetes Tipo
spite the expensive nature of cancer diagnostic tests, incorporating 1 (DT1) was created to have a longitudinal T1D registry with
them into routine diabetes treatment protocol may be justifiable. real-life data in Mexico.
Methods: Descriptive analysis of 474 T1D patients registered
on RENACED DT1 in 10 Mexican States (12 different medical
units), until 10/5/2016.
089 Results: Forty-one percent of patients were men, mean age 25
years. Fifteen percent have family history of T1D and 58% of
2016 RESULTS OF THE MULTI-CENTRIC STUDY
T2D. Thirty-three percent performs SMBG (self monitor of
RENACED DIABETES TIPO 1
blood glucose) 4 times/day, 32% uses an insulin-pump and 67%
R.N. Faradji1, M. Valenzuela-Lara2, are on basal-bolus regimen. Patients that perform SMBG 4
J.F. Bustamante-Martinez3, N.E. De La Garza-Hernandez4, times/day, had lower HbA1c levels (8.04; CI95% 7.78.4) than
A.P. Diaz-Barriga-Menchaca5, A. Escobedo-Ortiz6, those that monitor less (8.6; CI 95% 8.48.9) (p < 0.05). A lower
A. Flores-Camargo5, L. Islas-Ortega7, C. Lopez-Miramontes8, HbA1c level (<0.05) was observed in patients that use a con-
A. Martinez-Ramos-Mendez9, S. Miracle-Lopez10, tinuous glucose monitor (CGM) (7.8; CI 95% 7.48.1) than in
L.A. Ramirez-Toscano8, E. Rodriguez-Sanchez11, those who do not (8.7; CI 95% 8.59.0). A total of 19% and 32%
M. Tavera-Hernandez12, J.C. Valenzuela-Montoya13, of patients had HbA1c level <7% and <7.5%, respectively. The

A-36
ATTD 2017 E-POSTER DISCUSSION ABSTRACTS A-37

Results: Usability by patients was evaluated at Oxford Uni-

versity (2011, published elsewhere), Great Ormond St Hospital
and Ramsey Group Practice (2015/6). Universal ease of use was
presence of mild/moderate hypoglycemia is high (68.8%), as observed, with high patient acceptance.
well as that of severe hypoglycemia (18%), the presence of At Swansea University (2015) Home OGTT demonstrated
chronic complications was low (6.5%). good correlation to laboratory glucose (R2 * 0.96). SmartSensor
Conclusions: According to the literature, the percentage of telemed has reprocessed (2015) Home OGTT data collected in an
patients with HbA1c at goal is lower than desired, even though they earlier comparative OGTT study; sensitivity and specificity are
are on state-of-the-art treatment. Performing SMBG 4 times/day above 97%.
and CGM use are associated with better glycemic control. This is Conclusions: Home OGTT is easy to use, acceptable to pa-
the first online system for T1D registry in Mexico. A larger number tients, and easy to administer in the clinic. With considerable
of cases will lead to better national representation. improvements in performance over alternatives, Home OGTT
could transform detection of glucose intolerance and diabetes.

090
091
EVALUATION OF A HOME OGTT TEST KIT: A
CONVENIENT, SENSITIVE AND SPECIFIC TEST INPATIENT AND INSULIN COSTS AMONG TYPE 1
FOR GLUCOSE INTOLERANCE AND DIABETES DIABETES PATIENTS TREATED WITH SENSOR-
AUGMENTED INJECTION VERSUS SENSOR-
J. Jackson1, S. Luzio2, A. Allinson3, C. Peters4
AUGMENTED PUMP THERAPIES
1
SmartSensor telemed Ltd, Executive Office, Didcot, United C. Zhu1, M. Bansal1, M. Gill1
Kingdom
2 1
Swansea University Medical School, Institute of Life Sciences, Medtronic plc, Health Economics and Reimbursement,
Swansea, United Kingdom Northridge, USA
3
Ramsey Group Practice, GP Surgeries, Ramsey, Isle of Man
4
Great Ormond St Hospital, Department of Endocrinology Background and Aims: Continuous glucose monitoring
and Diabetes, London, United Kingdom (CGM) has been proven to enhance the clinical outcomes of type
1 diabetes (T1D) patients. Sensor-augmented injection (SAI) and
Background and Aims: There is considerable controversy sensor-augmented pump (SAP) therapies use CGM with multiple
over the performance of HbA1c for screening and diagnosis, with daily injections and insulin pumps, respectively.
many studies showing poor sensitivity and specificity. OGTT Methods: Assessment of inpatient and insulin costs from a pri-
offers greater sensitivity and specificity, but is inconvenient for vate payers perspective among patients treated with SAI and SAP
patients to access and for health care professionals to provide. was performed. Data were extracted from Truven MarketScan
Additionally, there are several circumstances where OGTT is the database (2010-2014) for patients with a minimum 1 year follow-up
most or the only recommended test, such as detecting gestational time since study index date. Specific algorithms were applied to
diabetes. A new test kit aims to provide the performance of define T1D patients on treatment regimens (SAI vs. SAP). The an-
OGTT with the convenience of a home test; Home OGTT. alyses were restricted to patients aged 18-64 years with continuous
Aims: To establish usability and acceptability of a recently enrollment in the same health plan with pharmacy benefits and
optimised version of Home OGTT and to determine its perfor- continuously receiving therapy from the analysis index date.
mance for detecting dysglycaemia. Results: Two cohorts (N = 644 for both) were matched based on
Methods: Home OGTT (SmartSensor telemed Ltd, Didcot, patient characteristics: age, gender, Charlson Comorbidity Index
UK) uses a novel disposable device to perform an OGTT pro- (CCI), and macrovascular and microvascular complications. Mat-
cedure and record the test data for analysis by a healthcare pro- ched annual costs, adjusted to 2015 dollars, showed that SAP patients
fessional. The latest version allows the patient to scan the test had approximately 32% lower inpatient cost and 30% lower insulin
data using a smartphone, with results immediately available in an cost. SAI therapy was associated with an increased readmission rate
electronic record. and average length of stay in hospitals compared to SAP usage.
A-38 ATTD 2017 E-POSTER DISCUSSION ABSTRACTS

1
Conclusions: According to this matched cohort analysis, payer University of Cambridge, Wellcome Trust-MRC Institute
costs were higher for inpatient admissions and insulin among of Metabolic Science, Cambridge, United Kingdom
2
T1D patients using CGM technology with injections versus pump Inselspital- Bern University Hospital, University of Bern,
therapy. SAP therapy should be considered as a means to reduce Department of Diabetes, Endocrinology, Clinical Nutrition
healthcare expenditures and improve resource utilization. & Metabolism, Bern, Switzerland
3
Cambridge University Hospitals NHS Foundation Trust,
Department of Diabetes & Endocrinology, Cambridge,
United Kingdom
4
092 University of Cambridge, Department of Paediatrics,
Cambridge, United Kingdom
FROM HUMAN MESENCHYMAL STEM CELL 5
Medical University of Graz, Department of Internal Medicine-
TO INSULIN-PRODUCING CELLS Division of Endocrinology & Diabetology, Graz, Austria
A. Refaie1, M. Gabr2, M. Zakaria2, S. Khater3, S. Ashamallah3,
A. Moustafa4, M. Ghoneim5 Background and Aims: None of the home closed-loop
1 studies so far focused specifically on well controlled type 1 di-
Urology & Nephrology Center, Mansoura University, abetes. In this population, we aimed to establish if the risk of
Nephrology, Mansoura, Egypt hypoglycaemia could be alleviated and glucose control improved
2
Urology & Nephrology Center, Mansoura University, by the application of closed-loop insulin delivery.
Biotechnology, Mansoura, Egypt Methods: In an open-label randomised crossover study design,
3
Urology & Nephrology Center, Mansoura University, seventeen type 1 diabetes adults with HbA1c<7.5% on insulin
Pathology, Mansoura, Egypt pump therapy (7 male, age 45 9 years, HbA1c 6.9 0.4%, diabetes
4
Urology & Nephrology Center, Mansoura University, duration 29.7 12.0 years) underwent a period of day-and-night
Immunology, Mansoura, Egypt closed-loop insulin delivery, and a period of usual pump therapy, the
5
Urology & Nephrology Center, Mansoura University, latter reflecting current clinical practice. Each period lasted four
Urology, Mansoura, Egypt weeks. The study was carried out during free-living without su-
pervision. During closed-loop, model predictive control algorithm
Background and Aims: We could produce insulin-producing directed insulin delivery; prandial insulin delivery was applied using
cells (IPCs) from human mesenchymal stem cells (MSCs) by di- standard bolus wizard. Analyses were by intention to treat.
rected differentiation. Although the yield of the formed IPCs was Results: Time when glucose was in target range 3.9 to 10mmol/l,
modest, yet transplantation of these cells in diabetic mice resulted in primary outcome, was significantly increased during closed-loop
their cure. We have tried to provide an explanation for this obser- compared to usual therapy (p < 0.001; Table). Mean glucose, time
vation. Differentiated MSCs, were transplanted under the renal spent above and below target range were significantly lower during
capsule of diabetic mice. The kidneys were harvested after 1, 2, 4 closed-loop compared to control (p < 0.001 to p = 0.007). Hypogly-
and 12 weeks. IPCs were counted at each period. The proportion of caemia burden measured by AUC <3.5mmol/l was reduced during
IPCs increased to reach a maximum of *20% at 4 weeks. closed-loop (p = 0.001). Glycaemic variability as measured by
Aim: To study the efficiency of these cells in treatment of standard deviation of glucose was also reduced (p = 0.002). Total
larger animals and identify their functional longevity. daily insulin was comparable (p = 0.29). No serious adverse events
Methods: we have induced diabetes in 6 dogs (15 20 Kg) by occurred.
a mixture of alloxan and streptozotocin. Differentiated human Conclusions: Four-week home use of unsupervised day-and-
cells (5 million/kg) were encapsulated and transplanted beneath night hybrid closed-loop under free-living in well-controlled
the rectus sheath. adults with type 1 diabetes is safe, improves glucose control and
Results: Six dogs are currently under follow up. Three had reduces hypoglycaemia burden. Larger and longer studies are
completed a 6 months follow up. Two became euglycemic with warranted.
normal glucose tolerance curve. The third is on the hypergly-
cemic side although the profile of its glucose tolerance resembles
a normal one. A harvested capsule after 6 months form trans-
plantation was examined. By immunoflourescence, IPCs were
seen and co-expression with c-peptide was confirmed. The pro-
portion of IPCs was again in the range of *20%.
Conclusions: In conclusion, IPCs can be formed by directed
differentiation from of MSCs. These cells undergo further differ-
entiation in vivo. Evidence was provided that these cells can cure
chemically induced diabetes in small as well as large animals.

093
CLOSED-LOOP INSULIN DELIVERY IN WELL-
CONTROLLED ADULTS WITH TYPE 1 DIABETES:
A FREE-LIVING DAY-AND-NIGHT RANDOMISED
CROSS-OVER TRIAL
L. Bally1,2,3, H. Thabit1,3, S. Hartnell3, M. Tauschmann1,4,
J.K. Mader5, H. Kojzar5, M.E. Wilinska1,4, T.R. Pieber5,
M.L. Evans1,3, R. Hovorka1,4
ATTD 2017 E-POSTER DISCUSSION ABSTRACTS A-39

094
ACCURACY OF CONTINUOUS GLUCOSE
MONITORING DURING POSTPRANDIAL PERIOD
AND ITS INFLUENCE ON CLOSED-LOOP
PERFORMANCE
L. Biagi1,2, A. Hirata Bertachi1,2, I. Conget3, C. Quiros3,
M. Gimenez3, F.J. Ampudia-Blasco4, P. Rossetti5, J. Bondia6,
J. Veh 1
1
Universitat de Girona, Institut dInformatica i Aplicacions,
Girona, Spain inserting and calibrating new sensors and with sensor accuracy
2
Federal University of Technology, Parana, COMIN, and reliability throughout its functional life. We developed a new
Guarapuava, Brazil sensor for our next-generation system and tested its accuracy
3
Hospital Clnic i Universitari, Endocrinology Dpt, Barcelona, versus reference venous (YSI) values collected during clinic
Spain visits throughout 10 days of wear.
4
Hospital Clnico Universitario de Valencia, Endocrinology Methods: Fifty subjects (49 with type 1 diabetes, 1 with type 2
and Nutrition Dpt, Valencia, Spain diabetes, 20 male, mean SD age 32.5 18.7 years) enrolled and
5
Hospital Francesc de Borja, Endocrinology Dpt, Gandia, 1 subject withdrew early. Subjects wore 1 sensor each, and used
Spain SMBG values (CONTOURNEXT meter) for once-daily cali-
6
Universitat Politecnica de Valencia, Instituto Universitario brations. Clinic visits were conducted on days 1, 2, 4, 7, and 10 of
de Automatica e Informatica Industrial, Valencia, Spain sensor wear; sensor readings were compared to venous YSI
values for up to 8 hours (ages 6-17) or 12 hours (adults). A total
Background and Aims: Postprandial period (PP) represents a of 994 paired CGM-YSI values were analyzed for mean absolute
challenge for closed-loop (CL) systems. The development of more relative difference (MARD) and within-15% and within-20%
accurate sensors will contribute to increase the performance of CL agreement rates (matched pairs within 15 or 20 mg/dL for ref-
control. In this study we analyzed the accuracy of the Medtronic erence values 100 mg/dL or within 15 or 20% for reference
Paradigm VeoTM EnliteTM continuous glucose monitoring sys- values >100 mg/dL).
tem used in a CL study aimed to PP glucose control and the impact Results: The overall MARD was 8.1% (95% confidence in-
of sensors accuracy in the CL performance. terval, 7.6 to 8.5%). The Table shows accuracy metrics overall
Methods: Twenty subjects underwent a standardized mixed and for each in-clinic day. Thirty-nine (79.6%) of the 49 de-
meal test on 4 occasions wearing two glucose sensors in parallel ployed sensors were functional at the end of Day 10; adhesive
(randomised, prospective, crossover). On 2 occasions glycemic failure accounted for most of the early attrition.
control was made by CL and in another two, usual pump therapy Conclusions: This new CGM systems accuracy throughout
was used. During 8 hours plasma glucose (PG) was measured its 10-day functional life, together with the convenience associ-
every 15 min. Sensor accuracy was assessed through the mean ated with once-daily calibrations, supports its nonadjunctive use
absolute relative difference (MARD) in different glucose ranges. in diabetes management and closed-loop insulin delivery systems.
Times in each glycaemic range were compared between the trials
with the 10 sensors with best and worst MARD.
Results: Overall, MARD was 12.0 7.5%; 18.9 11.9% and
10.3 + 8.7% in <70 mg/dl and >180 mg/dl glucose ranges, re- 096
spectively. In CL studies, the 10 best sensors displayed a MARD PRECISE II PIVOTAL TRIAL OF A LONG TERM
of 4.5 0.9% in comparison to 19.8 7.5% in trials with the IMPLANTABLE CGM SYSTEM: 90 DAYS
10 worst sensors. Studies performed with the 10 best sensors OF SUSTAINED ACCURACY AND STRONG SAFETY
showed less time in hypoglycaemia (PG <70mg/dl) (2 7 vs. PROFILE
32 38 min; p = 0.03) and needed less rescues using carbohy-
drates, (1 vs 11). M. Christiansen1, L. Klaff2, R. Brazg2, A. Chang3, C. Levy4,
Conclusions: Under postprandial circumstances, Medtronic D. Lam4, D. Denham5, G. Atiee6, B. Bode7, S. Walters8,
Paradigm VeoTM EnliteTM system showed a numerical accu- T. Bailey9
racy in different glucose ranges close to previously reported. The 1
Diablo Clinical Research, Clinical, Walnut Creek, USA
performance of closed-loop algorithm to control postprandial 2
Rainier Clinical Research Center, Clinical, Renton, USA
period was related to sensor accuracy. 3
John Muir Physician Network Clinical Research Center,
Endocrinology and Metabolism, Concord, USA
4
095 Mount Sinai Diabetes Center, Endocrinology- Diabetes and
Metabolism, New York, USA
5
ACCURACY AND STABILITY OF A NOVEL Clinical Trials of Texas, Clinical, San Antonio, USA
6
SUBCUTANEOUS GLUCOSE SENSOR Worldwide Clinical Trials, Clinical, San Antonio, USA
7
J. Leach1, J. Welsh2, D. Le3, J. Hughes1 Atlanta Diabetes Associates, Clinical, Atlanta, USA
8
Senseonics, Clinical, Germantown, USA
1 9
Dexcom Inc., Research & Development, San Diego, USA AMCR Institute, Clinical, Escondido, USA
2
Dexcom Inc., Clinical Affairs, San Diego, USA
3
Dexcom Inc., Biometrics, San Diego, USA Background and Aims: A pivotal trial studying the efficacy
and safety of an implantable continuous glucose monitoring
Background and Aims: Users of continuous glucose moni- (CGM) system (Eversense) was investigated throughout a 90-
toring (CGM) systems are concerned with the frequency of day period.
A-40 ATTD 2017 E-POSTER DISCUSSION ABSTRACTS

Methods: Blinded, single-arm, 90-day prospective multi-center Conclusions: One per day calibration maintained system
study assessing safety and accuracy of a long-term CGM sensor. accuracy throughout the 90-day period for the Eversense CGM
Enrollment included 82 subjects in the Primary Effectiveness Anal- system, which is equal to or better than the current two per day
ysis Population for accuracy assessment by comparing matched calibration CGM systems. The Eversense CGM system pro-
CGM glucose values to YSI measured venous glucose values and vides the benefit of a long term implantable that is accurate
an additional 8 site training subjects. The primary safety objective while having the potential to lessen the burden of daily diabetes
was the incidence of device-related or sensor insertion/removal management.
procedure-related adverse events. All sensors were inserted in the
upper arm and 15 subjects were inserted with 2 sensors, bilaterally. 098
Results: The accuracy assessment found an MARD (40-
400 mg/dL) for the Primary Effectiveness Population of 8.8% THE EFFECT OF THE MOBILE APPLICATION
(n = 16,653, SD 8.6%, 95% CI 8.6%8.9%). A Kaplan-Meier EUGLYCA ON GLYCEMIC CONTROL
survival analysis found 91% of the sensors functioning through OF CHILDREN AND ADOLESCENTS WITH DIABETES
90 days. Fourteen (14) device or procedure related adverse MELLITUS TYPE 1
events occurred in 7 subjects over the 9,773 sensor in vivo days. C. Chatzakis1, D. Floros2, K. Tsiroukidou1, A. Vamvakis1,
The most common adverse events were mild bruising or ery- K. Kosta1, I. Tsanakas1, M. Papagianni1
thema. There was one SAE where a second procedure was
1
needed for a sensor removal due to issues of locating the sensor. Hippokrateion General Hospital of Thessaloniki- Aristotle
There were no incisional infections either at insertion or removal. University of Thessaloniki, Pediatric Endocrinology
Unlike other CGM systems, there were no complaints of skin Unit, Third Department of Pediatrics, Thessaloniki, Greece
2
irritation or reaction to the adhesive patch. Aristotle University of Thessaloniki, Electrical Engineering,
Conclusions: The Eversense CGM demonstrates both an ex- Thessaloniki, Greece
cellent safety profile and sub 9.0% MARD accuracy during
continuous use for up to 90 days. Background and Aims: Euglyca is a mobile application that
we developed for patients with diabetes. It calculates the amount
of carbohydrates and lipids that a patient consumes during a meal
097 and by taking into consideration eight more parameters (pre-
meal blood glucose, targeted blood glucose, insulin/carbs and
ACCURACY ASSESSMENT FOR A LONG TERM insulin/lipids ratio, insulin sensitivity, active insulin, physical
IMPLANTABLE CGM SYSTEM WITH ONE PER DAY activity and illness) calculates the required bolus dose of insulin.
CALIBRATION IN THE PRECISE II STUDY Aim of this study is to evaluate the efficacy of this application on
M. Christiansen1, L. Klaff2, R. Brazg2, A. Chang3, C. Levy4, patients glycemic control.
D. Lam4, D. Denham5, G. Atiee6, B. Bode7, X. Chen8, T. Bailey9 Methods: 36 children and adolescents with T1DM were in-
cluded in the study. 18 of them were appointed to use the appli-
1
Diablo Clinical Research, Clinical, Walnut Creek, USA cation for three months and the other 18 were matched controls.
2
Rainier Clinical Research Center, Clinical, Renton, USA At the baseline and three months later, we calculated the amount
3
John Muir Physician Network Clinical, Endocrinology of hypoglycemias, hyperglycemias and normoglycemias for each
and Metabolism, Concord, USA patient and their HbA1c as well, in order to proceed with the
4
Mount Sinai Diabetes Center, Endocrinology, Diabetes appropriate comparisons.
and Metabolism, New York, USA Results: In our target group we found an increase in the
5
Clinical Trials of Texas, Clinical, San Antonio, USA number of normoglycemias by 18% while in our control group
6
Worldwide Clinical Trials, Clinical, San Antonio, USA there was a decline by 5% (P < 0.001). Furthermore, we found
7
Atlanta Diabetes Associates, Clinical, Atlanta, USA 6.7% and 12% decline in the number of hypoglycemias and
8
Senseonics, Engineering, Germantown, USA hyperglycemias respectively while in our control group they
9
AMCR Institute, Clinical, Escondido, USA were elevated by 0.2% and 5.5%. In addition, in the target group
it was revealed a reduction in HbA1c by 1% while in the control
Background and Aims: The PRECISE II pivotal trial was a group there was a 0.25% increase (P < 0.001).
blinded, single-arm, 90 day prospective multi-center study of the Conclusions: Mobile application Euglyca improved the
fully implantable sensor based Eversense CGM System. Parti- glycemic control of children and adolescents with T1DM, how-
cipants used the system per the manufacturers instruction by ever further research is needed.
performing two per day calibration. The impact of one per day
calibration on sensor accuracy was evaluated with the 82 subjects 099
in the Primary Effectiveness Analysis Population.
Methods: For the analysis of one per day calibration, only one PERFORMANCE AND ACCURACY CAPABILITY
of the two calibrations was used to assess the calibration stability OF A NEW, WIRELESS-ENABLED BLOOD GLUCOSE
throughout 24 hrs. When possible, the SMBG measurement MONITORING SYSTEM THAT LINKS TO A SMART
taken the night before a next day 12 hour in-clinic visit was used MOBILE DEVICE: LABORATORY AND CLINICAL
to enable accuracy assessment. SAMPLE REAPPLICATION STUDIES
Results: The overall 90 Day accuracy measured by >16,000 YSI B. Harrison1, D. Brown1, M. Takeshima2
glucose values found an MARD (40-400 mg/dL) of 9.5% (SD 9.3%,
1
95% CI 9.3% 9.6%) for one per day calibration compared to 8.8% Ascensia Diabetes Care, Parsippany NJ, USA
2
(SD 8.6%, 95% CI 8.6% 8.9%) for two per day calibration. Similar Panasonic Healthcare Co. Ltd., Toon City Ehime, Japan
stability in performance was maintained with 90.5% and 93.3% of
matched CGM glucose within 20/20% of YSI glucose for the one per Background and Aims: To evaluate performance and accu-
day calibration and two per day calibration, respectively. racy of the ContourNext ONE blood glucose monitoring system
ATTD 2017 E-POSTER DISCUSSION ABSTRACTS A-41

(BGMS) during Second-Chance sampling. The BGMS features group (V group D (C-peptide) = 1.3 + 3.6 ng/ml, M group D (C-
an easy-to-use, wireless-enabled BG meter that links to a smart peptide) = -0.3 + 1.5 ng/ml, p < 0.05, V group D (HOMA-
mobile device via Bluetooth technology and syncs with the b) = 22.0 + 33.8%, M group D (HOMA-b) = 8.1 + 55.8%,
Contour Diabetes app. p < 0.05).
Methods: In the laboratory study, samples were tested at 3 Conclusions: The addition of V to M for a period of six
temperatures (16C, 22C, 34C) with blood adjusted to 3 BG months had no effect on AS in drug nave T2DM patients but V
levels (70, 300, 500 mg/dL) at 3 hematocrit levels (20%, 42%, treatment improved glycemic control (HbA1c) and b-cell func-
55%). Two sample reapplication methods were used (initial tion (C-peptide, HOMA-b).
volume, 0.28 and 0.46 lL); each sample was tested with 3 delay
times between initial and second inoculation (5, 30, 55 sec-
onds). For each sample and condition, 10 replicate BGMS
readings were obtained with each of 3 test strip lots. In the 101
clinical study, 52 subjects with diabetes performed self-tests
with an intentionally insufficient initial blood application to PHARMACOKINETICS, PHARMACODYNAMICS,
produce a number of Second-Chance sampling opportunities. AND SAFETY FOLLOWING SINGLE OR REPEATED
Results were compared with YSI reference results and assessed 3 MG NASAL GLUCAGON DOSES IN ADULTS
per the following acceptance criterion: 95% of results within WITH TYPE 1 OR TYPE 2 DIABETES (T1D OR T2D)
15 mg/dL (BG <100 mg/dL) or 15% (BG 100 mg/dL) of H. Dulude1, E. Sicard2, M. Rufiange3, C. Piche1, C.B. Guzman4,
reference result. S. Zhang5, T. Shen6, J.G. Jacobson7, X.M. Zhang8
Results: In the laboratory study, results met protocol- 1
specified acceptance criteria. In the clinical study (BG range, Locemia Solutions, Clinical Development, Montreal,
65-347 mg/dL; hematocrit range, 35%-55%), 100% (85/85) of Canada
2
subject fingertip self-test results were within 15 mg/dL or Altasciences/Algorithme Pharma, Principal Investigator,
15% of YSI reference result. Moreover, 92.9% (79/85) of Montreal, Canada
3
self-test results were within 10 mg/dL or 10% of YSI ref- Altasciences/Algorithme Pharma, Scientific and Regulatory
erence result. Affairs, Montreal, Canada
4
Conclusions: In the laboratory and in a clinical setting used by Eli Lilly and Company, Diabetes Business Unit. Development,
subjects with diabetes, BGMS sample reapplication results met Indianapolis, USA
5
acceptance criteria. Eli Lilly and Company, Statistics, Indianapolis, USA
6
Eli Lilly and Company, PK/PD and Pharmacometrics,
Indianapolis, USA
7
Eli Lilly and Company, Diabetes, Indianapolis, USA
8
100 Eli Lilly and Company, Diabetes Business Unit, Development,
Toronto, Canada
EFFECT OF VILDAGLIPTIN ON ARTERIAL
STIFFNESS IN DRUG NAIVE PATIENTS Background and Aims: Examine the pharmacokinetics (PK),
WITH TYPE 2 DIABETES MELLITUS pharmacodynamics (PD), and safety of single or repeated 3-mg
I. Zografou1, P. Doukelis1, T. Griva1, E. Spentzou1, nasal glucagon (NG) doses given in randomized sequence in a
P. Anyfanti1, D. Papadopoulou1, C. Sampanis1 4-period, cross-over study.
Methods: Subjects (insulin-using adults with T1D or T2D,
1
Hippokration General Hospital, Diabetes Center, BMI 18.5-35.0 kg/m2) received 4 NG treatments (Trts) 1 wk
Thessaloniki, Greece apart. Trts were given 4 hrs after a low-carbohydrate breakfast.
Trts were: 1) Single 3-mg NG; 2) 3-mg NG plus another 3-mg
Background and Aims: Arterial Stiffness (AS) is a predictor NG 15 minutes later (same nostril); 3) 3-mg NG plus another
of cardiovascular (CV) events and mortality in patients with
diabetes mellitus. We assessed the effect of a DPP-4 inhibitor
vildagliptin (V) on AS in drug nave patients with type 2 diabetes
mellitus (T2DM).
Methods: Sixty-four drug nave subjects with T2DM and
inadequate glycemic control participated in this randomized,
open-label study. Half of the patients received metformin ( M)
1700 mg/d ( M group) and the other half M 1700 mg/d plus
vildagliptin 100 mg/d (V group) for 6 months. AS (carotid
femoral pulse wave velocity, cfPWV), body weight (BW),
body mass index (BMI), systolic blood pressure (SBP), dia-
stolic blood pressure (DBP), glycosylated hemoglobin
(HbA1c), Albumin/Creatinine ratio (A/C ratio), C-peptide,
HOMA-IR and HOMA-b were assessed at baseline and after
6 months.
Results: cfPWV, BW, BMI, SBP, DBP remained unchanged
after 6 months in both groups (p = NS). V decreased HbA1c
more effectively than M alone (V group: D (HbA1c) =
-1.7 + 0.7%, M group: D (HbA1c) = -1.2 + 1.2%, p < 0.05).
Moreover C-peptide and HOMA-b raised significantly in V
A-42 ATTD 2017 E-POSTER DISCUSSION ABSTRACTS

3-mg NG 15 minutes later (opposite nostril); 4) 2 concurrent 3-

mg NG doses (both nostrils).
Results: 32 subjects were enrolled (T1D: 23, T2D: 9).
Number of subjects who received Trts 1, 2, 3, 4 were 27, 28, 25
and 29, respectively. Baseline (BL) blood glucose was 40-
181 mg/dL. For Trts 1-4, PK parameters of change from BL for
glucagon were mean area under the curve 0-3 hrs: 2471, 4097,
4639, and 3611 hr  pg/mL, median Tmax: 0.17, 0.33, 0.50, and
0.33 hrs; PD parameters of change from BL for glucose were
mean area under the effect concentration 0-3 hrs: 157, 168, 190,
and 194 hr  mg/dL, median Tmax: 0.75, 1.00, 1.00, and 1.00 hrs.
Repeated NG doses resulted in higher glucagon concentrations,
but gave glucose responses comparable to single dose (Figure).
The only serious adverse event (AE; cellulitis) was not drug-
related. Most drug-related AEs were transient and resolved
within 2 hours.
Conclusions: Although repeat dosing resulted in greater
systemic glucagon exposure, it did not result in a meaningful discussion topics, given more time for discussing self-
increase in observed glucose response. All NG treatments were gathered data.
well-tolerated. Results: Individuals with Type 1 (n = 15) and Type 2 (n = 9)
completed the survey (Figure 1). While motivation was a
common challenge, stress management (Type 1), BG variability,
exercise and mood (Type 2) were the most wanted consultation
102
topics. The workshops revealed large differences in preferred
DIFFERENTIATING PRESENTATION OF PATIENT- ways of data presentation, e.g. GPs preferred quick ways of re-
GATHERED DATA BETWEEN TYPE 1 AND 2 viewing data during consultation, while specialist preferred to
DIABETES DURING CONSULTATIONS receive and review more detailed data before consultations
(Table 1).
M. Bradway1, A. Holubova2, R. Joakimsen3, E. Arsand1 Conclusions: Due to numerous factors associated with dia-
1
University Hospital of North Norway, Norwegian Centre betes, it is necessary to not only consider a patients diabetes type
for E-Health Research, Troms, Norway and clinicians specialties but also patients personal challenges,
2
Czech Technical University in Prague, Faculty of Biomedical resources, and capacity when preparing patient-gathered data for
Engineering, Prague, Czech Republic consultations. The Full Flow project (2016-19) will study use of
3
University Hospital of North Norway, Institute of Clinical dynamic data-formats for presenting patient-gathered data dur-
Medicine, Troms, Norway ing consultations.

Background and Aims: The integration of mobile health

(mHealth) technology within medical practice is a discussion
riddled with debate and unresolved questions. In addition to 103
privacy and security is the question of how to present patients THE IMPLEMENTATION OF MYSTAR CONNECT
mHealth self-gathered data during consultations. For chronic and CLINICAL INFORMATION SYSTEM IMPROVES
resource-heavy illnesses, such as diabetes, it is necessary to INTERMEDIATE OUTCOMES IN 42498 PEOPLE
consider adaptive formatting for the presentation of patients WITH DIABETES
sensor/app data to medical personnel, with the aim to determine
sound strategies and formats for presentation during consulta- V. Provenzano1, D. Brancato1, G. Picone1, A. Scorsone1,
tions. M. Fleres1, G. Saura1, V. Aiello1, F. Provenzano1, A. Di Noto1,
Methods: Both an anonymous online patient survey as well L. Spano1
as three workshops involving clinicians involved in diabetes 1
Hospital of Partinico, Regional Reference Center
treatment, EHR-vendors, patients and researchers were con- for Diabetology and Insulin Pumps, Palermo, Italy
ducted. Diabetes patients in Norway answered questions re-
lated to experienced challenges and desired consultation Background and Aims: The Diabetes Network of Palermo
District links all the public healthcare diabetes resources
of Palermo District. Since 2012, DNPD implemented a Chronic
Care Model (CCM) mainly based on the full sharing of
clinical data between primary care resources and special-
ized diabetes centers, through a single clinical information
system ( MyStar Connect). Aim of the present study is to
assess if the implementation of MyStar Connect was fol-
lowed by an improvement of the management of people with
diabetes.
Methods: Assessment of the trend of the proportion
of people with diabetes who yearly (from 2012 until to
2015) achieved some target values of intermediate (primary)
outcomes: HbA1C <7.0%; LDL <100 mmHg; systolic blood
ATTD 2017 E-POSTER DISCUSSION ABSTRACTS A-43

pressure (SBP) <140 mmHg; a triple target including all the 105
previous outcomes. Secondary outcomes were defined by dif-
A RANDOMIZED TRIAL COMPARING CONTINUOUS
ferent target values of the primary outcomes. A sub-analysis for
SUBCUTAENOUS INSULIN INFUSION VERSUS
diabetes type was also performed.
CONTINUING MULTIPLE DAILY INSULIN
Results: MyStar Connect implementation from 2012 until
INJECTIONS IN PATIENTS WITH TYPE 1 DIABETES
to 2015 was followed by a significant increase, on 42,498 people
USING CONTINUOUS GLUCOSE MONITORING
with diabetes, of those who achieved HbA1c <7.0% (from 39.8
to 45.9%; p < 0.001), LDL <100 mmHg (from 51.8 to 57.4%; E. Toschi1, T. Riddlesworth2, K. Ruedy2, C. Kollman2,
p < 0.001), SBP <140 mmHg (from 53.0 to 57.5%; p < 0.001). D. Price3, R. Beck2
Triple target was achieved by an increasing rate of 4,254 people 1
with diabetes (from 37.2 to 51.6%; p < 0.001). Joslin Diabetes Center, n/a, Boston, USA
2
Conclusions: Though we are not able to exclude the influence Jaeb Center for Health Research, n/a, Tampa, USA
3
of other factors, such as the introduction of new drugs and of Dexcom Inc., n/a, San Diego, USA
other tools of the CCM, MyStar Connect implementation was
followed by a significant improvement of the health of people Background and Aims: To assess the incremental benefits of
with diabetes. changing the insulin delivery method from multiple daily injec-
tion (MDI) to continuous subcutaneous insulin infusion (CSII)
in patients with type 1 diabetes (T1D) already using continuous
glucose monitoring (CGM).
104 Methods: A randomized trial was conducted at 21 endocri-
nology practices in the US. Subjects were in the CGM Group in a
THE UNIVERSITY OF VIRGINIA/PADOVA TYPE 1 preceding 24-week randomized trial comparing CGM versus
DIABETES SIMULATOR GOES SINGLE DAY usual care in adults >25 years old with T1D using MDI. In ad-
R. Visentin1, E. Campos Nanez2, M. Schiavon1, D. Lv2, dition, it was necessary for eligibility for this trial that the subject
M. Breton2, A. Facchinetti1, C. Dalla Man1, B. Kovatchev2, used CGM at least 75% of the time in the last month of the
C. Cobelli1 preceding trial and had a total daily insulin dose <100 units of
1
insulin/day. Eligible subjects were randomly assigned to con-
University of Padova, Department of Information Engineering, tinue MDI or start CSII (OmniPod) plus continuing CGM
Padova, Italy (Dexcom G4 Platinum CGM System with software 505).
2
University of Virginia, Center for Diabetes Technology, The primary outcome was CGM-measured time in range 70 to
Charlottesville, USA 180 mg/dL over 24 weeks.
Results: Mean age of the 75 subjects was 46 + 14 years (range
Background and Aims: A new version of the UVA/Padova 26 to 72 years), 47% were female, and 88% nonHispanic white.
Type 1 Diabetes (T1D) Simulator has been recently developed in Mean T1D duration was 21 + 14 years and total daily insulin dose
order to provide more realistic testing scenarios. Here we present was 59 + 21 units/day. Baseline mean HbA1c was 7.6 + 0.8%
these upgrades to the previous simulator, which was accepted by with 20% <7.0%, 23% 7.0-7.4%, and 57%> 7.5%. Mean time 70
the Food and Drug Administration in 2013. to 180 mg/dL was 12.2 hours/day with median time >180 mg/dL
Methods: Intra-day variability of insulin sensitivity (SI) 10.6 hours/day and time <60 mg/dL 15.1 minutes/day.
has been modeled, based on clinical experiments in T1D Conclusions: The final 24-week outcome visit will be com-
(Hinshaw et al., Diabetes2013), accounting for intra-subject pleted in November 2016. Outcome results will be presented.
variability of daily SI and high inter-subject variability. As
a result, time-varying distributions of both subjects basal in-
sulin infusion and insulin-to-carbohydrate ratio were calculated
and included. A model of dawn phenomenon has been in- 106
cluded, based on clinical findings in T1D subjects (Mallad
et al., DTT2015). Moreover, using a rich clinical T1D dataset BUDGET IMPACT OF CONTINUOUS SUBCUTANEOUS
(Schiavon et al., ATTD2013 and ATTD2016), the model of INSULIN INFUSION COMPARED WITH MULTIPLE
subcutaneous insulin delivery has been updated with com- DAILY INSULIN INJECTIONS FOR THE TREATMENT
mercially available fast-acting insulin analogs. Finally, a model OF TYPE 1 DIABETES IN SPAIN
of intra-dermal insulin delivery (Lv et al., JDST2015) and a M. Gimenez1, I. Elas2, M. Alvarez2, C. Quiros1, I. Conget1
new CGM error model (Facchinetti et al., MBEC2015) have 1
been added. Hospital Clnic i Universitari de Barcelona, Diabetes Unit.
Results: One hundred in silico adults, adolescent, and Endocrinology Department. IDIBAPS, Barcelona, Spain
2
children have been generated accounting for the above de- Medtronic Iberica, Health Economics & Outcomes Research,
scribed modifications. The new simulator captures the intra- Madrid, Spain
day glucose variability observed in real experiments, also well
reproducing the nocturnal glucose increases, and the simu- Background and Aims: Objective: To estimate the budget
lated insulin profiles reflect the kinetics observed in clinical impact (BI) of continuous subcutaneous insulin infusion (CSII)
data. compared with multiple daily insulin injections (MDI) for the
Conclusions: The new modifications introduced in the T1D treatment of patients with type 1 diabetes (T1D) presenting re-
simulator allow to extend its domain of validity from single- current severe hypoglycaemic events in Spain.
meal to single-day scenarios, thus enabling a more realistic Methods: A BI model was developed to estimate direct
framework for in silico testing of continuous glucose monitors, healthcare for T1D patients over a four-year period from the Spanish
closed-loop algorithms, and new insulin formulations. healthcare system perspective. Target population was defined based
A-44 ATTD 2017 E-POSTER DISCUSSION ABSTRACTS

on a retrospective observational study evaluating the efficacy of cerned with systems affecting specific social situations such as
CSII at Hospital Clnic i Universitari de Barcelona (20032008), school and friendships; adolescents were most interested in the
where one of the main indications for switching to CSII was physical features of the system, wearability, and time-savings;
recurrent severe hypoglycaemia episodes; in this study, the mean whereas parents were most concerned about device safety, trade-
number of episodes per year in the last 2 years before starting CSII offs, and the bumpy process of developing trust in a system.
was 1.33, being 0.08 in the last 2 years of follow up (p = 0.003). Conclusions: Incorporating stakeholder perspectives on use
Total cost included therapy costs (insulin and CSII) and major of automated insulin delivery may improve adoption and health
hypoglycaemic events costs. Unitary costs (e,2016) were ob- and quality of life indicators among users. Efforts are needed to
tained from national databases. build trust in systems, optimize device interactions, and provide
Results: The average total cost of treating a patient with T1D clear guidance on use in order to optimize update and sustained
presenting recurrent severe hypoglycaemic episodes over a four- device use.
year period was 13,274e and 23,094e with CSII and MDI, re-
spectively. During this time horizon, therapy costs per patient were
higher with CSII (11,902e vs 2,393e) while major hypoglycaemic
events costs were less with CSII compared with MDI (e1,371 108
versus e20,701). The total BI was e9,821 ( e2,455 per patient THE INFLUENCE OF PSYCHOSOCIAL FACTORS
per year). ON INTEREST IN AND USE OF AUTOMATED INSULIN
Conclusions: The higher therapy costs associated with CSII DELIVERY SYSTEMS: PERSPECTIVES OF ADULTS
compared with MDI were totally offset by the reduction of severe WITH TYPE 1 DIABETES (T1D) AND PARTNERS
hypoglycaemic events and result in cost savings.
K. Barnard1, S. Suttiratana2, D. Naranjo2, K. Hood2, L. Laffel3,
J. Weissberg-Benchell4
1
107 BHR Limited, Research Management, Nr Portsmouth,
United Kingdom
2
INFLUENCE OF PSYCHOSOCIAL FACTORS Stanford University Medical School, Psychology
ON INTEREST IN AND USE OF AUTOMATED INSULIN & Behavioural Science, San Francisco, USA
3
DELIVERY SYSTEMS: PERSPECTIVES OF YOUTH Harvard University, Joslin Diabetes Center, Boston, USA
4
WITH TYPE 1 DIABETES (T1D) AND THEIR PARENTS Ann and Robert H. Lurie Childrens Hospital of Chicago,
K. Barnard1, D. Naranjo2, S. Suttiratana2, L. Laffel3, Northwestern Universitys Feinberg School of Medicine,
J. Weissberg-Benchell4, K. Hood5 Chicago, USA
1
BHR Limited, Research Management, Nr Portsmouth, Background and Aims: To explore the influence of psy-
United Kingdom chosocial factors on interest in and uptake of automated insulin
2
Stanford University, Psychology, San Francisco, USA delivery among adults with T1D and partners.
3
Harvard University, Joslin Diabetes Center, Boston, USA Methods: A qualitative study consisting of focus groups and
4
Ann and Robert H. Lurie Childrens Hospital of Chicago, semi-structured interviews held across four sites in the United
Northwestern Universitys Feinberg School of Medicine, States and United Kingdom with adults with T1D and partners.
Chicago, USA All data were transcribed verbatim and analyzed with content
5
Stanford University, Psychology & Behavioural Science, and thematic analyses methodology.
San Francisco, USA Results: 113 adults (mean age 39.5yrs, range 18-77; mean
A1c 7.5%, range 5.0-11.8%) and 55 partners participated. We
Background and Aims: Aim: To explore the influence of conducted 23 focus groups [n = 16 adult; n = 7 partner] and 48
psychosocial factors on interest in and uptake of automated in- individual interviews [n = 38 adult; n = 10 partner]. Adults and
sulin delivery systems among children/adolescents with T1D and partners were most concerned about the accuracy, adaptability,
their parents. features and algorithm quality of these systems alongside ex-
Methods: A qualitative study consisting of focus groups and pectations that systems would stabilize blood glucose levels and
semi-structured interviews held across four sites in the United reduce long-term complications. Concerns included device
States and United Kingdom with children/adolescents with T1D safety and trust and control for adults while partners were rela-
and their parents. Data were transcribed verbatim and analyzed tively more concerned about the potential disappointment asso-
with content and thematic analyses methodology. ciated with devices given past device experiences. Invasiveness
Results: Participants were 16 children (mean age 10.3 yrs, in life generally and specifically in social situations were raised
range 9-11; mean A1c 6.4%, range 6.4-13.0%), 35 adolescents by both groups.
(mean age 15.1, range 12-20.8; mean A1c 8.7%, range 8.0%- Conclusions: Incorporating stakeholder perspectives on use
13.0%), and 65 parents (mean youth age = 12.8; mean A1c of of automated insulin delivery may improve adoption of devices
youth = 8.1%, range 6.4%-13%). We conducted 23 focus groups and health and quality of life indicators among persons with di-
[n = 8 children; n = 3 adolescent; n = 12 parent] and 41 individual abetes. Efforts are needed to set realistic expectations and provide
interviews [n = 7 children; n = 16 adolescent; n = 18 parent]. Key education on device safety and component features. Continued
themes were burden and benefits of the systems, impact on attention should be paid to psychosocial factors of system use and
quality of life and relationships, and how much the system people affected beyond those diagnosed with T1D along with
controlled diabetes versus the youth. Children were most con- family members.
ATTD 2017 E-Poster Viewing Abstracts

109 Methods: 22 men with diabetes mellitus type 1 (the average

age of 27.1 7.38 years) and 30 healthy men of the same age
CSII AND CGM IN REAL CLINIC PRACTICE (control group) (the average age of 28.0 4.3 years) were in-
IN MOSCOW REGION volved in the study. The main and control groups were formed on
I. Barsukov1, A. Dreval1, T. Shestakova1, M. Taisumova1, the principle of copy pair. Standard clinical examination
O. Medvedev2, M. Panteleev2, A. Sveshnikova2 methods were used: history taking, physical examination and
1 consultation of specialists to identify complications of diabetes.
Moscow Regional Research and Clinical Institute named after Spectrophotometric and fluorometric methods were used.
M.F. Vladimirsky, Endocrinology, Moscow, Russia Results: In men with diabetes mellitus type 1, there is increase
2
Moscow State University, Department of Basic Medicine, in the level of active products reacting with thiobarbituric acid
Moscow, Russia compared to the same indicator in the control group. The level of
antioxidant defense in men with diabetes mellitus type 1 unlike
Background and Aims: CSII and CGM are generally used the control was characterized by decrease in total antioxidant
in Moscow Region but advantages of these methods are not activity. However, stability of processes of superoxide dismutase
obvious yet. activity and the glutathione system in comparison with control is
Aim: To estimate effectiveness of CSII in patient with dia- observed against the backdrop of intensity increase of oxidative
betes mellitus and compare different basal regimes processes.
Methods: 187 patients with diabetes mellitus: 175 (93.6%)- Conclusions: Activation of lipid peroxidation is observed in
type 1, 12 (6.4%) - type 2. All used CSII (Accu-check Spirit men with diabetes mellitus type 1.
Combo, Rosh) and SGM was performed in the first week after
the pump therapy was introduced (iPro2, Medtronic). We
compared two basal regimes: Renners scale with different
basal rate during the day (n = 82) and constant basal rate 111
(n = 105).
Results: Change from MII to CSII lead to the reduce in daily LOW POWER LASER ACCELETRATES WOUND
insulin dose by 30.8% and cut down HbA1c by 1.2% in three HEALING IN DIABETIC RATS
months. Number of hypoglycemia decreased from 2.0 0.8 per M. Shabani1, M. Aminforghani1, M. Babaei1, S. Haghayegh1
day to 0.8 0.5 per day.
1
Patients with different basal rate had more hypoglycemia Iran University of Medical Sciences, Biochemistry, Tehran,
comparing with constant basal rate: AUC less goal was consid- Iran
erably bigger in the group of patients with different basal rate
(0.17 0.27 and 0.07 0.16, p = 0.016). Background and Aims: Introduction: Diabetic wounds
Conclusions: Change from MII to CSII lead to the decrease in have been the area of challenge since many years with different
HbA1c; constant basal rate of insulin is safer because it induces approaches to improve the problem. Nitric Oxide (NO) has been
less hypoglycemia. shown to play a crucial role in wound healing. In addition, ap-
plication of laser on wound healing has already been examined.
Thus, this study was designed to investigate the efficacy of low
110 power laser irradiation for dermal wound healing of diabetic rats.
Materials and Methods: 36 male SD rats were used in this
LIPID PEROXIDATION SYSTEM IN MEN WITH study. Diabetes was induced by IP injection of streptozotocin. A
DIABETES MELLITUS TYPE 1 full-thickness circular wound was made on the back of all rats.
L.I. Kolesnikova1, L.V. Belenkaya1, L.F. Sholokhov1, Rats were selected to be irradiated directly upon their wound
M.A. Darenskaya1, I.M. Mikhalevich1 with a combination of 670 nm (100 mw, 2J/cm2) and 810 nm (50
mw, 1 J/cm2) every other day. Wound imaging was performed on
1
Scientific Centre for Family Health and Human Reproduction days 0, 7, 12, 16, 20 and 22. The wounds margin and context
Problems, Department of Pathophysiology, Irkutsk, Russia were scored pathologically. NO was measured by NO analyzer.
Results: Percent open wound area (POWA) was significantly
Background and Aims: Oxidative stress induced by hyper- lower in the Diabetic Laser group in comparison to the Diabetic
glycemia and subsequent cellular damage is thought to be one of Non-Laser group in all measurement days. Also the POWA de-
the major pathophysiological factors causing complications in crease in DML group was quicker than DMNL group (P = 0.021;
diabetes. The objective of this study was to identify features of mean difference = 19.7%). and P = 0.013, respectively). NO
the lipid peroxidation-antioxidant defense system in men of re- production was increased in DML group as compared to DMNL
productive age with diabetes mellitus type 1. group.

A-45
A-46 ATTD 2017 E-POSTER VIEWING ABSTRACTS

Conclusions: Our study showed that the irradiation of diabetic (GDM) is are considered to be risk factors for the onset of dia-
wounds with a combination of low dose 670 nm and 810 nm betes in later life. The purpose of this study was to identify the
lasers accelerates wound healing process possibly by stimulation risk factors associated with abnormal glucose tolerance on the
of macrophage-derived NO. first postpartum 75-g oral glucose tolerance test (OGTT) among
Japanese women with GDM.
Methods: Forty-four women completed a 75-g OGTT at 12
weeks postpartum to diagnose abnormal glucose tolerance, in-
112 cluding impaired glucose tolerance and type 2 diabetes (T2DM)
CORRELATIONS OF PLASMINOGEN ACTIVATOR in women with GDM. Dietary habits by a self-administered di-
INHIBITOR-1 IN CHILDHOOD OBESITY etary history questionnaire were examined at 28 gestational
weeks and 12 weeks postpartum. All the women diagnosed with
B. Virgolici1, O. Timnea2, H. Virgolici3, L. Popescu4 GDM received nutrition counseling and routine antenatal care.
1
Carol Davila University of Medicine and Pharmacy, Results: None of the participants had insulin treatment during
Biochemistry, Bucharest, Romania pregnancy. Of 44 women with GDM, 35 (79.5%) had normal
2
Ecologic University, Physiology, Bucharest, Romania glucose tolerance, 7 (15.9%) had impaired glucose tolerance, and
3
Carol Davila Univ of Medicine and Pharmacy, Marketing, 2 (4.5%) had overt T2DM. No significant difference was found
Bucharest, Romania between the criteria of GDM at diagnosis and the abnormal
4
Carol Davila Univ of Medicine and Pharmacy, Physiology, glucose tolerance at the postpartum. The significant risk factor
Bucharest, Romania associated with postpartum AGT was pre-pregnancy BMI over
25 kg/m2 (P < 0.05). Total weight gain, age, parity, and nutri-
tional status were not associated with postpartum AGT.
Background and Aims: Plasminogen activator inhibitor-1
Conclusions: Identification of antepartum risk factors in ad-
increases the risk of atherothrombosis and promotes fibrosis and
vance is crucial for preventing impaired glucose tolerance and
has high levels in obesity. Obese children have high risk for
the onset of T2DM. Maintaining an appropriate BMI before
cardiovascular diseases in early childhood.
pregnancy may reduce the risk of future T2DM in reproductive-
The aim of this study was to calculate the correlations of PAI-
aged women.
1 in childhood obesity and to observe the metabolic differences
between obese children with normal or high PAI-1 levels.
Methods: Sixty obese children (10-16 years) and thirty lean
children were enrolled. The obese children were divided in two
groups according to the cutoff value of 43 ng/mL for PAI-1. The 114
group with high values (n = 43) had the PAI-1average 73.7 ng/mL DOES THE ARTIFICIAL PANCREAS REDUCE
and the second obese group (n = 17) had PAI-1average 30.3 ng/mL. DIABETES DISTRESS AND HYPOGLYCEMIA
Spectrophotometric and immunoenzymatic methods were used. WORRY IN INDIVIDUALS WITH TYPE 1 DIABETES?
Results: The obese children had higher PAI-1 level than the
lean children ( p < 0.001). The obese group with high PAI-1 level C. Berget1, K.A. Driscoll1, S. Hanes2, T. Marcal3, L.S. Towers1,
versus the obese group with normal PAI-1 level had lower HDL- D.M. Maahs3, D.P. Howsman4, S. Deshpande5,
c (p < 0.03), plasma magnesium (p < 0.01), plasma A immuno- R. Gondhalekar5, G. Forlenza1, B.W. Bequette4, F. Cameron4,
globulins (p < 0.02), TSH (p < 0.05) and higher levels for white E. Dassau6, S.D. Patek7, T. Ly2, F.J. Doyle III5,
blood cells and lymphocytes (p < 0.01) and higher activity for B. Buckingham2, K.K. Hood2
ALT and AST (p < 0.01). In obese children PAI-1 was negatively 1
University of Colorado Denver, Barbara Davis Center,
correlated (p < 0.05) with magnesium (r = -0.31), with TSH (r = Aurora, CO, USA
-0.33) and positively correlated with lymphocytes (r = 0.40) and 2
Stanford University, Pediatric Endocrinology, Palo Alto, CA,
AST (r = 0.27). USA
Conclusions: In conclusion, in childhood obesity, high PAI-1 3
Stanford University, Pediatric Endocrinology, Palo Alto, CA,
acts in cluster with low magnesium, with dyslipidemia, with USA
inflammation and hepatic cytolysis increasing the risk for hepatic 4
Rensselaer Polytechnic Institute, Chemical and Biological
fibrosis and atherosclerosis. Engineering, Troy, NY, USA
5
Harvard, John A. Paulson School of Engineering and Applied
Sciences, Cambridge, MA, USA
6
113 Harvard, John A. Paulson School of Engineering and Applied
Sciences, Cambridge, MA, USA
7
ABNORMAL GLUCOSE TOLERANCE AFTER University of Virginia, Center for Diabetes Technology,
GESTATIONAL DIABETES MELLITUS Charlottesville, VA, USA
IN JAPANESE WOMEN
H. Watanabe1, S. Nomachi2, Y. Honda3, S. Fukuda3, Y. Sato3 Background and Aims: The purpose of this study was to
examine the impact of artificial pancreas (AP) use on diabetes
1
Osaka University Graduate School of Medicine, Children distress and worry about hypoglycemia in adolescents and adults
and Womens Health, Suita, Japan with type 1 diabetes (T1D).
2
University of Hyogo, Nursing Art & Science, Akashi, Japan Methods: Twenty-five adolescents and adults (mean age =
3
Sato Hospital, Department of Obstetrics and Gynecology, 25.9 10.5 years; 52% male) with T1D (mean duration = 14.5 8.0
Takasaki, Japan yrs) used the Diabetes Assistant (DiAs) AP system with the Zone
Model Predictive Control (Zone-MPC) algorithm for 1-2 weeks
Background and Aims: Various maternal and pregnancy (adolescents 1 week, adults 2 weeks). The Diabetes Distress Scale
characteristics among women with gestational diabetes mellitus and Worry scale of the Hypoglycemia Fear Survey were completed
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-47

at baseline and immediately following study completion with higher 116

scores indicating greater distress and worry. Focus groups were
AN UNANNOUNCED MEAL DETECTION MODULE
conducted at study completion to assess human factors.
FOR ARTIFICIAL PANCREAS CONTROL SYSTEMS
Results: At study entry, scores on both diabetes distress
(mean = 46.1 19; range = 28-102) and hypoglycemia worry A. El Fathi1, S. Mohamed Raef2, A. Haidar3
(mean = 8.0 7.6; range = 0-24) were low. Repeated measures 1
ANOVA revealed no change in distress (mean = 46.3 21.2; McGill University, Department of Electrical and Computer
F = 0.31, p > 0.58) or worry (mean = 9.9 8.8; F = 1.1, p > 0.30) at Engineering, Montreal, Canada
2
study completion. Focus group responses revealed variability in McGill University, Department of Biomedical Engineering,
experience with the system. Participants experienced less worry Montreal, Canada
3
about nocturnal hypoglycemia and lows in general and flexibility McGill University, Department of Biomedical Engineering,
when eating. Connectivity issues and charging batteries daily Montreal, Canada
were considered burdensome.
Conclusions: Using the DiAs AP system with Zone-MPC Background and Aims: Most artificial pancreas systems rely
algorithm did not add distress and hypoglycemia worries for on patient announcement of upcoming meals and deliver accom-
study participants. Future research is needed to better understand panying insulin boluses in order to achieve better postprandial
both the overall psychosocial impact of AP use as well as the glucose levels. However, the omission of prandial insulin boluses
specific impact for highly distressed and worried individuals is common in patients with type 1 diabetes, especially adolescents.
with T1D. We developed a novel module for detection of unannounced meals
during the operation of our artificial pancreas.
Methods: From real-time glucose measurements and a meal
digestion model, the novel module constructs, i) a probability
density function characterizing the amount of carbohydrate
115 content of a hypothetical missed meal and ii) the time of the
MONITORING AND SUPERVISION OF ARTIFICIAL hypothetical missed meal. The module will flag a meal if the
PANCREAS OPERATION WITH MULTIVARIABLE variance of the density function is smaller than a threshold.
STATISTICAL ALGORITHMS Results: The detection module is evaluated in a four-week
virtual experiment using our metabolic simulator that runs our
A. Cinar1, K. Turksoy2, I. Hajizadeh1 artificial pancreas control algorithm. We tested the module on 12
1
Illinois Institute of Technology, Chemical and Biological virtual type 1 diabetes patients who fail to announce the meal
Engineering, Chicago, USA to the artificial pancreas three times per week. Out of the 144
2
Illinois Institute of Technology, Biomedical Engineering, non-announced meals, 138 meals are detected (Sensitivity of
Chicago, USA 95.83%). The detection time is 40 (30-50) min, and the precision
of the estimated mealtime is 10 (0-20) min. Our algorithm reports
Background and Aims: Detection and mitigation of the ef- a false positive detection once every 56 days. These findings
fects of faults in an artificial pancreas (AP) is critical for its safe were also confirmed by running our module on clinical data of a
operation. Faults in glucose sensors, insulin pump and control 72-hour artificial pancreas outpatient study.
system can yield insulin infusion rates that can cause hypogly- Conclusions: Our meal detection algorithm performed well
cemia or hyperglycemia. The multivariable AP system devel- on simulation data, and validated with clinical data. Control
oped at IIT enables the use of additional variables in monitoring actions based on this meal detection module has to be developed.
the AP operation, detection abnormalities in its operation, and
diagnosis of faults. The algorithms developed focus on real-time
detection of unexpected changes in signals that may be caused by 117
faults or disturbances.
Methods: This novel system monitoring approach uses dy- THE ARTIFICIAL PANCREAS (AP) IMPROVES
namic time warping for synchronization of several signal tra- GLYCEMIC CONTROL DURING EXTENDED
jectories, Savitzky-Golay filter and multiway principal component EXERCISE AT SKI CAMP IN ADOLESCENTS
analysis. WITH TYPE 1 DIABETES (T1D)
Results: Data from 14 subjects that participated in 60-hour G. Forlenza1, D. Chernavvsky2, M. DeBoer2, J. Robic2,
closed-loop AP experiments with variations in meal amounts B. Kovatchev2, R.P. Wadwa1, L. Messer1, D. Maahs1,
and physical activity levels and times are used. Glucose mea- M. Breton2
surements from a CGM sensor and biometric variables such as
1
heat flux, skin temperature, galvanic skin response, energy University of Colorado Denver, Barbara Davis Center,
expenditure and acceleration from a physical activity armband Aurora, USA
2
are monitored for fault detection. The unexpected changes in University of Virginia, Center for Diabetes Technology,
signals are detected successfully. The results illustrate that the Charlottesville, USA
proposed method is able to detect various types of faults with
high accuracy. No faults were missed in the fault scenarios Background and Aims: Intense exercise often leads to ab-
tested. normal glucose control in patients with T1D. Adjusting man-
Conclusions: The algorithm detects faults or unexpected agement for exercise remains one of the major burdens of T1D
dynamic changes and labels them according to predefined rules, therapy. AP systems can improve glycemic control during and
and it can be used to warn about sensor and AP faults. This will after exercise, but exercise studies have been limited. Skiing
be a new module in our multivariable AP that can be integrated camps present a unique mix of prolonged physical activities of
with our controller performance assessment and data recon- varied intensities, with metabolic effects compounded by cold,
ciliation modules. altitude, and stress/fear/excitement.
A-48 ATTD 2017 E-POSTER VIEWING ABSTRACTS

Background and Aims: Zone-MPC permits patients to remain

fully in closed-loop for prolonged, multi-month, periods of time.
As long-term, fully autonomous, sparsely supervised AP operation
becomes routine, consideration of automatic fault detection and
mitigation becomes vital. We conducted a randomized-crossover
study of home use of the Zone-MPC system versus sensor aug-
Methods: In a randomized controlled clinical trial, 32 ado- mented pump (SAP) in which insulin infusion set (IIS) and con-
lescents with T1D (ages 10-16; 17 males) participated in a 5-day tinuous glucose monitor (CGM) sensor failures were provoked via
ski camp (*5h skiing/day), with N = 16 fitted with the UVA AP prolonged wear.
system (vs. sensor augmented pump), at two sites: Wintergreen Methods: Monitored home use of an Android phone-based
VA and Breckenridge CO. Control and treatment groups were AP system was conducted in 19 adults (58% female, median 23
matched by age and HbA1c; ski/snowboard proficiency was years-old, 11 years of type 1 diabetes, 8% A1c) in a randomized
balanced overall but with a strong site effect. All subjects were crossover study. Participants wore their IIS for up to 7 days and
physician monitored 24h/day using the UVA Diabetes Web their CGM for up to 14 days to increase the probability of sensor
Monitoring (DWM) system, allowing for strict enforcement of and infusion set failures.
safety protocols on and off the slopes. Results: Use of the Zone-MPC system (SAP vs closed loop)
Results: Percent time between 70-180mg/dL was signifi- improved percent time 70-140 mg/dL (41 vs 48%; p = 0.02), time
cantly better in the AP group (67% vs 73%; p = 0.03) overall 70-180 mg/dL (66 vs 73%; p = 0.004), and decreased median
with maximum effect in the second half of the night. Daytime CGM glucose (150 vs 141 mg/dL; p = 0.02) and glycemic vari-
% <70 mg/dL was also significantly decreased (3.5% vs. 1.8%; ability (SD 56 vs 51 mg/dL; p = 0.03) while decreasing percent
p = 0.006). Effects were site dependent with VA showing re- time <70 mg/dL (3.1 vs 1.7%; p = <0.001) and time <50 mg/dL
duced hypoglycemia and similar percent in range and CO (0.3 vs 0.1%; p = 0.01) [Table 1]. CL also greatly improved
showing improved time in range without hypoglycemia reduc- overnight control as measured by mean CGM glucose MN-6AM
tion (Figure 1). Post-study qualitative feedback was highly (158 vs 141 mg/dL; p = 0.01).
positive regarding the AP system and study experience. Conclusions: Zone-MPC significantly improved glycemic
Conclusions: During the first winter-sport AP camp, we re- control in a home-use environment despite prolonged CGM and
port significantly improved glycemic control without any device- IIS wear. This project represents the first home use AP study
related adverse events and with overwhelmingly positive patient attempting to provoke and detect component failure while suc-
feedback. cessfully maintaining safety and glucose control.

118 119
APPLICATION OF ZONE MODEL PREDICTIVE FULLY CLOSED LOOP MULTIPLE MODEL
CONTROL (ZONE-MPC) ARTIFICIAL PANCREAS (AP) PROBABILISTIC PREDICTIVE CONTROLLER
DURING EXTREME FAULT DETECTION TESTING (MMPPC) ARTIFICIAL PANCREAS (AP)
CONDITIONS: A RANDOMIZED CROSSOVER- PERFORMANCE IN ADOLESCENTS AND ADULTS
CONTROLLED HOME-USE TRIAL IN A SUPERVISED HOTEL SETTING
G. Forlenza1, S. Deshpande2, T. Ly3, D. Howsmon4, G. Forlenza1, F. Cameron2, T. Ly3, D. Lam4, D. Howsmon2,
F. Cameron4, N. Baysal4, L. Schulhof-Towers1, T. Marcal3, N. Baysal2, L. Messer1, P. Clinton3, C. Levister4, S. Patek5,
B.W. Bequette4, R. Gondhalekar2, F.J. Doyle III2, D. Maahs1, C. Levy4, R.P. Wadwa1, D. Maahs1, B.W. Bequette2,
B. Buckingham3, E. Dassau2 B. Buckingham3
1 1
University of Colorado Denver, Barbara Davis Center, University of Colorado Denver, Barbara Davis Center,
Aurora, USA Aurora, USA
2 2
Harvard University, John A. Paulson School of Engineering Rensselaer Polytechnic Institute, Chemical and Biological
and Applied Sciences, Cambridge, USA Engineering, Troy, USA
3 3
Stanford University, Pediatric Endocrinology, Palo Alto, USA Stanford University, Pediatric Endocrinology, Palo Alto, USA
4 4
Rensselaer Polytechnic Institute, Chemical and Biological Mount Sinai Ichan School of Medicine, Department of
Engineering, Troy, USA Endocrinology- Diabetes and Metabolism, New York, USA
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-49

5
University of Virginia, Center for Diabetes Technology, 120
Charlottesville, USA
DIABELOOP CLOSED-LOOP DOES BETTER THAN
SENSOR-AUGMENTED PUMP ON BLOOD GLUCOSE
Background and Aims: The goal of AP systems is to reduce
CONTROL DURING 3 DAYS WITH EITHER
complications and burdens of T1D. Early systems will be hybrid
INTENSIVE PHYSICAL EXERCISES, GASTRONOMIC
closed-loop, requiring prandial meal announcements. Such sys-
DINNERS OR REST IN T1D
tems do not eliminate the burden of pre-meal insulin dosing and
will provide suboptimal benefits if patients forget to bolus. Y. Reznik1, N. Jeandidier2, P.Y. Benhamou3, S. Borot4,
MMPPC is a fully closed-loop system which uses probabilistic M. Doron5, S. Franc6,7, H. Hanaire8, E. Huneker9,
estimation of meals to allow for automated meal detection. Here E. Renard10, P. Scaepelynck11, I. Xhaard7, A. Penfornis6,
we describe the performance of the MMPPC system with adap- G. Charpentier6,7, B. Guerci12
tive hypoglycemia minimization in a supervised hotel setting. 1
Methods: The Android phone-based AP system with remote CHU de CAEN, Endocrinologie, Diabetologie, CAEN,
monitoring was tested for 72 hours in 6 adults and 4 adolescents France
2
(30% female, 23 years-old, 10.6 years of T1D, 8.1% enrollment CHU STRASBOURG, Endocrinologie, Diabete et des maladies
A1c) across 3 clinical sites with daily exercise and meal chal- de la Nutrition, STRASBOURG, France
3
lenges involving both announced and unannounced meals. CHU GRENOBLE, Endocrinologie, GRENOBLE, France
4
Controller aggressiveness was adapted daily based on prior hy- CHRU BESANCON, Service dEndocrinologie, Metabolisme
poglycemic events. et Diabetologie, Nutrition, BESANCON, France
5
Results: Mean 24-hour CGM glucose was 157 mg/dL, with CEA_Leti, Technology Research Institute, Micro Technologies
63.3% of readings between 70-180 mg/dL, 2.9% of readings Department for Biology and Health, GRENOBLE, France
6
<70 mg/dL, and 9.0%> 250 mg/dL [Table 1]. Moderate hyper- CHSF, Service Diabetologie, CORBEIL ESSONNES, France
7
glycemia was relatively common with 24.5% of readings be- CERITD, Research, EVRY, France
8
tween 180-250 mg/dL, primarily within 3 hours of a meal. CHU TOULOUSE, Service de Diabetologie, Maladies
Overnight mean CGM glucose was 140 mg/dL, with 75.0% be- metaboliques et Nutrition, TOULOUSE, France
9
tween 70-180 mg/dL, 4.0% <70 mg/dL, 16.8% between 180- DIABELOOP SAS, Industrial Development, GRENOBLE,
250 mg/dL, and 3.9%> 250 mg/dL. Breakfast glycemic peak was France
10
higher for unannounced meals with values normalizing by 2- CHU MONTPELLIER, Service des maladies endocriniennes,
3 hours post-meal [Figure 1]. MONTPELLIER, France
11
Conclusions: Adaptive MMPPC was effective in a supervised HOPITAL DE LA CONCEPTION, Service de Nutrition,
setting despite meal and exercise challenges. Further studies are Endocrinologie, Maladies Metaboliques, MARSEILLE, France
12
needed in a less supervised environment with additional miti- CHU NANCY, Service de Diabetologie, NANCY, France
gations to prevent exercise induced hypoglycemia.
Background and Aims: Variability of way of life, is an issue
for good BG control in type 1 diabetes.(T1D) Diabeloop is a
Closed-loop (CL) system with a MPC algorithm reinforced by
a decisional matrix, uploaded on an android smartphone linked
to Dexcom CGM and a Cellnovo insulin patch-pump. The
primary aim of the study was to compare BG control with
Diabeloop CL versus sensor-augmented insulin pump therapy
(SAP), in well-educated patients practicing intensive physical
exercises repeatedly, or gastronomic dinners or rest, during
3 days.
Methods: Thirty-eight T1D patients on insulin pump ther-
apy (49.9 14.5 years old, diabetes duration:25.1 13.6
years, HbA1c:7.8 0.7%) participated in randomized nine-
center crossover study, including two 72-hour periods under
conditions of various exercises for the 1st group, gastronomic
dinners for the 2nd group, and rest for the 3rd one, either with SAP
or CL. Meals and exercises were announced in CL period.
Results: Percent time in 80-140mg/dL range during the
3 nights, was significantly higher in CL compared with
SAP (61.1[56.1;66.7] vs. 27.4 [19.2;39.0], P < 0.0001) as was
%time in 70-180mg/dL target range during the whole 72hrs:
79.0[75.6;82.5] vs. 60.8[55.2;67.0] P < 0.0001). Three-day
mean BG(mg/dL) was significantly lower in CL than in SAP
(138.7[133.5;144.2] vs. 156.5[148.5;165.0], P<0.0001), as was
the %time spent in hypoglycemia. Satisfaction of treatment
(DTSQ) was higher with CL (31.0 5.5) compared to SAP
(26.0 5.5), P = 0.0003.
Conclusions: In various situations such as intensive exercises,
gastronomic dinners or rest, Diabeloop was able to double the
%time spent in 80-140mg/dl BG range during the night, to re-
duce the time in hypoglycemia and to decrease average BG level,
compared to SAP.
A-50 ATTD 2017 E-POSTER VIEWING ABSTRACTS

121 %error = (calcBolus - idealBolus)*100/idealBolus, where calc-

Bolus is from the calculator and idealBolus is the dose assuming
HYSTERESIS-LIKE CURVE OF THE GLUCOSE/
perfect ICR, CHO and CF. The ICR and CF inputs were set at
PROTEIN BINDING AND DISSOCIATION: WHY VERY
12g/U and 50mg/dL/U respectively. Meal size estimates (CHO)
STRICT BS CONTROL FAILS TO REDUCE LONG-
ranged from 10 to 100g, and (BGcurrent - BGtarget) ranged
TERM SIDE EFFECTS OF DIABETES
from 50 to 300mg/dL.
K. Khodabakhshi Pirkalani1, Z. Talaeerad1 Results: With inputs ICR, CHO and CF set at +/-10%, 20%,
1 and 30%, the calcBolus %error ranges were [-17%,20%] (SD
Mehr Medical Group, Internal Medicine/Gynecology, 0.17), [-32%,47%] (SD 0.13), and [-44%,80%] (SD 0.19) re-
Tehran, Iran spectively. Maximum %errors occurred when CHO was the
maximum (100g) and BGcurrent - BGtarget was the smallest
Background and Aims: Mathematical modelling explaining (50mg/dL).
why very strict BS control cannot reduce long-term side effects A sensitivity analysis showed that for the 20% uncer-
in large cohort studies. tainty case, when CHO, ICR and CF were individually set to zero
Methods: Normal range of blood glucose has traditionally the calcBolus %error went down to [-20%,20%](SD 0.10),
been chosen arbitrarily or at most statistically. Both patients and [-20%,24%](SD 0.10) and [-29%,44%](SD 0.11) respectively;
physicians are happy if FBS is under 90-110 and PP BS under making calcBolus more sensitive to CHO and ICR.
180. But if we regard 70 as a the healthy BS value based on Conclusions: More research is needed to help PWD make
childhood samples, possible BS of the prehistoric mankind and better estimates of CHO, ICR and CF.
related animals with a different meal composition140mg/dl will
face us with twice as fast reversible glycosylation and 100%
more irreversible glycosylation. In this way the patient must cope
with three problems:
123
1. Degrade the surplus malfunctioning protein
2. Replace it with functioning proteins ON THE SENSITIVITY OF CONTINUOUS GLUCOSE
3. Renew the regular cooperation of different proteins in a MONITORING TO INSULIN INFUSION FAULTS
pathway that are basically glycosylated by their structure K. Kolle1,2,3, A.L. Fougner1,2,3, . Stavdahl1,3
and not function. As in a factory aberrant function of a 1
worker affects the whole pathway. Norwegian University of Science and Technology NTNU,
Department of Engineering Cybernetics, Trondheim, Norway
2
Results: Based on complex mathematical calculations, a very Central Norway Regional Health Authority, RHA, Trondheim,
strict blood glucose control with conventional treatments or even Norway
3
by closed loop systems must target a fasting BS at any time Norwegian University of Science and Technology NTNU,
around 70 and an HbA1C lower than 4.5-5. Though this cannot Artificial Pancreas Trondheim APT, Trondheim, Norway
be tolerated early in the treatment, strategy techniques are sug-
gested on how to achieve this goal in middle to long term. Background and Aims: The artificial pancreas is subject to
Control in the range of 5.5-7 is not much different from higher faults that the automated system must deal with while keeping
ranges in regard to long term side effects. the user safe. Fault detection is important in order to distinguish
Conclusions: New definition of BS target can solve the recent between disturbances, physiological variations and component
controversies. Novel methods to increase compliance for lower faults. Whereas the first two are handled by the nominal con-
BS values are presented. troller, the latter requires dedicated fault mitigating actions,
possibly even a system shut-down. Current body-worn systems
provide continuous glucose monitoring (CGM) data from the
subcutaneous tissue as the only standard measurement. Any
122 change of blood glucose concentration appears delayed in these
sensor readings.
UNCERTAINTY ANALYSIS OF INSULIN DOSES Methods: The dynamic model by Dalla Man et al. (2007)
FROM BOLUS CALCULATORS (DOI:10.1177/193229680700100303) is the basis for a for-
R. Kircher1, R. Mauseth2, D. Matheson3 ward sensitivity analysis using the Sundials CVODES package
in MATLAB. An insulin infusion fault, a meal and decreased
1
Engineering & Regulatory Affairs, Seattle, USA insulin sensitivity are simulated. All of these changes eventu-
2
Seattle Childrens Hospital, Seattle, USA ally lead to increased blood glucose concentrations. The sen-
3
Dose Safety, Inc., Seattle, USA sitivity of CGM to these glucose changes is investigated for
different sensing dynamics, using the intravascular glucose
Background and Aims: Accurate calculation of insulin bo- concentration as reference.
luses is a challenge for people with diabetes (PWD), even with a Results: The sensitivity of glucose measurements to insu-
hybrid closed loop artificial pancreas pump. The objective is to lin infusion faults is low, and notably lower than the sensi-
quantify errors of insulin bolus calculators as a function of pa- tivity to normal variations of insulin sensitivity. A meal has
tient uncertainties in their insulin to carb ratio (ICR), correction the strongest impact and possibly even masks the onset of
factor (CF) and estimating carbs (CHO). The formula is: calc- glucose excursions caused by infusion faults before they be-
Bolus = CHO/ICR + (BGcurrent - BGtarget)/CF where ICR, come apparent.
CHO and CF are patient specified inputs. Conclusions: An early distinction between physiologi-
Methods: An uncertainty analysis was performed on the cal variations and an insulin infusion fault is hardly possi-
calculator bolus, given selected % uncertainties of input pa- ble based exclusively on deviations in glucose concentration
rameters CHO, ICR and CF. Doses were characterized as +/- without more advanced pattern recognition that reveals subtle
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-51

differences. Slow glucose dynamics from plasma to the current models accuracy seldom allow to predict beyond 60
sensing site potentially obscures fault detection and diagnosis minutes. Increasing PH values may play a significant role to
based only on CGM. improve AP performance. The extension of the concept of
seasonality in time-series models for glucose prediction is
evaluated here in a proof-of-concept study to improve pre-
diction accuracy.
124 Methods: The main idea is that seasonality may emerge
OVERNIGHT GLUCOSE CONTROL WITH DUAL- AND after the pre-processing of CGM data collecting periods with
SINGLE-HORMONE ARTIFICIAL PANCREAS IN TYPE similar glycemic patterns (e.g. post-prandial periods for
1 DIABETES WITH HYPOGLYCEMIA UNAWARENESS similar meals). Then a familiy of seasonal stochastic models
VERSUS HYPOGLYCEMIA AWARENESS: can be built. Data covering 7 post-prandial periods for a same
RANDOMIZED CONTROLLED TRIAL 60g CHO meal was used, both in open-loop (OL) and closed-
loop (CL) scenarios. SARIMA and SARIMAX models were
R. Rabasa-Lhoret1, A. Abitbol2, V. Messier1, L. Legault3, identified and evaluated following a one-leave-out procedure.
M. Smaoui4, M. Ladouceur5, A. Haidar4 To evaluate goodness-of-fit, residual analysis, Akaike infor-
1
Institut de recherches cliniques de Montreal, Metabolic mation criterion (AIC) and mean squared error ( MSE) were
diseases, Montreal, Canada compared. Forecast accuracy was evaluated via root mean
2
LMC Diabetes & Endocrinology, Diabetes, Toronto, Canada squared error (RMSE) and mean absolute percentage error
3
Montreal Childrens Hospital, Endocrinology, Montreal, (MAPE).
Canada Results: RMSE in OL study was larger than in CL study due
4
McGill University, Biomedical engineering, Montreal, Canada to the larger fluctuations in OL when SARIMA models were used
5
Centre de recherche du Centre Hospitalier de lUniversite de (35.04 vs. 25.47). SARIMAX model further improved perfor-
Montreal, Biostatistics, Montreal, Canada mance in CL, by using insulin infusion and amount of meals as
exogenous variables. In this case, PH could be extended to up to
180 min for a MAPE below 10%.
Background and Aims: The dual-hormone (insulin and
Conclusions: Seasonality improved model accuracy allowing
glucagon) artificial pancreas may be justifiable in some but not
the extension of the PH.
all patients, among them are hypoglycemia unaware patients. We
conducted a randomized crossover trial comparing dual- and
single-hormone artificial pancreas over one night in 18 adult
patients with hypoglycemia unawareness and 17 patients with
hypoglycemia awareness. 126
Methods: All patients had documented nocturnal hypogly- MID-TERM BLOOD GLUCOSE PREDICTION:
cemia during two weeks of screening. Analysis was performed A HYBRID APPROACH USING GRAMMATICAL
using plasma glucose. EVOLUTION AND PHYSIOLOGICAL MODELS
Results: In patients with hypoglycemia unawareness, the
time spent below 4 mmol/L was 0% [010] on single-hormone S. Oviedo1, J. Veh1, I. Contreras1, R. Visentin2, M. Vettoretti2
nights and 0% [00] on dual-hormone nights (P = 0.28). In 1
Universitat De Girona, Institute of Informatics and
patients with hypoglycemia awareness, the time spent below Applications, Girona, Spain
4 mmol/L was 0% [00] on single-hormone nights and 0% [0 2
Padova University, Department of Information Engineering,
0] on dual-hormone nights (P = 0.79). In hypoglycemia un- Padova, Italy
aware patients, there were 3 hypoglycemic events (< 3.3 mmol/
L with symptoms of hypoglycemia or <3.0 mmol/L without Background and Aims: In this work a proposal of a hybrid
symptoms of hypoglycemia) on single-hormone nights and 2 model using Grammatical Evolution (GE) and physiological
events on dual-hormone nights. In hypoglycemia aware pa- model for personalized mid-term (120 minutes) blood glucose
tients, there was 1 event on single-hormone nights and none on predictions is introduced. The inclusion of physiological models
dual-hormone nights. in the overall GE model is to our knowledge, the first approach of
Conclusions: We conclude that the single-hormone artificial this kind.
pancreas might be sufficient for hypoglycemia-free overnight Methods: Insulin-on-Board and Glucose Rate of Appearance
control in patients with hypoglycemia unawareness. Day and models were included into a hybrid predictive model that uses
night studies in this population are needed. symbolic regression through Grammatical Evolution (GE). Four
personalized models per patient were developed, corresponding
125 to 6-hour periods.
The personalized models were evolved for 10 days of con-
STOCHASTIC SEASONAL MODELS FOR GLUCOSE tinuous data generated from 100 in-silico patients simulated
PREDICTION IN TYPE 1 DIABETES
E. Montaser1, J. Luis Diez1, J. Bondia1
1
Universitat Politecnica de Valencia, Systems Engineering and
Automation, Valencia, Spain

Background and Aims: An important feature of any arti-

ficial pancreas (AP) is its ability to predict glucose along a
given prediction horizon (PH), either as part of the control
algorithm or the patient supervision subsystem. However,
A-52 ATTD 2017 E-POSTER VIEWING ABSTRACTS

AS (GT3X+, ActiGraph) were acquired, while the subjects per-

formed a set of 24 5-min activities of various intensity (rest to
using the UVA-Padova simulator to predict BG 120 minutes
vigorous) under lab-conditions. The proposed model is compared
ahead. A glucose-specific fitness function (gRMSE) that penal-
with 1 linear model trained on the same data and 1 classical linear
izes the deviations according to their clinical harmfulness was
model for converting AS (counts per minutes from Actilife) into
used. Four additional days were used for validation.
EE. A leave-one-subject-out approach and the classical perfor-
Results: For the validation days, gRMSE and RMSE were
mance indicators were chosen for comparison purposes.
10.89 mg/dL and 11.8 mg/dL respectively during the nocturnal
Results: Table 1 shows performance obtained by each model.
period while average gRMSE and RMSE during the day were
Conclusions: Results displayed in Table 1 demonstrate that
26.93 mg/dL and 24 mg/dL respectively. The percentage in Clarkes
the proposed model yields better performances than the two
zones A and B was 99.37% at night and 97.95% during the day.
linear models. This model allowed EE computation for activities
Tables 1 and 2 show detailed results for the 4 time periods.
where linear models failed (cycling activity). Performance of the
Conclusions: Grammatical Evolution enhanced with physi-
AP developed by our research group may be further improved by
ological models is a feasible strategy for mid-term Blood Glu-
including other non-linear structural equations to obtain more
cose prediction.
accurate EE estimations.

127
A NONLINEAR MODEL FOR CONVERTING 128
ACCELEROMETERS SIGNALS INTO ENERGY
EXPENDITURE: TOWARDS THE ARTIFICIAL FACTORS THAT INFLUENCE THE PERFORMANCE
PANCREAS WHICH WILL CONSIDER PHYSICAL OF A HYBRID CLOSED-LOOP SYSTEM
ACTIVITY
A. Roy1, B. Grosman1, N. Parikh1, D. Wu1, N. Kurtz1,
1 2
H.M. Romero Ugalde , M. Garnotel , I. Xhaard , 3
R. Slover2, S. Garg3, T. Bailey4, R. Bergenstal5, B. Bode6,
G. Charpentier4, S. Franc4, E. Huneker5, C. Simon2, S. Bonnet1, R. Brazg7, J. Ilany8, B. Buckingham9, S. Anderson10,
on behalf of the Diabeloop consortium S. Weinzimer11, F. Kaufman12
1 1
Univ. Grenoble Alpes- F-38000 Grenoble- France. CEA LETI Medtronic Diabetes, R&D, Northridge, USA
2
MINATEC Campus- F-38054 Grenoble, France Barbara Davis Center for Childhood Diabetes, Pediatric
2
CARMEN INSERM U1060/Universite de Lyon 1/INRA U1235 Diabetes, Aurora, USA
3
Lyon- France. CRNH-Rhone-Alpes- Lyon, France Barbara Davis Center for Childhood Diabetes, Adult Diabetes,
3
CERITD - BIOPARC GENOPOLE Campus 3 batiment 5 - F- Aurora, USA
4
91058 Evry, France AMCR Institute, Advanced Metabolic Care Research,
4
CERITD - BIOPARC GENOPOLE Campus 3 batiment 5 - F- Escondido, USA
5
91058 Evry, France; and Centre Hospitalier Sud-Francilien Park Nicollet Clinic, International Diabetes Center,
CHSF-116- Bd Jean Jaures - 91106 Corbeil-Essonnes Cedex, Minneapolis, USA
6
France Atlanta Diabetes Associates, Endocrinology & Metabolism,
5
DIABELOOP SAS F-38000 Grenoble, France Atlanta, USA
7
Rainier Clinical Research Center, Metabolism and
Background and Aims: Physical activity (PA) has strong Endocrinology, Renton, USA
8
effects on glucose dynamics. Therefore, current research is Institute of Endocrinology, Sheba Medical Center,
conducted for including PA information on artificial pancreas Endocrinology, Tel Hashomer, USA
9
(AP) to improve glucose regulation performance. Among a va- Stanford University, Pediatric Endocrinology,
riety of PA indicators, energy expenditure (EE), computed from Stanford, USA
10
accelerometer signals (AS), is the preferred physiological vari- UVA Center for Diabetes Technology, Endocrinology and
able in AP applications. The most popular models for converting Metabolism, Charlottesville, USA
11
AS into EE are linear models. The aim of this work is to dem- Yale University School of Medicine, Pediatric Endocrinology,
onstrate that a nonlinear model improves EE estimation and may New Haven, USA
12
thus improve AP performance. Medtronic Diabetes, Clinical Affairs, Northridge, USA
Methods: A nonlinear model (sigmoid neural network) is
proposed for converting AS into EE. This model is obtained from Background and Aims: A multi-center pivotal trial of the
a database composed of 53 1955-year-old subjects in which MiniMed 670G hybrid closed-loop (HCL) system revealed a
indirect calorimetry (SERVOPRO 4100, Servomex, UK) and strong correlation between initial A1C and A1C delta, with a larger
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-53

Background and Aims: We apply formal specification and

verification techniques for checking correctness properties of
artificial pancreas control algorithms. Formal specification and
verification techniques systematically and exhaustively explore
the behaviors of closed loop in silico models to discover the
worst case effects of disturbances such as meals, external bo-
luses, sensor noise and set failures on the predicted blood glucose
levels of a virtual patient.
Methods: Our study examines two artificial pancreas con-
trollers: (a) PID-based hybrid closed loop controller (Steil et al.,
2011) and (b) Kalman filter-based predictive pump shutoff sys-
tem (Cameron et al.2012) using the Dalla-Man model for the
drop for higher initial A1C levels. Pivotal trial data was analyzed to virtual patient (Dalla-Man et al., 2008).
investigate several potential factors on the outcome of the trial. We formulate 10 different properties for the closed loop sys-
Methods: A total of 51 subjects (of 124 subjects in the tems. Examples include (a) Are insulin infusions at/above the
trial) had initial A1C between 6.57.5% and were divided into 3 basal rates possible when the patient is under hypoglycemia? (b)
groups: those who didnt experience any change (No-change), Is the pump ever shut down, when the patient is under hyper-
those who had a rise (Rise), and those who had a drop (Drop) in glycemia?
A1C levels. The reason to select such a tight A1C-range was to The tool S-Taliro (Fainekos et al., 2011) is used to systemat-
minimize the influence of initial-A1C on the analysis. ically search over a large range of disturbance inputs to discover
Results: The table lists data from the three groups. There were potential violations. The figure below shows a situation where
no significant differences between the Rise and Drop groups the Kalman-filter based system (Cameron et al., 2012) allows
for the initial A1C, average total daily dose of insulin, and av- insulin delivery to resume while the virtual patient is still under
erage daily carbohydrate input. However, study phase data hypoglycemia.
showed the percentage of daily insulin from meal-bolus was Results: Our analysis discovered potential issues and miti-
significantly higher for the Drop group (p = 0.0082), with a gations for each algorithm: specifically, we analyze the sensi-
lower carb-insulin ratio (p = 0.08). Furthermore, the daytime tivity of the PID controller to choices of the gain parameters, and
mean sensor glucose values for the Drop group were signifi- the sensitivity of the Kalman filter algorithm to sensor noise.
cantly lower than those of the Rise group, likely due to more Conclusions: In conclusion, we demonstrate how formal veri-
appropriate mealtime boluses. There was no difference in the fication techniques can be useful in identifying possible defects and
overnight mean glucose levels, indicating equivalent effective- their mitigations for artificial pancreas control algorithms.
ness of the system in the absence of meal disturbances.
Conclusions: The results of the HCL pivotal trial demonstrate
the positive effect of automating insulin delivery, particularly at 130
night, and the need to continue to effectively manage postpran-
dial glucose levels. GLYCAEMIC MANAGEMENT IN PATIENTS WITH
TYPE 2 DIABETES IN GERIATRIC CARE FACILITIES
A. Libiseller1, K.M. Lichtenegger1, A. de Campo2,
129 T. Wiesinger2, N. Stolletz2, G. Cuder1, B. Holl3,
P. Beck3, T.R. Pieber1
MODEL-BASED IN-SILICO VERIFICATION 1
OF ARTIFICIAL PANCREAS CONTROL Medical University of Graz, Endocrinology and Diabetology,
ALGORITHMS Graz, Austria
2
Geriatric Health Centers of the city of Graz, Acute Geriatric
S. Sankaranarayanan1, F. Cameron2, B.W. Bequette2, Department, Graz, Austria
D. Maahs3 3
Joanneum Research Forschungsgesellschaft mbH, HEALTH
1
University of Colorado- Boulder, Computer Science, Boulder, Institute for Biomedicine and Health Sciences, Graz, Austria
USA
2
Rensselaer Polytechnic Institute, Chemical and Biological Background and Aims: Diabetes guidelines recommend in-
Engineering, Troy, USA dividualized treatment strategies in older patients with diabetes
3
Stanford University, Pediatric Endocrinology, Stanford, USA with the main focus on avoiding hypoglycaemia. The aim of this
A-54 ATTD 2017 E-POSTER VIEWING ABSTRACTS

study was to assess glycaemic management in patients with type 2 132

diabetes with and without insulin therapy in geriatric care facilities.
SEVERE CRITICAL-ILLNESS INSULIN RESISTANCE
Methods: 170 medical records of geriatric patients with type 2
TREATED WITH U-500 INSULIN INFUSION:
diabetes were analysed retrospectively. Based on main diabetes
A CASE SERIES
therapy 46 patients were allocated to the insulin group (52.2%
female, 78 8 years, HbA1c 60 22 mmol/mol, BMI 28.2 5.4 kg/ L. Igneri1, A. Schorr
m2) and 124 patients to the no-insulin group (69.4% female, 80 9 1
years, HbA1c 48 12 mmol/mol, BMI 27.8 6 kg/m2). St. Mary Medical Center, Pharmacy, Langhorne, USA
2
Results: The overall blood glucose (BG) was 201 47 mg/dl St. Mary Medical Center, Endocrinology, Langhorne, USA
(insulin) compared to 160 38 mg/dl (no-insulin). 39.8% versus
59.3% of BGs were between 100-180 mg/dl, 3.9% versus 4.6% Background and Aims: Insulin resistance is a marker of
below 100 mg/dl, and 0.2% versus 0.2% below 60 mg/dl, re- disease severity in critically-ill patients and is associated with
spectively. In the insulin group 76% received premixed, 22% increased morbidity and mortality.
basal, and 2% bolus insulin as main diabetes therapy. In the no- Methods: We report two cases of severe insulin resistance
insulin group 42% received no diabetes medication, 33% bi- treated with regular insulin (U-500) infusion.
guanids, 17% sulfonylureas, 6% DPP4-inhibitors, 1% glinides, Results: Case 1: 66-year-old male with DM 2, CAD, and atrial
and 1% biguanids combined with sulfonylureas. fibrillation was diagnosed with acute pancreatitis. On day 3, he
Conclusions: Despite complex insulin therapies, overall BG in was intubated for acute respiratory distress syndrome. Insulin
the insulin group was higher than in the no-insulin group. Basal aspart (U-100) intravenous infusion protocol was initiated on day
insulin with its low hypoglycaemic risk is underutilized although 4 in response to BG levels above 200 mg/dL. On days 6-8, the
recommended by guidelines, whereas biguanids/sulfonylureas patient developed acute kidney injury and BG ranged 280-
are commonly used although they should be used with caution in 389 mg/dL despite maximum insulin infusion at 40 units/hr. He
older patients with diabetes. Therefore the individualization of was transitioned to U-500 infusion at 43 units/hr with titration
diabetes therapy is an issue, which could be solved in part by per physician order. BG trended from 389 to 120 mg/dL on day 9.
implementing electronical decision support systems considering U-500 was discontinued and the patient was transitioned back to
geriatric needs. insulin aspart infusion. The patient expired from PEA cardiac
arrest on day 10. Case 2: 51-year-old male with DM 2 and hy-
pertension presented with ST-elevation myocardial infarction.
131
ACCURACY AND USABILITY EVALUATION
OF COMMERCIALLY AVAILABLE BLOOD
GLUCOSE MONITORING SYSTEMS
IN A UNIVERSITY HOSPITAL IN GREECE
A. Tentolouris1, I. Eleftheriadou1, P. Grigoropoulou1,
T. Nikolaos1
1
Medical School- National and Kapodistrian University of
Athens, First Department of Propadeutic and Internal
Medicine, Athens, Greece

Background and Aims: Self-monitoring of blood glucose

(BG) is essential for achieving glycaemic control in individuals
with diabetes mellitus. The aim of our study was to evaluate the
accuracy according to the ISO15197:2013 requirements (95% of
the BG results within 15mg/dl or 15% of the reference method
at BG concentrations <100mg/dl and at BG concentration
100mg/dl, respectively) as well as the usability of 6 different
commercially available self-monitoring blood glucose monitor-
ing systems (SMBG).
Methods: A total of 120 adult individuals with both types of
diabetes mellitus, were recruited. Capillary BG was measured on
6 SMBG, while the YSI 2300 STAT PLUS was used as a ref-
erence instrument. Usability testing of the SMBG was performed
by Likert Scale Questionnaire.
Results: Four out of 6 SMBG fulfilled the minimum accept-
able performance according to the ISO15197:2013 requirements
in BG values <100mg/dl (n = 40), while 5 out of 6 in BG values
100mg/dl (n = 80) (Table 1). Among the 3 SMBG with the
highest performance (>99%) in the overall BG range (29-475mg/
dl), the one with an innovative strip insertion technique was less
favored by the participants.
Conclusions: Across the overall BG range tested, the majority
of SMBG satisfied the ISO requirements. A user-centered design
could enhance usability of the SMBG devices.
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-55

Insulin aspart (U-100) infusion protocol was initiated on day 1 Background and Aims: The purpose of this thesis was to
for BG above 500 mg/dL. Despite insulin aspart 40 units/hr, BG assess the impact of Ramadan fasting on lipid profile and gly-
trended only from 504 to 387 mg/dL. On day 2, he was transi- cemic monitoring among a group of patients with type 2 diabetes
tioned to U-500 infusion at 40 units/hr with titration per physi- before and during the holy month of Ramadan.
cian order, and after 14 hours BG reduced to 144 mg/dL. The Methods: The survey included 80 patients with T2D. The
patient remained on U-500 infusion until care was withdrawn on average age was 56 8 years. The body weight, height, waist
day 5. circumference, BMI were measured. Total cholesterol, HDL
Conclusions: Differences in metabolism of concentrated in- cholesterol, LDL cholesterol, triglycerides, apolipoprotein A1,
sulin may account for the attainment of glycemic control and apolipoprotein B and fasting serum glucose were assessed.
may be considered for refractory insulin resistance. Results: No significant change was noticed for body weight,
waist circumference and BMI during Ramadan. A significant
decrease in HDL-c levels during T2 (0.35 0.08 g/L) compared
to T1 (0.38 0.11 g/L) was found. No significant change in TC,
133 LDL-c and TGs was observed during the two periods. However,
EFFECT OF CHROMIUM SUPPLEMENTATION ON Apo A-1 showed a significant decrease, while Apo B was sig-
CHROMIUM STATUS AND BLOOD SUGAR AMONG nificantly higher during Ramadan. The calculation of lipid
TYPE 2 DIABETES PATIENTS ATTENDING THIKA fractions ratio showed a significant difference between T1 and
HOSPITAL, KENYA: A RANDOMIZED PLACEBO T2 on TG/HDL-c unlike the other ratios (CT/HDLc, LDLc/
CONTROLLED STUDY HDLc, apoB/apoA1). The daily energy intake between the two
periods decreased not significantly during Ramadan. The Iftar
M. Judith1 meal provided the largest part of the daily energy intake during
1 Ramadan. Ramadan fasting induced a non-significant decrease in
Kenyatta University, Food- Nutrition and Dietetics, Nairobi,
carbohydrate amounts, fats and proteins. However, the decrease
Kenya
in caloric intake was associated with an increase of saturated
fatty acids intake during Ramadan.
Background and Aims: Type 2 diabetes is characterized by
Conclusions: This study suggests that fasting during Rama-
insulin resistance and relative insulin deficiency. Studies show
dan could be beneficial for some patients with T2D who are well
contradictory findings on beneficial effects of chromium sup-
controlled and balanced.
plementation on blood sugar (FBS and HbAlc) that warrant ad-
ditional similar studies for conclusive evidence. Studies in Kenya
report a challenge with management of tight blood sugars re-
sulting to amputation of lower limbs among other complications
while chromium supplementation is not a component of man- 135
agement of blood sugar. CHARACTERIZATION OF GASTROINTESTINAL
Methods: This was a double blind randomized placebo con- MOTILITY IN EXPERIMENTAL MODELS
trolled trial with 180 participants on hypoglycemic drugs, ran- OF HYPERGLYCEMIA
domized into two study groups at a ratio of 1:1. Participants in the
intervention group received chromium picolinate (500mcg/day) J.F. Matos1, M. Americo2, Y. Sinzato3, D. Damasceno3,
and control group received placebo for a period of 4 months. J.R. Miranda1
Data was collected at baseline and end of month 4. Blood sam- 1
Institute of Biosciences of Botucatu, Physics and Biophysics,
ples were analyzed for chromium levels, fasting blood sugar and Botucatu, Brazil
HbAlc. T-test, Chi-square and Wilcoxon tests were used to 2
Institute of Biological Sciences and Health Barra do Garcas-
compare data between the study groups. Multiple regressions MT, Parasitology, Barra do Garcas, MT, Brazil
were used to determine predictors of elevated HbAlc at p 0.05. 3
Botucatu Medical School-SP-UNESP, Gynecology, Botucatu,
Results: At baseline, elevated fasting blood sugar with low Brazil
serum chromium levels were observed in both study groups.
Chromium supplementation significantly reduced HbAlc (DID:
Background and Aims: Acute changes in glucose concen-
intervention 1.44% versus control -0.79%; p = 0.001). No sig-
tration in the blood have important effects on motor and sensory
nificant difference in chromium status and fasting blood sugar
function of the upper gastrointestinal tract (TGI), and at the same
between the study groups. The predictors for elevated HbAlc
time, the TGI plays an important role in the regulation of post-
were age and fasting blood sugar.
prandial blood. The aim is to characterize the profile of gastric
Conclusions: Chromium supplementation should be a com-
motility in rats with severe diabetes model and evaluate the in-
ponent of treatment and management of blood sugar among type
fluence of glycemic variations in TGI.
2 diabetes patients.
Methods: Biosusceptometry of Alternating Current (BAC) is
a fairly simple technique, low cost and versatile in research re-
lated to the human TGI. BAC is composed of magnetic sensors
134 based in inductions coil. An electrode was inserted for mea-
surement of electromyography (EMG), and a ferrite bead to
DYSLIPIDEMIA AND GLUCOSE MONITORING measure the BAC in the rats. Severe diabetes was induced by a
AMONG TYPE 2 DIABETES PATIENTS DURING beta-cytotoxic agent (streptozotocin 40 mg/kg, ip) at adult rats.
RAMADAN: A PROSPECTIVE STUDY They were evaluated contractility and gastric emptying of seri-
M.B. Khaled1, G. Tiboura2 ous diabetic and control animals. Concomitant to this experiment
was also performed the hyperglycemic clamp in non-diabetic
1
university, Sidi Bel Abbes, Algeria animals to evaluate the hyperglycemic condition. P < 0.05 as
2
Djillali Liabes University, Biology, Sidi Bel Abbes, Algeria statistical significance limit.
A-56 ATTD 2017 E-POSTER VIEWING ABSTRACTS

Results: There was a significant decrease in contraction fre- Background and Aims: The Ambulatory Glucose Profile
quency and increase in gastric emptying time of the diabetic animals (AGP) was introduced to reveal underlying patterns in glucose
compared with control. As for the hyperglycemic clamp experiment by time of day ( Mazze et al., 1987). Key features included
no significant differences were gastric contraction rate during hy- display of the median (50th percentile), 25th and 75th percen-
perglycemic period. However, morphologically there was a sign of tiles, with smoothing to minimize influence of random fluctu-
irregularity obtained gastric motility during hyperglycemic time. ations. We propose several ways to improve the clinical utility
Conclusions: There is an influence of glycemic status on the of the AGP.
GI tract and may have alterations of frequency and gastric Methods: I proposed the use of 25th, 50th and 75th percentiles
emptying. and a complete Box plot for 8 point glucose profiles (1986).
Acknowledgment: FAPESP 2015/05045-8. This evolved into the original AGP (Mazze et al., 1987) allowing
one to examine the scattergram, the percentiles or both, super-
imposed by time of day. The method was adapted to display
CGM data with addition of 10th and 90th percentiles (Mazze
136 et al., 2008, 2009). The AGP can be augmented by simultaneous
RELATIONSHIP BETWEEN THE DEGREE display of IQR (75th 25th percentiles) or SD, and of estimated
OF IMPROVEMENT IN GLUCOSE METABOLISM insulin pharmacodynamic activity profiles by time of day. One
AND OXIDATIVE STRESS IN TYPE 2 DIABETES can display results for two days to facilitate evaluation of the
PATIENTS overnight period. One can synchronize and normalize data at
the onset of meals, bedtime and awakening, to focus on patterns
M. Ohara1, H. Nagaike1, N. Maeki1, S. Yamamoto1, during these time segments (Rodbard, 2013). Logarithmic
H. Kushima1, M. Hiromura1, T. Yamamoto1, T. Hayashi1, transformation provides better symmetry of the glucose distri-
T. Fukui1, T. Hirano1 bution and improved balance of hyper-, target-, and hypoglyce-
1 mic ranges (Rodbard, 2009).
Showa University, Diabetes Metabolism & Endocrinology,
Results: Several new modifications of the AGP will be dis-
Shinagawa-ku, Japan
played to illustrate their utility and advantages.
Conclusions: The AGP is useful, well accepted, easy to learn
Background and Aims: To evaluate whether glucose vari-
and use both by patients and physicians. These several new
ability and markers of diabetic control improve oxidative stress
modifications are expected to further enhance the effectiveness
in type 2 diabetes (T2DM).
of the AGP.
Methods: The study subjects were thirty-six T2DM outpatients
(mean age, 61.6 11.9 years; HbA1c level, 8.4 1.6%). The pa-
tients underwent 72 hours of continuous glucose monitoring
(CGM) and the diacron-reactive oxygen metabolites (d-ROMs) test
before and after a 6-month intervention (target: HbA1c<7%). The 138
following variables were calculated from the CGM data: fasting DATA DRIVEN BLOOD GLUCOSE PREDICTION:
plasma glucose (FPG), mean glucose level (MGL), mean ampli- INTERVAL VS. POINT BLOOD GLUCOSE
tude of glycemic excursions (MAGE), mean of daily differ- PREDICTION?
ences (MODD), and area under the postprandial plasma glucose
curve (AUCPP). The relationship between glucose metabolism A.Z. Woldaregay1, E. Arsand2,3, G. Hartvigsen1,3
improvement and oxidative stress improvement was evaluated. 1
UiT The Arctic University of Norway, Computer Science,
Results: MGL, FPG, HbA1c, MAGE, AUCPP, and d-ROMs Troms, Norway
were significantly improved in the second determination com- 2
UiT The Arctic University of Norway, Clinical Medicine,
pared to the first, but no significant difference was observed in Troms, Norway
MODD. The changes in FPG (r = 0.574), MGL (r = 0.439), and 3
Norwegian Centre for E-health Research, University Hospital
HbA1c (r = 0.403) were significantly correlated with the change of North Norway, Troms, Norway
in d-ROMs but not the changes in MAGE, MODD, or AUCPP.
The changes in d-ROMs and FPG were still significantly related
Background and Aims: Since diabetes patients are chal-
after adjustment for the other markers of diabetic control in
lenged to maintain their blood glucose (BG) levels as close as
multiple linear regression analysis (multiple R2 = 0.309).
possible to the normal range, BG prediction has a great potential
Conclusions: Improved glucose metabolism reduces oxida-
in improving decision support tools. Recently, due to more
tive stress in T2DM. In contrast to earlier cross-sectional findings
widespread use of glucose sensors and patient-gathered data in
showing a strong correlation between oxidative stress and gly-
general, research on data driven BG prediction are now more
cemic variability in T2DM, reduced oxidative stress in T2DM
relevant than ever. We present an overview of the current prac-
was strongly associated not with improved glucose variability or
tices and address the knowledge gap within these practices.
postprandial glucose, but with improved FPG.
Methods: We searched various electronic databases, see
Figure 1. The search targeted peer reviewed journal articles and
conference proceedings.
137 Results: 31 articles were critically analyzed, see Figure 1.
Predicting an interval or a point is dependent upon the choice
NEW ENHANCEMENTS TO THE AMBULATORY whether to include possible realization of uncertainties or not.
GLUCOSE PROFILE (AGP) Interestingly, our review identified inclination towards ma-
D. Rodbard1 chine learning approaches. BG, insulin and diet are the most
used input parameters, see Figure 2. Most of the studies have
1
Biomedical Informatics Consultants, Clinical Research used significant number of participants. Furthermore, the re-
and Biostatistics, Potomac MD, USA view found that point BG prediction is the most used form
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-57

139
A PHYSIOLOGICAL MODEL OF TYPE 2 DIABETES
FOR SIMULATING GLYCAEMIC VARIANCE AND
EFFECT OF INSULIN DOSE ADJUSTMENTS
T.B. Aradottir1,2, B. Van Orden3, D. Boiroux1, J.B. Jrgensen1,
H. Bengtsson2
1
Technical University of Denmark, Department of Applied
Mathematics and Computer Science, Kongens Lyngby,
Denmark
2
Novo Nordisk A/S, Technology Explorations, Hillerd,
Denmark
3
Novo Nordisk A/S, Digital Health, Seattle, USA

Background and Aims: More than 60% of type 2 diabetes

(T2D) patients on insulin treatment in USA do not reach re-
commended goals. The reasons reported include lack of em-
powerment, sub-optimal dosage, fear of hypoglycaemia and
complexity. Automatic dose adjustment algorithms use self-
monitored blood glucose (SMBG) measurements. SMBG is
prone to high variance due to intra-individual physiological
fluctuations and when done incorrectly. However, pharmacoki-
netic variance decreases with new ultra-long acting insulins.
With continuous glucose monitoring (CGM) becoming cheaper,
more accurate, and convenient in the near future, we see op-
Figure 1: Flow diagram of reviewed literature. portunities in using CGM data to address the barriers to reaching
treatment goals.
Methods: We present an adjusted model of the insulin-
glucose dynamics in type 1 diabetes (T1D) that also describes
endogenous insulin production in T2D. We adjust insulin sen-
compared to interval. Despite its ability of capturing un-
sitivity, body weight and distribution volume of the T1D model
certainties, interval prediction has got little attention.
to represent a cohort similar to a clinical trial in T2D.
Conclusions: Recently, data driven models have seen success
Results: We use the model to simulate CGM data in T2D
in terms of simplicity and accuracy. However, the accuracy de-
basal insulin treatment and compare the results with clinical
grades after certain prediction horizon possibly due to stochastic
trial results. We investigate sources of variance and through
phenomena involved in BG dynamics. In this regard, the current
simulations we evaluate the contribution of adherence, erroneous
research efforts lack use of interval prediction mechanisms with
measurements and ultra-long acing insulins. The results suggest
all its flavors in coping uncertainties.
that the simulation model is sufficient for testing dose adjustment
algorithms.
Conclusions: We find that dose adjustment algorithms
should compensate for adherence levels and high variance
for safety through methods including adherence detec-
tion and support, and implicit elements in dose adjustment
calculations.

140
A MINIMAL MATHEMATICAL MODEL
FOR INTERPRETATION OF DATA FROM
CONTINUOUS GLUCOSE MONITORING
I. Barsukov1, A. Dreval1, T. Shestakova1, O. Medvedev2,
M. Panteleev2, A. Sveshnikova2
1
Moscow Regional Research and Clinical Institute named after
M.F. Vladimirsky, Endocrinology, Moscow, Russia
2
Moscow State University, Department of Basic Medicine,
Moscow, Russia

Background and Aims: Aim. Development of a minimal

mathematical model of the interstitial glucose dynamics
Figure 2: Types of algorithm and input parameters used suitable for interpretation of data from continuous glucose
by the identified prediction models monitoring.
A-58 ATTD 2017 E-POSTER VIEWING ABSTRACTS

Methods: The data of CGM of 6 pregnant women with dia- 142

betes mellitus type 1, age 27.2 3.,4 years, 15.3 2.7 weeks of
GLYCEMIC VARIABILITY ASSESSED BY
pregnancy. The data of two patents with diabetes mellitus (one
TEMPORALLY-ADJUSTED, REGULARIZED
with MII and one with CSII) with detailed diary. In diary dose
GLYCEMIC EXCURSION TARGETS (TARGET):
insulin, amount of carbohydrates, physical activity was checked
THINKING OUTSIDE THE BOX
in. The mathematical model consists of six ordinary differential
equations, describing dynamics of blood and interstitial glucose R. Carter1, R. Basu2, A. Basu2, Y. Kudva2
level and the hormones and drugs concentration in blood. The 1
model integration and parameter estimation were performed in Mayo Clinic, Health Sciences Research, Rochester, MN, USA
2
COPASI software (http://www.copasi.org/). Mayo Clinic, Endocrinology, Rochester, MN, USA
Results: The developed model described experimental data
well. For patients with type 1 diabetes mellitus generally insulin Background and Aims: Widespread use of continuous glucose
production was zero and parameters of insulin elimination higher monitors (CGMs) has enabled insight into glycemic variation.
than those for healthy donors. Patient-specific parameters in- Numerous summary measures exist to reduce a CGM profile into a
cluded glucose uptake efficiency and rate of blood-interstitial single, clinically-meaningful patient-centered summary; however,
glucose interconversion. the commonly used summary measures do not account for expected
Relative objective value is 3% for the patients with accurate variation during the day (Fig 1A) and provide little information
data and 1% for the patients without exactly data. about specific patient-centered interactions that may be considered
Conclusions: The constructed mathematical model could be to improve glucose control. We propose a novel glycemic variation
utilized for assessment of patient-specific parameters from con- measure entitled Temporally-Adjusted, Regularized Glycemic
tinuous glucose monitoring data. Excursion Targets (TARGET) that accounts for expected variation.
Methods: Each 24-hour period is divided into seven mutually
exclusive and exhaustive bins (Fig 1B). These bins represent periods
where one would expect relative homogeneity of glucose values
141 within the bin and heterogeneity between bins. As conceptualized,
the binning follows a daily routine consisting of three meals per
SIMULATION FRAMEWORK TO TEST ALGORITHMS day at regular times and sleep overnight; however, the method is
FOR THE OPTIMISATION OF INSULIN BOLUS flexible in that it can be adjusted to fit individual daily habits.
PARAMETERS IN TYPE 1 DIABETES THERAPY TARGET is structured on a 0-100 point scale, with higher scores
G. Cappon1, M. Vettoretti1, A. Facchinetti1, G. Sparacino1, indicating a greater alignment of the observed CGM data stream
C. Cobelli1 with the values determined to represent expected glycemic variation.
Results: TARGET removes the single box that is formed
1
University of Padova, Department of Information Engineering, by standard in range calculations (e.g., 70 140 mg/dl for the
Padova, Italy

Background and Aims: Type 1 diabetes (T1D) therapy is

based on basal+bolus insulin administrations. The key parameter
for calculating meal boluses is the carbohydrate-to-insulin ratio
(CR), which is normally tuned by trial-and-error procedures and,
as such, suboptimal. The aim is to create a realistic simulation
framework to test algorithms for CR optimization from CGM
data.
Methods: The simulation framework is composed by the T1D
patient decision-making model of (Vettoretti et al., Proc. IEEE
EMBS, 2015), where we integrate the UVA/Padova T1D simu-
lator (2013 release) with insulin sensitivity variability (Visentin
et al., IEEE TBME 2015) and models of CGM error (Facchinetti
et al., MBEC 2015), and patients behaviour, thus allowing to
generate reliable multiple-day scenarios. This framework is used
to assess two CR optimization algorithms: the run-to-run (R2R)
approach described in (Herrero et al., CMPB, 2015) and a new
method (R2R+CBR), where the R2R is integrated with case-
based reasoning (CBR (Herrero et al., IEEE J-BHI, 2014), with
the main difference of employing as case parameters the carbo-
hydrate load and pre-meal glycemic level, in addition to meals
time, instead of exercise intensity.
Results: We simulated four 30-day sessions on 100 virtual
patients with R2R, R2R+CBR and no optimization. On av-
erage, in the fourth session, time spent in euglycemia and
risk index result, respectively, 82.68% and 3.84 with R2R,
75.87% and 4.90 with R2R+CBR, 67.37% and 7.02 with no
optimization.
Conclusions: We developed a simulation framework that al-
lows generating realistic scenarios to test existing and develop
new algorithms for CR optimization.
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-59

1
entire day). CGM data (Fig 1C-D) can be used to summarize University of Western Australia, School of Population Heath,
TARGET values to represent overall glucose control as well as Perth, Australia
2
control within particular epochs during the day. Princess Margaret Hospital for Children, Diabetes and
Conclusions: Trends and profiles informed through TARGET Endocrinology, Perth, Australia
3
will serve as a means to describe personalized management Curtin University, School of Public Health, Perth, Australia
4
strategies to improve glycemic variation. Telethon Kids Institute, Centre for Research Excellence in
Type 1 Diabetes, Perth, Australia

Background and Aims: To apply consumer choice theory to

143 quantify lifestyle benefits of insulin therapies to adolescent pa-
tients with type 1 diabetes and their parents in Western Australia,
NOVEL MEDTRONIC TURNING POINT PROGRAM generating additional quality-of-life measures to complement
IMPROVES COMPLIANCE AND HBA1C IN AT-RISK standard measures used by policymakers in subsidisation deci-
PATIENTS WITH TYPE 1 OR TYPE 2 DIABETES sions. A Discrete Choice Experiment survey will be developed to
D.S. Schwartz1, D.E. Dick2, C. McMahon3, D.O. Cohen3, determine what factors influence treatment preferences and
J. Knoulton4, S. Arunachalam3 how important they are in patient decision-making to evaluate
1
whether improved access to the costly gold-standard treatment
DGD Clinic, Chief Clinical Consultant, San Antonio, USA (continuous glucose monitoring) will result in cost-savings for
2
Methodist Healthcare Ministries, Medical Director, the Australian Government.
San Antonio, USA Methods: Focus groups were held with patients and parents to
3
Medtronic Diabetes, Diabetes Service & Solutions, identify appropriate treatment attributes for inclusion in the
Northridge, USA survey. Comprehension of the survey was tested during clinic
4
Methodist Healthcare Ministries, Director of Nursing, San appointments. A sample of 200 adolescent patients (and parents)
Antonio, USA with type 1 diabetes in Western Australia will be recruited to
complete the survey. A multinomial logit analysis will determine
Background and Aims: The Medtronic Turning Point Pro- the relative importance of each treatment attribute for male and
gram was introduced at the Methodist Healthcare Ministries of female adolescent patients and a willingness-to-pay analysis will
South Texas, Inc. (San Antonio, Texas) to demonstrate the fea- be conducted with data from the parent survey to determine how
sibility of a novel service to enhance compliance and improve financial burden influences final treatment decisions.
outcomes in high-risk patients with diabetes. Results: Focus groups were held with male (n = 2) and female
Methods: From January-August 2016, 50 patients ages 50 11 (n = 6) adolescent patients and parents. Final results from the
(Mean SD) years with uncontrolled type 1 or type 2 diabetes Discrete Choice Experiment are expected by November 2016.
(A1C = 10.3 1.2%) were enrolled in the program. All patients Conclusions: This project will generate new health economic
were instructed to measure blood glucose (SMBG) 3-4/day. data to improve understanding of patient preferences for insulin
Follow-up A1Cs were done in 28 subjects who used a smartphone therapies and providing quantified lifestyle benefits to policy-
mobile application that captured physical activity and nutrition makers in addition to quality-adjusted life-year measures.
information; provided patient self-management education, med-
ication reminders and observation-driven feedback from a health
coach (HC) to a Nurse or a primary care physician (PCP) or an
endocrinologist. Forty six percent (13/28) transiently wore a 145
continuous glucose monitoring (CGM) system (iPro 2, Med-
tronic, Northridge, CA). Data and mobile application informa- SMART INSULIN DOSING FROM THE REAL-TIME
tion, in addition to standard of care algorithms were analyzed by ASSESSMENT OF INSULIN EFFECTIVENESS
the health care team approximately every 2-4 weeks. IN TYPE I DIABETES
Results: At a mean of 94 42 [IQR = 63-122] days, the program
C. Fabris1, M. Breton1
had a retention rate of 74%, and demonstrated a mean change of
1
-2.8 1.5% in A1C (paired t-test, p < 0.001). Mean compliance with University of Virginia, Center for Diabetes Technology,
SMBG measures (2.0 0.9/day) was observed at a rate of 71 25%. Charlottesville, USA
Conclusions: The Program showed a significant reduction in
A1C, at 3-6 months. Extending PCP access through weekly patient- Background and Aims: Imperfect insulin replacement is a
HC interactions; and ongoing mobile or remote data monitoring, cause of hypo/hyperglycemia and increased glucose variability
HC-PCP coordination and endocrinologist-PCP engagement help in T1D. Insulin sensitivity (SI) is a primary factor mediating BG
to provide high-risk patients an effective and feasible solution to responses to administered insulin. We propose a novel method
achieve health goals. for optimal insulin dosing in T1D leveraging the real-time esti-
mation of Insulin Effectiveness (IE), an index related to SI, from
CGM and insulin pump data.
144 Methods: We propose to modulate insulin boluses using real-
time estimates of IE compared to the patients usual IE at the
CONSUMER CHOICE THEORY AND HEALTH
same time of day, while systematic circadian fluctuations are
TECHNOLOGY ASSESSMENTS IN AUSTRALIA -
managed by retrospective tuning of the insulin therapy parame-
A DISCRETE CHOICE EXPERIMENT TO ELICIT
ters (e.g., insulin-to-carbohydrate ratio and insulin-sensitivity
PREFERENCES AMONG ADOLESCENT PATIENTS
factor profiles). At the time of a meal, IE is estimated from
WITH TYPE 1 DIABETES
CGM and pump data using a 3-states Kalman filter which de-
N. Barnett1, E. Davis2, T. Jones2, E. Geelhoed1, M. De Bock2, scribes glucose-insulin dynamics in a logarithmic risk-space; the
R. Norman3, G. Smith4 patients usual IE is estimated from a month of historical data.
A-60 ATTD 2017 E-POSTER VIEWING ABSTRACTS

8
HGR 110 IMSS Guadalajara-Jal, Endocrinologia,
Guadalajara, Mexico
9
Hospital Espanol, Endocrinologia Pediatrica, Mexico City,
Mexico
10
Hospital Angeles Lomas, Endocrinologia, Mexico City,
Mexico
11
Hospital Rovirosa-Secretaria de Salud, Diabetes,
Villahermosa, Mexico
12
Hospital Angeles Lomas / Centro Medico ABC Santa Fe,
Endocrinologia Pediatrica, Mexico City, Mexico
13
Hospital De Gineco-Pediatra No. 31 IMSS, Endocrinologia
Pediatrica, Mexicali, Mexico
14
Centro para la Prevencion y Atencion Integral del VIH/SIDA
del Distrito Federal- Secretaria de Salud, Jefatura de Sistemas
de Informacion, Mexico City, Mexico
15
UNIVERSIDAD IBEROAMERICANA, EDUCACION,
Mexico City, Mexico

Background and Aims: Worldwide there has been an in-

crease in type 1 diabetes (T1D) incidence and prevalence,
leading to the creation of T1D registries. In Mexico, there is a
knowledge gap regarding these variables, treatment practices and
The smart bolus then consists in a standard bolus modulated by degree of diabetes control in patients with T1D. Our aim is to
the ratio of usual and real-time IEs. The algorithm was tested in describe the creation of a national T1D registry in Mexico:
the UVa/Padova T1D Simulator (1-meal scenario with varying RENACED Diabetes Tipo 1.
SI) and a clinical trial (NCT02558491). Methods: We developed an online system for longitudinal
Results: In simulation, the IE-informed bolus calculator al- data capture based on LAMP (Linux, Apache, MySQL, PHP)
lowed to reduce hyperglycemic (hypoglycemic) risk in case of technology, on a secure website, following the national laws of
decreased (increased) IE, limiting the impact of SI fluctuations, privacy and confidentiality (informed consent and notice of
compared to standard therapy (see Figure). In-vivo, preliminary privacy). The system allows for data capture relevant to T1D,
results (N = 5) show decreased postprandial time-below-70mg/dl based on St. Vincent Declaration. The data capture is performed
(14.1% to 1.9%) and time-above-250mg/dl (2.1% to 1.7%). by registered attending physicians with online access. The in-
Conclusions: We propose a new, data-oriented, paradigm for formation can be analyzed at clinic, state or national level.
insulin dosing in T1D, based on the real-time assessment of the Results: The online system development version 1.0 was
patients insulin need from sensor and pump data. finished in June 2014. 9% of patients were captured in 2014, 22%
in 2015 by one medical unit. Multiple improvements were done
on the database for ease of capture, addition of graphics and
reports. In 2016 69% of patients were captured in 12 medical
146 units in 10 different states. As of October 5th, 2016 a total of 494
patients have been captured, 474 with T1D, that live in 18 dif-
THE CREATION OF A NATIONAL TYPE 1 ferent states of the Republic.
DIABETES REGISTRY IN MEXICO: RENACED Conclusions: This is the first online T1D national registry in
DIABETES TIPO 1 Mexico. More support and diffusion will lead to more physicians
R.N. Faradji1, M. Valenzuela-Lara2, J.F. Bustamante-Martinez3, registering their patients, and will provide more knowledge of the
N.E. De La Garza-Hernandez4, A.P. Diaz-Barriga-Menchaca5, T1D reality in Mexico.
A. Escobedo-Ortiz6, A. Flores-Camargo5, L. Islas-Ortega7,
C. Lopez-Miramontes8, A. Martinez-Ramos9, S. Miracle-
Lopez10, L.A. Ramirez-Toscano8, E. Rodriguez-Sanchez11, 147
M. Tavera-Hernandez12, J.C. Valenzuela-Montoya13, SCREENING OF OBSTRUCTIVE SLEEP APNEA
M. Vidrio-Velazquez8, R. Nino Vargas14, SYNDROME WITH RESPECT TO THE INCIDENCE
E. Sainz-De La Maza-Viadero15, C. Magis-Rodriguez2 OF MACROVASCULAR COMPLICATIONS
1 AND IMPAIRMENT OF MICROCIRCULATION
Instituto Tecnologico de Monterrey / Centro Medico ABC,
Endocrinologia, Mexico, Mexico IN PATIENTS WITH THE DIABETIC FOOT
2
Centro Nacional para la Prevencion y el Control del VIH y el V. Fejfarova1, M. Klementova1, J. Polak2, A. Jirkovska1,
Sida-Secretaria de Salud, Epidemiologia, Mexico City, Mexico R. Bem1, M. Dubsky1, V. Woskova1, A. Nemcova1,
3
Hospital General de Tepic, Departamento de Medicina M. Krzova1, J. Skibova1
Interna, Tepic, Mexico
4 1
CEMEDIN, Endocrinologia, Monterrey, Mexico Institute for Clinical and Experimental Medicine, Diabetes
5
Instituto Tecnologico de Monterrey, Ciencias Medicas Basicas Center, Prague, Czech Republic
2
y Clnicas, Mexico, Mexico II. Internal department, FNKV, 3.LFUK, Prague,
6
Hospital General Dr. Miguel Silva- Secretaria de Salud, Czech Republic
Endocrinologia, Morelia, Mexico
7
Hospital DIF de la Ninez Hidalguense, Endocrinologia Background and Aims: The aims of our study were to assess
Pediatrica, Pachuca, Mexico subjectively described sleep disorders, the possible occurrence of
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-61

OSAS in patients with diabetic foot (DF) and its possible asso- will also measure hypoglycaemia, glycaemic variability,
ciation with macrovascular complications and the impairment of patient-reported outcomes, acceptability to health profes-
microcirculation. sionals and patients, and cost-effectiveness. Baseline and 12
Methods: We included into our study 105 patients with the month r-CGM trace will be undertaken for all patients (blind
DF completing screening questionnaires (Berlin, STOP-Bang to control group).
questionnaires, ESS), questionnaires evaluating the quality of Sample size: 50 clinics with 6 participants per clinic (150 in
sleep, and performing anthropometric examinations.The pres- each arm), will allow 10% clinic attrition and 20% patient at-
ence of cardiovascular heart disease (CHD) was present in 28.2% trition and give 80% power to detect a difference in mean
of all patients with the DF, strokes (9.7%), peripheral arterial HbA1c of 0.5% with a standard deviation of 1.3 and an alpha
disease (PAD in 58.3%), and the values of transcutaneous oxy- of 0.05.
gen pressure (TcPO2) determining the microcirculation were Conclusion: Our trial will generate cost effectiveness evi-
examined in all study patients. Patients were divided into 3 dence about the potential of this technology to drive personalised
groups (category 1- positive both screening questionnaires, high treatment intensification to achieve glycaemic targets.
risk of OSAS, category 2- positive one questionnaire, category 3-
both questionnaires negative).
Results: 29.8% of patients with DF (31/105) belonged to cate-
gory 1, 63.5% (67/105) to category 2 and 6.7% to category 3 (7/ 149
105). Only category 1 correlated significantly with awakening
(p = 0.03), poor quality of sleep (p = 0.0064), lack of sleep VIRTUAL PATIENTS DERIVED
(p = 0.001), tired feeling (p = 0.0001) and with higher excessive FROM THE CARELINK DATABASE
daytime sleepiness (ESS; p = 0.0039). Selected anthropomet- B. Grosman1, A. Roy1, D. Wu1, N. Parikh1, S. John2, S. Lee3,
ric measurements positively correlated with category 1- BMI K. Francine4
(p = 0.0012), neck (p = 0.001), waist (p = 0.0004) and hip circum- 1
ferences (p = 0.0099). Category 1 correlated significantly only with Medtronic Diabetes, Closed Loop, Northridge, USA
2
the presence of PAD (p = 0.023) but not with CHD and strokes. Medtronic Diabetes, Clinical Research Data, Northridge, USA
3
Compared to category 3 (0% of patients), significantly more patients Medtronic Diabetes, Clinical Research Administration,
with DF had TcPO2 below 40mmHg in category 1 (62.5%; Northridge, USA
4
p = 0.017), marginally significantly in category 2 (45.6%; p = 0.068). Medtronic Diabetes, Medical Affairs, Northridge, USA
Conclusions: The incidence of OSAS is probably higher in
risk group of patients with DF. Screening OSAtests correlated Background and Aims: The CareLink database has been
significantly in patients with DF with subjectively described used to generate a large number of virtual patients to be used for
sleep disorders, selected anthropometric parameters, PAD in silico testing.
and with the impairment of microcirculation. Supported by Methods: A mathematical model was developed with a tai-
00023001. lored parametric optimization method. Each virtual subject was
trained from 20 days of CareLink data. Using data from the
ASPIRE Clinical Trial and from a hybrid closed-loop feasibility
study, the virtual patients were validated. Insulin action speed
148 and the meal absorption rates used in the mathematical model
GENERAL PRACTICE OPTIMISING STRUCTURED were validated with published literature.
MONITORING TO IMPROVE CLINICAL OUTCOMES Results: Six-hundred sixty-three virtual patients were gener-
IN TYPE 2 DIABETES: THE GP-OSMOTIC STUDY ated. Circadian rhythm in daily insulin sensitivity was identified
PROTOCOL in 590 virtual patients. In addition, each virtual patient was as-
sociated with specific meal absorption rates, an age group, and a
J. Furler1 total daily dosage of insulin. The Table shows the overall con-
1
University of Melbourne, General Practice, Parkville, cordance between the glucose levels of the virtual and the
Australia CareLink patients.

Background and Aim: Our aim is to trial the (cost) effec- Table. Glucose Levels for CareLink and Virtual Patients
tiveness of retrospective continuous glucose monitoring (r- Mean Glucose
CGM), in an inclusive population of people with type 2 diabetes glucose 70-180 Glucose
managed in primary care. We will also gather robust hypogly- mg/dL mg/dL <70mg/dL
caemia prevalence data in this population.
Methods: OSMOTIC is an individually randomised con- CareLink Patients 154 69 1.2
trolled trial set in General Practices in Victoria testing the effect
of r-CGM (applied 14 days, 4 occasions per year), compared with Virtual Patients 157 72 2.9
usual care. A collaborative educational consultation with a
Credentialed Diabetes EducatorRegistered Nurse (CDE-RN)
will be provided to all patients.
Patients with T2D (>12 months), aged 18-80 years, diabetes Conclusions: The Medtronic CareLink database was dem-
duration >1 - <20 years and most recent HbA1c (in the previous 6 onstrated to be a rich database for producing a large number of
months) >7mmol/mol (0.5%) above their individualised target virtual patients with a variety of time dependent insulin sensi-
on maximum oral therapy and/or injectable therapy (insulin and tivities, variable pharmacokinetics, and meal absorption rates.
GLP1 agonists). These virtual patients can be used to advance further refinements
Results: Evaluation at baseline and 12 month will include in the development of artificial pancreas algorithms and auto-
HbA1c (primary), time in target, and diabetes distress. We mated insulin delivery systems.
A-62 ATTD 2017 E-POSTER VIEWING ABSTRACTS

150
SAFETY LAYER FOR AN INSULIN DELIVERY
SYSTEM
P. Herrero1, P. Peter1, R. Monika2, E.S. Mohamed1,
T. Christofer1, O. Nick2, G. Pantelis1
1
Imperial College London, Electrical and Electronic
Engineering, London, United Kingdom
2
Imperial College London, Medicine, London, United Kingdom
tions and metrics for hypoglycemia and hyperglycemia estimated.
Background and Aims: Safety is paramount in (semi-) The metrics used are Minutes below 70 mg/dL (MB70) and Minutes
automatic insulin delivery systems such as a closed-loop, or a de- above 240 mg/dL (MA240). Agreement of metrics (mean = 11.53
cision support, system for insulin dosing. Therefore, a safety layer is +/- 2.12 days per subject of sensor glucose) was analyzed by linear
desired in order to minimise any risk of adverse events (i.e. hypo- regression and Bland-Altman analyses. It was observed that esti-
and hyperglycaemia) due to any malfunctioning of the system (e.g. mates showed strong correlation with the measured MB70 (Gamma:
dosing algorithm miscalculation or sensor/pump failure). R-square = 0.95, slope = 0.80, intercept = 2.96; Lognormal: R-
Methods: This work presents a novel safety layer for an in- square = 0.96; slope = 0.82; intercept = -4.71) and MA240 (Gamma:
sulin delivery system which includes: a robust dynamic insulin R-square = 0.99; slope = 0.96; intercept = 3.12; Lognormal: R-
constraint using an interval-based bolus calculator; a predictive square = 0.98; slope = 0.90; intercept = 20.72) values. The bias
low-glucose insulin suspension system embedding an innovative against measured values was analyzed using a Bland-Altman plot
glucose forecasting algorithm using a physiological model of (Figure 1).
glucose dynamics; an adaptive carbohydrate recommender; a Results: The lognormal distribution slightly underestimated
model-based fault detection algorithm; and an alarm manager for MB70 and slightly overestimated MA240 compared to the
patients and carers. The system has been designed to be im- gamma. Though both show large bias at high MB70 values
plemented on a mobile platform (e.g. insulin pump handset or (>150 min/day), the distributions are generally used for estima-
smartphone) and is planned to be clinically tested, together with tion at MB70 levels less than 120 min/day to detect transitions
an intelligent insulin recommender, within the framework of the from low to high hypoglycemia.
EU-funded PEPPER project. Conclusions: Both distributions are a reasonable choice for
Results: In silico results with the UVa-Padova simulator on estimating degrees of hypoglycemia risk; however, the gamma
an adult population (n = 10) using a bolus calculator showed distribution is preferred since it has less estimate bias.
that the dynamic insulin constraint is able to eliminate severe
hypoglycaemia (<50mg/dL), while the predictive low-glucose
suspend completely eliminates hypoglycaemia (<70mg/dL) 152
without increasing hyperglycaemia. Using retrospective clini- HOW MUCH DO THE RELATIVES OF DIABETES
cal data, the implemented glucose forecasting algorithm out- PATIENTS KNOW ABOUT DIABETES?
performs a state-of-the-art forecasting algorithm - i.e.
autoregressive model (order 3) with recursive identification of P. Karakaya1
the coefficients- (rRMSE: 32.6% vs 24.2% - prediction horizon: 1
Bakirkoy Dr Sadi Konuk Training and Research Hospital,
90min). Department of Endocrinology, ISTANBUL, Turkey
Conclusions: A novel dynamic insulin constraint and a pre-
dictive low-glucose suspension algorithm have the potential to Background and Aims: The relatives of diabetic patients have
minimise hypoglycaemia in an adult type 1 population. an important role in the care, follow-up, treatment, and glycemia
control of the patients. The aim of this study was to investigate the
151 knowledge and experience of relatives of diabetic patients re-
garding diabetes mellitus (DM), its treatment and follow-up.
COMPARISON OF STATISTICAL DISTRIBUTIONS Methods: A survey was filled out in face-to-face meetings with
TO ESTIMATE HYPOGLYCEMIA AND relatives of 550 diabetic patients visited the internal medicine clinic
HYPERGLYCEMIA METRICS of our hospital, majority of whom were elderly. Socio-demographic
S. Jangam1, T. Dunn1, G. Hayter1 characteristics, knowledge about DM, patient treatment and follow-
up of the participants were questioned.
1
Abbott Diabetes Care, Research and Development, Results: A total of 550 patient relatives, 64.5% of which were
Alameda, USA female, were included in the study.
Results of a rough comparison based on the level of education
Background and Aims: With the availability of continuous showed that while 50.8% of those holding a university degree were
glucose (CGM) data, statistical distributions have been used to trained and knowledgeable about DM, only 14% knew the name
estimate hypoglycemia and hyperglycemia metrics when data is and the dosage of the medications. Among those with an elemen-
sparse and to reduce the effect of outliers. Two distributions have tary school degree, while only 24.6% had received training on DM,
been used: the lognormal distribution and the gamma distribu- 44.9% knew the names and the dosage of the medications.
tion. The latter is used in the FreeStyle Libre Glucose Pattern Conclusions: The patient relatives have roles just as sig-
Insights Report for determination of zones of low, moderate and nificant as physicians in follow-up and treatment of patients,
high hypoglycemia. their medications, diet, and exercise compliance, an accurate
Methods: Here, CGM data from the baseline period of a study glycemia control; and follow-up and prevention of acute and
(n = 87, T1DM = 42, T2DM = 45) has been fitted to both distribu- chronic complications. Therefore, it is important to train the
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-63

patient relatives as it is important to train the patients them- acuity, polypharmacy, peripheral neuropathies and insulin ther-
selves to ensure control DM and to prevent complications. This apy in elderly with DM.
study demonstrates that the patient relatives are still not suf- The aim of this study was the investigation the effects of fear
ficiently trained on this matter. of falling on time-distance parameters of gait in elderly with
type-2 diabetes mellitus.
Methods: The study included 31 Type-2 DM patients who
153 were diagnosed by a specialist doctor and 29 healthy control
group over aged of 55 years. Fear of falling was assessed by
COMPARISON OF HOMA-IR AND ANP VALUES International Fall Efficacy Scale. Time-distance parameters of
IN CONTROLLED AND UNCONTROLLED gait were measured with Zebris FDM-2.
DIABETES PATIENTS Results: In diabetics, as the fear of falling increased, the
P. Karakaya1, M. Mert1 length of step (r: -.660, p < 0.01) and cadance (r: -.401, p < 0.01)
decreased and duration of the step (r: .390, p < 0.05) increased.
1
Bakirkoy Dr Sadi Konuk Training and Research Hospital, There was no statistically significant relationship between the
Department of Endocrinology, Istanbul, Turkey fear of falling and the time-distance parameters of gait in the
control group (p > 0.05).
Background and Aims: Increase of plasma Atrial Na- Conclusions: Fear of falling affected time-distance pa-
triuretic Peptide (ANP) level is very important in the response rameters of gait negatively. In the management of Type-2
to hyperglycemia in diabetic patients due to the insulins so- DM patients, health professionals provide individualized
dium sparing effect that is known as a chronic stimulator of gait training.
ANP. We aimed to compare fasting and postprandial HOMA-
IR and ANP levels in patients with regulated and unregulated
diabetes 155
Methods: Sixty obese or non-obese patients with type 2 dia- TARGET POPULATION FOR INNOVATIVE
betes aged 30-70 years were included in the study. Systemic TECHNOLOGY ON DIABETES IN BRAZIL:
chronic diseases like functional thyroid disorder, liver failure, EVERYBODY NEEDS TO KNOW ABOUT
kidney failure., insulin use, as well as pregnancy were the PREVALENCE
exclusion criteria. Blood samples were drawn from all patients
for the routine biochemical tests, lipid profile, fasting and post- F. Laranjeira1, G. Brito1, G. Tannus1, M. Fonseca1
prandial glucose and insulin levels and high sensitive CRP (hs- 1
AXIA.BIO Life Science Consulting Group, Research,
CRP) and ANP. Sao Paulo, Brazil
Results: Demographic and laboratory data of all type 2 diabetes
patients with regulated (HbA1c<7) and unregulated (HbA1c> = 7) Background and Aims: Worldwide diabetes (DM) prevalence
blood glucose levels are presented in Table 1. Blood glucose, is around 9%, affecting approximately 415 million people. In
fasting and postprandial HOMA and K levels were significantly Brazil, although there are several epidemiological studies on this
higher in patients with unregulated diabetes. ANP and hs-CRP subject, there are still uncertainties about how many people have
levels were similar in both groups. Laboratory and demographic diabetes.
data of patients with fasting and postprandial HOMA levels >2.7 Methods: We performed a systematic review on the prevalence
and < = 2.7 were statistically similar. of diabetes in Brazil. Potentially relevant studies were identified
Conclusions: ANP levels are expected to be increased in through an exhaustive search in MEDLINE, SciELO, LILACS,
unregulated diabetes but our results did not support this. This Ministry of Health and hand-search. Eligibility criteria were cross-
result may be due to the number of patients included in the
study.

154
INVESTIGATION ON THE EFFECTS OF FEAR
OF FALLING ON TIME-DISTANCE PARAMETERS
OF GAIT IN ELDERLY WITH TYPE 2 DIABETES
MELLITUS
_xntasx ark1, K. Onbasx2, H. Akkan1
H. Kiloatar1, M. Is
1
Dumlupinar University, physical therapy and rehabilitation,
Kutahya, Turkey
2
Dumlupinar University, Internal Diseases, Kutahya, Turkey

Background and Aims: Diabetes mellitus (DM) is a chronic

metabolic disease which is deterioration of carbohydrate, fat and
protein metabolisms due to lack of insulin secretion or decreased
insulin sensitivity of tissues and requires continuous medical
care. Patients with DM fall more than their peers. The factors that
increase the risk of falls are previous falling tale, decreased lower
extremity function, impaired balance, presence of coronary ar-
tery disease, musculoskeletal pain, depression, decreased visual
A-64 ATTD 2017 E-POSTER VIEWING ABSTRACTS

sectional population-based studies, censuses, surveys or cohorts Conclusions: The presence of IAD in patients with T1DM
which reported the prevalence of diabetes, regardless if type 1 or leads to higher variability of HbA1c values and greater proba-
type 2, in the Brazilian population. We used meta-analysis methods bility of DR development and progress.
to achieve a more precise estimate of the prevalence proportions.
Descriptive analysis of studies was performed.
Results: Between 968 located references, we included 28 157
studies, reporting DM prevalence. Among them, thirteen are PERSONALISED CLINICAL DECISION SUPPORT
population-based studies with national representation and 15 FOR DIABETES MANAGEMENT USING
are studies with local/regional representation. In addition, 15 REAL-TIME DATA
studies were self-reported and 13 relied on blood glucose tests
(BGT) to confirm diagnosis. On self-reported studies, the mean C. Martin1, A. Aldea1, D. Brown1, D. Duce1,
prevalence was 6.25% (95% CI 5.75% - 6.74%; I2 50%) and on J.M. Fernandez-Real2, P. Gay3, P. Georgiou4, R. Harrison1,
studies confirmed by BG, the mean prevalence was 10.05% P. Herrero4, B. Lopez3, Y. Leal2, L. Nita5, P. Pesl4,
(95% CI 6.41% - 13.68%; I2 64%) (Table 1). People (24.0% to R. Petite3, M. Reddy6, J. Shapley7, F. Torrent-Fontbona3,
50.4%) were unaware of their diabetes status. M. Waite8, M. Wos2, N. Oliver6
Conclusions: There is difference between the prevalence 1
Oxford Brookes University, Computing and Communication
stated by self-reported studies and studies confirmed by BGT Technologies, Oxford, United Kingdom
tests. Increase the awareness about DM is a critical need. 2
Institut dInvestigacio Biomedica de Girona Dr. Josep Trueta,
Girona, Spain
3
University of Girona, Electrical Engineering, Electronics and
Automation, Girona, Spain
156 4
Imperial College London Institute of Biomedical
INTERCURRENT AUTOIMMUNE DISEASES Engineering, Dept. of Electrical and Electronic Engineering,
AS FACTOR CONTRIBUTING TO DIABETIC London, United Kingdom
5
RETINOPATHY PROGRESS IN PATIENTS Romsoft SRL, Iasi, Romania
6
WITH TYPE 1 DIABETES MELLITUS: 20-YEAR Imperial College London, Division of diabetes-endocrinology
FOLLOW-UP STUDY and metabolism, London, United Kingdom
7
Cellnovo Ltd, Bridgend, United Kingdom
L. Bolotskaya1, T. Niconova1, V. Lavrenova2, E. Bessmertnaya1, 8
Oxford Brookes University, Faculty of Health and Life
A. Ilin1, A. Zolotychin1 Sciences, Oxford, United Kingdom
1
FGBY, Moscow, Russia
2 Background and Aims: PEPPER (Patient Empowerment
Lomonosov Moscow State University, Biochemistry
Department, Moscow, Russia through Predictive PERsonalised decision support) is an EU-
funded research project to develop a personalised clinical deci-
Background and Aims: Polyautoimmunity can be an essen- sion support system for Type 1 diabetes self-management. The
tial factor contributing to the progress of type 1 diabetes mellitus tool provides insulin bolus dose advice and carbohydrate rec-
(T1DM) and its complications which significance has been un- ommendations, tailored to the needs of individuals. The former is
derestimated. According to this fact we decided to monitor determined by Case-Based Reasoning, an artificial intelligence
DHbA1c values within two cohorts of T1DM patients with and technique that adapts to new situations according to past expe-
without intercurrent autoimmune diseases (IAD). We also rience. The latter uses a predictive computer model that also
compared the frequency and severity of diabetic retinopathy promotes safety by providing glucose alarms, low-glucose in-
within two groups. sulin suspension and fault detection.
Methods: The investigated group included 155 patients with Methods: The user-centered design methodology aims to
T1DM onset occurred in 1994. HbA1c values in every patient were ensure that the tool meets patient needs and improves clinical
measured once in 3 months for 22 years (1994-2016). DHbA1c outcomes. A dual architecture accommodates insulin dosing
values were estimated using the relative frequency method with either by insulin pen or via the Cellnovo patch-pump. Data are
significance level a 0.05. Examination of eye-ground was per- gathered wirelessly in real-time from multiple sources includ-
formed twice a year. ing a continuous glucose monitor, capillary glucose monitor
Results: 25 from 155 monitored patients received a diagno- and physical activity monitor. The design ethos is to offer
sis of IAD. From 25 patients with IAD 15 (60%) had non- maximum benefit for minimum effort, so additional manual
proliferative DR, 8 (32%) had proliferative DR, and 2 patients data entry is strictly limited.
(8%) suffered from unilateral blindness. From 130 patients Results: The first prototype system has been designed, using
without IAD 60 (46%) had nonproliferative DR and 70 had not feedback from patients and clinicians, and tested using the UVA/
any symptoms of DR. Padova Type 1 diabetes simulator. Three subsequent phases of
Average annual oscillations of HbA1c value (DHbA1c, %) in clinical tests are planned. The first two will study safety, feasibility
patients with IAD were (data is presented as min-max scatter): and usability in situ; the last is a randomised control trial, in 2018.
18.3-26.7 in patients with nonproliferative DR, 24.3-30.3 in Conclusions: The first milestones have been reached towards
patients with proliferative DR, 50.4-56.3 in patients with blind- the integration of multiple types of real-time data into a mobile
ness. DHbA1c values, %, in patients without IAD had not sig- decision support system that uses artificial intelligence and pre-
nificant differences and were 7.3-11.4 in patients with dictive modelling to adapt its advice according to the needs of the
nonproliferative DR, 7.2-11.1 in patients without complications. individual.
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-65

158
DIABETIC FOOT 3D ANGIOGRAPHY SETS MAJOR
AMPUTATION RISK IN HEEL WOUND
F. Mercier1,2, B. Herve1, I. Jungfer1, L. Potier3,
M. Kamel3, M.P. Lacomme2, G. Trouillet2, D. Nuytens4,
V. Labbe Gentils5
1
Clinique Internationale Parc Monceau, 75, PARIS, France
2
Hopital Europeen de Paris, Hopital Europeen de Paris,
Aubervilliers, France
3
Hopital Bichat, 75, Paris, France
4
Institut National Podologie, 75, Paris, France
5
Hopital Jean Verdier, 93, Bondy, France

Background and Aims: Heel diabetic foot lesion (H.D.F.L.)

leads to major amputation if the external plantar arch artery
(E.P.A.A.) is missing. Collaterals dont supply this small ter-
minal artery. Revascularisation technics are hopeless.
Osteo-myelitis progress as antibiotics cannot reach the heel,
calcaneum fracture eventually occurs.
3D angiography is a new foot imaging technic.
Methods: Diabetic patients (n = 132) were evaluated for feet
wound during 24 months period, starting September 2014.
In 23 heel wound missing the E.P.A.A. arch artery was sus-
pected.
Mean age 67 (41 89) yld, sex ratio 81 men / 51 women,
diabetes type 2 (n = 99), type 1 (n = 33).
Chronic wound evolution was at least 6 weeks. 159
Wound excision, off loading, adjusted antibiotics, revascular-
isation if possible (not for E.P.A.A.), alginate dry dressing were EFFECT OF BODY MASS INDEX ON DIFFERENT
systematically proposed. GLYCEMIC RESPONSES TO PSYCHOLOGICAL
3D Angiography was performed in Inova GE hybrid cathlab STRESS IN PATIENTS WITH TYPE 1 DIABETES
suite, with 4F introductor, 4F straight catheter, 30 cc of non ionic
B. Ozaslan1, S.D. Patek1, J.H. Grabman2, J.A. Shepard2,
iodine.
E. Dassau3, M. Breton4, Y.C. Kudva5, S.A. Brown4,
Results: 23 patients had heel lesions, for which 18 patients
A. Basu5, J.E. Pinsker6, B.P. Kovatchev4,
had E.P.A.A. but 5 patients missed E.P.A.A.
F.J. Doyle III3, L. Gonder-Frederick7
None patient had bilateral missing E.P.A.A.
1
Conclusions: 3 D angiography is a key evaluation for heel University of Virginia, Systems and Information Engineering,
wound. Center for Diabetes Technology, Charlottesville, USA
2
E.P.A.A. missing leads to amputation in our experience. University of Virginia, Behavioral Medicine Center,
Charlottesville, USA
3
Harvard University, John A. Paulson School of Engineering
and Applied Science, Cambridge, USA
4
University of Virginia, Center for Diabetes Technology,
Past medical
history Charlottesville, USA
wound of bilateral 5
Mayo Clinic, Endocrine Research Unit, Rochester, USA
>6 Calcaneum Calcaneum wound le- 6
age sexe dialysis weeks osteomyelitis fracture sion DECEDED William Sansum Diabetes Center, Pediatric Diabetes Clinical
Service, Santa Barbara, USA
56 f no yes yes yes no no 7
62 h yes yes yes no yes yes University of Virginia, Center for Diabetes Technology,
62 f no yes yes yes yes no Behavioral Medicine Center, Charlottesville, USA
76 h yes yes yes yes yes yes
69 f no yes yes yes no no
Background and Aims: Psychological stress is understood to
be an important consideration in the management of type 1 di-
abetes (T1D), though the factors that contribute to variation in
glycemic responses under stress are largely unknown. This study
investigated the role of body mass index (BMI) on glycemic
control under stress.
Methods: Continuous glucose monitoring (CGM) and insulin
pump (continuous subcutaneous insulin infusion (CSII)) data
were collected from 37 patients with T1D over a one-week pe-
riod, along with meal records and daily self-assessment of psy-
chological stress levels (0 [none] - 4 [extreme]). Daily reported
carbohydrates (CHO), total insulin bolus amounts, and average
blood glucose (BG) from CGM were compared among patients
A-66 ATTD 2017 E-POSTER VIEWING ABSTRACTS

based on (i) BMI and (ii) the presence of stress. In addition, daily
averages of a model-based effectiveness index (quantifying
the combined impact of insulin and carbohydrates on glucose
levels) were compared in these groups.
Results: Figure 1 presents daily carbohydrate intake, insulin
bolus, average blood glucose, and average effectiveness index
for BMI in different ranges (above and below 26), with and
without stress. We observed a trend of reduced average BG in
higher BMI individuals under stress despite trends of increased
carbohydrate intake and reduced insulin dosing. Linear mixed
effects analysis of the effectiveness index demonstrated a sig-
nificant interaction between BMI and stress (p < 0.01), with both
treated as continuous variables.
Conclusions: Preliminary results suggest that the glycemic
effect of daily stress in T1D can be influenced by BMI, specifi-
cally with a possible increase in effectiveness index under stress
with increasing BMI.

160
AWARENESS OF TYPE 2 DIABETES THERAPY GOALS
collected at baseline using the SF-12 (GP n = 188, DSP n = 223) and
AND PERCEPTION OF PATIENTS HEALTH STATUS:
PAID (GP n = 184, DSP n = 220) questionnaires were analyzed.
PRELIMINARY BASELINE RESULTS
Results: Patients enrolled (n = 969) were by majority >55 years
OF THE PDM-PROVALUE STUDIES
of age (GP 82.0%, DSP 77.7%) with >10 years diabetes duration
K. Kulzer1, W. Daenschel2, I. Daenschel3, C. Parkin4, (63.5%, 62.5%). Baseline HbA1c levels, initial insulin therapy and
I. Vesper5, D. Messinger6, J. Weissmann5, L. Heinemann7 presence of comorbidities were similar in both studies. The ma-
1 jority of patients rated their general health status as good to
Forschungsinstitut Diabetes Akademie Bad Mergentheim, excellent (Figure 1A), whereas, about 2/3 (GP 69.0% and DSP
Clinic, Bad Mergentheim, Germany 63.6%) stated they had no or a minor problem regarding the clarity
2
Arztlicher Leiter Medizinisches Versorgungszentrums and concreteness of their diabetes therapy goals (Figure 1B).
am Kuchwald GmbH, Research, Chemnitz, Germany Conclusions: No notable differences were found in health
3
Hausarztpraxis, Clinic, Lunzenau, Germany status of the PDM-ProValue populations at baseline. The results
4
CGParkin Communications - Inc., Boulder City, USA reported as to patients problems regarding awareness of therapy
5
Roche Diabetes Care GmbH, Roche Diabetes Care GmbH, goals emphasize the importance of appropriate agreements be-
Mannheim, Germany tween patients and their physicians. Final results of the PDM-
6
IST GmbH, Biometrics, Mannheim, Germany ProValue studies will be reported in late 2017.
7
Profil Institute for Metabolic Research- Ltd, Research,
Dusseldorf, Germany

Background and Aims: Patients awareness of their diabetes 161

therapy goals is not only an important pre-condition but also an
indicator of quality of medical care. We explored awareness of QUALITY OF OUTPATIENT CARE OF PATIENTS
therapy goals and perception of health status among insulin-treated WITH TYPE 2 DIABETES: PRELIMINARY BASELINE
type 2 diabetes (T2D) patients at 53 general medical practices (GP) RESULTS OF THE PDM-PROVALUE STUDIES
and 47 diabetes-specialized practices (DSP) in Germany. L. Heinemann1, W. Daenschel2, C. Parkin3, I. Daenschel4,
Methods: Two 12-month, prospective, cluster-randomized I. Vesper5, D. Messinger6, J. Weissmann5, B. Kulzer7
studies (PDM-ProValue-GP, PDM-ProValue-DSP) are underway to
1
determine if use of a 6-step approach to personalized diabetes Profil Institute for Metabolic Research, Research, Dusseldorf,
management (PDM) improves glycemic control and other clinical Germany
2
and psychosocial measures. Preliminary self-reported data from Arztlicher Leiter Medizinisches Versorgungszentrums
>400 insulin-treated T2D adults with an enrollment HbA1c 7.5% am Kuchwald GmbH, Research, Chemnitz, Germany
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-67

3
CGParkin Communications, Inc., Boulder City, USA Conclusions: Whereas, no notable difference was found in
4
Hausarztpraxis, Clinic, Lunzenau, Germany HbA1c status of the PDM-ProValue populations at baseline, the
5
Roche Diabetes Care GmbH, Medical Affairs, Mannheim, HbA1c target goal in the DSP study is more ambitious, which is
Germany likely due to patients initial age, HbA1c and insulin therapy
6
IST GmbH, Biometrics, Mannheim, Germany patterns. Final results of the PDM-ProValue studies will be re-
7
Forschungsinstitut Diabetes Akademie Bad Mergentheim, ported in late 2017.
Bad Mergentheim, Bad Mergentheim, Germany

Background and Aims: One pivotal clinical goal of an ef-

fective therapy of type 2 diabetes (T2D) is to optimize patients 162
glycemic control. We explored the long-term blood glucose level BETTER LIFESTYLE TOOLS FOR BETTER HEALTH
(HbA1c), the 12-month target HbA1c and other diabetes treat-
ment measures among insulin-treated T2D patients at 53 general N. Patel1, S. Steed2, S. Manson-Smith3
medical practices (GP) and 47 diabetes-specialized practices 1
Guys and St. Thomas NHS Foundation Trust & Kings
(DSP) in Germany. College London, Diabetes, LONDON, United Kingdom
Methods: Two 12-month, prospective, cluster-randomized 2
Mumoactive, Chief Information Officer, London, United
studies (PDM-ProValue-GP, PDM-ProValue-DSP) are under- Kingdom
way to determine if use of a 6-step approach to personalized 3
Mumoactive, Mumoactive, London, United Kingdom
diabetes management (PDM) improves quality of outpatient di-
abetes care. Preliminary baseline data of 969 enrolled adults with Background and Aims: To develop an app that enables an
an enrollment HbA1c 7.5% were analyzed to describe the initial individual to track their diabetes and to use it as a 2 way com-
study situation as to HbA1c levels, agreement of 12-month HbA1c munication portal with their health care professional.
target goals and utilization of insulin regimens. Methods: Qualitative research to understand the current need
Results: Most patients were >55 years of age (GP 82.0% and for a digital app to support an individual with diabetes and to
DSP 77.7%) with relatively high percentage of patients age >75 then take this information to develop a user friendly app to that
years (GP 21.8%, DSP 13.0%). No notable between-group dif- enables individuals with diabetes to track their diabetes and to
ferences were seen in initial HbA1c levels (Mean: GP 8.50% and enable them to use the same app as a 2 way communication
DSP 8.34%) or insulin therapy, however, the distribution of portal.
BOT/CT vs. ICT differs (Figure 1A). As to the 12-month HbA1c Results: The Needs Identified by both healthcare profes-
target goal, the diabetes specialists tend toward more rigorous sionals and individuals with diabetes
glycemic control compared to GPs (Figure 1B).
- Easy and direct access to patient data
- Direct communication AS NECESSARY between health-
care professional and individual with diabetes to support
self-management
- development of a tool that enables empowerment, inde-
pendence and self-management
- a simple tool for entering values
- being able to write notes
Patients responses to the app
So this is truly a life saver! Thank you, thank you! This
app deserves awards!
Fast easy and convenient
Why hasnt anyone done this before?
Conclusions: Mumoactive is an app that has been designed
based on the needs of both the healthcare professionals who work
within in the NHS and individuals with Diabetes. It enables the
individual to track their values, add notes and messages, syncs
across devices, is fast secure and simple. This tool has the ability
to support self management and enables health care teams to
promote, encourage and support self management.

163
EVALUATION OF AN AUTOMATED WEB-BASED
GLYCAEMIC PATTERN DETECTION TOOL
M.J. Comellas1, J. Weissmann2, N. Casamira1, E. Albinana3,
R. Corcoy4, D. Fernandez-Garca5, J. Garca-Aleman5,
B. Garcia Cuartero6, C. Gonzalez Blanco4, M.T. Rivero7
1
Roche Diabetes Care Spain, Medical Affairs, Sant Cugat del
Valles, Spain
A-68 ATTD 2017 E-POSTER VIEWING ABSTRACTS

2
Roche Diabetes Care Deutschland GmbH, Medical Affairs,
Mannheim, Germany
3
Vithas Hospital Internacional Medimar, Pediatrics, Alicante,
Spain
4
Hospital de la Santa Creu i Sant Pau, Endocrinology,
Barcelona, Spain
5
Hospital Universitario Virgen de la Victoria, Endocrinology,
Malaga, Spain
6
Hospital Universitario Ramon y Cajal, Pediatrics, Madrid,
Spain
7
Complejo Hospitalario de Orense, Endocrinology, Orense,
Spain

Background and Aims: Emminens eConecta is a web-

based platform which includes an automated pattern detection
tool: eDetecta. It uses algorithms and alerts that can be easily
customized by health care professionals (HCP) for the person-
alized management of diabetes mellitus (DM).
The aim of this evaluation was to assess eDetecta performance
in data analysis measured as 1) pattern detection, concordance
with HCP and required time and 2) HCP satisfaction. Methods: Values from Dexcom G4 PLATINUM sensors with
Methods: International multicentre web-based evaluation in Software 505 were compared to temporally-matched reference
which endocrinologists were asked to analyse 4 anonymised re- venous (YSI) values collected during clinic visits on days 1, 4,
ports of real-life data downloads (2CSII and 2MDI) and indicate and 7 of sensor wear. Data from previous studies of 79 children
the presence/absence of up to 19 previously defined glycaemic and adolescents (2262 paired points) and 90 adults (2263 paired
patterns. Responses and required time were recorded and compared points) were used. Matched pairs were mapped to the SEG.
with automated analysis by eDetecta. Questionnaires focused on Results: The Figure shows good agreement between CGM and
data analysis and on eDetecta satisfaction were answered. YSI values for adult and pediatric cohorts on each of the 3 in-
Results: 37 HCP were enrolled and 145 pattern evalua- clinic days. In the pediatric and adult studies, 1990 (88.0%) and
tions were recorded (71CSII; 74MDI). Mean time needed by 2056 (90.9%) of the paired points were in the no risk zone,
HCP per report was 15.1 minutes (12.3-17.8 minutes) vs less than respectively. Performance was best on Day 4, when 92.7% and
1 minute by eDetecta. Concordance between patterns detected by 93.3% of points from the pediatric and adult studies, respectively,
HCPs and eDetecta was >65% in 16 patterns. One of the highest were in the no risk zone. Fewer than 0.5% of the points were in
concordances was observed on Hyperglycaemia in Time Block zones corresponding to moderate or more severe risk levels.
in case ISCI-1 (97.1%). 94.6% of participants considered that Conclusions: SEG analysis suggests that using CGM data
eDetecta was a reliable or very reliable tool, and 78.3% agreed or to guide diabetes management decisions poses minimal clinical
highly agreed it allowed them to make faster decisions. risk. Additional CGM-derived information such as trends and
Conclusions: In the evaluation, eDetecta automated analysis threshold crossings may further mitigate risks of nonadjunctive
was performed in less than 1 minute. Concordance between HCP use.
and eDetecta was good, non-concordant cases requiring further
analysis. HCP indicated it may be helpful in making clinical
decisions, allowing them to dedicate more time to the patient.
165

164 TESTING HEMOGLOBIN A1C IN A PEDIATRIC

OUTPATIENT CLINIC: A FINANCIAL ANALYSIS
RETROSPECTIVE ANALYSIS OF CONTINUOUS
M. Yafi1, B. Cross2, R. Yetman2
GLUCOSE MONITORING DATA WITH THE
1
SURVEILLANCE ERROR GRID SUPPORTS University of Texas at Houston Health Science Center,
NONADJUNCTIVE DIABETES MANAGEMENT Pediatric endocrinology, Houston, USA
2
DECISIONS University of Texas at Houston Health Science Center,
T. Walker1, J. Welsh1, D. Le2, D. Price1 Pediatrics, Houston, USA
1
Dexcom Inc., Clinical Affairs, San Diego, USA Background and Aims: Hemoglobin A1c (HbA1c) is one of
2
Dexcom Inc., Biometrics, San Diego, USA the most common measurements used in the assessment of gly-
cemic control and is thought to reflect the average glycemic
Background and Aims: Performance characteristics of control over past 3 months and predict the occurrence of diabetes
current-generation continuous glucose monitoring (CGM) sys- related complications. Testing HbA1c can be used to screen for
tems may be sufficient for use in diabetes management decisions diabetes as well in population at risk.
without requiring confirmatory self-monitored blood glucose Methods: A cost analysis was performed to compare the price
(SMBG) values. The surveillance error grid (SEG) assigns a of HbA1c kits purchased in our pediatric endocrinology clinic
numerical value and a corresponding risk category to each pair over a period of 2 years and compared to the total collections
of CGM/reference glucose values. To assess the risks resulting from payers within the same time. HbA1c testing was performed
from inaccurate CGM values prompting erroneous diabetes on existing patients with diabetes as well as patients at risk
management decisions, SEG analysis was performed. (obesity, existing symptoms).
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-69

Results: The total cost of HbA1c kits in 2 fiscal years was 167
$14,485.86. The total collections from testing HbA1c in the same
ONLINE-TUNED MODEL BASED COMPOUND
period was $28595. This shows a positive balance of $14,109.14
CONTROLLER FOR BLOOD GLUCOSE REGULATION
(Average of $7,000 per year). When adjusted for personnels time
IN TYPE 1 DIABETES PATIENTS
of handling the kits, preparing the machine daily, testing patients
and reporting results (20% of total time multiplied by total salary A. Bhattacharjee1, A. Easwaran1, M.K.S. Leow2, N. Cho3
of $29,000), the profit is estimated to be around $ 1,200 per year. 1
Conclusions: Testing HbA1c in the clinic setting offers im- Nanyang Technological University, School of Computer
mediate feedback to both the clinician and patients about dia- Science and Engineering, Singapore, Singapore
2
betes control. It can also be a parameter to provide an instant Tan Tock Seng Hospital, Department of Endocrinology,
diagnosis for diabetes when combined with analysis of clinical Singapore, Singapore
3
symptoms and other laboratory tests. Nanyang Technological University, School of Materials
Testing HbA1c during the clinic visit seems to be profitable Science & Engineering, Singapore, Singapore
even after adjusting for personnels expenses.
Background and Aims: A major concern of fully-automated
artificial pancreas system is the prediction of optimal insulin dose
without patients intervention for blood glucose regulation in type 1
diabetics. This paper deals with the design of a fully-automated
166 compound controller based on online-tuned internal model control
ESTIMATION OF OXYGEN CONSUMPTION BASED (IMC) and enhanced IMC (eIMC) along with meal detection
ON EXERCISE PATTERNS FOR PEOPLE WITH TYPE 1 module.
DIABETES Methods: Both IMC and eIMC are developed using Volterra
model without any prior knowledge of patient parameter. Volterra
A. Cinar1, S. Samadi1, N. Frantz2, M. Park3, K. Turksoy2, model estimates patient states online in the presence of unan-
M. Sevil2, I. Hajizadeh1, J. Feng1, C. Lazaro Martinez4, nounced meal disturbances and parameter variations. The time-
L. Quinn3 domain Volterra kernels computed online using recursive least
1
Illinois Institute of Technology, Chemical and Biological squares algorithm are converted into frequency-domain to obtain
Engineering, Chicago, USA Volterra transfer function (VTF). Both IMC and eIMC are devel-
2
Illinois Institute of Technology, Biomedical Engineering, oped using VTF. The compound model based controller is de-
Chicago, USA signed in such a way that eIMC will operate only when the glucose
3
University of Illinois at Chicago, Biobehavioral Health rate increase detector (GRID) of meal detection module is positive,
Science- College of Nursing, Chicago, USA otherwise conventional IMC will take the necessary control action.
4
Illinois Institute of Technology, Electrical and Computer Results: The compound controller is evaluated in the UVA/
Engineering, Chicago, USA Padova metabolic simulator for 10 adult subjects and compared
with IMC using scenarios that are used for validation and ro-
Background and Aims: A new model with two submodels is bustness analysis. Patients experience hypoglycemia only aver-
proposed for estimation of oxygen consumption in aerobic age of 1.81% and hyperglycemia 16.76% of the time using
treadmill and stationary bike exercises, with various patterns of compound controller, when the insulin sensitivity is increased by
exercise intensities. +20% and -20% respectively, whereas these values obtained
Methods: The first submodel uses the speed an d inclination using IMC are 4.2% and 39% respectively.
of the treadmill or the speed and resistance of the bike as the Conclusions: The combination of compound control strategy
inputs to estimate heart rate during the exercise and recovery and meal detection module is able to compensate the short-
time. It is an extension of a nonlinear model [1] with the speed of comings of slow subcutaneous insulin action and delayed peak of
treadmill as its only input. However, changes in grade (inclina- action that causes both hyper- and hypoglycaemia.
tion) of treadmill affects the heart rate remarkably. The second
submodel predicts oxygen consumption based on the heart rate 168
estimated by the first submodel.
Data to build and evaluate the submodels are obtained from REAL-TIME DETECTION OF LOSSES IN INFUSION
exercise studies conducted for young adults with T1D. Subjects SET ACTUATION (LISAS) IN A CLOSED-LOOP
wore Zephyr BioHarness and performed two episodes of 30- ARTIFICIAL PANCREAS
minute training with treadmill and bike at different intensities. D.P. Howsmon1, N. Baysal1, B.A. Buckingham2,
Oxygen consumption and heart rate are measured during car- G.P. Forlenza3, T.T. Ly2, D.M. Maahs2, T. Marcal2, L. Towers3,
diopulmonary treadmill stress test. The data set of eleven sub- S. Deshpande4, R. Gondhalekar4, F.J. Doyle III4, E. Dassau4,
jects are used. J. Hahn1,5, B.W. Bequette1
[1] TM Cheng, AV Savkin, BG Celler, SW Su, L Wang.
1
Nonlinear modeling and control of human heart rate response Rensselaer Polytechnic Institute, Department of Chemical
during exercise with various work load intensities, IEEE Trans. & Biological Engineering, Troy, NY, USA
2
Biomed. Eng., 55: 24992508, 2008. Stanford University, Lucile Salter Packard Childrens
Results: Mean average error of estimated oxygen consump- Hospital, Palo Alto, CA, USA
3
tion for speed, inclination or resistance as model inputs is 0.32 L/ University of Colorado Denver, Barbara Davis Center,
min for training and 0.51 L/min for testing data. Aerobic energy Aurora, CO, USA
4
expenditure is linearly proportional with oxygen consumption. Harvard University, John A. Paulson School of Engineering
Conclusions: The model gives good estimates of energy ex- and Applied Sciences, Cambridge, MA, USA
5
penditure based on time and intensity of aerobic exercise for Rensselaer Polytechnic Institute, Department of Biomedical
people with T1D. Engineering, Troy, NY, USA
A-70 ATTD 2017 E-POSTER VIEWING ABSTRACTS

Background and Aims: As evidence emerges that artificial

pancreas systems may improve clinical outcomes for patients
with type 1 diabetes, the burden of this disease will hopefully
begin to be alleviated for many patients and caregivers. However,
a decrease in burden potentially means patients will be slower to
act when devices stop functioning appropriately. One particular
device failure involves a loss in infusion set actuation (LISA),
where the apparent injected insulin fails to affect the patients
glucose as expected. Alerting patients to these events in real-time
has the potential to reduce hyperglycemia following a LISA.
Methods: A LISA detection algorithm was implemented in a
randomized cross-over study with closed-loop and sensor-
augmented-pump arms on 19 subjects. Each arm lasted two
weeks. Participants wore infusion sets for up to 7 days to provoke
failures. A clinician contacted patients to confirm and repair
LISAs detected by the algorithm. LISAs were determined by (i)
failed correction doses, (ii) ketones >0.6, (iii) insulin leaking at 170
insertion sites, and (iv) pump-occlusion alarms.
CORRELATING THE ESTIMATED INSULIN ACTION
Results: The LISA detection algorithm achieved a sensitivity
PEAK EFFECT WITH PATIENT CHARACTERISTICS
of 85% (n = 27) while issuing only 0.22 false positives per day.
IN THE 670G PIVOTAL TRIAL
Furthermore, the artificial pancreas using zone model predictive
control with LISA detection limited the mean hyperglycemia B. Grosman1, A. Roy1, D. Wu1, N. Parikh1, J. Shin2, S. Lee3,
exposure (> 250 mg/dL) in the four hours preceding a LISA to F. Kaufman4
46.4 minutes, compared to 103.0 minutes during sensor- 1
augmented-pump therapy (p-value 0.01). Medtronic Diabetes, Closed Loop, Northridge, USA
2
Conclusions: As patient burden is reduced by each generation Medtronic Diabetes, Clinical Res Data, Northridge, USA
3
of advanced diabetes technology, fault detection algorithms will Medtronic Diabetes, Clinical Research, Northridge, USA
4
help ensure that patients are alerted when they need to manually Medtronic Diabetes, Medical Affairs, Northridge, USA
intervene.
Background and Aims: The data set from the pivotal trial
for the Medtronic MiniMed 670G hybrid closed loop system
was used to estimate the insulin action peak effect (Teff) for
169 individual subjects using a mathematical model. Teff was then
OPEN TO CLOSED-LOOP TRANSITION SCHEMES correlated with a variety of patient variables and measures of
FOR IN VIVO GLUCOSE CONTROL glycemia.
Methods: Teff was determined in 61 subjects (out of the 124
F. Garelli1, N. Rosales1, E. Fushimi1, H. De Battista1 pivotal trial participants) to be 211 30 minutes (median STD).
1
LEICI, EE-UNLP, La Plata, Argentina In the Table, the Pearsons correlation-coefficient (R) between
Teff and the subjects characteristics is given.
Background and Aims: Continuous Subcutaneous Insulin In- Results: The results show a strong positive correlation be-
fusion (CSII) treatments for people with type 1 diabetes can combine tween Teff and A1C after 3 months of the use of the 670G hybrid
the use of a smart insulin pump with a Continuous Glucose Monitor closed loop system. There was essentially no correlation between
(CGM) and a control algorithm to improve the regulation of the Teff and age, duration of diabetes, or total daily dose of (TDD)
insulin delivery. New control algorithms need to be tested in clinical after 3 months of 670G use.
trials under strictly safety conditions for the patients.
Methods: In this work, a safety layer called the SAFE loop Table. Correlation coefficient (R) between Teff
(Revert et al., IEEE-TBME 2013) is reformulated to be em- and A1C, Diabetes Duration, Total Daily
ployed during clinical trials in two different ways: the Time Dose of Insulin and Age
Enable mode, which gives a criterion for the transition between Measurement R
open and closed-loop therapy in hybrid configurations, and the
Amplitude Enable mode, which diminishes the risk of an hypo- A1C measurement after 3 months 0.40
glycemic event when testing fully closed-loop algorithms by of using the 670G
delimiting the Insulin-On-Board (IOB) profile of the new control
algorithm with the IOB profile corresponding to the usual open- Duration of diabetes -0.17
loop therapy of the patient (or a factor of it). TDD after 3 months of 670G use -0.10
Results: In-silico trials show how using the Time Enable Age 0.07
method the transition from open (bolus) to closed-loop is made
automatically and gradually, resulting in a bumpless mechanism
which improves the controller response. In the case of the Am-
plitude Enable mode, the results show how the patient is protected Conclusions: These data suggest that the speed of insulin
from hypoglycemia even when controller gain is disproportionally action is an indicator of the effectiveness of intensive insulin
large, as shown in the figure below. therapy. Since the direction of the relationship is not known, it is
Conclusions: Two IOB constraint profiles are proposed for its also possible that glucose toxicity can increase Teff. However,
use in clinical trials, giving rise to a safe mechanism for in vivo Teff is likely one of many factors that accounts for insulin
testing both hybrid and fully closed-loop controllers. therapy effectiveness.
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-71

171
PERSONALIZED INSULIN LIMITS - MEDTRONIC
HYBRID CLOSED-LOOP SAFEGUARD DURING
AUTOMATED INSULIN DELIVERY
B. Grosman1, D. Wu1, A. Roy1, N. Parikh1, J. Shin2, S. Lee3,
F. Kaufman4
1
Medtronic Diabetes, Closed Loop, Northridge, USA
2
Medtronic Diabetes, Clinical Res Data, Northridge, USA
3
Medtronic Diabetes, Clinical Research Administration,
Northridge, USA
4
Medtronic Diabetes, Medical Affairs, Northridge, USA

Background and Aims: The algorithm used in the Medtronic

hybrid closed-loop system requires patients to administer insulin
for meals, while the system delivers insulin automatically every
5-minutes. The maximal amount of insulin given automatically
is determined by the algorithm, which adapts every 24 hours. In
early feasibility trials that were conducted to assess the safety of
insulin limits, 8 patients set their pumps to their personalized Background and Aims: Higher glycaemic variability (GV) is
insulin limits and glycemia was evaluated at 4 hours at which associated with severe hypoglycaemia and may contribute to
point there was no hypoglycemia. During the pivotal trial of the development of diabetes complications independent of HbA1c.
Medtronic hybrid closed-loop system, there was a significant Objectives of the current study was to explore the relationship of
decrease in the mean sensor glucose values <50 mg/dL and GV with hypo and hyperglycaemia outcomes during SAP and CL
<70 mg/dL during closed-loop compared to open loop, again insulin delivery.
suggesting that the insulin limits provided a safe constraint on Methods: Post-hoc, exploratory analysis of published, day
automated insulin delivery. The pivotal data suggested that in- and night hybrid CL study. During an open-label, three-centre,
sulin limits could be extended beyond 4 hours. multi-national randomised two-period crossover study, 33 adults
Methods: Extension of the 4 hour insulin limit time-out occurs with T1D (age 40 9years, duration of diabetes 20 9years, and
when the patient measures a blood glucose value. The appearance duration of pump therapy 7.8 5.9years) treated with continuous
of hypoglycemia after resetting insulin limits was evaluated. subcutaneous insulin infusion with HbA1c between 7.5% and
Results: During the trial, there were 628 cases where the in- 10% underwent automated CL and SAP for 12 weeks. GV was
sulin limit time-out was reset; 74% of the cases were automati- assessed using coefficient of variation (CV) of glucose over the
cally resolved in <60 minutes (total time delivering at insulin entire three-month period.
limit between 4-5 hours) and 12% continued on insulin limits for Results: During SAP, GV was positively associated with time
a total time of 6 to 8 hours. There was not a single case of spent in hypoglycaemia (below 3.9mM: r = 0.6, p < 0.001; below
hypoglycemia after the extension of insulin limits. 3.5mM: r = 0.6, p < 0.001; below 2.8mM: r = 0.6, p < 0.001). In
Conclusions: This analysis suggests that insulin limit contrast, there was no relationship between GV and any level of
time-out can likely be extended without increasing the risk of hypoglycaemia during CL insulin delivery (figure). Higher GV
hypoglycemia. during SAP was associated with larger reduction in hypogly-
caemia <3.9mM (r = 0.4, p = 0.013), as well as smaller reduction
of hyperglycaemia >10mM during CL (r = 0.6, P = 0.001). There
was no difference in GV between SAP and CL (39 4% vs.
172 39 4%, p = 0.5).
Conclusions: Higher glycaemic variability during SAP is
RELATIONSHIP BETWEEN GLUCOSE VARIABILITY
strongly associated with more time spent in hypoglycaemia
AND GLYCAEMIC CONTROL DURING SENSOR
and larger reduction in time spent in hypoglycaemia during
AUGMENTED PUMP THERAPY (SAP) AND CLOSED-
CL. The association of GV to time spent in hypoglycaemia was
LOOP (CL) INSULIN DELIVERY IN ADULTS
not observed during CL. Further work is required to understand
WITH TYPE 1 DIABETES (T1D)
the clinical relevance of these observations.
L. Leelarathna1, H. Thabit2, S. Hartnell3, M. Willinska2,
S. Dellweg4, C. Benesch4, J. Mader5, M. Holzer5, M. Evans2,
S. Arnolds4, T. Pieber5, R. Hovorka2
1 173
Central Manchester University Hospital NHS Foundation
Trust, Manchester Diabetes Centre, Manchester, A NOVEL MEAL DETECTION ALGORITHM
United Kingdom FOR AN ARTIFICIAL PANCREAS
2
University of Cambridge, Wellcome Trust-MRC Institute C. Ramkissoon1, P. Herrero2, J. Bondia3, J. Vehi1
of Metabolic Science, Cambridge, United Kingdom
3 1
Cambridge University Hospitals NHS Foundation Trust, Informatica i Aplicacions, Enginyeria Electrica Electronica i
Department of Diabetes & Endocrinology, Cambridge, Automatica, Girona, Spain
2
United Kingdom Centre for Bio-Inspired Technology, Biomedical Engineering,
4
Profil, Profil, Neuss, Germany London, United Kingdom
5 3
Medical University of Graz, Department of Internal Medicine, Automatica e Informatica Industrial, Ingeniera de Sistemas y
Graz, Austria Automatica, Valencia, Spain
A-72 ATTD 2017 E-POSTER VIEWING ABSTRACTS

Background and Aims: Many studies have reported that a Results: Assuming constant variables in a subject across the
high number of missed meal boluses occur, especially in adoles- four visits and using thresholds of 15% MPE and 20% MAPE,
cents during insulin pump therapy. It is predicted, that this behavior we accepted at least one and at most four PD model test-fits per
will carry over to artificial pancreas therapy and therefore, a means subject. Thus, we successfully validated the PD model by 4-fold
to reduce poor outcomes due to announced meals must be im- leave-one-out cross-validation.
plemented. The aim is to implement an algorithm to detect meals Conclusions: The PD model accurately simulates glucose
using data from a continuous glucose monitor (CGM) and the excursions based on plasma insulin and glucagon concentrations.
insulin delivered to the subject. The reported PK/PD model including equations and fitted pa-
Methods: An Unscented Kalman Filter is employed to predict rameters allows for in silico experiments and were used for in-
the states of a composite Bergman-Hovorka model altered to vestigations of future study design. Simulation studies may help
include an auxiliary disturbance parameter. Then, an algorithm improve diabetes treatment involving glucagon for prevention of
checks the cross-correlation between the disturbance parameter hypoglycemia.
and continuous glucose monitor levels and employs a threshold to
detect an abnormal event. At this time point, a positive distur-
bance parameter value indicates a rise in glucose due to a meal.
This methodology was tested using meals simulated in silico with 175
10 adult patients over a period of ten days (30 meals per subject). INTEGRATION OF SCREENING, MONITORING
Results: Carbohydrate amounts tested were 85 17 g. The true AND FOLLOW UP TOOLS FOR EFFECTIVE FOOT
positive rate (sensitivity) and false positive rate were 90 12.7% CARE: EXPERIENCE FROM A DIABETES CARE
and 5 1.8%, respectively. The accuracy and specificity were CENTRE
94 1.5%, and 95 1.8%, respectively. The change in glucose
experienced at detection was 4 14 mg/dl and the detection time D. A. Padhye1
was 28 3 min. 1
Conquer Diabetes, Foot, Mumbai, India
Conclusions: This algorithm provides meal detection with
minimal change in blood glucose levels. When used in con-
Background and Aims: The diabetic patients in India are
junction with a postprandial control strategy, it may improve
routinely screened for target end organ damage which includes
postprandial outcomes.
retinopathy, cardiomyopathy, and nephropathy. Currently the
practice does not involve routine check for early diabetic neu-
174 ropathy and vasculopathy. The prevalence of diabetic neuropa-
VALIDATION OF A SIMULATION MODEL thy is low in Indian population; we believe this could be because
DESCRIBING THE GLUCOSE-INSULIN-GLUCAGON of lack of active early screening of asymptomatic patients.
PHARMACODYNAMICS IN PATIENTS WITH TYPE 1 Methods: Our retrospective study of 1500 patients over a
DIABETES period of 10 months looked for diabetic foot related complica-
tions. This was done by using Biothesiometer to assess vibration
S.L. Wendt1, A. Ranjan2, J.K. Mller3, S. Schmidt2, perception for neuropathy, Doppler to measure ankle brachial
C.B. Knudsen1, J.J. Holst4, S. Madsbad2, H. Madsen3, index (ABI) for vasculopathy and Podiascan for abnormal planter
K. Nrgaard2, J.B. Jrgensen3 pressure points. The patients were educated about foot care and
1
Zealand Pharma A/S, Bioanalysis and Pharmacokinetics, appropriate treatment was initiated
Glostrup, Denmark Results: An abnormal vibration perception was present in
2
Copenhagen University Hospital Hvidovre, Endocrinology, 46.5% (n = 698/1500) of study population, with 25.9% of patients
Hvidovre, Denmark had severe impairment of vibration sense. The Ankle Brachial
3
Technical University of Denmark, Applied Mathematics and Index (ABI) of >1.3 in right leg, left leg and bilaterally was re-
Computer Science, DTU Compute, Kongens Lyngby, Denmark corded in 70, 42 and 40 patients respectively. Twenty seven pa-
4
University of Copenhagen, Faculty of Health and Medical tients recorded midfoot collapse and 338 patients had 1stmetatarsal
Sciences, Copenhagen, Denmark high pressure points. The follow up was maintained to look for
development of complication
Background and Aims: Currently, no consensus exists on a Conclusions: The study demonstrates the substantial burden
model describing endogenous glucose production (EGP) as a of neuropathy and vasculopathy. An integrated approach that
function of glucagon concentrations. Reliable simulations to de- includes neuropathic and vascular screening along with other
termine the glucagon dose preventing or treating hypoglycemia or routine screenings will help achieve an objective of detecting
to tune a dual-hormone artificial pancreas control algorithm need a early diabetes related complications. The early intervention,
validated glucoregulatory model including the effect of glucagon. education, treatment and follow up tools are important for an
Methods: Eight type 1 diabetes patients each received a effective preventive management.
subcutaneous (SC) bolus of insulin on four consecutive days to
induce mild hypoglycemia followed by a SC bolus of saline or
100, 200 or 300 lg of glucagon. Blood samples were analyzed 176
for concentrations of glucagon, insulin and glucose. We fitted
pharmacokinetic (PK) models to insulin and glucagon data using SELECTED FACTORS ASSOCIATED WITH THE
maximum likelihood and maximum a posteriori estimation SENSE OF RESPONSIBILITY FOR THE HEALTH
methods. Similarly, we fitted a pharmacodynamic (PD) model to IN PATIENTS WITH TYPE 2 DIABETES
glucose data. The PD model included multiplicative effects of M. Jaworski1, M. Adamus1
insulin and glucagon on EGP. Bias and precision of PD model
1
test-fits were assessed by mean predictive error (MPE) and mean Medical University of Warsaw, Department of Medical
absolute predictive error (MAPE). Psychology, Warsaw, Poland
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-73

Background and Aims: Preliminary studies have shown a

significant role sense of responsibility for the health in patients
with diabetes in term of health-related behaviour. Responsibility
for health construct is defined as self-perceptions relating to
motivation and use of various behavioral strategies to keep good
health. Two aspects characterize this variable: active involve-
ment and adequate behavior. Aim of this study was analyze se-
lected factors which could influence on the level of sense of
with T1D wore an investigational device and continuous glucose
responsibility for health in patients with diabetes.
monitoring (CGM; Dexcom G4) concurrently (both masked) at
Methods: The study was conducted among patients with
home for 14 nights. A data subset was used to train the ANN; the
diabetes aged between 18 and to 71. Three research tools were
remaining data were then used to assess the efficacy of the device
used in the study: Health Suggestibility Scale (HSS), Health
to detect NH (<4.0 mmol/L).
Sense of Responsibility Scale (HSRS), Life Orientation Test
Results: Fifty adults (24 women; 12/50 Gold score 5;
(LOT-R). The number used drugs, the current age, body weight
mean SD age 46 14.8 years; BMI 27.8 5.5 kg/m2; T1D du-
and growth were analyzed too. The method of structural
ration 24.0 11.2 years; HbA1c 7.8 1.0%) participated. Of 714
equation modeling (SEM) was using to develop models ana-
study nights, n = 187 did not meet pre-specified quality criteria
lyzed the relationship between psychological characteristics
and were excluded, n = 401 were used to train the ANN algo-
(suggestibility and optimism) and sense of responsibility for
rithm, and n = 126 were used to assess efficacy. Analysis during
health in diabetic patients with complications and accompa-
the training phase indicated that micro-tremor at 19Hz and 5Hz
nying diseases.
had greatest sensitivity for NH detection. Other frequency ranges
Results: The SEM analysis showed that there are four main
modulated this relationship. There was negligible association
factors which influence on the formation of sense of responsi-
between skin temperature and NH. During 31/126 (24.6%) as-
bility for the health in diabetic patients with complications and/or
sessment nights, CGM glucose reached <4.0 mmol/L; see Table
accompanying diseases. There are: health suggestibility, dispo-
for device alerts by glucose nadir. No alerts occurred on 77/95
sitional optimism, duration of therapy and number of used drugs.
(81.1%) nights when CGM remained >4.0 mmol/L.
Conclusions: The sense of responsibility for the health is a
Conclusions: Increasing sensitivity for NH detection paral-
very important factor in work with patients with diabetes. These
leled the glucose nadir, supporting a physiological basis for the
results could have a clinical implications. It could be used to
association with micro-tremor. This non-invasive prototype de-
create a new therapys form in context of the biopsychosocial
vice shows initial promise for diabetes management, though
model.
further refinement is required.

177 178
THE FEASIBILITY AND EFFICACY OF EMPLOYING AN EARLY INFECTIOUS DISEASE OUTBREAK
MICRO-TREMOR AND SKIN TEMPERATURE DETECTION SYSTEM BASED ON SELF-RECORDED
TO ALERT FOR NOCTURNAL HYPOGLYCAEMIA DATA FROM PEOPLE WITH DIABETES
IN ADULTS WITH TYPE 1 DIABETES
A.Z. Woldaregay1, T. Chomutare2, D. Albers3, L. Mamykina3,
1 2 3 3 3
S.A. McAuley , N. Cohen , W. Fifield , A. Ratcliff , M. Smith , O. Hejlesen4, C. Lovis5, G. Demiris6, P.J. Toussaint7,
B. Robinson4,5, A.J. Jenkins1,6,7, D.N. ONeal1,7 C.F. Basse8, A. Horsch1,9, F. Godtliebsen2, T. Botsis1,
1 E. Arsand2,10, G. Hartvigsen1,2
St Vincents Hospital Melbourne, Department
1
of Endocrinology and Diabetes, Melbourne, Australia UiT The arctic University of Norway, Computer Science,
2
Baker-International Diabetes Institute, Diabetes, Melbourne, Troms, Norway
2
Australia Norwegian Centre for E-health Research, University Hospital
3
Grey Innovation, Research and Development, Melbourne, of North Norway, Troms, Norway
3
Australia Columbia University, Biomedical Informatics, New York, USA
4 4
Royal North Shore Hospital, Department of Endocrinology, Aalborg University, Department of Health Science
Sydney, Australia and Technology, Aalborg, Denmark
5 5
Royal North Shore Hospital, Kolling Institute of Medical University of Geneva & Geneva University Hospitals, Medical
Research, Sydney, Australia Informatics, Geneva, Switzerland
6 6
University of Sydney, NHMRC Clinical Trials Centre, Sydney, University of Washington, Biomedical Informatics
Australia and Medical Education, Seattle, USA
7 7
St Vincents Hospital Melbourne, University of Melbourne NTNU, Department of Computer and Information Science,
Department of Medicine, Melbourne, Australia Trondheim, Norway
8
University of Bergen, Philosophy, Bergen, Norway
9
Background and Aims: To determine if nocturnal hypogly- TU Munchen, Medical Statistics and Epidemiology,
caemia (NH) can be detected in type 1 diabetes (T1D) using Munchen, Germany
10
changes in micro-tremor and skin temperature as inputs into a UiT The arctic University of Norway, Clinical Medicine,
prototype device. Troms, Norway
Methods: The investigational wrist- and finger-worn device
(Firefly Health, Australia) combines accelerometers, gyroscopes, Background and Aims: During infections, many people with
and skin and environmental temperature sensors. Data were diabetes experiences high blood glucose (BG) levels. We pro-
processed by an artificial neural network (ANN). Participants pose to utilize patient-gathered information and develop an
A-74 ATTD 2017 E-POSTER VIEWING ABSTRACTS

179
CGM-BASED GLYCEMIC VARIABILITY INDICES
ALLOW ACCURATE CLASSIFICATION OF IGT
AND T2D SUBJECTS
E. Longato1, G. Acciaroli1, A. Facchinetti1, A. Maran2,
G. Sparacino1, L. Hakaste3, T. Tuomi3, C. Cobelli1
1
University of Padova, Information Engineering, Padova, Italy
2
University of Padova, Department of Medicine, Padova, Italy
3
University of Helsinki, Diabetes and Obesity Research
Program, Helsinki, Finland

Background and Aims: Tens of Glycemic Variability (GV)

indices are available in the literature to characterize the dynamic
properties of glucose concentration profiles measured by CGM
sensors, in both healthy and disease states. Whether GV indices
are usable for classifying patients is, however, still controversial.
Recently (Acciaroli et al., Diab Tech Meeting, 2016), we dem-
onstrated that for the quite simple task of distinguishing healthy
from and IGT/T2D subjects, a pool of 25 CGM-based GV indices
can be successfully used. Here, we move a step forward by as-
sessing if a GV indices-based classifier can further distinguish
IGT from T2D subjects.
Methods: The dataset consists of 62 subjects classified by the
OGTT test, 36 IGT and 26 T2D subjects. Each subject was mon-
itored by either the Guardian Real Time or the iPro CGM systems
(Medtronic MiniMed, Inc., Northridge, CA), from which 37 GV
indices were extracted. In addition to GV, age, sex, BMI and waist
circumference were also used. To classify each patient a Support
Vector Machine classifier with polynomial kernel was developed.
Results: IGT subjects were distinguished from T2D subjects
with 87.1% accuracy. The addition of some basic clinical pa-
rameters to GV indices improved performance by 12.9%.
Conclusions: GV indices, combined with basic clinical pa-
rameters, can be used to accurately distinguish IGT from T2D
patients. Further work will concern the extension of the database
and the identification of the minimal set of GV indices needed for
Electronic Disease Surveillance Monitoring Network, which can classification purposes.
possibly detect infectious disease outbreaks during incubation
period. The system incorporates data gathered from diabetes
diaries, CGM devices, and other appropriate physiological in- 180
dicators from people with type 1 diabetes. The system will detect
a cluster of people with elevated BG levels on a spatio-temporal CHOLINE DEPRIVATION MODIFIES CRUCIAL
basis. In this regard, we aim to develop mathematical models on HEPATIC MARKERS OF THE INSULIN SIGNALING
individual level that can track and detect any unreasonable BG PATHWAY IN STREPTOZOTOCIN-INDUCED
deviations. DIABETIC RATS
Methods: The models combine a novel approach for BG
H. Al-Humadi1
tracking and outlier detection, which is based on set of auto-
1
regressive models and predicts the expected BG intervals for an Babylon University/Pharmacy College, Pharmacology
individual. The actual BG value is compared against the predicted and Toxicology, Hilla, Iraq
intervals, which is generated using auto-regressive and Auto-
regressive moving average methods. The models were developed Background and Aims: The evaluation of induction of
using one-month BG data, sampled using Dexcom CGM devices, choline-deprivation CD in rats model with or without diabetes
based on two type 1 diabetes patients. DM depending on the changes of hepatic enzymes & insulin
Results: The models were capable of accurately monitoring signaling pathway.
individuals normal BG fluctuations and detecting statistically Methods: A 30-day dietary CD with or without DM induced
significant BG deviation. However, these models were found to by intraperitoneal streptozotocin administrated on adult male
be affected by the variability of the individuals BG evolution, rats; the assessment of changes in insulin level IN, blood sugar
quality of diabetes self-management behavior, which needs to be BS and gene expression GEX of some hepatic insulin signalling
taken into account in the next phase of the study. pathway (IR, IRS-1, Glut-2 except Glut-4) in addition to the
Conclusions: We demonstrated a novel model for using BG liver histopathological HST sections, enzymes (ALT, AST, ALP,
levels as source of information for a disease surveillance purpose. c-GT) and lipid profile (TC,TG, HDL, LDL).
Further analysis and results will be presented at ATTD 2017 re- Results: In combination group, there was a significant in-
garding the developed models, next steps, and challenges. crease in BS with a significant decrease in IN; serum liver
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-75

enzymes were demonstrated a significant increase except in ALT. Conclusions: The sensors possibly measure the flux of glu-
Non-HDL lipid profile assay revealed also a significant increase cose more closely to the peritoneal membrane and thereby correlate
while HDL revealed no significant changes in comparison with better to the blood glucose excursions than glucose measured in
control. HST examination of liver sections revealed micro- and samples of peritoneal fluid.
macrovesicular steatosis with mild degenerative changes. The Whether measurements of glucose in the upper abdomen
mRNA GEX of IR level in rat hepatic tissue was significantly correlate better to blood glucose fluctuations than measurements
increased. On the other hand, IRS-1 mRNA GEX level was im- in the lower abdomen, will be one of the questions investigated in
paired as well as in DM group. For glucose transporters, GLUT 2 future experiments.
mRNA GEX level revealed a significant increasing but GLUT 4
mRNA GEX was reacted positively and significantly more earlier. 182
Conclusions: Our new design of combined CD with DM
explore that CD increase the sensitization of the hepatic IN LABORATORY AND CLINICAL SAMPLE
signaling pathways by minimizing the hyperglycemic HYG re- REAPPLICATION STUDIES: PERFORMANCE
sponse with an additional anti-diabetic effect as augmenting the AND ACCURACY CAPABILITY OF A NEW,
inhibitory action of HYG on GLUT-2 but with hepatotoxic effect WIRELESS-ENABLED BLOOD GLUCOSE
and progressive dyslipidemia. MONITORING SYSTEM THAT LINKS
TO A SMART MOBILE DEVICE
B. Harrison1, D. Brown1, M. Takeshima2
181 1
Ascensia Diabetes Care, Parsippany NJ, USA
INTRAPERITONEAL GLUCOSE MEASUREMENTS 2
Panasonic Healthcare Co. Ltd., Toon City Ehime, Japan
USING AMPEROMETRIC GLUCOSE SENSORS
AND INTRAPERITONEAL FLUID SAMPLES Background and Aims: To validate ContourPlus ONE
IN AN ANIMAL MODEL blood glucose monitoring system (BGMS) performance and
M.K. Am1,2,3, A.L. Fougner2,3,4, I. Dirnena-Fusini1,2, accuracy capability during Second-Chance sampling. This new
P.C. Bosch2,4, K. Kolle2,3,4, D.R. Hjelme2,5, R. Ellingsen2,5, BGMS features an easy-to-use, wireless-enabled BG meter that
. Stavdahl2,4, S.M. Carlsen1,2,6, S.C. Christiansen1,2,6 links to a smart mobile device via Bluetooth technology and
syncs with the Contour Diabetes app.
1
Norwegian University of Science and Technology NTNU, Methods: In the laboratory study, testing was conducted at 3
Department of Cancer Research and Molecular Medicine, temperatures (16C, 22C, 34C) with blood adjusted to 3 BG
Trondheim, Norway levels (70, 300, 500 mg/dL) at 3 hematocrit levels (20%, 42%,
2
Norwegian University of Science and Technology NTNU, 55%). Two sample reapplication methods were used (initial
Artificial Pancreas Trondheim APT, Trondheim, Norway volume, 0.28 and 0.46 lL); each sample was tested with 3 delay
3
Central Norway Regional Health Authority, RHA, Stjrdal, times between initial and second inoculation (5, 30, 55 seconds).
Norway For each sample and condition, 10 replicate BGMS readings
4
Norwegian University of Science and Technology NTNU, were obtained with each of 3 test strip lots. In the clinical study,
Department of Engineering Cybernetics, Trondheim, Norway 52 subjects with diabetes performed self-tests with an inten-
5
Norwegian University of Science and Technology NTNU, tionally insufficient initial blood application to produce a number
Department of Electronics and Telecommunication, Trondheim, of Second-Chance sampling opportunities. Results were com-
Norway pared with YSI reference results and assessed per the follow-
6
St. Olavs University Hospital, Department of Endocrinology, ing acceptance criterion: 95% of results within 15 mg/dL (BG
Trondheim, Norway <100 mg/dL) or 15% (BG 100 mg/dL) of reference result.
Results: In the laboratory study, results met protocol-specified
Background and Aims: Fast, accurate and reliable glucose acceptance criteria. In the clinical study (BG range, 65-347 mg/
sensing is crucial for a well-functioning artificial pancreas (AP). dL; hematocrit range, 35%-55%), 97.7% (84/86) of subject fin-
The glucose dynamics in the peritoneal space are faster than in the gertip self-test results were within 15 mg/dL or 15% of YSI
subcutaneous tissue, making it a preferable sensing site for an AP. reference result. Moreover, 90.7% (78/86) of self-test results were
The peritoneal space also offers advantages by being less within 10 mg/dL or 10% of YSI reference result.
influenced by temperature variations and mechanical forces re- Conclusions: The BGMS sample reapplication results met
ducing vascular perfusion, thus potentially providing a more acceptance criteria in the laboratory and in a clinical setting used
stable environment for glucose sensing. by subjects with diabetes.
Methods: Two anaesthetised, non-diabetic pigs (27.0 and
31.4 kg) were implanted with four glucose sensors (Abbott 183
Freestyle Libre Flash), one in each quadrant of the peritoneal
space, and two sensors of the same type placed subcutaneously IMPACT OF SENSOR-AUGMENTED INSULIN PUMP
on the abdominal wall. Meal glucose excursions were simulated THERAPY ON GLYCAEMIC CONTROL IN CHILDREN
by intravenous glucose infusions. Frequent blood and intraperi- AND ADULTS WITH TYPE 1 DIABETES: A CLINICAL
toneal (IP) fluid samples were drawn and analysed in a blood gas PRACTICE EXPERIENCE IN SPAIN
analyser (BGA) (Radiometer ABL 725) for comparison.
P.I. Beato-Vibora1, E. Gil-Poch2, L. Galan-Bueno2,
Results: Intravenous glucose challenges gave a rise in IP glu-
F. Morales-Perez1, F.J. Arroyo-Dez2
cose levels, measured with both amperometric sensors and BGA.
1
A potential difference in glucose dynamics was observed Badajoz University Hospital, Department of Endocrinology,
between IP fluid samples from the upper (cranial) part compared Badajoz, Spain
2
to the lower (caudal) part of the peritoneal space, but this was not Badajoz University Hospital, Department of Paediatrics,
apparent in the corresponding amperometric sensor signals. Badajoz, Spain
A-76 ATTD 2017 E-POSTER VIEWING ABSTRACTS

Background and Aims: SAPs outcome depends on patients

being educated in using RT-CGM data. We analysed the effect
on glycaemic variability of a specific training path.
Methods: 21T1DM patients (16M/21F, age 42.2 1.3 years;
disease duration 19.0 9.4years, starting 640G therapy by 5.8
4.3months) attended a structured SAP education (individual and
collective) focused on the correct use and interpretation of CGM
data and therapy adjustment. We analysed HbA1c and CGM data
considering the date of the course as Day 0, the whole follow up
(from Day60 to Day120) and subintervals (Day30 Day 0),
(Day0 Day30), (Day30 Day60) and (Day60 Day90). We
evaluated: Mean of Sensor Glucose (SG); Coefficient of Varia-
tion (CV); Conga-1; Conga-2; Conga-4; J-index; M-value and
Lability Index.
Results: In the whole follow up (-60 - 120) the time effect
has a significant descending trend for the following variables:
SGM; Conga-4M; J-IndexM and M-valueM and for SGM;
CVM; Conga-1M and LIM in the interval (60 90). The
comparison of the median values of indexes calculated for each
patient in the interval (from -30 to 0) vs all subsequent, were
not statistically significant. The Means Number of Suspen-
sions, in the whole follow up has a significant increasing. The
Background and Aims: To evaluate the effect of sensor- median values of HbA1c, before, 1 month and 3 months after
augmented pump therapy (SAP) on HbA1c in routine clinical the course showed a trend improvement that was not statisti-
practice in adult and paediatric patients with type 1 diabetes cally significant.
(T1DM). Conclusions: These results suggest how a structured educa-
Methods: We performed a retrospective evaluation of the tional intervention is important to acquire competences. The
effect of SAP on glycaemic control in all T1DM patients treated significant lack for subintervals suggests this process is slow and
in the adult and paediatric Diabetes Departments at a referral continuous and it depends on the basic preparation of patients.
hospital in Badajoz (southwestern Spain).
Results: 85 T1DM patients (14 patients 18 years-old), age
35 14 (2-61), 62% female, diabetes duration 21 12 years,
were on SAP (53 Medtronic-MiniMed-640G with predic- 185
tive low-glucose suspend feature, 29 Medtronic-MiniMed- SET-UP AND PROCEDURE FOR INTRAPERITONEAL
Paradimg-VEOTM with low-glucose suspend feature, 3 GLUCOSE MONITORING IN ANAESTHETISED
Animas vibe+DexcomG4). Median time on SAP was 16 ANIMALS
months (range 1-153). 35 patients were using pumps and self-
monitoring of blood glucose (SMBG) before starting SAP, 8 P.C. Bosch1,2, M.K. Am2,3,4, . Stavdahl1,2, R. Ellingsen2,5,
were on multiple daily insulin ( MDI) injections and sensors, A.L. Fougner1,2,4, S.M. Carlsen2,3,6, D.R. Hjelme2,5
39 were on MDI and SMBG and 3 children had started SAP at 1
Norwegian University of Science and Technology NTNU,
diabetes onset (< 20 months-old). Percentage of patients with
Department of Engineering Cybernetics, Trondheim, Norway
HbA1c 7% increased from 35% to 54% at the end of follow- 2
Norwegian University of Science and Technology NTNU,
up (p = 0.001). HbA1c at end of follow-up was significantly
Artificial Pancreas Trondheim, www.apt-norway.com,
lower than HbA1c before the start of SAP in the whole group
Trondheim, Norway
(7.5 0.9% vs 7.0 0.7%, p = 0.001) and in the group with 3
Norwegian University of Science and Technology NTNU,
baseline HbA1c >%7 (8.0 0.1% vs 7.3 0.1%, n = 52,
Department of Cancer Research and Molecular Medicine,
p = 0.001). In this group, 81% of the patients improved their
Trondheim, Norway
HbA1c and end-of-follow-up Hba1c was significantly lower 4
Helse Midt-Norge, Central Norway Regional Health Authority,
when different groups of age; time on SAP and previous and
Stjrdal, Norway
current treatments were analysed (Table 1). 5
Norwegian University of Science and Technology NTNU,
Conclusions: Sensor-augmented pump therapy provides a
Department of Electronics and Telecommunication, Trondheim,
sustained improvement in glycaemic control in real-life clinical
Norway
practice in children and adults with T1DM. 6
St. Olavs Hospital, Unit for applied clinical research,
Trondheim, Norway

Background and Aims: The peritoneum is a promising lo-

184 cation for an artificial pancreas (AP). To develop an intraperi-
toneal (IP) AP, it is crucial to understand the dynamics and
PATIENTS EDUCATION FOR USING RT-CGM DATA distribution of IP glucose. To verify the performance of proto-
IN SAP THERAPY MANAGMENT types and map IP glucose dynamics two ways of measuring IP
A. Girelli1, S. Bonfadini1, E. Zarra1, U. Valentini1 glucose were developed.
Methods: Two proposed ways of obtaining IP glucose data
1
UO Diabetologia, ASST Spedali Civili, Brescia, Italy have been developed and tested on two male pigs (27.0 &
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-77

31.4kg). One procedure is continuous and uses a commercially meters. Sensor performance was assessed by the daily median
available, wireless, amperometric glucose sensor (Abbott Free- absolute relative difference (ARD). Sensor failures were defined
style Libre Flash (FLF)) in combination with an open source as primary sensor failure (??? for more than two hours or per-
readout unit (LimiTTer by JoernL @ GitHub), which reads sistent 20% difference between MBG and SG) or failure of the
and transmits the sensor data to an Android app (xDrip by sensor adhesive.
stephenblackwasalreadytaken @ GitHub), from where it was Results: Sensors were worn for a median (IQ range) of
downloaded for further processing. The other is a manual, dis- 14.8 days (12.0-19.6). 90% were functioning at end of week 1,
crete sampling, where IP-fluid samples are taken with a spe- 45% at end of week 2, and 20% at the end of week 3. The Median
cially designed device that allows sampling from the location of ARD for week 1 was 10%, for week 2 was 11%, and for week 3
the FLF. These samples are then analysed with a blood gas was 13%. The figure provides the overall sensor survival curve,
analyser (Radiometer ABL 725). with curves for sensor failure due to adhesive failures and sensor
Results: Both sampling procedures worked reliably al- failures. All but one sensor failure was due to ??? message on
lowing a new and unique glucose value to be acquired every the sensor for more than 2 hours.
minute. Conclusions: This study showed a significant decrease in
Conclusions: The quality of the preliminary data obtained sensors surviving into the third week of wear. Most failures were
seems good; however, further analysis is needed to confirm due to two hours of ???. Sensor accuracy was slightly de-
this. These novel ways of monitoring IP glucose allow com- creased in the third week.
parison of a continuous measurement device to a gold standard.
Once the reliability of the IP FLF is verified, FLFs may be
utilised for future trials. This might provide a relatively inex-
pensive way for medium- or long-term continuous IP glucose 187
monitoring in animals.
USER PERFORMANCE EVALUATION OF A NEW
WIRELESS-ENABLED BLOOD GLUCOSE
MONITORING SYSTEM THAT LINKS TO A SMART
186 MOBILE DEVICE IN SUBJECTS WITH AND WITHOUT
DIABETES
CONTINUOUS GLUCOSE MONITOR SENSOR
SURVIVAL AND ACCURACY OVER 21 CONSECUTIVE M. Christiansen1, C. Greene2, S. Pardo2, R. Morin2, T. Bailey3
DAYS OF WEAR 1
Diablo Clinical Research, Walnut Creek CA, USA
2
D. Tinti1, T. Marcal2, L. Towers3, J. Doiev2, T. Ly1, L. Messer3, Ascensia Diabetes Care, Parsippany NJ, USA
3
G. Forlenza3, D. Maahs3, B. Buckingham1 AMCR Institute Inc., Escondido CA, USA
1
Stanford University, Pediatric Endocrinology, Palo Alto, USA
2 Background and Aims: To evaluate accuracy of the new
Stanford, Pediatric Endocrinology, Palo Alto, USA
3 ContourNext ONE blood glucose monitoring system (BGMS)
University of Colorado Denver, Pediatric Endocrinology,
in clinical settings used by subjects with and without diabetes.
Aurora, USA
The BGMS features an easy-to-use, wireless-enabled blood
glucose meter that links to a smart mobile device via Blue-
Background and Aims: As part of a study to develop algo- tooth technology and can sync with the Contour Diabetes
rithms to detect sensor failure while in closed-loop we asked app.
subjects to wear a sensor for 3 weeks. This abstract describes the Methods: This 2-center clinical study enrolled 375 subjects
causes of sensor failure and sensor accuracy during this 21 day with (n = 332) or without (n = 43) diabetes, who had never
study. used this BGMS. Secondary objectives included accuracy per
Methods: A total of 20 subjects (mean age = 28 years) wore FDA Draft SMBG Guidance 2014 Section C for all subjects
Dexcom G4 sensors for 3 weeks in an outpatient setting on 40 with or without diabetes (ie, 95% of results within 15% and
occasions. Meter blood glucose (MBG) measurements were 99% within 20% of laboratory method across entire tested
performed at least four times per day on Accu-Check Aviva range).
Results: Considering both subjects with and without diabe-
tes, 99.5% (370/372) of subject fingertip self-test results were
within 15% and 99.7% (371/372) were within 20% of YSI
reference results. For study staff tests of subject fingertip blood,
99.7% (374/375) of results were within 15% and 100% (375/
375) were within 20% of YSI reference results. At least 95%
of BGMS results for all subjects were within 9.5% and 8.9% of
YSI reference result for subject and study staff-obtained fin-
gertip tests, respectively. Regression analysis of results for
all subjects demonstrated a strong correlation between BGMS
and YSI reference results (adjusted R2 > 0.98 for subject and
study staff-obtained fingertip results). By Parkes-Consensus
Error Grid analysis, 100% (372/372) of all subject-obtained and
100% (375/375) of study staff-obtained fingertip results were
within Zone A.
Conclusions: The BGMS exceeded FDA Draft 2014 cri-
teria in a clinical setting used by subjects with and without
diabetes.
A-78 ATTD 2017 E-POSTER VIEWING ABSTRACTS

6
188 Bambino Gesu- Childrens Hospital, Unit of Endocrinology
and Diabetes, Rome, Italy
GLYCEMIC VARIABILITY IS ASSOCIATED 7
University of Padua, Unit of Metabolic Diseases- Department
WITH ORTHOSTATIC HYPOTENSION IN PATIENTS of Internal Medicine-DIMED, Padua, Italy
WITH TYPE 2 DIABETES MELLITUS
I. Eleftheriadou1, K. Makrilakis1, S. Kalopita1, C. Stathi1, Background and Aims: Pediatric patients often use Dexcom
C. Kapelios1, P. Thomakos1, S. Liatis1 G4AP (i.e. G4 Platinum Share with software 505) CGM sensor
1 on their arm, although it was assessed and approved only to be
Medical School- National and Kapodistrian University inserted on the abdomen or gluteus area.
of Athens, First Department of Propadeutic and Internal Here we compare the accuracy of this sensor in these three
Medicine, Athens, Greece sites: abdomen, gluteus and arm.
Methods: A total of 30 T1D children, 5-9 years old, wore the
Background and Aims: Orthostatic hypotension is a com- Dexcom G4AP sensor for 2 sessions of 4 days each, during a
mon manifestation of cardiac autonomic neuropathy (CAN) in closed-loop study (Del Favero et al., Diabetes Care 2016). Fre-
patients with type 2 diabetes (T2D). Recent studies have re- quent fingerstick blood glucose measurements (SMBG) were
ported an association between CAN and glycemic variability obtained. Sensor was inserted in different sites according to the
assessed by continuous interstitial tissue glucose monitoring patient habit.
(CGM). The aim of our study was to investigate the association We compared absolute deviation (AD) and absolute relative
between orthostatic hypotension (decrease in systolic blood deviation (ARD) of CGM with respect to SMBG in the three sites
pressure >20mmHg or in diastolic blood pressure >10 mmHg with ANOVA. If the test detected a difference between the sites,
within 3 min of standing) and indices of glycemic variability in a a further pair-wise comparison was performed.
cohort of patients with T2D. Results: Results are reported in Tab 1. Overall, no accuracy
Methods: A total of 90 participants were examined (age difference was detected neither in ARD (p = 0.491) nor in AD
62.8 9.9 years, diabetes duration 9.0, 4.0-15.0 years). Diagnosis (p = 0.315) among the three sites. Similarly, no difference
of CAN was based on the 4 standard cardiovascular tests pro- was detected in the hypo- and eu-glycemic regions. In hyper-
posed by Ewing. A commercially available device was used for glycemia a significant difference was detected between the two
the 24-h CGM. Mean interstitial glucose, standard deviation of approved sites, i.e. abdomen and gluteus (DARD = -2.38%
the mean glucose (SDMG), M-value and mean amplitude of [CI:-4.55%;-0.22%], p = 0.035), whereas the comparisons arm-
glycemic excursions (MAGE) were calculated. abdomen and arm-gluteus were not conclusive.
Results: Twenty two participants (24.4%) had CAN, while
45 patients (50%) had orthostatic hypotension. Patients with
orthostatic hypotension had higher mean interstitial glucose val-
ues (152.2 33.0 vs. 136.9 26.0 mg/dl, p = 0.021) and higher
MAGE (118.0, 85.9-159.8 vs. 92.1, 74.6-112.5, p = 0.026) than
patients without orthostatic hypotension. No significant differ-
ences were observed in SDMG and M-value. Multivariate logistic
regression analysis, after controlling for age, diabetes duration,
HbA1c and mean interstitial glucose values, demonstrated that
the odds of orthostatic hypotension increased significantly with
higher MAGE (OR: 1.012, 95%CI: 1.001-1.023, p = 0.037).
Conclusions: Blood glucose fluctuations as assessed by
MAGE are associated with orthostatic hypotension, irrespec-
tive of mean interstitial glucose values and long term glycemic
control in patients with T2D.
Conclusions: These results suggest that the accuracy of the
Dexcom G4AP sensor placed on the arm is comparable with
189 the accuracy in the other approved insertion sites (abdomen and
gluteus). Larger randomized trials are needed to draw final
ACCURACY OF DEXCOM G4 ON PEDIATRIC
conclusions.
PATIENTS IN THREE INSERTION SITES:
ARM, ABDOMEN AND GLUTEUS
S. Faccioli1, S. Del Favero1, R. Bonfanti2, D. Iafusco3,
I. Rabbone4, A. Sabbion5, R. Schiaffini6, D. Bruttomesso7, 190
C. Cobelli1, on behalf of PedArPan study group1
ASSOCIATION OF HYPOGLYCEMIA AND GLUCOSE
1
University of Padua, Department of Information Engineering, VARIABILITY MEASURED BY DIFFERENT INDICES
Padua, Italy IN PATIENTS WITH TYPE 2 DIABETES
2
Scientific Institute, Hospital San Raffaele, Pediatric A.M. Gomez1, O.M. Munoz Velandia2, A. Marn Sanchez1,
Department and Diabetes Research Institute-, Milan, Italy M.C. Fonseca Galvis1, M. Rondon2, M.A. Garca Jaramillo3,
3
Second University of Naples, Department of Pediatrics, F.M. Leon Vargas4
Naples, Italy
4 1
University of Turin, Department of Pediatrics, Turin, Italy Hospital Universitario San Ignacio, Endocrinology,
5
Azienda Ospedaliera Universitaria Integrata of Verona, Bogota D.C, Colombia
2
Regional Center for Pediatric Diabetes- Pediatric Diabetes Hospital Universitario San Ignacio, Epidemiology
and Metabolic Disorders Unit, Verona, Italy and Biostatistics, Bogota D.C, Colombia
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-79

3
Universidad Antonio Narino, Applied Sciences, Bogota, Methods: Accu-Chek Aviva Expert [1], FreeStyle InsuLinx
Colombia [2], FreeStyle Precision Neo [3] and MyStar DoseCoach [4] were
4
Universidad Manuela Beltran, Applied Sciences, Bogota D.C, evaluated. Blood glucose levels in capillary blood samples from
Colombia 100 subjects were determined with three lots of each BGMS and
two different comparison methods (hexokinase [HK]- and glu-
Background and Aims: In the last decade multiple studies cose oxidase [GOD]-based laboratory methods) and deviations
have shown the association between glucose variability (GV) between the results were calculated.
and hypoglycemia. Recent publications constantly introduce Results: When evaluated against the comparison method in-
new GV indices, however there is still a lack of evidence dicated by the respective manufacturer, BGMS 1 had 92%99.5%
supporting the use of one GV parameter in particular, specially of the individual lots within the allowed limits, BGMS 2, 97%
in clinical practice. 99.5%; BGMS 3, 86%96% and BGMS 4, 98%99%. All BGMS
Methods: In this study, Continuous Glucose Monitoring had 100% of the results within Consensus Error Grid zones A
(CGM) data for 657 days from 140 type 2 diabetes patients (4,69 and B.
average days per patient) were analyzed. Mean glucose (MG), Conclusions: In this study, only two of the four tested BGMS
standard deviation (SD), 1, 2 and 4 hour Continuous Overlapping (BGMS 2 and 4) with built-in insulin dose advisors fulfilled
Net Glycemic Action (CONGA 1, 2 and 4), Mean Amplitude of the accuracy requirements of ISO 15197:2013 when compared
Glucose Excursions (MAGE), M Value, J Index, Interquartile against the manufacturers comparison method.
Range (IQR) and both High and Low Blood Glucose Index
(HBGI, LBGI) were estimated.
Results: Hypoglycemia was present in 103 patients, 433 192
events in 259 days of recording (0.66 events per patient/day). In
the multivariate analysis both SD and MG proved a statisti- DIFFERENCE BETWEEN SCANNED DATA
cally significant association with hypoglycemia, OR for AND CONTINUOUS DATA OF THE FREESTYLE
MG 0.87 (CI 0.83-0.92 p = 0.000) and OR for SD 1.22 (CI 1.12- LIBRE SYSTEM
1.33 p = 0.000), these results remained consistent for intersti- S. Pleus1, U. Kamecke1, M. Link1, C. Haug1, G. Freckmann1
tial glucose values below 60 and 50mg/dL. A strong correlation 1
was found between SD and other GV parameters including r Diabetes-Technolgie Forschungs, und
Institut fu
CONGA 1,2 and 4 (r = 0.84), (r = 0.89), (r = 0.95) respectively, Entwicklungsgesellschaft mbH, an der Universitat Ulm, Ulm,
M Value (r = 0.77), J Index (r = 0.82), IQR (r = 0.95) and MAGE Germany
(r = 0.95).
Conclusions: This analysis concludes that given the strong Background and Aims: The FreeStyle Libre (FSL, Abbott
correlation of SD with other GV parameters, the significant as- Diabetes Care Inc.) system is labeled to replace blood glucose
sociation with hypoglycemia and considering the ease of their measurements in many situations. Upon scanning the FSL sensor
calculation, MG and SD could be recommended as the preferred unit with the FSL reader, it shows current glucose readings
parameter of GV for use in clinical practice. (scanned data) and a graph displaying glucose history of the
last 8 hours. Additionally, all sensor data are stored in the reader.
This analysis aimed at investigating whether FSL scanned data
191 differs from continuous data.
Methods: Each of 20 participants wore two FSL systems,
MEASUREMENT ACCURACY OF FOUR BLOOD one on each upper arm. Each system continuously measured
GLUCOSE MONITORING SYSTEMS WITH INSULIN tissue glucose for up to 14 days and stored 1 value every 15
DOSE ADVISORS FOLLOWING ISO 15197:2013 minutes. To obtain current glucose readings and to store con-
tinuous data on the reader, the FSL sensor unit was scanned
T. Leucht1, N. Jendrike1, A. Baumstark1, S. Pleus1, C. Haug1,
with the FSL reader approximately 16 times daily. Individual
A. Beer2, F. Flacke2, G. Freckmann1
relative differences (RDs) between scanned data points and
1
Institut fur Diabetes-Technologie Forschungs, und individual continuous data points having identical timestamps
Entwicklungsgesellschaft mbH an der Universitat Ulm, Institut
fur Diabetes-Technologie Forschungs, und
Entwicklungsgesellschaft mbH an der Universitat Ulm, Ulm,
Germany
2
Sanofi, Global Diabetes Division, Frankfurt, Germany

Background and Aims: Self-monitoring of blood glucose is

a key element in the therapy of people with diabetes treated
with insulin. Some blood glucose monitoring systems (BGMS)
have built-in insulin dose advisors that provide recommenda-
tions for the next insulin dose. Adequacy of the suggested dose
depends strongly on the measurement quality of the BGMS.
Accuracy of four such BGMS was evaluated following ISO
15197:2013 where 95% of blood glucose results measured
with the BGMS have to be within 15 mg/dl from compari-
son method results at glucose concentrations <100 mg/dl and
within 15% at concentrations 100 mg/dl. Furthermore, 99%
of the results have to be within Consensus Error Grid zones
A and B.
A-80 ATTD 2017 E-POSTER VIEWING ABSTRACTS

(n = 525) were calculated. Additionally, RDs between scanned 194

data points and corresponding data points linearly interpolated
COST-EFFECTIVENESS OF REAL-TIME
from continuous sensor data (n = 8824) were calculated.
CONTINUOUS GLUCOSE MONITORING (CGM)
Results: The distribution of individual relative differences is
WITH SELF MONITORING OF BLOOD GLUCOSE
shown in the figure. The middle 95% (interquantile range 2.5th to
(SMBG) COMPARED TO SMBG ALONE FOR TYPE 1
97.5th quantile) of values were within -9.9% and +10.2% for
DIABETES (T1DM) IN FRANCE
non-interpolated data and within -10.6% and +13.6% for inter-
polated data. B. Klinkenbijl1, S. Tunis2, C. Graham3, S. Chaugule3
Conclusions: Scanned data are readily available; however, in 1
this evaluation differences between scanned data and corre- Dexcom, International Access, Essertines sur Rolle,
sponding continuous glucose data were observed. More detailed Switzerland
2
information would be helpful to understand the background and Consultancy, Pharmacoeconomics and Outcomes Research,
possible impact on therapeutic decisions. Indianapolis, IN, USA
3
Dexcom, Global Access, San Diego, CA, USA

Background and Aims: To explore the cost-effectiveness of

193 real-time CGM in France, in T1DM using Multiple Daily In-
jections (MDI), compared to current SMBG standard practice.
FREEDOM FROM ACETAMINOPHEN Methods: The IMS CORE Diabetes Model (CDM) (v. 9.0)
INTERFERENCE WITH A LONG TERM was used to assess the lifetime (50 year) cost-effectiveness of
IMPLANTABLE CGM SYSTEM CGM (G4Platinum) plus SMBG, compared to SMBG alone for
C. Lorenz1, W. Sandoval1, M. Mortellaro1, L. Kelley2 a T1DM cohort. All model assumptions were based on published
research. Base case (BC) assumptions included a) starting
1
Senseonics, Chemistry, Germantown, USA HbA1c 7.6%; b) change in HbA1c -0.50% for CGM+SMBG,
2
Senseonics, Clinical, Germantown, USA 0.02% for SMBG alone; c) 50% reduction in severe hypogly-
caemic events (SHEs) and non-severe hypoglycaemic events
Background and Aims: Interference with acetaminophen (NSHEs) when using CGM; d) dis-utilities of -0.0142 for
(paracetamol), one of the most commonly used over the counter NSHEs and SHEs not requiring medical intervention, and -0.047
medications, has been shown to falsely elevate glucose read- for SHEs requiring medical resources. Assumed direct costs for
ings in electrochemical CGM sensors. Acetaminophen is oxi- these events were e0, e216.27, and e4156.33 respectively.
dized by the CGM applied electrode potential thereby creating Results: The BC incremental cost-effectiveness ratio (ICER)
a biased sensor reading. The new fluorescence-based Eversense was e11460. Sensitivity analyses (SAs) showed that results were
CGM System was subjected to standard test methods to de- minimally impacted by changes in baseline HbA1c, direct costs
termine the effect on performance in the presence of acet- assumed for SHEs, and shorter time horizons. SAs assuming
aminophen. 25% reduction in hypoglycaemic events for CGM+SMBG, or
Methods: A paired-difference test adapted from ISO 50% reduction in dis-utilities resulted in ICERS of e27682 and
15197:2013 was used to screen for interfering effects. Potential e22606. Parallel results were found when direct costs for CGM,
interferents were tested in simulated interstitial fluid media at SMBG, complications, and hypoglycaemic events were based on
concentrations in excess of physiological or therapeutic con- recent data for Germany, Netherlands and the UK.
centrations to simulate worst case concentrations. Each sub- Conclusions: Use of CGM is cost effective within the popu-
stance was tested at two glucose concentrations within the ranges lation of MDI-using T1DM, assuming a e30,000 willingness-to-
specified by ISO15197: 2013; one low concentration of ap- pay threshold per Quality-Adjusted Life Year (QALY). Results
proximately 4 mmol/L and one high concentration of ap- demonstrate the direct and substantial effect of hypoglycaemia
proximately 18 mmol/L. clinical and quality-of-life assumptions for simulations that in-
Results: The effect of acetaminophen on Eversense sensors clude diabetes cohorts at high risk for hypoglycaemia.
exposed to high (far above therapeutic) dose of acetaminophen
are tabulated below.
Conclusions: Eversense sensor glucose measurements in
media containing a toxic concentration of acetaminophen met
ISO 15197:2013 specified acceptance values of measurement 195
bias <0.55 mmol/L and <10%. Unlike short duration transcuta- INCORPORATION OF MODELS OF CGM SENSOR
neous CGM systems, the Eversense sensor does not use elec- ERROR AND FAULTS AFFECTING CGM SENSORS
trochemical reactions to measure glucose and thus is not affected IN THE UVA/PADOVA TYPE-1 DIABETIC
by compounds such as acetaminophen that may be oxidized by SIMULATOR: ASSESSMENT ON CLINICAL DATA
a sensing electrode. The Eversense CGM sensor retained the
ability to accurately sense glucose concentrations in the pres- Y. Leal1, S. Del Favero2, M. Vettoretti2, R. Visentin2,
ence of acetaminophen providing a potential safety advantage A. Facchinetti2, C. Cobelli2
to patients. 1
Institut dInvestigacio Biomedica de Girona, Service
of Diabetes- Endocrinology and Nutrition UDEN Josep Trueta
Hospital, Girona, Spain
2
University of Padova, Department of Information Engineering,
Padova, Italy

Background and Aims: Recently, a new model of the mea-

surement errors in continuous glucose monitoring (CGM) sensor
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-81

was identified for Dexcom G4TM Platinum [Facchinetti et al. dictable calibration and correlation factors for glucose concen-
MBEC 2014]. Furthermore, a model for the faults affecting CGM tration using standardized units.
sensors (e.g. disconnections and compression artifacts) was Conclusions: Data shows good correlation between mea-
proposed [Facchinetti et al. DTT 2016]. The inclusion of these surements obtained using the spectroscopy sensor and the ref-
two components in the UVA/Padova Type-1 diabetic simulator is erence analyzer as shown in the data gathered from extracted
critical for accurate in silico testing of CGM-based applications burn blister fluid. The study demonstrates that a minimal invasive
like the artificial pancreas. In this work, both models are incor- spectroscopy-based glucose sensor system could be a reliable,
porated into the most recent version of the simulator and simu- pain-free, method for glucose measurements. Clinically, this
lated data are compared against clinical data. system may serve to improve patient compliance with regard to
Methods: 108 traces of subjects wearing the Dexcom G4TM self-monitoring and thus improved glucose control.
Platinum (DG4P) and undergoing an 1 day hospital admission
are available. Blood glucose samples were collected every
15 + 5 min using YSI. The accuracy of these CGM measure-
ments was compared with the accuracy of the simulated CGM in 197
108 traces obtained replicating the clinical protocol. MARD,
MAD and CEG-Zone-A were used for assessment. Frequency EFFECT OF FLASH GLUCOSE MONITORING
and duration of real and simulated disconnections were com- ON METABOLIC CONTROL AND SELF ESTEEMED
pared. Finally, frequency, duration and amplitude of real and TREATMENT SATISFACTION IN PEOPLE
simulated compression artifacts were compared. WITH TYPE 1 DIABETES
Results: Overall median (IQR) MARD was 12.5(6.9)% for M. Londahl1, K. Filipsson1, E. Lindholm2, P. Katzman1
real data vs 12.7(6.4)% for simulated data ( p = 0.63). Similarly 1
MAD was 19.8(10.2)% vs 19.0(9.5)% ( p = 0.94), and CEG- Skane University Hospital, Endocrinology, Lund, Sweden
2
Zone-A was 81.2(22.2)% vs 80.0(20.4)% ( p = 0.12). 997 dis- Skane University Hospital, Endocrinology, Malmo, Sweden
connections occurred on real data vs. 1056 on simulated ones.
90.0% vs 89.1% of them lasted <20 min. Background and Aims: The use of CGM is limited by high
Conclusions: Results suggest that the UVA/PADOVA sim- costs and in some cases disadvantages of continuous presentation
ulator equipped with the two models are able to reproduce the of blood glucose levels. Flash glucose monitoring (FGM) re-
clinical trial observations. quires an active initiative from the patient to present the present
glucose level and the glucose trend. The aim of this study was to
evaluate the effects on HbA1c and diabetes treatment satisfaction
after introduction of FGM in people with type 1 diabetes.
196 Methods: Patients with type 1 diabetes received FGM if their
individual metabolic goal was not met or if they had problems
A FIRST CLINICAL VERIFICATION OF A RADIO with blood glucose fluctuations or repeated hypoglycaemic
FREQUENCY-BASED SPECTROSCOPY SENSOR events. Their HbA1c was measured before initiating of FGM
INTENDED FOR GLUCOSE DETECTION and every third month thereafter. A DTSQs questionnaire was
IN INTERSTITIAL FLUID completed before FGM initiation and a DTSQs follow-up
questionnaire three months later.
V. Gonzalez1, M. Lindblad1, M. Renlund2, P. Rangsten2,
Results: 803 patients with a follow-up time between 3 and 21
F. Huss1
months and a baseline HbA1c level of 66.5 14.5 mmol/mol were
1
Uppsala University Hospital, Burn Center- Dept. of Plastci included. HbA1c has decreased to 59.0 11.2 mmol/mol after
and Maxillofacial Surgery, Uppsala, Sweden 3 months. The improvement sustained over time (58.2 10.9
2
Ascilion AB, R&D, Kista, Sweden mmol/mol). The mean decrease in HbA1c was 7.3 9.2 mmol/
mol. In those with HbA1c>70 mmol/mol the change was -12.8
Background and Aims: A novel, minimal invasive, chipset 10.4 (min:max -62:+10) mmol/mol. Their self-estimated treat-
that combines extraction of interstitial fluid (ISF) and glucose ment satisfaction was 2.5 (+3 to -3) and rating for continued
sensing is under development. The chipset combines several hun- FGM use was 2.7. Unacceptable high blood glucose was less
dred hollow micro needles, and a radio frequency spectroscopy often present (-0.2) as were unacceptable low blood glucose
sensor element. This abstract presents the first clinical verification levels (-0.4). Self-estimated treatment satisfaction scores were
of the sensor using fluid from burn blisters as substitute for ISF. similar over the whole HbA1c range.
Methods: Blister fluid was collected from acute burn patients. Conclusions: Use of FGM adds clinical significant advantage
Each sample was divided into two parts. One portion was mea- to individuals with 1 diabetes in terms of HbA1c reduction and
sured using the radio frequency spectroscopy sensor. The second improved self-estimated treatment satisfaction.
portion of blister fluid was used as reference and analyzed with
respect to glucose content using a clinical chemistry ana-
lyzer (Abbott Diagnostics Architect Plus C16000). Two different
teams conducted the measurements separately. The measure- 198
ments were subsequently organized in pairs and compared using
an accuracy plot. REALIZATION OF BGM WITHIN 10% ACCURACY
Results: The spectroscopy sensor showed adequate accuracy BASED ON INNOVATIVE OPTICAL TRANSMISSION
with 92% of the measurements being within accepted range ABSORBANCE SYSTEM
when compared to the clinical chemistry analyzer half way T. Moriuchi1, H. Satou1, Y. Komata2, R. Aikawa3, K. Sode4
through the study. Acceptance criteria from the ISO15197-2013
1
standard was used. The spectroscopy sensor showed a linear Terumo Corporation, DM and Consumer Healthcare Division-
correlation to glucose concentrations within the range with pre- General Hospital Business Group, Nakakoma-gun, Japan
A-82 ATTD 2017 E-POSTER VIEWING ABSTRACTS

2
Terumo Corporation, DM and Consumer Healthcare Division- from the registry database from July 2015 till Oct 2016. Sta-
General Hospital Business Group, Tokyo, Japan tistical analysis was done using Wilcoxon signed rank test and
3
Terumo Corporation, DM and Consumer Healthcare Division- Mann-Whitney test through the GraphPad Prism 7. 26 non
General Hospital Business Group, Ashigarakami-gun, Japan evaluatable data sets and patients were excluded from the
4
Tokyo University of Agriculture & Technology, Department analysis.
of Biotechnology & Life Science, The Graduate School Results: The glycemic variability data was evaluated in 86
of Engineering, Tokyo, Japan patients. Mean duration of diabetes was 13.66 years (SD 9.26
1.12, min 0, max 50 yrs; 95% CI 11.41-15.9, p < 0.0001). Mean
Background and Aims: Since the blood glucose monitor- age was 57.61 years (SD 10.97 1.3, Min 31, Max 81 yrs; 95%
ing (BGM) is essential in diabetes care and management, fur- CI 54.95-60.27, p < 0.0001). The HbA1c reductions before
ther improvement of the accuracy of BGM is mandatory to (mean 8.5 2.2, Min 5.1, max 15 yrs; 95% CI 7.96-9.07) and
achieve the better glycemic control of diabetics. In this paper after 3 months (mean 7.8 1.7, min 5, max 16; 95% CI 7.41-
we report our current results of our new BGM using an inno- 8.25) were numerically important (-0.68), but did not achieve
vative single optical transmission absorbance system, to mea- statistical significance (p = 0.1365, NS). The comparative re-
sure glucose level and hematocrit simultaneously. In addition, a ductions in HbA1c in patient group <60 & >60 yrs did not
simple and inexpensive test strip was developed, in order to achieve statistical significance (p = 0.7898, NS). 16 patients had
measure the blood glucose level of the whole blood accurately all four drug time changes, drug choice changes, dose changes,
and rapidly. diet modification
Methods: To realize the high-accuracy measurement of the Conclusions: The utility of the novel FreeStyle Libre Pro(TM)
blood glucose level, we have developed highly sensitive re- translates into a physician led and patient enabled empowerment
agent, with brand new enzyme and original absorption dye, and tool which helps physicians customise the therapy and empowers
accurate hematocrit detection technology. We employed a new patients through the visual snapshots to sensitively adapt to the
generation GDH(FAD) enzyme, which shows high catalytic prescribed regimen.
efficiency with low Km value, to measure the reaction end
points. Together with the original high absorption dye, the optical
transmission absorbance principle achieved the increase in the
resolution at a low glucose level. Additionally, high-accuracy 200
hematocrit compensation was achieved using multi wave-
DETERMINATION OF THE OPTIMAL TIMING
lengths detection.
FOR POSTPRANDIAL GLUCOSE MEASUREMENT
Results: Accuracy of our new BGM was evaluated with in-
IN PREGNANT WOMEN WITH GESTATIONAL
house blood samples, adjusted to three blood glucose lev-
DIABETES EVALUATED BY THE CONTINUOUS
els (0mg/dL, 100mg/dL, 400mg/dL) with various hematocrit
GLUCOSE MONITORING SYSTEM IPRO
(Hct20%, Hct40%, Hct60%). From the measurement results of
N = 270, 93.7%(253/270) of results were within 5% (5mg at R. Radermecker1, J.C. Philips1, M. Franck1
0mg/dL) accuracy and 100% (270/270) of results were within 1
10%(10mg at 0mg/dL) accuracy. University of Liege, Diabetes, Nutrition and Metabolic
Conclusions: We have developed a novel BGM based on an disorders, Liege, Belgium
innovative optical transmission absorbance system, and achieved
accuracy within 10%. Background and Aims: Using the Continuous Glucose
Monitoring System (IPro; Medtronic Minimed) for a group
of pregnant women with gestational diabetes (GD) based on
IADSPG criteria (fasting, 5.0 mmol/L; 1-h, 10.0 mmol/L; 2-h,
199 8.6 mmol/L).We attempted to answer the following question:
when does the physiological peak of postprandial glucose
EVALUATION AND UTILITY OF A NOVEL occur in real life for the 3 meals and after a standardized
GLUCOSE MONITORING SYSTEM IN INDIAN breakfast.
ADULTS WITH DIABETES Methods: We included 13 pregnant women in our study
Y. Munjal1, G. Pangtey2, J. Chowdhary3, N. Wadhwa4, (31 3 years old, 6/13 primipara and 2/13 with GD history).
N. Kashyap5 Patients received an IPro CGMS for use over 5 days after an
educational program and dietary advises. This was calibrated at
1
Banarsidas Chandiwala Institute of Medical Sciences, least 4 times a day. No problem of intolerance occurred during
Chief Consultant and Director, Delhi, India this trial. The last day of the trial, patients received a standardized
2
Lady Hardinge Medical College, Medicine, Delhi, India breakfast at hospital.
3
Banarsidas Chandiwala Institute of Medical Sciences, Results: The postprandial glucose peak was reached after
Junior Consultant, Delhi, India 68 23 min, 79 30 min and 109 28 for breakfast, lunch
4
AUW Global, Medical Sciences, Delhi, India and dinner respectively. The postprandial glucose peak was
5
Banarsidas Chandiwala Institute of Medical Sciences, observed earlier with a standardized breakfast (70 g carbohy-
Dietician, Delhi, India drates): 52 11 min.
Conclusions: Our results show that the optimal time for
Background and Aims: Flash glucose monitoring is a testing in pregnant women with GD is between 45 and 120 min
novel glucose sensing technique that estimates interstitial glu- postprandial. Based on a practical approach, it seems to be easier
cose levels for up to 14 days and does not require any calibration. to advise a 60-min interval. In real life, results depend on the type
Methods: The FreeStyle Libre Pro(TM) sensor was utilised of meals but the earliest peak remains always at breakfast which
in 112 patients for 14 days. We did a real world evaluation could be more discriminant.
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-83

201 Results: HbA1c reduction from baseline to 24 weeks was

greater in the CGM Group than Control Group (-0.9 + 0.7%
FLASH GLUCOSE MONITORING IN NONCOMPLIANT
versus -0.5 + 0.7%, adjusted difference in mean change =
CHILDREN AND ADOLESCENTS WITH TYPE 1
-0.4 + 0.1%, 95% confidence interval -0.7% to -0.2%,
DIABETES
P < 0.001). More participants in the CGM Group had a relative
E. Tirelli1, G. Frontino1, V. Favalli1, C. Bonura1, reduction in HbA1c >10% (56% versus 26%, odds ratio 4.3, 95%
A. Rigamonti1, F. Meschi1, R. Bonfanti1 confidence interval 1.8 to 10.5, P = 0.001). CGM-measured time
1 above 250 mg/dL (P = 0.006) and glycemic variability (P = 0.01)
San Raffaele Scientific Institute, Pediatrics, Diabetes Research were lower in the CGM Group. Among the 61 in the CGM Group
Institute, Milan, Italy completing the trial, 97% used CGM >6 days/week in month 6.
There were no severe hypoglycemic or diabetic ketoacidosis
Background and Aims: Recent data involving adults with events in either group.
type 1 diabetes (T1D) have shown that Flash Glucose Mon- Conclusions: In adults >60 years of age with T1D and T2D
itoring (FGM) is effective. However, data regarding children using MDI, CGM use was high and associated with improved
and adolescents who are noncompliant with regular capillary HbA1c and reduced glycemic variability. Therefore, CGM should
blood glucose monitoring (BGM) are lacking. Our aim was to be considered for older adults with diabetes using MDI.
assess whether FGM could represent a empowering tool and
improve glucose control in children and adolescents who
are noncompliant to regular BGM and treated with multiple
injections. 203
Methods: We evaluated 13 T1D patients (7 males, 6 females)
with the following characteristics: mean age of 14 years (SD 4.1 QUALITY OF LIFE BENEFITS ACCRUING FROM CGM
years), mean T1D duration of 4.8 years (SD 2.2 years), under- USE IN PATIENTS WITH TYPE 1 DIABETES
going multiple injection therapy. HbA1c readings was evaluated ON MULTIPLE DAILY INJECTIONS: RESULTS
at baseline (FGM insertion) and after 3 months. Average daily FROM A PROSPECTIVE, RANDOMIZED
BGM readings were evaluated at FGM placement and average CONTROLLED TRIAL
number of daily FGM scans were evaluated at 3 months. At FGM W. Polonsky1, D. Hessler2, R. Beck3
insertion the patients were educated on how to interstitial glucose
1
monitoring management and trend arrow evaluation. Patients University of California, San Diego, and Behavioral Diabetes
were required to measure at least one BGM to assess FGM ac- Institute, Psychiatry, San Diego, USA
2
curacy. A paired t-test was used to compare HbA1c means. University of California, San Francisco, Family Community
Results: Mean HbA1c at FGM placement and after 3 months Medicine, San Francisco, USA
3
were 81 28mmol/mol and 66 19mmol/mol respectively. The Jaeb Center for Health Research, n/a, Tampa, USA
decrease in mean HbA1c of 15mmol/mol was statistically sig-
nificant (p < 0.001). Average number of daily BGM readings in Background and Aims: CGM trials have not consistently
the 3 months prior to FGM placement: 1.7 1.3. Average number demonstrated positive quality of life (QOL) benefits in adults
of daily FGM scans in the 3 months: 10.7 6.6. No episodes of with type 1 diabetes (T1Ds). We examined pre-post QOL data
severe hypoglycemia or DKA were declared. from the DIAMOND trial, which demonstrated glycemic bene-
Conclusions: FGM use may improve glucose control and fits with CGM use in adults withT1D on multiple daily injections
motivation in noncompliant children and adolescents with T1D. (MDI).
Methods: DIAMOND, a 24-week randomized clinical trial,
randomly assigned adults with T1D (mean age 48 + 13 years,
202 T1D duration 21 + 14 years, HbA1c 8.6 + 0.6%) to CGM (Dex-
Com G4 Platinum CGM System with software 505, N = 105)
CGM IN OLDER ADULTS WITH TYPE 1 AND TYPE 2 or SMBG only (N = 53). To assess QOL change, subjects com-
DIABETES USING MULTIPLE DAILY INJECTIONS pleted, at baseline and week 24, a series of diabetes-related
OF INSULIN: RESULTS FROM THE DIAMOND TRIAL measuresdiabetes distress (DDS), hypoglycemic fear (HFS-W)
K. Ruedy1, C. Graham2, T. Riddlesworth1, C. Kollman1, and hypoglycemic confidence (HSC)and generic measures
D. Price2, R. Beck1 overall well-being (WHO-5) and health status (EQ-5D). Multiple
regression analyses adjusting for clustering by clinic site and
1
Jaeb Center for Health Research, n/a, Tampa, USA baseline QOL examined group differences over time.
2
Dexcom Inc., n/a, San Diego, USA Results: While there were no significant group differences in
WHO-5 or EQ-5D over time, CGM subjects did evidence greater
Background and Aims: To determine the effectiveness of improvement than SMBG subjects in all three diabetes-related
CGM in MDI-using adults >60 years of age with type 1 (T1D) or QOL outcomes, with greater reductions in DDS (p = .002) and
type 2 (T2D) diabetes. HFS-W (p = .037), and a larger HSC rise (p = .009). Among DDS
Methods: Multicenter, randomized trial conducted in the US subscales, CGM subjects displayed significantly larger reductions
and Canada in which 116 individuals >60 years (mean 67 + 5 in diabetes-related emotional burden (p = .046), regimen distress
years) with T1D (N = 34) or T2D (N = 82) using MDI therapy (p = .017) and interpersonal distress (p = .001). Reductions in total
were randomly assigned to either CGM (Dexcom G4 Platinum DDS and regimen distress subscale scores were associated with
CGM System with software 505, N = 63) or continued man- decreases in HbA1c (both p < .05). Adjusting for participant
agement with SMBG (N = 53). Mean diabetes duration was demographic factors did not alter the pattern of effects.
23 + 12 years and mean baseline HbA1c was 8.5 + 0.6%. The Conclusions: CGM use lowers diabetes distress and hypo-
primary outcome, central-lab HbA1c at 24 weeks, was obtained glycemic fear and raises hypoglycemic confidence in T1D adults
for 114 (98%) participants. using MDI.
A-84 ATTD 2017 E-POSTER VIEWING ABSTRACTS

204 evaluated in pregnant patients with diabetes. We assessed the

clinical safety and accuracy of the system compared to self-
IMPROVEMENT OF GLYCEMIC CONTROL
monitoring of blood glucose (SMBG).
AND HYPOGLYCEMIA PREVENTION OF SENSOR-
Methods: 74 patients were enrolled into the study and ap-
AUGMENTED PUMP THERAPY WITH LOW-
plied a masked sensor, they were diagnosed with type 1 diabetes
GLUCOSE MANAGEMENT SYSTEM IN TYPE 1
(n = 24), type 2 diabetes (n = 11) or gestational diabetes (n = 39).
DIABETIC PATIENTS: OMEGA STUDY OUTCOMES
Average gestation was 26.6 6.8 weeks (meanSD), average age
L. Schoumacker-ley1, L. Duchesne2, T. Cuny3, P. Bohme1, was 30.5 5.1 years, duration of diabetes was 13.1 7.3 years for
O. Ziegler1, B. Guerci1 T1DM and 3.2 2.5 years for T2DM, 49/74 (66.2%) used insulin
1 to manage their diabetes. The study was conducted across 9 UK
CHRU Nancy Brabois, Diabetology, Vandoeuvre-Les- sites and 4 in Austria.
Nancy, France Results: Consensus Error Grid analysis showed 88.1%
2
CHR MERCY, Diabetology, Mercy, France (4433/5031) of paired FreeStyle Libre sensor and SMBG read-
3
CHRU Nancy Brabois, Endocrinology, Vandoeuvre- ings were within zone A (clinically accurate) and 99.8% (5022/
Les-Nancy, France 5031) were within zones A&B (clinically acceptable). 87.1% of
results were within 1.1mmol/L (20mg/dL)/20% of SMBG
Background and Aims: The MiniMed 640G sensor- values. Mean absolute relative difference (MARD) was 11.8%.
augmented pump system (SmartGuard system, Medtronic) is Regression analysis demonstrated good agreement with SMBG
the first system which can automatically suspend insulin delivery (r = 0.92, slope = 1.12, intercept = -0.84 mmol/L [-15.1 mg/dL]).
in advance of predicted hypoglycemia and restart it upon re- Clinically accurate results were obtained for each type of dia-
covery. This observational study analysed this system on gly- betes.
cemic control, hypoglycemia incidence and quality of life in User satisfaction questionnaires indicated high levels of ac-
uncontrolled T1DM already treated by CSII. ceptance for sensor wear and ease of use of the device. There
Methods: Twenty-one T1D patients (9W12M) needing CGM were no unanticipated device-related adverse events.
according to the French consensus statement, were recruited. Conclusions: Results show good agreement between the
Baseline characteristics: age = 44 10 yrs, diabetes dura- FreeStyle Libre System and SMBG for pregnant women with
tion = 28 12 yrs, BMI = 26.0 4.9 kg/m2, HbA1c = 8.1 0.9%; diabetes, indicating the device is safe and accurate for use by this
symptomatic hypoglycemia incidence was 3.0 2.2 times per population.
week, severe hypoglycemia (SH) from 0 to 12 events during the
last 6 months. HbA1c levels, hypoglycemia and SH incidences
were recorded at inclusion, at 3 months (M3) and 6 months (M6).
The performance of the system (suspend before low [SBL] and
suspend low [SL] events) was measured with downloaded pump 206
and sensor data. The quality of life was evaluated using TRIMD USE OF ACCURACY PROBABILITY CURVES:
(Treatment Related Impact Measures Diabetes) and TRIMD- A NEW METHOD FOR DISTINGUISHING
Device questionnaires. DIFFERENCES BETWEEN BLOOD GLUCOSE
Results: A significant reduction in symptomatic hypogly- MONITORING SYSTEMS MEETING ISO 15197:2013
cemia (-1.5 1.6, p = 0.001) at M6 was observed and no epi- ACCURACY REQUIREMENTS
sode of SH or ketoacidosis was reported. No significant change
in HbA1c level (-0.1 0.7%, p = 0.35) at M6 was found. The S. Pardo1, R. Shaginian1, D. Simmons1
frequencies of SBL and SL events for SmartGuard were 1
Ascensia Diabetes Care, Parsippany NJ, USA
1.5 0.9 events/day and 0.1 0.1 events/day, respectively. Fur-
thermore, we found a significant improvement in TRIMD Background and Aims: While blood glucose monitoring
(+16.9 22.2% p = 0.006), and in TRIMD-Device questionnaires systems (BGMSs) are required to meet ISO 15197:2013 accu-
(+11.2 17.4%, p = 0.013) at M6. racy criteria (95% of results within 15 mg/dL [BG <100 mg/
Conclusions: The SmartGuard system reduces significantly dL] or 15% [BG 100 mg/dL] of the reference result), not all
hypoglycemia and SH incidences without any glycemic control BGMSs that meet the ISO 15197:2013 criteria have the same
deterioration, increasing comfort and quality of life in a selected probability of producing accurate results. Here we present a new
T1D population. methodology for revealing the differences between BGMSs that
meet ISO 15197:2013 accuracy criteria.
205 Methods: Regression models were fit to BG data (BGMS on
the y-axis; YSI on the x-axis). Weighted least squares was used to
ACCURACY EVALUATION OF FREESTYLE LIBRE account for the proportional nature of standard deviations (ie,
FLASH GLUCOSE MONITORING SYSTEM WHEN standard deviation is proportional to the lab value of glucose
USED BY PREGNANT WOMEN WITH DIABETES concentration). Regression coefficients with statistical estimates
E. Scott1, A. Kautzky-Willer2 of coefficients of variation (CV) were used to compute the prob-
ability that for any given BG value, the BGMS result would be
1
St Jamess University Hospital, Department of Diabetes within 15% of the lab value; 95% lower confidence limits were
and Endocrinology, Leeds, United Kingdom also computed.
2
Medical University of Vienna, Department of Internal Results: As shown in Figures 1A-1C, all 3 BGMSs met ISO
Medicine III Division of Endocrinology and Metabolism, 15197:2013 accuracy criteria (ie, 95%; BGMS A, 99.63%;
Vienna, Austria BGMS B, 99.63%; BGMS C, 99.81%). However, BGMS A
shows the highest accuracy, as all values fall above the 95%
Background and Aims: Accuracy of the FreeStyle LibreTM probability line indicating accuracy across all BG ranges; BGMS
Flash Glucose Monitoring System has not previously been B and C show decreased accuracy at low BG ranges.
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-85

207
PERFORMANCE OF MEDTRONICS
FOURTH-GENERATION GLUCOSE SENSOR
SYSTEM
A. Sullivan1, M. Johnson1, J. Hall1, K. Nogueira1, T. Engel1
1
Medtronic, Sensor R&D, Northridge, USA

Background and Aims: Medtronics fourth-generation con-

tinuous glucose monitoring (CGM) system is designed to reliably
support an insulin-delivering hybrid closed-loop system, with
flexibility for a standalone CGM system.
This novel 7-day system is comprised of a subcutaneous
glucose sensor, transmitter platforms communicating to insu-
lin pumps via proprietary RF technology or to a mobile display
app via standard Bluetooth low energy (BLE) technology, and
an innovative algorithm designed to drive exceptional sensor
performance and insulin dosing. New features include proac-
tive diagnostics and preventing the display of unreliable sensor
glucose (SG) values.
The improved sensor and algorithm design optimizes signal
stability and consistency, enhancing the performance and min-
imizing variability. The superior sensor performance is main-
tained during extended use of these technologies.
Methods: Clinical studies were conducted with subjects wear-
ing real-time systems with an insulin pump, mobile display app,
and/or a recorder. The studies evaluated sensor performance
during hypoglycemic and hyperglycemia challenges, as well as
continued use in an in-home environment.
Results: Sensor accuracy data were evaluated against YSI
reference points. The mean absolute relative difference
( MARD) between sensors with minimum calibrations and
YSI reference values was 10.45% (23,709 evaluation points)
for the abdomen location and 9.09% (10,526 evaluation
points) for the arm. The MARD for sensors with the re-
commended 3-4 calibrations/day was 9.54% (22,601 evalu-
ation points) for the abdomen and 8.68% (10,771 evaluation
points) for the arm.
Conclusions: Clinical results have demonstrated that the new
sensor and algorithm system is reliable with superior perfor-
mance that is consistent over extended use and periods of gly-
cemic variability.

208
IMPACT OF BLOOD GLUCOSE MEASUREMENT
ERROR ON THE ACCURACY OF MEDTRONICS
4TH GENERATION CGM SYSTEM
A. Varsavsky1, C. Palerm1, N. Parikh2, P. Weydt1
1
Medtronic - Inc., Intensive Insulin Management, Northridge,
USA
2
Medtronic - Inc., Advanced Research, Northridge, USA

Background and Aims: CGM systems that require periodic

blood glucose (BG) measurements for calibration rely on the
integrity of the BG value to ensure accurate sensor glucose
tracings. However, BG meter performance can vary signifi-
Conclusions: Results from these probability curves high- cantly, and even the most accurate available BG meter contains
light the importance of BGMS accuracy across the entire measurement error. We present an evaluation of the impact of
BG range to achieve optimal glycemic control while avoiding BG errors on the performance of Medtronics 4th Generation
hypoglycemia. CGM system.
A-86 ATTD 2017 E-POSTER VIEWING ABSTRACTS

Methods: Gaussian errors were added to YSI measurements 210

and used to calibrate 4th Generation sensors. CGM performance
CONTINUOUS GLUCOSE MONITORING (CGM)
was then evaluated relative to the uncorrupted YSI (*22,500
ASSOCIATED WITH LESS DIABETES-SPECIFIC
evaluations).
EMOTIONAL DISTRESS AND LOWER A1C
To simulate the worst case BG error for which the meter would
AMONG ADOLESCENTS WITH TYPE 1 DIABETES
still comply with the ISO15197:2003 standard, the errors stan-
dard deviation was 10% for YSI 75mg/dL, and 7.5mg/dL for A. Vesco1, J. Weissberg-Benchell1
YSI <75mg/dL. For worst case using ISO15197:2015, the errors 1
standard deviation was 7.5% for YSI 100mg/dL, and 7.5mg/dL Ann & Robert H. Lurie Childrens Hospital of Chicago,
for YSI <100mg/dL. Department of Psychiatry, Chicago, USA
Results: For a BG meter that complies to the ISO15197:2015
standard, the worst case impact on CGM performance was small. Background and Aims: CGM psychosocial research is
Relative to a YSI-calibrated sensor (i.e., zero BG error), MARD limited. Aim of the current analysis was to assess association
increased by 1.1%; 20/20 and 40/40 agreements decreased between diabetes technology (CGM Alone; Pump Alone; CGM+
by 2.4% and 0.2%. For a BG meter that complies to ISO15197: Pump; or No Technology) and diabetes-specific emotional dis-
2003, worst case impact on CGM performance remained modest. tress and A1c.
MARD increased by 1.9%; 20/20 and 40/40 agreements de- Methods: 854 adolescents and 1038 parents completed
creased by 5.1% and 0.6%. In both cases average sensor bias was measures of diabetes-specific distress. ANOVA compared levels
unaffected. of technology use with distress and A1c. Cohens d values with
Conclusions: Performance of Medtronics 4th Generation Hedges correction provided effect sizes between groups.
CGM system is robust in the presence of large calibration BG Results: For adolescent-reported distress, CGM Alone con-
error. The data suggest that the worst performing BG meter that ferred moderate effect over No Technology (d = 0.59) and small
complies with ISO15197:2003 will have little impact on sensor effects over CGM+Pump (d = 0.26) and Pump Alone (d = 0.29);
performance. Pump Alone and CGM+Pump demonstrated small effects over
No Technology (d = 0.34 and 0.33 respectively). For parent-
reported distress, CGM Alone demonstrated small-medium
effects over No Technology (d = 0.41), over CGM+Pump (d =
209 0.28), and over Pump Alone (d = 0.23). Pump Alone, although
significant, was associated with small effect (d = 0.23) over No
IMPROVEMENT OF ACCURACY IN NON-INVASIVE Technology; CGM+Pump was associated with slight-no effect
GLUCOSE MEASUREMENT (NIGM) USING over No Technology (d = 0.18). For A1c, CGM Alone demon-
MID-INFRARED SPECTROSCOPY (MIR) strated small-medium effect size, although statistically non-
AND PHOTOACOUSTIC (PA) DETECTION significant, over No Technology (d = 0.48) and over Pump Alone
H.F. v. Lilienfeld-Toal1 (d = 0.37) and was comparable to CGM+Pump (d = 0.03). Pump
Alone and CGM+Pump were significantly associated with lower
1
Elte Sensoric GmbH, Gelnhausen, Germany A1c over No Technology (d = 0.24 and d = 0.49 respectively);
CGM+Pump conferred significant advantage over Pump Alone
Background and Aims: MIR offers the possibility of NIGM (d = 0.34).
that combines the advantages of high physical and physiological
specificity. In earlier experiments with PA detection in MIR we
showed that such a system is not significantly influenced by
disturbances such as fluid loss, fluid loads, physical activity and
age of the individual. Insufficient accuracy is a major limitation
in the general application of NIGM. We present data towards
sufficient accuracy of our system compared with current invasive
glucose measurement devices.
Methods: Introducing additional features to measurement
and the evaluation of the PA signals we were able to improve
the overall accuracy of the system. To prove the effect on the
in-vivo measurement, we compared the glucose values pre-
dicted by the MIR PA system to values simultaneously mea-
sured with HemoCue during an oral glucose load in tests
with 5 individuals. Datasets had been taken every 5 Min. The
data are calculated with PLS cross-validation analysis for
each test.
Results: For the resulting 174 datasets we found a good cor-
relation (r (pearson) = 0,92) and a MARD of 9%. 91% of the
points are in zone A of error grid.
Conclusions: The non-invasive MIR PA glucose sensor is Conclusions: Technology use is associated with lower ado-
work in progress. We have added refinements to the hardware lescent distress than lower parent distress. CGM alone is asso-
and software to overcome the limitations of the system and im- ciated with lower adolescent and parent distress than using pump
prove its accuracy. Although the mathematical approach to the or CGM+Pump. CGM use is associated with lower A1c than
data using PLS needs to be improved further, we believe that our pump use and has similar effect when used alone or with a
system will be able to achieve a precision comparable to that of pump. CGM use shows benefit for psychosocial and health
todays standard glucometers. outcomes.
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-87

211
THE TYPE 1 DIABETES PATIENT DECISION-MAKING
MODEL FOR ASSESSING THE INFLUENCE OF HYPO/
HYPER-ALERT SETTINGS ON CONTINUOUS
GLUCOSE MONITORING (CGM) NONADJUNCTIVE
USE
M. Vettoretti1, A. Facchinetti1, G. Sparacino1, C. Cobelli1
1
University of Padova, Department of Information Engineering,
Padova, Italy

Background and Aims: By using a type 1 diabetes patient

decision-making (T1D-DM) model (Vettoretti et al., Proc IEEE
EMBC, 2015) and in silico trials, we recently provided evidence
that CGM nonadjunctive use is safe (Edelman JDST, 2015). A
further step is assessing how customable high alert (HA) and low
alert (LA) thresholds can affect glycemic control when CGM is
used nonadjunctively.
Methods: In silico clinical trials, based on the T1D-DM
model, are designed to determine the impact of alerts settings on
Dexcom G5 Mobile (DG5M) nonadjunctive use. Simulations are can be compared (Figure). The intensity of hypoglycemia/hyper-
performed on 100 virtual subjects for two weeks with SMBG glycemia are the magnitude of the vectors of area-under-the curve
treatment and CGM nonadjunctive use, which is repeated for 21 (AUC) less than 70 mg/dL and time/day less than 70 mg/dL and
different alerts settings obtained by combining commonly used AUC greater than 180 mg/dL and time/day greater than 180 mg/dL,
thresholds for LA (80, 70 and 55 mg/dl) and HA (180, 200, 250, respectively. A prognostic glycemic risk parameter (GPR) based
300, 350, 400 mg/dl or none). Time in 70-180 mg/dl, below on the Diabetes Control and Complications Trial can be calculated.
70 mg/dl, below 50 mg/dl, above 180 mg/dl and above 250 mg/dl We applied baseline and end-of-study data from the low-glucose
metrics are assessed for the two treatments. suspend trial (ASPIRE In-Home study) (2) to the rGP and com-
Results: With DG5M, CGM nonadjunctive use, compared to pared the changes in rGP area/PGR to changes in HbA1c.
SMBG use, improves time in 70-180 mg/dl and time above (2) Bergenstal R et al. N Engl J Med 2013;369:224-232.
250 mg/dl if HA 300 mg/dl, improves time above 180 mg/dl if
HA 250 mg/dl, improves time below 70 and 50 mg/dl with all Results: We found that the rGP comprehensively describes
alert settings. Best performance is achieved with HA = 180 mg/dl overall glycemic control both numerically and visually. It cor-
and LA = 80 mg/dl. relates to both long-term and short-term risks and identifies in-
Conclusions: The in silico trials show that DG5M non- dividuals in whom the changes in HbA1c and rGP/PGR in the
adjunctive use is effective in reducing time in hypoglycemia ASPRIE in-Home study are discrepant.
regardless of the alerts setting and time in hyperglycemia pro- Conclusions: The rGP is a composite metric that provides a
vided that HA 250 mg/dl. more complete understanding of glycemic control than HbA1c. It
can be used to investigate the effects of a technology, drug or
psycho-educational intervention on glycemic control.
212
ASSESSMENT OF GLYCEMIC CONTROL
WITH A COMPOSITE METRIC: THE REVISED 213
GLUCOSE PENTAGON (RGP) SCREEN VIEW FREQUENCY AND ALERT SETTINGS
1 2 3 4
R. Vigersky , A. Thomas , J. Shin , B. Jiang , T. Sigmund , 5 AFFECT GLYCEMIC CONTROL IN CHILDREN AND
C. McMahon4 ADOLESCENTS USING REAL-TIME CONTINUOUS
GLUCOSE MONITORING
1
Medtronic Diabetes, Medical Affairs, Northridge, USA
2
Medtronic GmbH, Market Development, Meerbusch, Germany X. Liu1, J. Welsh2, T. Walker2
3 1
Medtronic Diabetes, Clinical Research, Northridge, USA Dexcom Inc., Biometrics, San Diego, USA
4 2
Medtronic Diabetes, Engineering, Northridge, USA Dexcom Inc., Clinical Affairs, San Diego, USA
5
Diabetes Private Practice, Private Practice, Munich, Germany
Background and Aims: Behavioral changes prompted by
Background and Aims: New ways to describe overall diabetes continuous glucose monitoring (CGM) screen views (SV) and
control are necessary since HbA1c does not reflect the severity/ alerts may improve outcomes. We examined whether the rate of
frequency of hypoglycemia and hyperglycemia nor glycemia SV and individualized alert settings were correlated with sensor
variability. We used continuous glucose monitoring data to update glucose (SG) values in children and adolescents.
the previously published composite metric - the Glucose Pentagon Methods: Data from 136 children ages 2-11 years and 40
(1) - creating a revised Glucose Pentagon (rGP). adolescents ages 12-17 years were extracted from an anon-
(1) Thomas A, et al. Diabetes Technol Ther 2009;11:399-409. ymized database of customers using Dexcom CGM systems who
requested technical assistance.
Methods: The five axes of the rGP describe the area of a pen- Results: Among children, SV numbered 33 17 per day and
tagon for non-diabetics (green) to which those with diabetes (red) the mean (SD) SG was 10.2 1.8 mmol/L. Among adolescents,
A-88 ATTD 2017 E-POSTER VIEWING ABSTRACTS

SV numbered 32 24 per day and the mean SG was

9.2 1.7 mmol/L. In both groups, lower-frequency SV (<16/day)
were associated with higher least-squares mean SG levels than
higher-frequency screen views (>36/day) (children, 10.4 vs.
9.7 mmol/L, respectively; adolescents, 10.0 vs. 9.0 mmol/L, re-
spectively). A separate analysis of 142 children and adolescents
who set hypoglycemia alerts showed an association between
lower (<4.4 mmol/L) thresholds and lower mean SG values (9.3
vs. 10.1 mmol/L). Similarly, those with hyperglycemia alerts set
to <11.1 mmol/L had lower mean SG values than those with
hyperglycemia alerts set to 11.1 mmol/L (8.8 vs. 10.0 mmol/L,
respectively).
Conclusions: In both children and adolescents, more-frequent
SV were correlated with lower mean SG values, suggesting that
real-time CGM data contribute to timely and appropriate treat-
ment decisions. Similarly, lower threshold alert levels for both
hypoglycemia and hyperglycemia were associated with lower Conclusions: We expect this concept to enable researchers to
mean SG values. The extent to which device interactions and more efficiently handle the administrational-, patient-, and data-
alert settings facilitate or reflect treatment decisions warrants related tasks of mHealth and eHealth interventions. By utilizing
further study. this concept in an ongoing clinical trial, we will demonstrate the
potential of evaluating mHealth technologies as disease and self-
management treatment interventions.
214
THE NEED FOR UPDATED EVALUATION
APPROACHES FOR EHEALTH AND MHEALTH 215
INTERVENTIONS A DYNAMIC CONCEPT
FOR MORE EFFICIENT TRIALS EFFECT OF TELEPHONE COACHING SUPPORT
1,2 1 1 1,3 ON GLYCEMIC CONTROL IN TYPE 2 DIABETES
E. Arsand , H. Blixgard , M. Muzny , A. Giordanengo ,
TREATED WITH GLARGINE INSULIN
M. Bradway1,2
1 V. Bellido1, C. Tejera2, F. Carral3, G. Ignacio1, A. Soto4,
University Hospital of North Norway, Norwegian Centre J.M. Garca Almeida5, C. Morales6, M. Lopez de la Torre7,
for E-health Research, Troms, Norway D. Bellido2
2
UiT The Arctic University of Norway, Department of Clinical
1
Medicine, Troms, Norway Hospital Universitario Cruces, Endocrinologia, Barakaldo,
3
UiT The Arctic University of Norway, Department of Computer Spain
2
science, Troms, Norway Complejo Hospitalario Universitario de Ferrol,
Endocrinologia, Ferrol, Spain
3
Background and Aims: Clinical trials are notorious for Hospital Universitario Puerto Real, Endocrinologia, Cadiz,
falling behind schedule and over budget. In fact, nearly 90% of Spain
4
clinical trials fail to reach intended outcomes on time. Today, Complejo Hospitalario Universitario de A Coruna,
mobile health (mHealth) technologies, e.g. apps for diabetes self- Endocrinologa, Coruna, Spain
5
management, provide additional challenges by developing faster Hospital Universitario Virgen de la Victoria, Endocrinologia,
than clinical trials are able to evaluate them. Therefore, ap- Malaga, Spain
6
proaches to assess mHealth self-management interventions, es- Hospital Universitario Virgen Macarena, Endocrinologia,
pecially randomized controlled trials, must adapt. Sevilla, Spain
7
Methods: A review of traditional eHealth and mHealth as- Hospital Universitario Virgen de las Nieves, Endocrinologia,
sessment studies revealed that tested interventions were often Granada, Spain
outdated by the time trials ended, thus reducing their quality and
potential impact. Major bottlenecks were identified and a holistic Background and Aims: eStar program is a telephone
platform for electronically supporting study management was support platform for type 2 diabetes (DM2) on glargine insulin.
proposed. Specific technological functionalities were designed This study aimed to evaluate whether this program was effective
through collaboration between researchers, patients, medical in helping patients reach their optimal insulin glargine dose
experts, and improved through consultation with our hospitals within 6 months.
research unit, the regional ethical board (REK) and Norwegian Methods: Both DM2 patients who were already receiving
Data Protection Authority (Datatilsynet). glargine and patients who started receiving it, included in eStar
Results: An electronic study-management concept was de- program from June/2013 to August/2015, were included in the
veloped. Modules with functionalities specific to major trial study. The primary outcome was whether a patient reached his/
stages (Figure 1), several of which have been separately tested her optimal glargine dose within 6 months.
with success, facilitate a new holistic system for evaluation of Results: 138 patients consented and were included in eStar
eHealth and mHealth trials. The concept is acknowledged by program. Of those, 43 discontinued the telephone follow-up
Datatilsynet and REK (ref. 2013/1906/REK sr-st B), with (control group). The median age was 63 12 years; 56.5% men;
planned implementation (autumn 2016) during the Tailoring mean duration of diabetes: 15.6 11.6 years. 34% were already
Type 2 Diabetes Self-Management project. on glargine (37% intervention group/36% control group). The
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-89

medicine for young adults may be a successful way to increase

mean baseline HbA1c was 8.9 1.7%. A significantly greater clinical care engagement during this transitional stage.
proportion of patients in the intervention group reached their
optimal insulin glargine dose (75.8%, 72/95 vs 23.3%, 10/43;
p < 0.001). The mean change in HbA1c between baseline and 6
months for the intervention group was -1.5 (-1.9 to -1.1, 217
p < 0.001) and -0.8 (-1.2 to -0.3, p < 0.001) for the control
group. Other differences between the groups at 6 months were A REVIEW OF MOBILE APPLICATIONS GENERAL
observed in insulin dose and fasting plasma glucose, in favor of PRACTICE: FORMING A QUESTIONNAIRE
the intervention. No differences in BMI were observed between TO EXPLORE CLINICIAN NEEDS
the groups. M. Smith1, M. Bradway2,3, E. Arsand2,3
Conclusions: The eStar program is an effective way to help
1
patients reach their optimal insulin glargine dose, improves University of Aberdeen, School of Medicine and Dentistry,
glycemic control and can be an effective support for diabetes Aberdeen, United Kingdom
2
education. University Hospital of North Norway UNN, Norwegian Centre
for E-health Research, Troms, Norway
3
UiT The Arctic University of Norway, Department of Clinical
216 Medicine, Troms, Norway

HOME TELEMEDICINE INCREASES THE NUMBER Background and Aims: Increasing patient-empowerment
OF TYPE 1 DIABETES CARE VISITS ATTENDED and participation in health-management is critical, especially in
BY YOUNG ADULTS preventative and chronic care domains. While this has tradi-
C. Berget1, K.A. Driscoll1, K. Ketchum1, C. Cain1, J. Raymond2 tionally been facilitated by primary care professionals, more
recently patients are turning toward mobile self-management
1
University of Colorado Denver, Barbara Davis Center, tools. While clinicians could better understand patients situa-
Aurora, USA tions through these, concerns exist, e.g. safety and efficacy.
2
Childrens Hospital Los Angeles, Center for Endocrinology- Therefore, clinicians, researchers and app-developers must iter-
Diabetes and Metabolism, Los Angeles, CA, USA atively collaborate to synthesize strategic and relatable app-use
in medical practice.
Background and Aims: To determine adherence to essential Methods: A literature review searched Embase, MEDLINE
components of T1D care: visit attendance, device downloads and and PubMed using terms related to key concepts of mobile self-
A1C measurement in young adults with T1D enrolled in a home management apps and primary care. The quality of studies was
telemedicine trial (CoYoT1). assed using the NHLBI tools. Data on efficacy, usability and
Methods: Visits occurred every 3 months for 1 year with 3 acceptability of apps was extracted. A brainstorming ses-
home telemedicine visits and 1 in-person clinic visit. Prior to sion, involving app developers and a GP with mHealth research
each visit, participants were instructed to download their T1D experience, was conducted to develop a questionnaire for
devices from home and obtain A1C at a local laboratory. clinicians.
Results: 45 young adults (Mage = 19.8 + 1.6 years; 56% fe- Results: In the review, the search resulted in 960 papers and
male) with T1D (Mdiabetes duration = 8.6 + 4.6 yrs.) participated. At 12 were finally included. Type 2 diabetes was the most com-
least 72% of participants downloaded their T1D devices and monly targeted disease-case, with apps reportedly impacting
>80% completed A1C (Table 1). In the year prior to study en- outcomes e.g. HbA1c or self-management domains. Clinicians
rollment, participants completed 2.5 + 1.2 in person T1D clinic and patients were positive about app-usage but, only 2 and 3
visits. With access to home telemedicine, participants completed studies respectively reported on these outcomes. Cited problems
significantly more T1D clinic visits (M = 3.3 + 1.1; t = -3.563; included: treatment adherence, technical difficulties and ap-
p < 0.001) compared to the number of T1D visits attended in the pointment duration. Brainstorming sessions resulted in a ques-
prior year. The proportion of participants who completed >4 T1D tionnaire (n = 18 questions) with the following sections: GPs
visits/year, as recommended by ADA, significantly improved status of app-use in practice, barriers to use, presentation of T2D
from 22% to 67%; McNemar c2 p < 0.001. apps to address GPs clinical needs.
Conclusions: Providing young adults with T1D care visits via Conclusions: Additional responses to the questionnaire are
home telemedicine increases visit frequency. Additionally, data needed to obtain actionable results. We encourage GPs to answer
needed for quality T1D care was successfully obtained for the the questionnaire (http://tinyurl.com/hs6kzt5) to further inform
majority of home telemedicine visits. Offering home tele- the society about how GPs can effectively use apps.
A-90 ATTD 2017 E-POSTER VIEWING ABSTRACTS

218
TELEHEALTH CARE MANAGEMENT FOR HIGH
RISK YOUTH WITH DIABETES MELLITUS
E. Burtman1, R. Rapaport1, A. Aluf1, H. Ciporen1, S. Lehrer1
1
Icahn School of Medicine at Mount Sinai, Pediatric
Endocrinology, New York, USA

Background and Aims: Telehealth Care Management

(TMC) is a program offered throughout the New York City
Health & Hospital system that connects patients with diabetes
mellitus to trained nurses using telecommunication technology
and provides remote, interactive healthcare services. Patients
transmit their daily blood glucose (BG) readings from their
OneTouch Ultra meter via wireless modem to TCM. BGs are
reviewed by case-specific nurses who provide care management
and contact the clinic based diabetes team as indicated. The
purpose is to extend the reach of the clinician beyond the medical
visit, enabling patients and their families to remain engaged and
supported in self-management. The TCM program is offered to
youth with poor control due in part to inadequate home super-
vision and who are at high risk for diabetic ketoacidosis.
Methods: The multidisciplinary diabetes care team at our
center selected individuals for TMC intervention. Long term data
were available in four individuals. Hemoglobin A1c (Hba1c) was
measured at enrollment, within the first year and after 1 year of
intervention.
Results: HbA1c decreased by an average of 2.5% in the first
year and remained 0.7% lower than baseline thereafter in those
with T1DM. No improvement in HbA1c was noted in the patient
with T2DM.

randomly varying real-life factors on glucose control for sub-

jects with Type 1 diabetes (T1DM). Results would confound
possible direct relationships to clinical outcomes as multiple
factors would occur daily to impact glucose control. We ex-
amined the independent effects of such factors on clinical
outcomes in silico.
Conclusions: Our preliminary data suggest that TMC plays a Methods: Within the UVA/Padova T1 Diabetes Metabolic
positive role in improving HbA1c in youth with T1DM. How- Simulation platform (T1DMS v3.2, 2013; www.tegvirginia
ever, the success of intervention is highly individualized and .com), we have modeled lifestyle variabilities in a virtual pop-
dependent on care-givers adherence to the recommended man- ulation of 100 adults with T1DM. Factors included carbohydrate
agement. Data in a larger group of TIDM and T2DM patients counting errors, insulin dosing errors, missed dosing, and blind
followed for longer duration would help validate the value blousing, as reported in the literature. All 100 subjects experi-
of TMC. enced the variabilities with identical timing and magnitude, and
all were treated based on readings from identically modeled
SMBG and CGM sensors with errors. Simulations were repeated
with factors selectively included or excluded, allowing assess-
219 ment of the individual contribution of each, in isolation and in
combination with others.
MODELING THE VARIABILITY OF DAILY LIFE Results: We successfully implemented lifestyle variability
IN-SILICO: EXERCISE, CARBOHYDRATE COUNTING models and validated each independently against reported results
ERROR, INSULIN DOSING ERROR AND METER in the literature. This series of case studies using our lifestyle
ERROR IN A T1DM POPULATION variability models demonstrates how in silico simulation allows
M. Weinheimer1, Y.Y. Chang1, S. Riddle1, G. Kongable1 examination of factors that disrupt glucose control in ways that
would be difficult to study in a clinical trial (Figure).
1
The Epsilon Group, Simulation and Modeling, Charlottesville, Conclusions: Since each varying real-life factor can be con-
USA trolled independently and is perfectly repeatable, the impact of
each factor can be examined either separately or interdepen-
Background and Aims: The use of clinical studies would be a dently, providing a more meaningful measure of their effects on
very costly and difficult approach for assessing the effect of glucose control.
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-91

220 betic foot syndrome. Every 15 minutes, the sensor stored the cur-
rent temperature. A pilot study revealed that a temperature 25C
SIGNIFICANT IMPROVEMENT OF BLOOD GLUCOSE
best predicts wearing. Per patient, wearing time and the percentage
CONTROL IN A HIGH RISK POPULATION OF TYPE 1
of days the shoe was never worn were calculated; observation time
DIABETES USING A MHEALTH APP,
was divided into quartiles to analyse the course of adherence.
A RETROSPECTIVE OBSERVATIONAL STUDY
Results: On average, data from 133.5 days per patient could
M. Hompesch1, K. Kalcher2, F. Debong3, L. Morrow1 be analysed. Patients wore their diabetic footwear on 4.2 3.6 h/
1 day. On 51% of days, patients did not wear their diabetic foot-
Profil Institute for Clinical Research, Inc., Scientific Services, wear at all. There was a significant gender*time interaction for
Chula Vista, USA the number of days the footwear was worn: While early adher-
2
Medical University of Vienna, Med Statistik, Vienna, Austria ence was similar between men and women, men showed a sig-
3
mySugr GmbH, Community Relations, Vienna, Austria nificant increase in the percentage of days they have worn their
shoes, while women showed a continuous decrease.
Background and Aims: To investigate the impact of mySugr Conclusions: Nearly every second day patients did not wear
on blood glucose (BG) control in a less well-controlled popula- their footwear at all. Men and women showed a comparable early
tion. mySugr (registered class-I medical-device application) adherence but women had more problems with late adherence as
was developed to make management and use of metabolic health they were more and more unwilling to wear their footwear.
data appealing; it is the market leading diabetes-app with over Feedback of the temperature sensors could be used to timely
900,000 registered users (11.2016). It has shown positive effects detect problems with adherence in order to enhance it.
in well-controlled populations1, when compared to the previ-
ously published impact of mHealth-tools2.
Methods: A randomly selected group of 440 highly engaged
users (logging 5 days/week for 6 months) with mean(t0) 222
183 mg/dl (representing eA1c 8%) were included; T1D, age
30.8 15.3 years, 47.3% female.Changes in BG-results (mean, CHARACTERIZATION OF THE EFFECT
standard deviation (SD)), HighBloodGlucoseIndex and Low- OF INTERSTITIAL FLUID PH IN PUMP INFUSION
BloodGlucoseIndex at baseline (t0), week 2-4 (t1) and month 3-6 SITE SUBCUTANEOUS TISSUE DUE TO
(t2) were analyzed. Baseline data (t0) was processed using an INFLAMMATORY RESPONSE USING ANIMAL
intercept of regression model based on data from first week of use. MODELS
Results: Baseline BG was 210.75 69.36 mg/dl, dropping J. Fusselman1, G. Zhang1, S. Chattaraj1, V. Patel1
to 173.08 63.26 mg/dl (t2) reduction in mean of 17.88%
1
(p < 1*10^-23), SD 8.79% (p < 0.005). HBGI dropped from 5.39 Medtronic Diabetes, Chemistry- Material and Microbiology,
(t0) to 3.36 (from High to Medium risk) (p < 1*10^-10), whereas Northridge, USA
LBGI rose from 0.22 (t0) to 0.44 (t2) (p < 1*10^-36).
Conclusions: The results demonstrate that logging alone may Background and Aims: No Delivery alarm is often
positively impact the quality of blood glucose control. These found in insulin infusion pumps and one possible source of No
findings need to be confirmed in a prospective, controlled clinical Delivery alarm is occlusion of the infusion set by insulin
study. We hypothesize that the additions of mySugr Bolus and aggregates. Insulin (Humalog or Novolog) is supplied at pH
CDE-led Coaching may result in further benefits. 7.0 - 7.8 and has an isoelectric point (pI) of approximately 5.6.
Insulin increasingly forms aggregate precipitates as the pH
decreases and approaches the pI. Insertion of the cannula of
infusion set into the subcutaneous tissue of patients can cause
221 an immune response inflammatory effect, which can lower
the tissue pH, hence an in-vivo test using an animal model was
USING TEMPERATURE SENSORS TO ASSESS designed to verify the hypothesis: inflammatory response caused
THE COURSE OF ADHERENCE TO CUSTOMISED the pH lowing effect that triggers the insulin aggregation/
DIABETIC INSOLES precipitation.
D. Ehrmann1, M. Spengler2, M. Jahn2, D. Niebuhr3, T. Haak1, Methods: Two in-vivo studies were performed to verify the
B. Kulzer1, N. Hermanns1 above hypothesis. Both studies used a Type 1 diabetic miniature
swine model.
1
Research Institute Diabetes FIDAM, Research Institute of the
Diabetes Academy Mergentheim, Bad Mergentheim, Germany 1. Study 1: In-vivo pH measurement of subcutaneous tissue/
2
IETEC foot orthotics GmbH, Research and Development, body fluid under various degree of injury and cannula in-
Kunzell, Germany sertion techniques.
3
University of Applied Sciences, Nursing and Health Sciences, 2. Study 2: Pump occlusion test designed with various man-
Fulda, Germany made pH decreasing event.
Results: The experiment has demonstrated that minimal to
Background and Aims: Adherence to customised diabetic mild sub-acute inflammation occurred in association with in-
footwear is crucial for a preventive effect regarding serious di- sertion of the cannula, and the lowing of tissue pH in the inflamed
abetic foot problems. However, adherence data often rely on self- tissue was also demonstrated in this study.
report or cover rather short time spans. The aim of this study was Conclusions: Although the lower pH due to the inflammatory
to objectively assess the course of patients adherence with a response did not precipitate insulin in this study, increased pump
temperature sensor. force for more than 30 minutes was observed on several occa-
Methods: Temperature sensors were incorporated into the sions, which is similar to the pump force obtained from the field
specialized footwear of 26 patients with type-2-diabetes and dia- returned pump traces.
A-92 ATTD 2017 E-POSTER VIEWING ABSTRACTS

223 224
LONG-TERM USE OF MHEALTH TECHNOLOGY IMPACT OF AN INSULIN/CARBOHYDRATE
IN PATIENT WITH TYPE 1 DIABETES LED ADJUSTMENT ALGORITHM APP ON INDIVIDUAL
TO IMPROVEMENT IN METABOLIC CONTROL: BLOOD GLUCOSE MANAGEMENT DURING
A CASE STUDY PLANNED EXERCISE
A. Holubova1,2, J. Muzk1,2, M. Muzny1,3, M. Polacek2, J. Hall1, K. Stephen1, A. Croall2, J. MacMillan3, L. Murray3,
D. Fiala1, E. Arsand3,4, J. Kaspar1,2, K. Hana2, P. Smrcka2, N. Wiratunga4, S. Massie4, S. MacRury1
D. Janckova Zdarska5, M. Kvapil5, J. Broz5 1
University of the Highlands and Islands, Division of Health
1
Charles University, Spin-off company and research results Research, Inverness, United Kingdom
2
commercialization center of the First Faculty of Medicine, Diabetes Scotland, Inverness, United Kingdom
3
Prague, Czech Republic University of the Highlands and Islands, Educational
2
Czech Technical University in Prague, Faculty of Biomedical Development Unit, Inverness, United Kingdom
4
Engineering, Prague, Czech Republic Robert Gordon University, School of Computing Science and
3
Norwegian Centre for E-Health Research, University Hospital Digital Media, Aberdeen, United Kingdom
of North Norway, Troms, Norway
4
UiT The Arctic University of Norway, Department of Clinical Background and Aims: Risk of hypoglycaemia is a signifi-
Medicine, Troms, Norway cant barrier to undertaking physical exercise among people with
5
Second Faculty of Medicine, Department of Internal Medicine, type 1 diabetes. An app based on a personalised prediction al-
Prague, Czech Republic gorithm is being developed for people with type 1 diabetes to
optimise blood glucose control associated with physical activity.
Background and Aims: Mobile Health (mHealth) tech- Incorporation of case-based reasoning and linkage with online
nologies for monitoring and evaluation of data, such as gly- structured education is being explored for personalisation.
caemia levels, carbohydrate intake, insulin doses, physical Methods: Physically active people with type 1 diabetes
activity, and others, in patients with type 1 diabetes enable (n = 10) were recruited using a Facebook forum to test a co-
faster orientation in the current glycaemia and factors influ- produced app based on an insulin/carbohydrate adjustment al-
encing them. At the same time, displaying health informa- gorithm. Participants wore Actigraph-GT3X activity monitors
tion and clear presentation of important parameters has the and Libre Freestyle blood glucose meters over the eight week
potential to enable more adequate response to glycaemia study period. Physical activity, blood glucose and app use time-
changes throughout the day and increase patients self-care lines for individual participants were analysed to assess the im-
interest. pact of the app on planned activity.
Methods: During a testing phase of the mHealth system a Results: A total of 119 app uses were logged, with an average
22-year old Type 1 diabetes patient was equipped with the of 3 uses per week per user across a range of activities. An
Diabetes Diary smartphone application, glucometer with a example of app use is shown. Once ready to exercise, blood
Bluetooth interface, smartwatch and activity tracker, for a glucose and activity/intensity were entered. The app indicated
period of 24 months. Values of measured glycaemia, carbo- that blood glucose was too low, recommended carbohydrate
hydrate intake, insulin doses and physical activity were au- consumption and retesting blood glucose before undertaking the
tomatically transferred to a secure server where which both selected exercise intensity. Advice was followed and 30 minutes
the patient and the clinician had access to the data through a later, with low blood glucose treated, the participant had a suc-
web application. cessful exercise session.
Results: During the period of 2 years prior to the com-
mencement, the mean HbA1c value was 80.7 2.3 mmol/mol.
During the period the patient used the devices, the mean
HbA1c value was 65.7 7.36 mmol/mol. The patient did
not observe any increase in frequency of hypoglycemia. He
associates the improved metabolic control with the use of
mHealth system.

Conclusions: When using the above mentioned devices

and applications, there was a significant improvement in the
mean HbA1c values (-18.6%) without a subjective increase
in hypoglycemia values. The patient adapted the technology Conclusions: Use of the app improved blood glucose levels
in his everyday life and expressed his desire to keep using during and after exercise for some participants. For other par-
them. ticipants it highlighted that one-size-fits-all advice may not be
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-93

appropriate, supporting the use of case-based reasoning, even Methods: SAP is changing the life in patients with type 1
for people already confident in managing their diabetes during diabetes, offering both insulin delivery and continuous glucose
physical activity. monitoring with automatic suspension (when low or before low
glucose value is achieved). Carelink is a software which allows
evaluation of SAP (basal/ bolus insulin, glucose values, carb
intake, automatic and manual suspension). This robust data re-
225 quires more time for evaluation, which overwhelms health care
USE OF AN AUTOMATED BOLUS CALCULATOR providers (HCP) and can underutilize the benefits of this system
BY A TELEMEDICINE SYSTEM FOR THE in patients. Time spent for interpreting the results is not recog-
MANAGEMENT OF INSULIN THERAPY nized by health care authorities. We allocate five segments on the
IN TYPE 1 DIABETES PATIENTS dashboard page of Carelink in a systematic approach: 1) Basic
statistics with hypo and hyper events; 2) Modal day with insulin
A.R. Maurizi1, R. Del Toro1, A. Naciu1, A. Lauria Pantano1, and glucose profile; 3) Postprandial data; 4) Basal/Bolus data;
E. Fioriti1, S. Manfrini1, P. Pozzilli1 5) Sensor and suspension statistics, where every segment is an-
1
University Campus Bio-Medico, Endocrinology and Diabetes, alyzed with correlation through others.
Rome, Italy Results: Our experience of this approach over 2 years in 142
patients on SAP shows less time spent for interpreting and
greater satisfaction by HCP with improvement of glucose control
Background and Aims: In Type 1 Diabetes (T1D) patients on
in type 1 diabetes patients using SAP.
multiple daily injections (MDI), adjustments of insulin dose at
Conclusions: Our approach is easy and practical and can
meal times must be made by taking into account several parameters
be used in everyday practice in type 1 diabetes patients using
as blood glucose levels, insulin/carbohydrate ratio, carbohydrate
SAP.
intake at each meal. A bolus advisor system (Accu-Chek Aviva
Connect) developed for the establishment of insulin doses to be
administer, takes into account all above parameters.
Aim of this randomized trial was to evaluate the efficacy of a 227
bolus advisor system on glycaemic control and patients com-
pliance to Self-Monitoring of Blood Glucose (SMBG), using a INSULIN PUMP AND CGM DATA TO BE USED
telemedicine system. FOR SEARCHING OF NONCOMPLIANT PATIENTS
Methods: 24 T1D patients were enrolled and HbA1c and AND THERAPY MISTAKES
patients compliance were recorded at entry into the trial and at 3 Y. Philippov1, A. Agadzhanyan1, L. Ibragimova1, E. Surkova1,
and 6 months follow-up. As secondary end-points Kovatchev A. Mayorov1
indices were evaluated. Paired t test (two tailed) and analysis of
1
variance were used to evaluate differences at different time Endocrinology Research Centre, Diabetes Institution,
points. Moscow, Russia
Results: HbA1c at entry was 7.65% 0.87 (SD) in patients
using bolus advisor system with bolus calculator and data trans- Background and Aims: In real-life CSII and CGM effec-
mission by App on a Smartphone activated and 7.55% 0.98 tiveness can be affected by several factors, such as low com-
(SD) in the control group with bolus advisor turned off. pliance, wrong pump/CGM settings, technical issues, knowledge
After 6 months of observation, a significant reduction in HbA1c deficiency, and so on.
was observed in active group vs. control subjects (7.32% 0.82 vs. The aim of our study was to use stored in pumps data for
8.32 1.38 P = 0.04). searching patients to be actively supervised (non compliant, with
A major compliance to SMBG assessed as mean number of significant mistakes in settings).
daily measurements (P = 0.03) and as total of the measurements Methods: Cross-sectional observational single-centre study
for each quarter (P = 0.02) was observed in active group vs. includes data from type 1 diabetes mellitus patient transferred to
control group. CSII in Endocrinology Research Centre. All participants were
Conclusions: This bolus advisor system is a friendly wire- treated in routine outpatient regimen within last year or longer.
lessly meter that helps to improve glycaemic control and patients We selected only CSII+CGM users (Medtronic Paradigm Real-
SMBG compliance. Time and Veo pumps) who use sensors in real-life for self-
monitoring.
Results: We included data from 75 patients with long history
of CSII using (3,6 1,7 years) and found several types of issues:
226
1) Technical issues: hight infusion set use duration; loss of
INTERPRETING THE DATA FROM SENSOR Fixed Primes; extreme Fixed Prime amount;
AUGMENTED PUMP IN TYPE 1 DIABETES: 2) CGM issues: extreme high/low pump alarms frequency;
A 5-STEP APPROACH IN 10 MINUTES manual stop pump due to hypo- exists; hypo- alarms
G. Petrovski1, M. Zivkovic1 without actions; sensor self-calibrations; calibrations due to
rapid BG changes;
1
University Clinic of Endocrinology, Center for Insulin Pump, 3) Bolus wizard issues: No of Carbohydrate ratios/day pro-
Skopje, FYR Macedonia file; unsuitable BG targets; incorrect Active insulin set-
tings;
Background and Aims: We are presenting a simple, sys- 4) General diabetes compliance: Carbohydrate under-/
tematic 5 step approach in 10 minutes to interpret the data ob- overcounting; low SMBG frequency;
tained from Sensor Augmented Pump (SAP) using Carelink Pro 5) Advanced pump functions issues: Dual/Square boluses
software (Carelink). use; Temp basal use;
A-94 ATTD 2017 E-POSTER VIEWING ABSTRACTS

6) Pump settings self-adjustment: Basal Rate profile chan- plication convenient and helpful. The largest drawback of the
ges; Bolus Wizard settings changes. application the patients found was the lack of food items in the
build-in food database (which included 2110 items).
We found 2 types of issues in all participants. The study was funded by Russian Science Foundation (project
Conclusions: Low compliance and technical mistakes can No 15-14-30012).
persist in CSII+CGM users very often. Searching and selecting
patients with such issues can be a tool in routine healthcare
system.
229
KALMAN SMOOTHING OF GLUCOSE DATA APPLIED
228 TO PARTIAL LEAST SQUARES MODELING
OF NON-INVASIVE NEAR-INFRARED
REMOTE MONITORING OF DIET FOR PATIENTS
MEASUREMENTS
WITH GESTATIONAL DIABETES USING A
SPECIALIZED MOBILE APP DIARY O.M. Staal1,2, T. Karstang1, S. Slid1, . Stavdahl2
E. Pustozerov1, P. Popova2, Y. Bolotko2, A. Tkachuk2, 1
Prediktor, Medical, Gamle Fredrikstad, Norway
A. Gerasimov2 2
Norwegian University of Science and Technology, Department
1 of Engineering Cybernetics, Trondheim, Norway
Saint-Petersburg Electrotechnical University, Department
of Bioengineering Systems, Saint Petersburg, Russia
2 Background and Aims: Blood and interstitial fluid glucose is
Federal Almazov North-West Medical Research Centre,
measured in people with diabetes using fingerprick and Continuous
Institute of Endocrinology, Saint Petersburg, Russia
Glucose Monitor (CGM) devices. We show that such data can be
homogenized, cleaned and interpolated using a preprocessing al-
Background and Aims: To maintain the appropriate control
gorithm. The algorithm is used to generate a suitable glucose re-
of patients with gestational diabetes throughout the course of
sponse variable from fingerpricks for Partial Least Squares (PLS)
treatment, a specialized software was developed and implemented
modeling of non-invasive near infrared (NIR) transcutaneous
into the clinical practice. It allows physicians to receive stan-
measurement data recorded with a prototype device.
dardized reports on patients diet and glucose monitoring plan in-
Methods: The algorithm is based on Kalman filtering and
between visits and provide patients with a remote feedback.
Rauch-Tung-Striebel smoothing. A discrete-time input sequence
Methods: Patients with diagnosed gestational diabetes were
of fingerprick/CGM data is converted to an interpolated stream
given an application, developed for both mobile devices and
of estimates of the blood/interstitial glucose. The algorithm is
desktop computers, which was used to keep track on the diet,
robust against measurement errors in the input data and can
blood glucose (BG), physical activity, sleep and ketones. The
correct erroneous readings when enough data is present. The
records collected onto a standardized spreadsheet were sent via
algorithm is compared to cubic spline interpolation in simulated
E-mail and stored remotely for further analysis.
and real glucose data sets, and the influence on PLS modeling of
Results: By October 2016, 138 patients were included in the
NIR datasets is investigated.
study. A total of 17820 measurements of BG (in average,
Results: The interpolated output signal from our algorithm
45.4 days of monitoring by patient), and 19527 meals (in aver-
describes the most likely glucose trajectory through the input
age, 36.7 days of monitoring by patient) recorded by patients
dataset of noisy glucose measurements, with a measure of the
were analyzed. The average percent of BG levels for a single
uncertainty (variance) of the estimate at any point along the
patient during fasting (n = 4956) exceeding 5.1 and 5.3 mmol/L
trajectory. This is useful in order to determine when the estimate
were 27.6% and 19.4%; after meal (n = 12864) exceeding 7.0 and
is trustworthy enough to be used in further processing. Use of the
7.6 mmol/L - 20.0% and 8.3% respectively. The average carbo-
algorithm in PLS modeling improves the cross validation cali-
hydrates consumption was distributed during breakfast, lunch,
bration performance and prediction performance compared to
dinner and additional food intakes in the proportion of 27.5%/
cubic spline interpolation.
29.5%/26.0%/17.0%.
Conclusions: The presented algorithm is a useful preproces-
Conclusions: According to the post-study survey, all the pa-
sing step in many applications of automated and manual glucose
tients, who provided the feedback, found the usage of the ap-
data processing, including PLS modeling of NIR data for non-
invasive glucose measurements.

230
SMARTPHONE APP FOR NOCTURNAL
HYPOGLYCEMIA PREDICTION FROM SMBG
MEASUREMENTS
P. Tkachenko1, Y. Sulema2, E. Renard3, S. Pereverzyev1
1
Johann Radon Institute for Computational and Applied
Mathematics, Inverse Problems and Mathematical Imaging,
Linz, Austria
2
National Technical University of Ukraine Igor Sikorsky Kyiv
Polytechnic Institute, Applied Mathematics Department, Kyiv,
Ukraine
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-95

3
Montpellier University Hospital, Department of Conclusions: The app is being developed in NTUU KPI,
Endocrinology, Diabetes, Nutrition, Montpellier, France Kyiv, Ukraine. Further improvements and extensions of the app
are planned to create a ready-to-use product.
Background and Aims: Nocturnal Hypoglycemia (NH) is the
most feared type of hypoglycemia in patients with diabetes
treated by insulin. However, there is still a lack of methods
aiming at the prevention of such cases, especially for patients 231
performing Self Monitoring of Blood Glucose (SMBG) from ACCURACY OF CONTINUOUS A1C ESTIMATE
finger sticks. USING ROUTINE MONITORING DATA
The aim is to develop a smartphone app DiaSafeNight
equipped with a special tool for predicting Nocturnal Hypogly- P. Vaisanen1, P. Lonnroth2, M. Saraheimo3
cemia (NH) from SMBG measurements. 1
Tampere University of Technology, Electronics and
Methods: DiaSafeNight incorporates the general function- Communications Engineering, Tampere, Finland
ality of commonly used logbook apps (data collection, storing, 2
Quattro Folia Oy, Research, Espoo, Finland
analysis and graphical representation) together with the novel 3
Neliapila Oy, Clinic, Helsinki, Finland
NH predictor. The new mathematical technique introduced by
the authors and staying behind the predicting algorithm allows
Background and Aims: The measurement of A1c is the most
an aggregation of the clinical prediction rules known from the
well established means to monitor glycemic control in persons
literature in such a way, that the resulting predictor performs at
with diabetes. Often A1c measurements are done 2-4 times per
the level of the best involved one, or even outperforms all of
year, although A1c levels can change substantially in 3-4 weeks.
them. DiaSafeNight has been tested on historical clinical data
This might lead to unmonitored gaps. To prevent this, we have
collected within the European FP7 project DIAdvisor and
introduced an algorithm for continuous A1c estimate (eA1c) to
H2020-MSCA-RISE-2014 project AMMODIT. The pre-
guide in efficient day-to-day diabetes management. Here, we
diction accuracy observed on these data confirmed the above
evaluate the accuracy and robustness of the algorithm by using
statement on the performance and is higher than conventionally
routine monitoring data.
considered as acceptable. The prediction is expressed in terms
Methods: The accuracy of the algorithm was analyzed with
of the probability of NH during the forthcoming night (in %)
30 subjects with diabetes. 9 subjects were excluded by having too
and categorized into 4 risk zones: low risk, moderate risk, high
little SMBG data or laboratory A1c measurements. In total, 9123
risk, and extreme risk.
SMBG and 76 A1c measurements were included. The algorithm
Results: The beta-version of the smartphone app DiaSafe-
adapts to users SMBG measurement habits and individual
Night has been created.
metabolic characteristics to estimate A1c. Retrospective analyze
with leave-one-out cross-validation was used for algorithm ac-
curacy evaluation. Mean absolute deviation (MAD), mean ab-
solute relative deviation (MARD) of eA1c from reference A1c,
and correlation (R) between eA1c and reference A1c were cal-
culated. In addition, an error grid analysis was made.
Results: MAD and MARD were 0.34% and 4.45%, respec-
tively. Correlation was strong (R = 0.87). Error grid analysis
showed that 90.14% of eA1c values were within 10% from ref-
erence A1c.
Conclusions: The results show that the adaptive algorithm
can accurately calculate continuous A1c estimate with routine
monitoring data that can be even biased or irregular. Further-
more, the introduced continuous A1c algorithm is one of the best
described in the literature and thus, it can be used to guide in day-
to-day diabetes management.

232
NO ESCAPE: BENEFITS OF DIGITAL RETINAL
SCREENING IN PRISON
C. Wallis1, O. Anderson2
1
Dorset Healthcare, Dorset Diabetic Eye Screening
Programme, Dorset, United Kingdom
2
The Royal Bournemouth and Christchruch Hospitals,
Ophthalmology, Bournemouth, United Kingdom

Background and Aims: To establish levels of diabetic reti-

nopathy in the prison population of Dorset from 2012 - 2016 and
compare to the normal Dorset population with diabetes.
To establish if screening in prisons is a cost effective way of
monitoring referable retinopathy not requiring treatment.
A-96 ATTD 2017 E-POSTER VIEWING ABSTRACTS

Methods: Any retinopathy (R1 background, R2 prepro-

liferative, R3 proliferative and M1 maculopathy using English
National Diabetic Eye Screening criteria) and referable reti-
nopathy rates (R2 preproliferative, R3 proliferative and M1
maculopathy) for the prisoner population (2012-2016) were
compared with normal population (2012-2015).
Results: 171 male prisoners were screened. 99 (57.9%) had no
diabetic retinopathy, 72 (42.1%) had any retinopathy and 18
(10.5%) had referable retinopathy. Of those with referable reti-
nopathy 8 were referred to hospital eye service (HES) and 10
were monitored using digital imaging.
78,112 of the normal diabetic population were screened.
52,951 (67.8%) had no diabetic retinopathy, 25,161 (32.2%) had
any retinopathy and 4,658 (6%) had referable retinopathy.
The prison population had a statistically higher rate of refer-
able retinopathy compared to the normal population (Fishers
exact test p = 0.0186).
Conclusions: The prison population showed higher levels of
referable retinopathy. The cost of taking a prisoner out of prison
for a medical appointment is 300 and an outpatient appointment
in HES is 120. Screening a prisoner in prison costs 33.98. The
total cost of 10 patients attending HES is 4,200 with associated
security risks vs 339.80 using digital imaging screening. Digital
imaging is a cost effective way of monitoring referable diabetic
retinopathy not requiring treatment.

233 The higher initial HbA1c was, the greater it dropped (Table 3).
INSULIN PUMP IN TYPE 2 DIABETES PATIENTS No severe hypoglycemia was observed (3 during the multi-
WITH INSULIN OPTIMIZATION FAILURE: injections year preceding T0).
RESULTS OF AN OBSERVATIONAL, The weight increased moderately with some heterogeneity 29
MULTICENTRIC TRIAL IN REAL LIFE patients remained stable, 21 lost weight (3 10%); 30 gained
3%, 18 5% et 5 10% of their initial weight.
J.P. Courreges1, F. Travert2, S. Clavel3, D. Huet4, Insulin dose dropped significantly by 0.25 U/kg/d (-20%) ;
E. Ghanassia5, K. Mohammedi2, O. Dupuy4, A. Poussier3 in 20 patients (22%) it decreased by 30%, in 6 by 50%
1
Hopital General, Service de Medecine interne et Diabetologie, (Table 5).
Narbonne, France Conclusions: In real life , multicentric results of a 1-year
2
CHU Bichat, Diabetology, Paris, France AIP-treatment confirm its benefit for T2D insulin requiring pa-
3
CH Hotel Dieu, Diabetology, Le Creusot, France tients with insulin optimization failure in spite of very high
4
CH Saint Joseph, Diabetology, Paris, France insulin doses. It produces a very significant drop in HbA1c
5
Clinique Sainte Therese, Diabetology, Sete, France (-1.55%) and in insulin needs (-20%) without any severe hy-
poglycemia or important weight gain.
Background and Aims: Can type 2 diabetic (T2D) patients
with conventional insulin optimization failure benefit from an
ambulatory insulin pump (AIP) in real life? 234
Methods: 110 T2D patients (diagnosed for 16.0 8 years, age
: 60.0 9.6 yrs, sex ratio M/F: 1.56) showing insulin optimization OPPORTUNITIES AND LIMITS OF AN AMBULATORY
failure (HbA1c8% in spite of 3 injections/d - 0.7 U/kg/d) INSULIN PUMP TREATMENT IN TYPE 2 DIABETIC
undergo AIP treatment in an observational multicentric trial for ELDERLY PATIENTS ( 65 YEARS) IN CASE OF
12 months. INSULIN OPTIMIZATION FAILURE
Results: There were 18 trial discontinuations among which 8 D. Huet1, S. Clavel2, F. Travert3, E. Ghanassia4, J.P. Le Berre5,
AIP removal. 92 patients (83.6%) reached the end and were H. Bonnaure6, V. Cosma7, J.P. Courreges8
analyzed. 1
With AIP, HbA1c results improve significantly at T0/T6/T12 CH Saint Joseph, Diabetology, Paris, France
2
(Table 1). At T0, T6 and T12, HbA1c 8% and 10% proportion CH Hotel Dieu, Diabetology, Le Creusot, France
3
evolves (Table 2). CHU BIchat, Diabetology, Paris, France
4
Clinique Sainte Therese, Diabetology, Sete, France
5
HIA Desgenettes, Diabetology, Lyon, France
6
CH Narbonne, Diabetology, Narbonne, France
7
CHU Nmes, Diabetology, Nmes, France
8
Hopital General, Service de Medecine interne et Diabetologie,
Narbonne, France
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-97

Background and Aims: Is ambulatory insulin pump (AIP) 235

a valuable opportunity (efficacy and safety) for type 2 dia-
WHO WOULD BE THE BEST CANDIDATE
betic (T2D) elderly patients in case of insulin optimization
FOR AN AMBULATORY INSULIN PUMP (AIP)
failure?
TREATMENT AMONG TYPE 2 DIABETIC (T2D)
Methods: 30 elderly T2D patients aged over 65 years (13 70
PATIENTS WITH INSULIN OPTIMIZATION FAILURE
years) and 75 years (diagnosed for 19 9 years, 69.6 4.8 years
old, sex ratio (M/F) 1.31), consecutively enrolled (E group), S. Clavel1, F. Travert2, D. Huet3, E. Ghanassia4, J.M. Andrieu5,
showing glycemic control failure in spite of insulin multi- L. Labbe1, R. Zaharia3, J.P. Courreges6
injections (3.7 1.1 injections/d) are treated with AIP for 12 1
months. CH Hotel Dieu, Diabetology, Le Creusot, France
2
Their results are compared with a 54-patient group (NE group) CHU Bichat, Diabetology, Paris, France
3
identical to the previous but younger (<65 years). CH Saint Joseph, Diabetology, Paris, France
4
Results: The comparative results of E vs NE are collected at Clinique Sainte Therese, Diabetology, Sete, France
5
T0, T6 and T12. CH Beziers, Diabetology, Beziers, France
6
Glycemic control evolves as follows: Hopital General, Service de Medecine interne et Diabetologie,
Narbonne, France

Background and Aims: Insulin optimization failure may lead

to examine the accuracy of an AIP in T2D patients. To whom
may it benefit the most in terms of metabolic results?
Methods: In a multicentric observational trial, 92 T2D pa-
tients (diagnosed for 16.0 8 years, aged 60.0 9.6 yrs, sex
ratio M/F: 1.56, BMI 33.6 4.0 kg/m2) with multi-injections
(n = 3.6 1.1/d) failure (HbA1c 9 .52 1.47%) underwent an AIP
treatment for 1 year.
Results: After 1 year of AIP, patients reached a better glycemic
control (HbA1c 7.96 1.0% - DT0/T12 = -1.56% -p <0.0001).
There were classified into 3 groups according to HbA1c re-
Weight and insulin doses evolution are not significantly dif- duction between T0 and T12:
ferent:

The 3 groups were compared on several parameters at T0.

No significant difference was noticed between the 3 groups in
terms of sex ratio, age, duration since diagnosis, weight/BMI,
insulin dose, micro and macro-angiopathy proportion and pre-
existing metformin treatment. Only initial HbA1c level was
significantly different between the groups.
HbA1c reduction according to initial HbA1c level was found
In the E group, 3 AIP treatments (10%) have been suspended linear until it reached 10% then went exponential.
on patient demand (+ 1 death not related with AIP) vs 5 (9.3%) in
the NE group.
No severe hypoglycemia has been noticed in both groups.
The burden of hospitalizations linked with diabetes has been
studied:

Conclusions: These results show the feasibility in real life of

an AIP treatment in elderly T2D patients ( 65 years) with
similar levels of efficiency and safety as in younger patients.
These identical results for both groups confirm the accuracy of an
AIP treatment in elderly T2D patients in case of multi-injections
failure.
A-98 ATTD 2017 E-POSTER VIEWING ABSTRACTS

Results: 468 patients were enrolled (female 257) 196 in SCII.

Average age SCII 25.59 3 y and MDI 30.61 4 y. Diagnostic
age 15.76 2 SCII and MDI 11.17 1 y; HbA1c SCII 7.5% 0.89,
MDI 8.32 1.54 (p < 0.0001) frequency of measurements SCII
4.36 2.4, MDI 3.49 1.64 (p < 0.0001) mean glycaemia
SCII162.18 27.3, MDI 179.9 40.88 (p < 0.0001); SD SCII
76.49 18.76, MDI 81.45 23.24 (p < 0.0001) % of Variation SCII
48.17 9.51, MDI 45.88 10.24 (p 0.09).
Conclusions: Based on our results type 1 DM patients on SCII
have better metabolic control, higher frequency of self-monitoring,
lower HbA1c and glycemic less variation than those receiving MDI.

237
BASAL INFUSION RATE SELF-MANAGEMENT IN CSII
AND METABOLIC CONTROL
U. Di Folco1, M.R. Nardone1, M. Timpano1, C. Tubili1
1
Patients have been classified into 4 groups according to S. Camillo Forlanini, Diabetology, Rome, Italy
composite endpoint (HbA1c reduction and weight evolution) to
assess more specifically the level of metabolic control they Background and Aims: Many CSII diabetic patients change
achieved (no severe hypoglycemia occurred): often Basal Infusion Rate (BIR) for short or long periods. It is not
Conclusions: 68.5% of highly insulin resistant patients, with clear if this self-management improves the metabolic control,
insulin optimization failure, improved their overall metabolic more specifically reducing hypoglycemic events.
control with an AIP, while 28 patients improved a composite Aim of this study was to evaluate the correlation between BIR
endpoint (HbA1c reduction 1% + stable or increasing weight self-management and metabolic control in CSII type 1 diabetes
3%). subjects in real life.
No clinical profile has been identified except initial high Methods: 109 outpatients were stratified in 3 groups ac-
HbA1c level both with single or composite endpoint. This may cording to HbA1c: 1) < 7% (n = 50; 45.8%); 2) 7.1- 7.9% (n = 29;
result from highly heterogenic enrolled patients. 26.6%); 3) >8.0% (n = 30; 27.5%). Total and basal insulin re-
quirements (U/day; U/kg); average basal rate (U/h), number of
basal rate daily changes; temporary basal speed frequent usage
( 3 times/week); severe hypoglycemic events (<50 mg/dl) in the
236 last 4 weeks were analyzed (t test).
DIFFERENCES IN THE METABOLIC CONTROL Results: Well controlled pts. (group 1) were younger than
IN PEOPLE WITH TYPE 1 DIABETES TREATMENT groups 2 and 3 (p < 0.005) ones and delivered less basal insulin
WITH SUBCUTANEOUS INSULIN INFUSION VS (55.5 7.7%, vs. 57.7 13.9% and 56.8 10.8% respectively).
SCHEME WITH MULTIPLE DAILY DOSES Their insulin requirement and average infusion rate were lower
(0.28 0.1; 0.29 0.09; 0.34 0.14 U/kg; p < 0.03 between
A. Dain1, L. Rista2, M.L. Ruiz Morosini3 groups 1 and 3; 0.82 0.3; 0.85 0.3; 1.09 0.5 U/h; p < 0.004
1
Catedra de Biologia Celular - Histologia y Embriologia, between groups 1 and 3).Severe hypoglycemic events were re-
Facultad de Ciencias Me, Cordoba, Argentina ported by 5 (10%) well controlled patients, 3 (10%) by moder-
2
CEDYN Rosario, Rosario, Argentina ately uncontrolled, 5 (18%) by severely uncontrolled ones. In
3
Hospital Nacional de Clnicas, Buenos Aires, Argentina group 1 the temporary basal rates usage was more frequent (46%)
than in the other two groups.
Conclusions: Basal rate self-management is an important
Background and Aims: Type 1 diabetes is a complex disease.
component of CSII and seems effective to maintain good HbA1c
Technology has been incorporated offering benefits in metabolic
levels and to prevent hypoglycemic events.
control and quality of life.
Aim: To analyze the degree of metabolic control in people
treated with SCII vs MDI, assessing glycosylated hemoglobin
(HbA1c) and other parameters of glycemic variability 238
Methods: We performed a multicenter restrospective analysis
and we included Type 1 DM patients, on SCII or MDI treatment LONG-TERM EFFECT OF CONTINUOUS
by 6 months, from 2 years old with at least 1 year from diagnosis SUBCUTANEOUS INSULIN INFUSION
between 2015-16. ON GLYCAEMIC CONTROL
We excluded pregnant woman, drug abuse, ketoacidosis, A. El-Laboudi1, A.A. Min1, J. Carling1, P. Hammond1
surgery in the last 3 months or eating disorder.
1
Data were anonymous and ethical requirements were met. Harrogate and District NHS Foundation Trust, Diabetes
Blood glucose data were collected through commercial soft- and Endocrinology, Harrogate, United Kingdom
ware (Glucose average, SD, number and frequency of mea-
surements over 30 days). The HbA1c levels were determined. Background and Aims: Reduction in HbA1c using contin-
We analyzed: t Test, Shapiro-Wilks, Wilcoxon and ANCOVA uous subcutaneous Insulin Infusion (CSII) needs to be sustained
(p < 0.05). to result in reduction of diabetes-related vascular complications.
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-99

4
D-medical Clinic, Diabetes Unit, Madrid, Spain
5
Metronic International Trading Sarl, Health Economics
& Outcomes Research, Tolochenaz, Switzerland
6
Heva Heor Sarl, 186 avenue Thiers, Lyon, France

Objective: To estimate the cost-effectiveness of sensor-

augmented pump (SAP) with low glucose suspend (LGS) versus
continuous subcutaneous insulin infusion (CSII) alone for the
treatment of patients with type 1 diabetes (T1D) and recurrent
severe hypoglycemia.
Methods: The CORE Diabetes Model was used to pro-
ject clinical and economic outcomes for the treatment of
hypoglycemic-prone T1D patients. Patient baseline character-
istics and treatment effects were taken from one published
randomized controlled trial, in which severe hypoglycemic
events were observed at 2.2 and 0 per 100 patient-months for
CSII and SAP+LGS, respectively. Unit costs (e,2016) and
utility values were obtained from published sources. The main
For example, the 0.5% reduction in HbA1c, achieved by CSII, in
analysis was conducted from the National Health Service
the meta-analysis of RCTs would need to be sustained for 10
(NHS) perspective over a lifetime horizon with costs and health
years to result in a 5% decrease in the number of patients de-
results discounted at 3%. Furthermore, the societal perspective
veloping retinopathy.
was considered in an alternative scenario. Univariate and
We conducted retrospective review of 207 case-notes of pa-
probabilistic sensitivity analyses (SA) were performed to assess
tients with diabetes who have been treated with CSII from 1998
the models robustness.
until 2015 in our centre to assess long-term effect of CSII on
Results: SAP+LGS was associated with improvements in
HbA1c.
quality-adjusted life years (QALYs) by 1.877, versus CSII.
Methods: HbA1c was collected at baseline (prior to starting
SAP+LGS total direct costs were e47,665 higher, leading to an
CSII), 3 months, 6 months then yearly after start of CSII.
incremental cost-effectiveness ratio (ICER) of e25,394 per
The changes since baseline for the HbA1c were compared for
QALY gained. Likewise, SAP+LGS total direct costs increased
each time interval by Wilcoxon signed-rank test or paired t-test
by e41,036 when the societal perspective was considered,
as appropriate.
leading to an ICER of e21,862 per QALY gained. SA indicated
Results: Notes of 178 CSII treated patients (104 females, 74
the model projections were robust to the various scenarios
males) were analysed with mean age of 43 (range 16-75) years.
tested.
Mean SD baseline HbA1c was 8.68 1.55%.
Conclusions: From the NHS and the societal perspective,
CSII was associated with a statistically significant reduction of
SAP+LGS represents a cost-effective option compared with CSII
mean HbA1c by 0.88% at three months (P < 0.001). Maximum
for the treatment of hypoglycemic-prone T1D patients in Spain
improvement in HbA1c was achieved at 6 months with HbA1c
when considering a willingness-to-pay threshold of e30,000 per
reduction of 0.9% (P < 0.001). Statistically significant reduction
QALY gained.
was maintained for the first three years following CSII initiation.
Analysis of patients with pre-pump HbA1c 8.5% (n = 86,
Mean SD baseline HbA1c = 9.88 1.87%) showed clinically
and statistically significant reduction in HbA1c, compared to pre- 240
pump HbA1c, that was maintained for 15 years (figure 1). MULTI-CENTRIC STUDY RENACED DIABETES TIPO
Conclusions: This study demonstrates the long-term meta- 1: METABOLIC DIFFERENCES BETWEEN INSULIN
bolic benefits of CSII by achieving sustained HbA1c reduction PUMP USERS AND THOSE ON BASAL BOLUS
over 15 years. BY INJECTION
R.N. Faradji1, M. Valenzuela-Lara2, J.F. Bustamante-
Martinez3, N.E. De La Garza-Hernandez4,
239 A.P. Diaz Barriga-Menchaca5, A. Escobedo-Ortiz6,
A. Flores-Camargo5, L. Islas-Ortega7, C. Lopez-Miramontes8,
COST-UTILITY ANALYSIS OF SENSOR AUGMENTED
A. Martinez-Ramos-Mendez9, S. Miracle-Lopez10,
PUMP THERAPY WITH LOW GLUCOSE SUSPEND
L.A. Ramirez-Toscano8, E. Rodriguez-Sanchez11,
FUNCTION FOR PATIENTS WITH TYPE 1 DIABETES
M. Tavera-Hernandez12, J.C. Valenzuela-Montoya13,
AND RECURRENT SEVERE HYPOGLYCEMIA
M. Vidrio-Velazquez8, R.S. Nino-Vargas14,
IN SPAIN
E. Sainz-De La Maza-Viadero15, C. Magis-Rodriguez2
I. Conget1, I. Elas2, F.J. Ampudia-Blasco3, 1
P. Martn-Vaquero4, M. Alvarez2, A. Delbaere5, Instituto Tecnologico de Monterrey/Centro Medico ABC,
S. de Portu5, S. Roze6 Endocrinologia, Mexico, Mexico
2
Centro Nacional para la Prevencion y el Control del VIH y el
1
Hospital Clnic i Universitari de Barcelona, Diabetes Unit. Sida- Secretaria de Salud, Epidemiologia, Mexico City, Mexico
3
Endocrinology Department. IDIBAPS, Barcelona, Spain Hospitale General de Tepic, Departamento de Medicina
2
Medtronic Iberica, S.A, Health Economics & Outcomes Interna, Tepic, Mexico
4
Research, Madrid, Spain CEMEDIN, Endocrinologia, Monterrey, Mexico
3 5
Clinic University Hospital, Diabetes Reference Unit, Instituto Tecnologico de Monterrey, Ciencias Medicas Basicas
Endocrinology & Nutrition Department, Valencia, Spain y Clnicas, Mexico, Mexico
A-100 ATTD 2017 E-POSTER VIEWING ABSTRACTS

6
Hospital General Dr. Miguel Silva, Secretaria de Salud, 241
Endocrinologia, Morelia, Mexico
7
Hospital DIF de la Ninez Hidalguense, Endocrinologia EVALUATION OF CORRELATION BETWEEN
Pediatrica, Pachuca, Mexico INCREASED BLOOD GLUCOSE LEVELS AND
8
HGR 110 IMSS Guadalajara, Jal, Endocrinologia, INCREASED PUMP FORCE (SYSTEM PRESSURE)
Guadalajara, Mexico MEASUREMENTS
9
HOSPITAL ESPANOL, Endocrinologia Pediatrica, J. Fusselman1, S. Chattaraj1, J. Shin2, B. Grosman3
Mexico City, Mexico 1
10
Hospital Angeles Lomas, Endocrinologia, Mexico City, Medtronic Diabetes, Chemistry- Material and Microbiology,
Mexico Northridge, USA
2
11
Hospital Rovirosa- Secretaria de Salud, Diabetes, Medtronic Diabetes, Clinical Research, Northridge, USA
3
Villahermosa, Mexico Medtronic Diabetes, Close-loop Development, Northridge,
12
Centro Medico ABC Santa Fe / Hospital Angeles Lomas, USA
Endocrinologia Pediatrica, Mexico City, Mexico
13
Hospital De Gineco, Pediatra No. 31 IMSS, Endocrinologia Background and Aims: The Medtronic insulin infusion
Pediatrica, Mexicali, Mexico pump is a positive displacement design, designed to overcome
14
Centro para la Prevencion y Atencion Integral del VIH/SIDA the tissue or system pressure for assured delivery of a known
del Distrito Federal, Secretaria de Salud, Jefatura de Sistemas volume with each pump stroke. Recently, several publications
de Informacion, Mexico City, Mexico have speculated a pump wearer pumping with a partial or
15
UNIVERSIDAD IBEROAMERICANA, EDUCACION, completely occluded infusion set might experience temporary
Mexico City, Mexico flow interruptions or silent occlusions. These publications
speculated that the silent occlusion was a potential factor for
Background and Aims: There is limited information re- unexplained hyperglycemia in insulin pump users. To date,
garding the differences in metabolic control in patients with Type none of these publications showed clinical evidence establish-
1 Diabetes (T1D) treated with insulin pump therapy (CSII) or a ing a correlation between rising system pressure and glycemic
basal-bolus (BB) regime by injections in Mexico. We developed effects. Therefore, the silent occlusion theory remains hypo-
an online system, RENACED Diabetes Tipo 1(DT1), to have a thetical, with no supporting data. However, these publications
longitudinal registry of real life data of T1D patients. Our aim is did trigger our interest to evaluate the potential correlation
to evaluate the differences in metabolic control between patients between increased blood glucose levels and increased pump
treated with CSII or BB. force (system pressure) measurements, therefore, a total of
Methods: Multi-centric study where a bivariate analysis 204 days of patient data from different clinical sites were an-
(alfa = 0.05) was performed in 363 T1D patients, that use CSII or alyzed for the pump force sensor data versus the continuous
BB, registered in the RENACED DT1 system up to 10/5/2016. glucose sensor data.
Results: Of the 363 patients, 121 are on CSII and 242 on BB. Methods: The effect of the mean normalized force on the
Patients on CSII had lower HbA1c levels (7.9;CI 95% 7.68.1) sensor glucose (SG) value was examined on a total of 204 days
than those on BB (8.8;IC 95% 8.59.1) (p < 0.05). The total in- of patient data during the study under both fasting and bolus
sulin daily dose was lower on CSII (0.60 IU/kg; CI 95% 0.5-0.6), conditions.
than on BB (0.76 IU/kg; CI 95% 0.7-0.8) (p < 0.05). (Table 1). Results: Both Fasting and Bolus evaluation results show no
CSII was associated with higher SMBG (self monitor of capillary significant correlation between pump force sensor data and
blood glucose) per day, as well as use of continuous glucose sensor glucose drop (DSG). In fact, the majority of SG drops less
monitor (CGM) (p < 0.01). A significantly higher event rate of than 50 mg/dL (potential hyperglycemia) are in the low nor-
mild/moderate hypoglycemia/week was observed in the CSII malized pump force region.
group (4.3 vs. 2.5; p = 0.02). An interesting finding is that those Conclusions: Both evaluation methods show that there is no
patients on CSII exercise more. correlation between the unexplained hyperglycemia and high
pump system pressure.

242
INSULIN PUMPS THE KENYAN EXPERIENCES
G.A.M. Gaman1, F. Sherdel1
1
Comprehensive Diabetes Centre, OPD, Nairobi, Kenya

Background and Aims: Information on pump usage is scarce

in Africa. Pump usage (CSII) is still rare due to the relatively high
costs and lack of trained personnel and support services. We
embarked to review our clinical experiences with patients on
sensor augmented pumps at the Comprehensive Diabetes Centre
in Nairobi, Kenya.
Conclusions: According to the literature, CSII use, higher Methods: A survey was carried out on 32 patients 27 type 1
number of SMBG/day and CGM is associated with better gly- and 5 type 2 previously on MDI and switched to Sensor aug-
cemic control. It is interesting that those on CSII exercise more, a mented pump therapy. Mean age of patients in the study was
finding that will need to be confirmed with higher number of 24 yrs. Basic Demographic data including HbA1c was collected,
patients in the registry. A questionnaire was administered to assess Quality of life
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-101

retrospectively pre and one year post pump usage. Follow up

HbA1c was thereby done for all patients and for up to 20 patients
HbA1c data was collected reaching 4 years.
Results:

Results: The potential of this method, registered on patent and

priority requests, are illustrated via in-silico trials including
intra-patient variability and mixed meals. The table below shows
the time in hyper and hypoglycemia reduction achieved with the
proposed automatic SB with respect to standard treatment for a
cohort of 30 patients facing a high glycemic index mixed meal of
25g, a mixed meal of 50g, and a 75g and 100g of CHO meal.
Conclusions: A novel automatic algorithm for open-loop
treatment is introduced, based on the super bolus treatment and
Conclusions: Our survey suggested improvement in HbA1c the IOB profile of the patient. This algorithm can be easily loaded
when comparing CSII to MDI with an greater QOL and im- into any medical device software and depends only on the DIA of
provement in treatment satisfaction while on pump therapy. the patients.
While the survey was basic with several limitation, they mimic
results of pump therapy that has been seen in countries with vast
insulin pump experience and more resources available. This
would suggest that these benefits can also be replicated in 244
countries with limited resources with small dedicated pump
EFFECT OF DIFFERENT TYPES OF INSULIN
training centres.
THERAPY ON TRANSPLANT STATE, METABOLIC
CONTROL IN PATIENTS WITH TYPE 1 DIABETES
AFTER KIDNEY TRANSPLANTATION
243 A. Glazunova1, E. Tarasov1, M. Shamkhalova1, G. Musaeva2,
M. Shestakova1,3, Y. Moysyuk4
AUTOMATIC SUPER-BOLUS AND BOLUS SHAPING
1
FEATURES FOR INSULIN PUMPS Federal Endocrinology Research Centre, Institution of
1 1 2
F. Garelli , N. Rosales , J. Veh , H. De Battista 1 Diabetes, Moscow, Russia
2
I.M. Sechenov First Moscow State Medical University,
1
LEICI, EE, La Plata, Argentina Institution of Ophthalmology, Moscow, Russia
2 3
UdG, EE, Girona, Spain I.M. Sechenov First Moscow State Medical University,
Institution of Diabetes, Moscow, Russia
4
Background and Aims: This work presents a method for Moscow Regional Research and Clinical Institute MONIKI,
automatic insulin bolus shaping based on residual insulin or Institution of Trasplantology, Moscow, Russia
insulin-on-board (IOB) estimation as an extra feature for com-
mercial insulin pumps. Among other potential applications, this Background and Aims: To evaluate the impact of different
methodology allows the pump to automatically generate the so- types of insulin therapy (continuous subcutaneous insulin infusion
called super-bolus for the compensation of high glycemic index (CSII) using insulin pump or multiple insulin injections (MII)) on
meals, which has been recently related to the best theoretical carbohydrate metabolism, the state of the transplant in patients with
basal-bolus combination for the reduction of glucose excursions type 1 diabetes (DM1) after kidney transplantation (KT).
under open-loop treatment. Methods: The study included two groups of patients with DM1
Methods: The method computes the individual extra amount after transplantation: 1) 21 patients treated with CSII; 2) 20 with MII.
of bolus and basal cutoff time from the duration of insulin Mean duration of diabetes in the first group was 25 years[20.5;34.5],
action (DIA) of the patient using a physiological compart- the second group - 24.5 years [20;30]. Posttransplantation period
mental model for subcutaneous insulin absorption. Then, by in both groups was comparable: 8.0[7.0;36.0] and 7.5 [7.0:19.0]
means of a simple switching law, it automatically reestablishes months.
basal insulin when IOB reaches its basal level, thus avoiding Results: The mean level of glycated hemoglobin (HbA1c)
unnecessary detrimental transients generated with manual or in groups before the study did not differ: 9.0%[8.0;9.6] and
a-priori computations. 9.0%[8.7; 9.8]. When patients were transfer into CSII-HbA1c
A-102 ATTD 2017 E-POSTER VIEWING ABSTRACTS

Background and Aims: Pregnancy in women with T1D is

associated with increased risk of maternal complications, neo-
natal morbidity and mortality. Optimizing glycemic control
improves all of these outcomes.
Objective: To assess the efficacy and safety of SAPT in
pregnant women with type T1DM and to describe maternofetal
outcomes
Methods: Observational prospective cohort from August
2009 until November 2015 of 34 pregnant women with T1D who
started SAPT before or during pregnancy. The main indication
was severe hypoglycemia and poor glycemic control.
Results: 34 pregnant women with T1D on SAPT were in-
cluded. Mean preconceptional A1c was 8.24% 2.02. 20% of
pregnancies were planned. 17 patients started SAPT at 17.6 8.3
weeks. Absolute reduction of A1c level from pre-pregnancy to
third trimester was -1.72% (p < 0.0001), without differences
between the women who initiated SAPT during the pregnancy.
52% of patients in second trimester and 66.6% in third trimester
achieved A1c <6.5% respectively. Maternal and fetal outcomes:
91.1% underwent cesarean section. The main reasons were it-
erative cesarean (30%), fetal distress (20%), and pre-eclampsia
(16%). 35.9 weeks was the mean gestational age at delivery
and 15 pregnancies resulted in preterm delivery. There were no
maternofetal mortalities and no severe hypoglycemic episodes.
Two patients had diabetic ketoacidosis (DKA), none of them
required hospitalization in ICU.
Conclusions: SAPT is a safe alternative of treatment in
pregnant patients with T1D, which allows achieving goals of A1c
without severe maternal hypoglycemia or increase in DKA, no
matter the time of the beginning of the therapy.

246
HISTOPATHOLOGICAL AND MOLECULAR
ANALYSIS OF ADIPOSE TISSUE AFTER CSII
was significantly lower in this group after 12 months: 7.3% [7.3; CATHETER IMPLANTATION: STEEL VERSUS
8.6] (p < 0.00001), while in the MII group: 8.1%[7.5;8.7] TEFLON
(p < 0.03). Recurrent diabetic nephropathy at the stage of
microalbuminuria was diagnosed in 20% in the MII-group J.R. Hauzenberger1, J. Munzker1, P. Kotzbeck1, M. Asslaber2,
with continuing poor glycemic control (ratio albumin/creatinine V. Bubalo3, J.I. Joseph4, T.R. Pieber1
>30mg/g). All patients, treated CSII had normal albuminuria. 1
Medical University of Graz, Internal Medicine, Graz, Austria
GFR(CKD-EPI): 86 [68.4;101]ml/min/1.73 m2 in CSII-group 2
Medical University of Graz, Pathology, Graz, Austria
107[91;130]ml/min/1.73 m2 - MII group (p < 0.05). Positive of 3
Medical University of Graz, Biomedical Research, Graz,
hemoglobin, parathyroid hormone and blood pressure dynamics Austria
did not differ in the groups after KT. 4
Thomas Jefferson University, Anesthesiology, Philadelphia,
Conclusions: The CSII using insulin pump allows to reach USA
target values faster and more efficiently, CSII seems to be more
effective than MII in reducing glycemic variability (Table 1), and Background and Aims: Choice of CSII catheter material is
this improving the control of complications and overall prog- mostly based on personal experience or preference, since pro-
nosis in patients with DM1 after KT. found molecular studies of tissue response are lacking. Previous
studies have used patient questionnaires to evaluate the ad-
vantages of Teflon over steel or vice versa. In this study we
245 evaluated the adipose tissue response over 7 days of wear time
to commercially available Teflon and steel insulin infusion
EFFICACY, SAFETY AND MATERNOFETAL catheters using standard histopathological staining methods
OUTCOMES IN 34 PREGNANT WOMEN WITH TYPE 1 and qPCR.
DIABETES (T1D) IN SENSOR AUGMENTED PUMP Methods: CSII catheters (Quick-set and SureT) were inserted
THERAPY (SAPT) into the subcutaneous adipose tissue of 10 swine for 7 days,
A.M. Gomez Medina1, D.C. Henao Carrillo1, L.M. Marn 4 days and 1 day of wear-time. Tissue surrounding catheters was
Carrillo1, M. Rondon1, J.L. Silva Herrera1 excised and stained to determine morphological changes, fibrosis
and mononuclear cell infiltration. Quantitative real-time PCR
1
Pontificia Universidad Javeriana, Hospital San Ignacio- was carried out to determine changes in cytokine and inflam-
Unidad de Endocrinologa, Bogota, Colombia matory cell marker gene expression.
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-103

Results: Steel causes substantially more bleeding and there- 248

fore fibrin deposition compared to Teflon. Infiltration of
USE OF TECHNOLOGICAL FEATURES OF INSULIN
neutrophils is significantly higher and the inflamed area is
PUMPS AS PREDICTORS OF GLYCAEMIC CONTROL
40% larger around steel than around Teflon. IL-6 is immedi-
IN PATIENTS WITH CSII THERAPY
ately upregulated inducing macrophage differentiation, that
constantly increase in number over 7 days independent of N. Hermanns1, B. Kulzer1, T. Haak1, D. Ehrmann1
material. 1
Conclusions: Our study design allowed for an adequate trend Research Institute Diabetes FIDAM, Research Institute of the
analysis beyond the clinically recommended wear-time of CSII Diabetes Academy Mergentheim, Bad Mergentheim, Germany
catheters. Data showed a trend towards worse immunological
outcomes for steel sets for all characteristics analyzed. However, Background and Aims: Technological features of CSII
more research is needed to confirm our findings. therapy comprise temporal basal rates, different basal rate pro-
files, use of bolus calculators, and bolus variants. In addition,
real-time CGM data as well as computerized analysis of glucose
data can be used to adjust insulin pump therapy. In this study, the
associations between the frequency of use of these technological
247 features and glycaemic control were analysed.
PSYCHOSOCIAL RISK AND INSULIN PUMP THERAPY Methods: Stepwise linear regression analysis with HbA1c as
IN CHILDREN WITH TYPE 1 DIABETES IN IRELAND dependent variable was performed. Independent variables were
the frequency of use of the technological features mentioned
E. Hennessy1,2,3, C. Cronin1, A. Bradfield1, O. Neylon4, above. Patients indicated the frequency of use per week of those
A. Khan5, D. Bux6, P. Leahy3, S. OConnell1, S. ORiordan1,2,3 features on a five-point Likert scale (0-not at all, 1-less than once
1
Cork University Hospital, Paediatrics and Child Health, Cork, per week, 2-at least once a week, 3-several time per week, 4-
Ireland every day).
2
HRB, Clinical Research Facility - Cork, Cork, Ireland Results: 150 patients participated in the study (99.3% with
3
University College Cork, Department of Paediatrics, Cork, type 1 diabetes, mean HbA1c: 8.3 0.8%). The most frequently
Ireland used feature was the bolus calculator with a mean score of
4
Sligo General Hospital, Department of Paediatrics, Sligo, 2.7 1.8, followed by use of bolus variants (1.6 1.5); temporal
Ireland basal rates (1.4 1.3) and basal profiles (0.6 1.5) were rarely
5
Tralee General Hospital, Department of Paediatrics, Tralee, used. The stepwise regression retained only use of basal rate
Ireland profiles as well as computerised analysis of glucose as significant
6
Waterford University Hospital, Department of Paediatrics, predictors of glycaemic control (b = -0.17, p = .04; b = -0.20,
Waterford, Ireland p = .02).
Conclusions: Patients used the technological features of their
Background and Aims: Psychosocial factors may be essen- insulin pump rather rarely. More frequent use of different basal
tial in explaining poor glycaemic control in children with Type 1 rate profiles and systematic analysis of glucose profiles were
diabetes (T1D). associated with better glycaemic control. Patients should be
Aim is to examine the psychosocial risk and the risk for better educated how to interpret their glucose data and make use
emotional distress in children with T1D. of the benefits of the technological features of their insulin pump.
Methods: The Risk Index for Poor Glycaemic Control
(RI-PCG) is the screening tool to assess psychosocial risk
(low risk score 0-1, moderate = 2, high risk 3), contains so- 249
cial and psychological subscales. The Paediatric Index of
Emotional Distress (PI-ED) was used for emotional distress DOSING ACCURACY AND BOLUS DELIVERY TIME
assessment. OF SIX DIFFERENT INSULIN PUMPS
Results: 103 children with T1D (53 males) aged 318 years U. Kamecke1, G. Freckmann1, C. Haug1, R. Ziegler2
(mean 12.3 3.4) were analysed. 63.5% of patients had low risk
1
on the RI-PGC, moderate - 15.7%, high - 20.8%. 8.7% of patients Institut fur Diabetes-Technologie Forschungs, und
were at high risk for emotional distress, mainly female (91.3%), Entwicklungsgesellschaft mbH, an der Universitat Ulm, Ulm,
p < 0.03. There was a significant correlation between higher RI- Germany
2
PGC scores and higher PI-ED scores, p < 0.002. Diabetes Clinic for Children and Adolescents, Diabetes Clinic
30% of patients were on continuous subcutaneous insulin in- for Children and Adolescents, Munster, Germany
fusion (CSII): age 11.5 4, duration of T1D 5.3 3.3, HbA1c
64.1 9.6. 72.4% of CSII patients had a low risk on the RI-PGC, Background and Aims: In insulin pump therapy, bolus doses
27.6% were at moderate and high risk. 8.3% of CSII patients are delivered to cover meals and to correct high glucose values.
were at high risk for emotional distress. Psychological subscale The aim of this study was to assess precision and trueness of a
of RIPGC showed lower risk in CSII patients compare to patients 10 U bolus delivery and to measure the delivery time of this
on multiple daily injections (p < 0.05). bolus.
Conclusions: High psychosocial risk is associated with Methods: In an experimental setting following EN 60601-2-
emotional distress. Risk for emotional distress is higher in 24, six insulin pumps with different insulin infusion sets (IIS)
females with T1D. Psychological risk for poor control is re- were evaluated, Accu-Chek Insight with Accu-Chek Insight
duced in children on CSII. The ability to predict higher risk of Flex [1] and Accu-Chek Insight Rapid [2], Accu-Chek Spirit
diabetes related complications and psychological distress Combo with Accu-Chek FlexLink [3] and Accu-Chek Rapid-
would allow for early intervention by trained clinical Psy- D Link [4], Animas Vibe with Inset II [5], MiniMed
chologist. 640G with MiniMed Quick-set [6], mylife OmniPod [7],
A-104 ATTD 2017 E-POSTER VIEWING ABSTRACTS

GGT were all elevated (265, 188 and 74 U/L respectively). Bone
age was delayed (-3.2 SD). GH was low (peak 1.48 ng/ml). GnRH
test was normal. After extensive education continuous insulin de-
livery was initiated. Hepatomegaly disappeared after 3 months and
CGM showed improved glucose measurements. After 6 months
HbA1c decreased to 7.1% and he gained 6 cm in height. cushingoid
appearance completely resolved.
Conclusions: Although the knowledge about the etiology of
Mauriac syndrome is limited, careful continuous insulin delivery
via insulin pump can be a solution. Close follow up of this patient
is warranted.

251
IDENTIFYING PATIENTS WITH TYPE 2 DIABETES
AT RISK FOR HYPOGLYCEMIA, AMONG THOSE
TREATED WITH LONG-ACTING INSULIN ANALOGS
B. Kovatchev1, Z. Meng2, M. Breton1, B. Leroy3, A. Cali3
Paradigm Veo with MiniMed Mio [8], MiniMed Quick-
set [9], and MiniMed Sure-T [10]. Insulin pumps were in- 1
Center for Diabetes Technology, University of Virginia,
stalled with the tip of the cannula in a water-filled beaker placed Charlottesville, VA
2
on an electronic balance. To avoid evaporation, an oil film was Sanofi, Bridgewater, NJ
3
applied. After a run-in period, 12 successive 10 U boluses were Sanofi, Paris, France
delivered and weighed individually. Each combination of insulin
pump and IIS was tested 9 times. In addition, time for insulin Hypoglycemia does occur in type 2 diabetes, particularly in
delivery was recorded with a stopwatch. For insulin pumps with patients using insulin. Twenty years ago we introduced the Risk
adjustable speed, all possible settings were tested. Analysis of blood glucose (BG) data1 to quantify the extent and
Results: The accuracy assessment showed deviations from the frequency of hypo- and hyperglycemia. Over the years, its utility
target dose from -4% to +8% for discrete single boluses. Time to has been repeatedly confirmed, and the Low BG Index (LBGI)
deliver 10 U bolus doses ranged from 19 sec to 6:39 min for the was established as a powerful predictor of severe hypoglyce-
different models. mia2. The objective of this reanalysis is to identify patients re-
Conclusions: For 10 U bolus doses, delivery accuracy was porting symptomatic hypoglycemia (SH) during pivotal trials of
reliable for all tested insulin pump systems. Delivery time, Toujeo (insulin glargine 300 Units/mL).
however, strongly varied among the different insulin pumps. Self-monitoring (SMBG) daily profiles were available from
two multicenter trials: EDITION 2 (N = 802 patients, 44,787
SMBG readings, 699 SH episodes) and EDITION 3 (N = 869
patients, 47,817 SMBG readings, 236 SH episodes). Both ran-
250 domized their participants to Toujeo or Lantus (insulin glar-
CONTINUOUS INSULIN DELIVERY IN A PATIENT gine 100 Units/mL); for this analysis, the data was pulled across
WITH MAURIAC SYNDROME the two types of insulin. The LBGI was computed for each person.
Among several measures of glucose variability (e.g. Standard
M. Kocova1, E. Sukarova-Angelovska1, L. Gigova1 Deviation, Coefficient of Variation), the LBGI was the only metric
1
University Childrens Hospital, Endocrinology & Genetics, significantly correlated with SH: r = 0.38, p1.1) experienced 6-fold
Skopje, FYR Macedonia more symptomatic hypoglycemia than those at minimal risk
(LBGI 1.1): 3.74 vs. 0.63 SH episodes per person.
While EDITION 2 and EDITION 3 had different frequencies
Background and Aims: Introduction. Mauriac syndrome is
of hypoglycemia, the LBGI detected patients at risk uniformly
rare in children with type 1 diabetes who are under inappropriate
well across the two studies. Thus, symptomatic hypoglycemia is
control. It consists of dwarfism, hepatomegaly and cushingoid
predictable during treatment with long-acting insulin.
appearance. We present a boy at 13.5 years with a typical clinical
presentation of Mauriac syndrome.
Methods: Case report. The boy was diagnosed with diabetes
at 5 years of age. He was started on multiple injection therapy 252
with insulin. However, due to the social circumstances, the boy
start early self- injecting insulin, receiving many additional PAQ, A SIMPLE, WEARABLE THREE-DAY BASAL/
doses in order to eat frequent meals. His visits to the diabetol- BOLUS INSULIN DELIVERY DEVICE, DESIGNED
ogist were rare. At the age of 13.5 years he was evaluated for the FOR DISCREETNESS
short stature falling from 25%o to -3SD. Moon face, prominent C. Kuerschner1, L. Lilly1, J. Warner1
abdomen, hepatomegaly (6 cm) was noticed. He was pre-
1
pubertal. CeQur, Marlborough, USA
Results: His HbA1c was 11.87% (106 mM/mol). CGM showed
large glucose variations. Fundoscopy revealed normal retina. Background and Aims: Peyrot, et. al. revealed that >50%
Thyroid hormones and antibodies as well as transglutaminase an- of patients on insulin reported skipping injections. Embarrassment
tibodies showed normal values. Hepatic enzymes AST, ALT and was one of the most common factors contributing to these
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-105

omissions. To decrease this, as well as other barriers to insulin devices on intended users to confirm these findings and their
therapy, CeQur designed a body worn device, PAQ, to deliver effect on user acceptance.
both basal and bolus insulin for 3-days. Feedback received from
adults with diabetes during PAQ development was the importance
of discreetness and a key variable to consider was non-detectability 253
under clothing. To this end different colors of the PAQ device were
tested for detectability when worn under clothing. WHOSE DECISION IS IT? DECISION COACHING
Methods: A white colored PAQ and a gray colored PAQ were WITH A PATIENT DECISION AID IN YOUTH AND
placed on a body form with a dark and a medium skin tone. A THEIR PARENTS TO IDENTIFY THE RIGHT INSULIN
thin, white t-shirt was placed on the body form covering the DELIVERY METHOD FOR THEM
devices. Participants were shown the forms and asked to choose M. Lawson1, A.L. Shephard2, B. Feenstra3, L. Boland3,
the most noticeable device color. The device labels were dif- N. Sourial4, D. Stacey5
ferent for each skin tone to prevent selection bias.
1
Childrens Hospital of Eastern Ontario, Endocrinology
and Metabolism, Ottawa, Canada
2
Childrens Hospital of Eastern Ontario Research Institute,
Clinical Research Unit, Ottawa, Canada
3
University of Ottawa, Faculty of Health Sciences, Ottawa,
Canada
4
McGill University, Epidemiology, Montreal, Canada
5
Ottawa Hospital Research Institute, Clinical Epidemiology,
Ottawa, Canada

Background and Aims: To evaluate the effect of decision

coaching with a decision aid on decisional conflict and satis-
faction for youth and parents considering a change in insulin
delivery method (standard regimen vs. MDI vs. CSII).
Methods: Pre-post test design. Pediatric diabetes social work-
ers were trained to provide decision coaching through an online
tutorial and skill-building workshop. Youth and parent(s) con-
sidering a change in the youths insulin delivery method were
referred for decision coaching by their physician. Decision
coaches used the Ottawa Family Decision Guide pre-populated
for insulin delivery options. Data were collected prior to decision
coaching (T1) and 10-14 days post (T2). Primary outcome was
decisional conflict measured with the Decisional Conflict Scale
(DCS), compared at T1 and T2 using a paired t-test. Other out-
comes: choice predisposition vs. actual choice (T1&T2) and
satisfaction with coaching questionnaire (T2).
Results: 45 families participated, each consisting of 1 youth and
1-2 parents. Youth (n = 45), 56% male, mean age 12.5 2.9 (SD)
years (range 6.3-17.7). Parents (n = 66), 38% fathers. Coaching
sessions averaged 54.8.2 8.8 minutes. Mean youth DCS de-
creased from 32.0 19.7 (T1) to 6.6 12.2 (T2) (p < 0.0001).
Parent DCS decreased from 37.6 20.7 (T1) to 3.5 7.4 (T2)
(p < 0.0001). Between T1 and T2, there were changes in choice
predisposition for parents and youth. 89.2% of youth and 94.3% of
parents rated the coaching session as balanced and 94.6-98.1%
would probably/definitely recommend it to others.
Conclusions: Decision coaching with a decision aid reduced
decisional conflict for youth and parents facing a preference-
sensitive insulin delivery decision with high satisfaction ratings
from youth and parents.

254
Results: Thirty adults with diabetes taking at least two in- PAQ INSULIN DELIVERY DEVICE SIGNIFICANTLY
jections per day of insulin participated, 55% male, mean age 52 INCREASES THE PERCENTAGE OF TIME IN TARGET
years with a BMI of 31.7 kg/m2. The majority of the participants RANGE
(28/30, 93%) chose the white PAQ as more noticeable on both L. Lilly1, J. Mader2, F. Aberer2, J. Warner1, T. Augustin3
the dark and medium skin tone.
1
Conclusions: A gray colored PAQ may provide a more dis- CeQur, Clinical Research, Marlborough, USA
2
creet insulin delivery option for adults with diabetes with dif- Medical University of Graz, Department of Internal Medicine,
ferent skin tones. Additional studies will be conducted with Graz, Austria
A-106 ATTD 2017 E-POSTER VIEWING ABSTRACTS

3
JOANNEUM RESEARCH Forschungsgesellschaft mbH,
HEALTH - Institute for Biomedicine and Health Sciences, Graz,
Austria

Background and Aims: Data suggest not only is HbA1c <

7.0% desirable, but increasing time in target range and mini-
mizing glycemic variability is of importance.
Methods: Twenty adults with type 2 diabetes with HbA1C
7.0 11.0% on established regimen of insulin therapy (2
injections/day) other glucose lowering agents were enrolled
into a single center, single arm, treat to target study evaluat-
ing PAQ. PAQ is a simple body worn 3-day insulin deliv-
ery device which provides continuous basal and bolus insulin
on demand. The study comprised three periods: baseline
(insulin injections), transition to PAQ and PAQ treatment
(12-weeks). Five subjects wore a continuous glucose moni-
toring device for 1 and 2 weeks during baseline and PAQ
treatment periods; respectively. Endpoints HbA1c, per-
centage of time (%oT) in target, and glucose variability (SD Results: Changing the insulin preparations and increasing the
and rate of change). insulin dosage up to 160 U/day failed to decrease the skin reac-
Results: Mean HbA1c decreased from 8 to 7.1% following 12 tions, nor improved glycemic control. Once intervention with an
weeks of PAQ. After Week 8 of PaQ the %oT in target range [70- insulin pump (MiniMed Paradigm VEO) started, the skin le-
180 mg/dL] increased from 50.8% to 70.4% (P < 0.001), the %oT sions disappeared. Also a dramatic improvement with HbA1c
<70 mg/dL was unchanged from 0.24% to 0.38% (P < 0.085), resulted with a decrease in insulin dosage (Fig. 2).
mean SD of all glucose readings decreased from 181 (17) mg/dL
to 161 mg/dL (14) (P = 0.041), and the rate of change in glucose/
hour (07:00-21:50) decreased from 18.50 (5.26) to 13.19 (2.10)
(P = 0.058).
Conclusions: Insulin delivery with PAQ resulted in a decrease
in HbAc, an increase in the percentage of time in target, reduced
glycemic variability and did not result in increased hypoglyce-
mia. Additional studies are needed to confirm this effect.

255
INSULIN SKIN REACTION AND SUBCUTANEOUS
INSULIN RESISTANCE SOLVED BY INSULIN PUMP
(CSII) IN A 63-YEAR-OLD FEMALE WITH TYPE 2
Conclusions: While the mechanisms are not fully understood,
DIABETES MELLITUS: A CASE REPORT
this case report findings suggest that CSII, by delivering smaller
P. Massucco1, L. Spadafora2, K. Bonomo1, O. Cohen3, intermittent boluses of insulin, may be more rapidly and con-
F. Cavalot1, A. Guerrasio1 tinuously absorbed increasing efficiency of the insulin and de-
1 creasing the allergic skin reaction. Another plausible explanation
S. Luigi Gonzaga University Hospital- Orbassano Turin,
is that pump therapy might decrease insulin degradation at the
Department of Internal Medicine- Diabetes Unit-, Orbassano
inflamed site, decreasing subcutaneous insulin resistance. Here,
Turin, Italy
2 CSII resulted in a dramatic and persistent reduction in HbA1c,
Internal Medicine School, University of Turin- Italy, Turin,
far beyond the expected improvement deriving only from
Italy
3 switching to CSII, indicating that this injection modality might
Institute of Endocrinology, Ch. Sheba Medical Center, Tel
have an effect both on the local inflammatory response and the
Hashomer, Israel
responsiveness to insulin.
Background and Aims: While insulin-induced skin reac-
tions have become less common, since the advent of recombinant
insulin, the use of modified insulins may still induce adverse 256
local/systemic reactions. CSII has been described in several case
reports and has shown many beneficial effects, therefore, be- ONLINE SURVEY ON PERCEPTIONS AND ISSUES
coming a recommended therapeutic option. The following de- WITH CONTINUOUS SUBCUTANEOUS INSULIN
scribes the effect intervention with an insulin pump on one Type INFUSION THERAPY IN ADULTS WITH TYPE 1
2 Diabetic mellitus (T2D) patients skin reaction and glycaemic DIABETES
control. N. Taleb1, V. Messier1, J. Rene1, S. Ott-Braschi2, J. Pickup3,
Methods: A delayed local cutaneous reaction at insulin in- J.L. Ardilouze4, R. Rabasa-Lhoret1
jection sites with concurrent central necrosis (Fig.1) is described
1
in a 63-year-old female with poorly controlled T2D on warfarin Institut de recherches cliniques de Montreal, Metabolic
treatment for atrial fibrillation. Diseases, Montreal, Canada
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-107

2
Centre Hospitalier de lUniversite de Montreal, Methods: Participants performed an afternoon 45-minute
Endocrinology, Montreal, Canada exercise at 60% of VO2peak on an ergocycle 3 hours after the last
3
Kings College London, Diabetes & Nutritional Sciences, meal. In a random order, insulin infusion rate was reduced at the
London, United Kingdom time of exercise, 20 minutes or 40 minutes before. Capillary
4
Universite de Sherbrooke, Medicine, Sherbrooke, Canada glucose levels were measured at the onset of exercise, 15 minutes
after the onset of exercise and then every 5 minutes until the end
Background and Aims: We conducted an online survey to of exercise.
identify the perceptions and problems experienced by adults with Results: There was no significant difference between the
type 1 diabetes (T1D) who use continuous subcutaneous insulin three strategies for time spent below 4 mmol/L, time spent be-
infusion therapy (CSII). tween 4 and 10 mmol/L and decrease in glucose levels. The
Methods: Participants (n = 86, women: 76%, age: 41 13 years, number of hypoglycemic events below 3.5 mmol/L was 2, 7 and
T1D for 22 12 years, CSII use for 6 4 years, mean catheter 5 when insulin infusion rate was reduced 40 minutes before
replacement every 3.3 days) completed a 39-item questionnaire. exercise, 20 minutes before exercise and at the time of exercise,
Results: With CSII usage, improved glucose control and de- respectively.
creased number/severity of hypoglycemic episodes were per- Conclusions: Earlier timing and/or larger insulin infusion rate
ceived by 79% and 68%/50% of patients, respectively. In the last reduction should be tested in a larger group of patients.
year: 1) most participants reported no increase in anxiety/worry
(89%), no negative impact on life or family schedule (95%) or on
time off from work/school (86%) related to CSII; 2) however,
participants reported numerous technical problems at infusion site 258
[pain (49%), irritation/itchiness (40%), build-up of fat/lumpiness
(34%), infection/inflammation (16%)] and with catheters [block- IMPROVING PATIENTS CARE IN DIABETES
age (39%), kinking (38%), bubbles/air pockets (26%), and leakage USING TECHNOLOGY-INSULIN PUMP
(11%)]. Only 11% of patients did not report any catheter related SERVICE, THE DARENT VALLEY
problems in the last year. HOSPITAL EXPERIENCE
Conclusions: Patients using CSII positively perceived key G. Mlawa1,2, D. Affam1, A. Ogunko1, C. Eboh1
aspects of diabetes control but also reported a high frequency of
1
infusion site/catheter issues. The frequency of CSII infusion site/ Darent Valley Hospital, Endocrinology & Diabetes, Kent,
catheter issues is probably underestimated and warrants further United Kingdom
2
attention by health professionals and manufacturers. Queens Hospital, Acute Medicine/Endocrinology & Diabetes,
London, United Kingdom

Background and Aims: Diabetes patients have to be offered

257 high quality care and evidence-based diabetes treatment in-
cluding insulin pumps.
TIMING OF PRE-EXERCISE BASAL INSULIN Insulin pump therapy is recommended for adults and children
INFUSION RATE REDUCTION TO PREVENT over the age of 12, with type 1 diabetes with recurrent disabling
EXERCISE-INDUCED HYPOGLYCEMIA IN ADULTS hypoglycaemia or poorly controlled diabetes despite being on
WITH TYPE 1 DIABETES USING INSULIN PUMP multiple daily injections.
THERAPY The aim of this study was to measure the practice in our
V. Messier1, A. Roy-Fleming1, C. Suppere1, Hospital in the use of insulin pumps for treatment of diabetes and
C. Cameli2, S. Elbekri3, M. Smaoui4, L. Legault5, whether it is in keeping with NICE guidelines.
R. Rabasa-Lhoret1 Methods: This was a retrospective study from November
2015 to January 2016.
1
Institut de recherches cliniques de Montreal, Metabolic The following demographic data were recorded: Age, Gender,
diseases, Montreal, Canada Duration on insulin pump. Patients commenced on insulin
2
Universite de Rennes 1, Medicine, Rennes, France pump at the preconception clinic, during pregnancy and paedi-
3
Universite de Montreal, Medicine, Montreal, Canada atric clinics.
4
McGill University, Biomedical engineering, Montreal, Canada INCLUSION CRITERIA:
5
Montreal Childrens Hospital, Endocrinology, Montreal, Disabling hypoglycaemia
Canada Poorly controlled diabetes with persistent high HbA1c
>8.5%(69mmol/mol) despite multiple daily injections
Background and Aims: Regular exercise should be encour- Insulin initiated by trained specialist team
aged in most patients with type 1 diabetes (T1D) but is mainly Structured education to patient
limited by fear of hypoglycemia. To reduce the risk of exercise- Results: 80 patients were identified, aged 16 or over
induced hypoglycemia, two types of adjustment may be con- Data source: patients notes, Diabeta3 (vectra) software
sidered: insulin dose reduction and/or carbohydrate supplements. available in the diabetes department
Insulin pump therapy (CSII) offers flexibility using temporary 98%(79) had type 1 diabetes, 1.25%(1) type 2 diabetes
insulin infusion rate reduction. Few studies have investigated the 13%(11) commenced insulin pumps from other European
pre-exercise optimal timing at which patients should implement countries, or other Hospitals in UK.
insulin infusion reduction to prevent exercise-induced hypogly- 12%(10) commenced insulin pumps in pediatric age for poor
cemia. The objective of this study was to compare the efficacy of glycaemic control.
different pre-exercise timing for an 80% insulin infusion rate 5%(4) started insulin pump during pregnancy (poor control/
reduction to prevent exercise-induced hypoglycemia in adults hypoglycaemia).
with T1D using CSII. 34 were females, and 46 males.
A-108 ATTD 2017 E-POSTER VIEWING ABSTRACTS

Conclusions: Insulin pump therapy at Darent Valley Hos- Methods: Observational prospective study conducted on a
pital was initiated by specialist team and in keeping with the group of 28 type 1 diabetic patients (14 men), with an average age
guidelines. of 42.6 9.4 years, treated with CSII, and with 5 years of follow up.
Insulin pump therapy resulted in improvement in hypogly- We compared the change in metabolic control, glycemic var-
caemic episodes in 50% of patients, while 36%% had sustained iability (defined by standard deviation), hypoglycemic events and
improvement in HbA1c. insulin dose at baseline, 1 and 5 years after initiation of treatment.
Insulin pump may be offered in type 2 diabetes patients with Descriptive statistic was expressed as mean SD. A p value
high insulin requirements. <0.05 defined the level of statistical significance.
Results:

259 Baseline 1 year Sig* 5 years Sig**

FOUR-YEARS BASAL BOLUS THERAPY HbA1c (%) 8,7 1,7 7,9 1,1 p = 0,05 7,8 0,9 ns
Mean Glucose 175,9 44,3 169,6 39,6 ns 171,3 36,6 ns
IN PEDIATRIC PATIENTS WITH TYPE 1 (mg/dl)
DIABETES MELLITUS Standard 97,4 40,5 75,1 19,9 p = 0,032 79,1 22,6 ns
Deviation
1 1 1 1
E. Mozzillo , P. Buono , A. Casertano , A. De Matteis , (mg/dl)
S. Mobilia1, V. Fattorusso1, A. Franzese1 Hypoglycemic 13,8 8,9 11,4 6,2 ns 9,1 6,2 ns
events
1
Federico II University of Naples, Translational Medical 1 month (%)
Severe 28 6 ns 2 ns
Science, Naples, Italy hypoglycemia
1 year (n)
Background and Aims: Basal-bolus insulin therapy simulates Total insulin 49,7 21,6 40,8 18,1 p = 0,015 43,7 17,9 ns
dose (IU)
physiological pancreatic secretion of the hormone. Aim of this
study is to analyze the effects of a four-years basal-bolus insulin- * Baseline 1 year ** 1 year 5 years
therapy in T1DM pediatric patients treated by insulin pump (CSII
group) or multiple daily injection therapy (MDI group: 3 injections
Conclusions: After 1 year of follow up we observed that CSII
of rapid/analogue insulin as bolus, glargine at bed-time as basal).
therapy has provided a significant improvement in metabolic
Methods: 82 T1DM outpatients (M 40, age 10.5 3.7 years, 41
control, glycemic variability and decrease insulin dose. Hy-
CSII-group), in Pediatric Diabetes Center of University of Naples
poglycemia, events have been reduced during the first year of
Federico II, were enrolled in the study; patients with onset of
treatment.
T1DM <1 year have been excluded. Glycated hemoglobin
The improvements obtained with CSII during first year remain
(HbA1c%), Body Mass Index (BMI) z-score, Units/Kg/day of in-
without significant changes after 5 years of follow-up.
sulin administered as basal and Units/Kg/day of insulin adminis-
tered as bolus were evaluated at the beginning of the study and after
1, 2, 3 and 4 years of the same therapy. No patient dropped.
Results: Mean HbA1c% value remains stable in both groups; 261
Mean BMI z-score increases during follow-up in both groups;
Units/Kg/day of basal insulin show statistically significant in- GLYCEMIC CONTROL AMONG CSII USERS
crease in MDI-group; while only during first two years of follow- AFTER 5 YEAR SUPERVISION IN ROUTINE
up in CSII-group; Units/Kg/day of bolus insulin remain stable CLINICAL PRACTICE
during follow-up in MDI-group and increase in CSII-group. L. Ibragimova1, Y. Philippov1, A. Mayorov1
Conclusions: In conclusion CSII-long term therapy doesnt
1
seem to show advantages compared to MDI therapy. Endocrinology Research Centre, Institute of Diabetes,
Moscow, Russia

Background and Aims: In clinical studies continuous sub-

260 cutaneous insulin infusion (CSII) leads diabetes mellitus patients
to have better glycaemic control. But in real-life routine clinical
SUBCUTANEOUS CONTINUOUS INSULIN INFUSION practice positive effects of CSII can disappear.
IN TYPE 1 DIABETIC PATIENTS: GLYCEMIC The aim of the study was to assess glycaemic control among
CONTROL AFTER 5 YEARS OF FOLLOW-UP CSII users in routine clinical practice.
M. Pazos-Couselo1, M. Gonzalez-Rodrguez1, F. Casanueva1, Methods: This 5 years observational study included type 1
J.M. Garca-Lopez1 diabetes mellitus users of CSII. We selected patients who were
1
transferred from multiple daily injections to CSII in our research
University Hospital Santiago de Compostela, Endocrinology centre with special education and long-term supervision by en-
and Nutrition Service, Santiago de Compostela, Spain docrinologist (during 4 months after transferring to CSII). The
next 5 years all patients were followed up by endocrinologists in
Background and Aims: Subcutaneous continuous insulin routine clinical practice and didnt come to our research centre or
infusion (CSII) therapy has been associated with an improvement insulin pump specialists. We have analyzed glycaemic control:
in the HbA1c, lowering glycemic variability and hypoglycemic before CSII initiation, after 4 months of supervision and after 5
events. This improvement is achieved approximately after 1 year years of routine real-life care.
of treatment. Results: In preliminary part of our study we have analyzed
The aim of our study is to measure if the improvement of the data from 17 patients. 4 months after transferring to CSII HbA1c
metabolic control remains after 5 years of treatment. was significantly lower compared to a basic level (7.8% [7.0;
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-109

8.1] vs 8.5% [7.5; 9.5]) p < 0.0001. 5 years later the HbA1c was Background and Aims: Insulin pumps, or Continuous Sub-
significantly higher compared HbA1c 4 months after transfer- cutaneous Insulin Infusion (CSII), have been cited as a tech-
ring to CSII (8.1% [7.7;9.1] vs 7.8% [7.0; 8.1], p < 0.01), and we nological advancement which can improve quality of life and
did not find significantly differences between basic HbA1c and biomedical outcomes of people with Type 1 diabetes. A concern is
HbA1c 5 years after transferring to CSII (p > 0,05). In contrast, that such devices may widen health inequality, particularly if up-
glucose variability (SD) didnt change significantly between 4 take is mostly seen in motivated patients who demonstrate good
months and 5 years points: 3.3[2.8; 3.7] vs 3.3[2.7;3.5], self-care behaviours. Understanding factors which enable people to
p > 0.05. incorporate CSII could provide an exemplar for the implementa-
Conclusions: Using CSII without regular re-education and tion of new technologies, with an impact on how people manage
special supervision can damage positive effects of CSII. their condition and the health service implications.
Methods: A systematic search of seven databases was con-
ducted to identify studies reporting patient experiences of living
262 with CSII. A critical interpretative synthesis of the evidence was
used to identify domains that are key to successfully incorpo-
ROUTINE USE OF CONTINUOUS SUBCUTANEOUS rating CSII.
INSULIN INFUSION IN A COHORT OF TYPE 1 Results: A total of 4,998 titles were identified. We viewed 274
DIABETES PATIENTS ATTENDED IN A DIABETES abstracts, reviewed 39 papers and included 21. Studies show that
REFERENCE UNIT CSII offers a level of choice, control and flexibility which is
C. Quiros1, C. Vinals1, V. Pane1, D. Roca1, I. Conget1, unmet by multiple daily injections. However, CSII is a complex
M. Gimenez1 technology where there is a leap in terms of expectations in
1 engagement, in comparison to previous insulin administration;
Hospital Clnic de Barcelona, Diabetes Unit. Endocrinology physiologically, practically, psychologically and socially.
Department, Barcelona, Spain Conclusions: Our findings suggest that there are many ben-
efits, but also complex issues to consider when CSII is intro-
Background and Aims: Continuous subcutaneous insulin duced. There is a potential encumbrance on self-care when
infusion (CSII) is an increasingly common effective option in balancing the burdens of a technologically-advanced, intensified,
Type 1 Diabetes (T1D) management. Data on its efficacy, safety regimen against its benefits. In order to successfully incorporate
and use comes frequently from clinical trials or retrospective CSII, people need to be able to utilise and access social support
controlled studies. We aimed to analyse the characteristics of the and navigate resources that are tailored to their needs.
routine use of CSII in a large cohort of patients attended in a
Diabetes Reference Unit and its relationship with glycaemic
control.
Methods: T1D patients using CSII with either a Veo or 264
640G Medtronic-Minimed pump linked to a glucometer (Con-
tour Next Link/2.0, Bayer) were included (10% using sensor- USE OF REGULAR U-500 INSULIN IN TYPE 2
augmented-pump). Data from 14 consecutive days were collected DIABETES TREATMENT WITH PUMP THERAPY:
from uploads in CareLink  software and HbA1c was obtained A FRENCH TWO CENTER LONG TERM EXPERIENCE
from medical records. IN 43 PATIENTS
Results: Data from 338 subjects with T1D were included (age D. Emilie1, H. Juliette1, P. Alfred2, M. Julia1, R. Anne1,
43.4 13.1 years; 64.5% females; diabetes duration 27.1 9.9 J. Michael1, R. Yves1
years, 9.3 4.8 years on CSII, HbA1c 7.7 1.0%). Average daily
1
basal/bolus insulin ratio was 52.5/47.5%, with a mean of 4.9 3.4 CHU de Caen, Endocrinology and Diabetes, Caen, France
2
bolus/day (78.9% advised bolus), 6.0 1.8 basal segments/day Hopital Sud-Francilien, Endocrinology and Diabetes, Corbeil
and 3.4 1.5 insulin/CH ratios/day. The number of capillary blood Essonnes, France
glucose readings/day (CBG) was 4.4 2.1 (37.9 15.6% <180mg/
dL and 11.4 9.2% <70 mg/dl). Those subjects with an HbA1c Background and Aims: Insulin therapy intensification by
7.5% (55%) performed more CBG (4.9 2.1 vs. 4.0 2.0; p < insulin pump is helpful in insulin resistant type 2 diabetes. The
0.001), more boluses (5.4 1.9 vs. 4.6 2.2; p < 0.05) and used use of concentrated insulin may help improving glycemic control
more basal segments/day (6.3 2.0 vs. 5.8 1.7; p = 0.002) but few studies are available.1 We report on the utilization of
Conclusions: CSII use by subjects with T1D in routine clin- regular insulin 500 U/ml (Eli Lilly, France) in a pump device
ical care is not far from expected and usually recommended. The over a 6-yr period.
frequency of CBG, bolus and number of basal segments/day Methods: Patients on pump therapy with U-100 insulin rapid
were associated with a better glycaemic control in terms of acting analogs were recruited in two circumstances: i)
HbA1c. HbA1c>8% and/or insulin dose >100 UI/d. After U-500 therapy
initiation, follow up was performed at 1 week, 3, 6, 12, 18, 24
months then annually for HbA1c, total daily dose (TDD), body
263 weight, incidence of hypoglycemia, treatment drop off.
Results: 43 patients were included, M/F ratio 24/19, age
WHAT ARE THE FACTORS THAT ENABLE PEOPLE 62.9 8.3, diabetes duration 19 8 yrs, TDD 205 61 UI/d, BMI
TO INCORPORATE THE INSULIN PUMP INTO THEIR 39 6 kg/m2, HbA1c 8.9 1.4%. Mean follow up on U-500 was
EVERYDAY LIVES? 21 15 months with 5 drop off. HbA1c decreased by -0.77%
C. Reidy1, A. Rogers1, M. Bracher1, A. Kennedy1, I. Vassilev1 (p = 0.0002), -1.07% (p < 0.0001), -0.94% (p = 0.0009), -0.84%
1
(p = 0.045) and -1.9% (p = 0.046) at 3, 6, 12, 18, 24 and 36 months.
University of Southampton, Health Sciences, Southampton, Body weight and TDD were stable. Non severe hypoglycemia oc-
United Kingdom curred in 23% subjects before U500 and in 48.6%, 45.5% and
A-110 ATTD 2017 E-POSTER VIEWING ABSTRACTS

1
70.6% subjects at 3, 12 and 18 months on U500. No severe hypo- IBIMA. Hospital General Universitario de Malaga. Ciberdem,
glycemia occurred. Diabetes Unit. Endocrinology and Nutrition, Malaga, Spain
2
Conclusions: When insulin resistant type 2 diabetes remains IBIMA. Hospital Regional Universitario de Malaga.
uncontrolled on pump therapy delivering TDD, U-500 regular Ciberdem, Malaga, Spain
3
insulin may durably improve glycemic control. Hypoglycemia IBIMA. Hospital General Universitario de Malaga. Ciberdem,
may be prevented by careful education of U-500 users. Diabetes Unit, Malaga, Spain
1. Lane WS. Endocr Pract 2010;16:181-6.
Background and Aims: To measure oxidative stress markers
in patients with type 1 diabetes (T1DM), type 2 diabetes
265 (T2DM) and healthy subjects and to compare these markers
in patients with T1DM according to insulin treatment (multi-
CONTRIBUTIONS OF BASAL AND POSTPRANDIAL ple daily injections MDI- vs. continuous subcutaneous insu-
HYPERGLYCEMIA IN TYPE 2 DIABETES PATIENTS lin infusion CSII-) and the presence of diabetes related
TREATED BY AN INTENSIFIED INSULIN REGIMEN: complications.
IMPACT OF PUMP THERAPY IN OPT2MISE TRIAL Methods: Cross-sectional study that included 205 subjects with
Y. Reznik1, A. Habteab2, J. Castaneda2, J. Shin3, M. Joubert1 similar age and sex: 123 with T1DM (64 on MDI and 57 on CSII),
1
39 with T2DM and 43 healthy subjects. Study variables in patients
Caen University Hospital, Endocrinology and Diabetes, Caen, with T1DM: insulin treatment, diabetes related complications, di-
France abetes duration, anthropometric measures, metabolic data. Total
2
Medtronic, Bakken Research Center, Maastricht, antioxidant capacity (TAC) (Cayman ELISA, ABTS method) and
The Netherlands oxidized LDL (ELISA kit) were measured in blood samples in all
3
Medtronic, Diabetes, Northridge, USA study subjects.
Results: Subjects with T1DM had lower TAC values than
Background and Aims: The Relative contribution of fasting subjects with T2DM and healthy controls (1.4 0.67 vs. 2.6 1.2
and postprandial hyperglycemia in type 2 diabetes subjects vs. 2.8 1.3 mM; p < 0.05). No significant differences in TAC
treated by an intensified insulin regimen using multiple daily levels were found in patients with T1DM according to insulin
injections (MDI) is poorly documented. Our aim was to measure therapy (MDI/CSII). We found lower TAC values in patients
fasting and postprandial hyperglycemia (HG) in type 2 diabetes who reported diabetes related complications than in those who
subjects on MDI before randomization in the OPT2MISE study. did not (1.2 0.6 vs. 1.5 0.7 mM; p = 0.05). There were no
We also analyzed the predictive value of these variables on the differences in oxidized LDL adjusted for LDL in any of the
metabolic response to CSII. groups.
Methods: We performed an analysis of continuous glucose Conclusions: Subjects with T1DM had lower TAC levels than
monitoring (CGM) recordings after 8-week run-in period in 259 those with T2DM and healthy controls. In subjects with T1DM
MDI patients. Area under curve (AUC) HG was calculated in the and any diabetes related complications TAC values were de-
basal (AUC-B), nocturnal (AUC-N) and postprandial (AUC-PP) creased. We did not find significant differences in TAC values
periods according baseline HbA1c level (Gr1 :<8, Gr2 :8-8.5, according to the use of CSII or MDI. Further studies are needed
Gr3 :8.5-9, Gr4 :9-9.5, Gr5 :>9.5%). Changes were analyzed in to confirm these findings.
131 subjects switched from MDI to CSII. Analysis of variance
was used for comparisons between groups.
Results: AUC-B was 27.5% to 121% higher in Gr5 vs Gr4 to
Gr1 (p = 0.0001). AUC-N was 18.5% to 93% higher in Gr5 vs 267
Gr4 to Gr1 (p = 0.0001). Conversely, AUC-PP did not differ SHORT TERM CONTINUOUS SUBCUTANEOUS
between groups HbA1c9.5% vs <9.5% (p = 0.72). HbA1c cor- INSULIN INFUSION THERAPY (CSII)
related with AUC-N (R = 0.32, p = 0.001) AUC-B (R = 0.31, SIGNIFICANTLY IMPROVES ERECTILE
p = 0.0001) and AUC-PP (R = 0.12, p = 0.048). After switch from DYSFUNCTIONS IN PATIENTS WITH TYPE-2
MDI to CSII, AUC-B, AUC-N and AUC-PP decreased signifi- DIABETES MELLITUS
cantly (-21%, -19% and -18.5%. When comparing responders
to non responders to CSII, the latter had higher AUC-B, AUC-N K. Singh1, P.E. Rai2, S. Karol1, P. Gupta1
and AUC-PP, but differences were not significant. 1
Fortis Hospital Mohali, Endocrinology, Chandigarh, India
Conclusions: Fasting and nocturnal HG are the major determi- 2
Fortis Hospital/ Pharmacovigilance Parexel- Chandigarh,
nants of poor glucose control in type 2 diabetes with MDI failure. Endocrinology, Chandigarh, India

Background and Aims: To determine the efficacy of 12

266 weeks CSII vs MSI (multiple subcutaneous injection) among
THE OXIDATIVE STRESS IN TYPE 1 DIABETES uncontrolled T2D patients with ED.
PEOPLE WITH CONTINUOUS SUBCUTANEOUS Methods: This is a 12 week comparative study on 46 T2D
INSULIN INFUSION VERSUS MULTIPLE DAILY patients (mean age 43.8 y) with uncontrolled hyperglycemia
INJECTIONS TREATMENT (HbA1C>10.0%) & ED and were on multiple oral hypoglycemics.
These patients had ED as graded on the Erectile Hardness Grading
M.S. Ruiz de Adana1, N. Colomo1, E. Rubio-Martin2, Scale 1- 4 (EHGS) [European Association of Urology]. All the
M.E. Dominguez-Lopez1, E. Garcia-Escobar3, J. Abuin- study patients had EHGS grading of 1 to 2 only. The patients were
Fernandez1, J. Morillas1, F. Sanchez-Torralvo1, M. Guerrero1, randomized into 2 treatment groups, 23 on MSI while another 23
G. Rojo-Martinez1 on CSII. Patients in both the groups were matched in terms of
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-111

duration of diabetes & complications, BMI, HbA1c, creatinine 269

clearance, S testosterone, FT4 TSH & prolactin levels. Stamp test
THE BENEFICIAL EFFECT OF THE PREDICTIVE
& International Index of Erectile Function (IIEF-5) were recorded.
LOW GLUCOSE MANAGEMENT SYSTEM
EHGS scale was recorded by patients in their diary at -7 day,
ON HYPOGLYCAEMIA INCIDENCE
baseline & weekly till the end of study.
AND GLYCAEMIC CONTROL IN TYPE 1
Results: Baseline HbA1c in CSII group was 11.2% and 8.7%
DIABETES MELLITUS
at the end of study while in MSI group it was 11.1% and 9.1%
respectively. In CSII group,4 patients achieved Grade 4 erec- P. Thomakos1, A. Mitrakou2, O. Kepaptsoglou1, I. Taraoune1,
tions,5 achieved Grade 3 erections (response rate 39.1%) while C. Baretto1, C. Zoupas1
in MSI, one achieved Grade 4 & three recorded Grade 3 re- 1
sponse(17.3%). Hygeia, Diabetes Center and Clinic, Athens, Greece
2
Conclusions: Short Term CSII therapy significantly im- Alexandra Hospital, Department of Clinical Therapeutics,
proved ED as achievement of Grade 4 erections was four times Athens, Greece
higher in this group (4/23 patients) compared to MSI (1/23) as
well as the overall response rate(Grade 3 and 4 erections) was Background and Aims: The aim of our study was to assess
2.25 times higher in CSII group. Marked reduction in HbA1c the effect of the Predictive Low Glucose Management System of
in CSII. the MiniMed 640GR Insulin Pump, compared to the Low Glu-
cose Suspend System of the MiniMed VeoR Insulin Pump, on
hypoglycaemia frequency and glycaemic control in people with
Type 1 Diabetes Mellitus (T1DM).
268 Methods: A cross-sectional study was conducted in 30 T1DM
patients using the MiniMed 640G vs 30 using the MiniMed
DIFFERENT STRATEGIES FOR PREVENTION
Veo sensor-augmented Insulin Pump for a minimum of 3
OF HYPOGLYCEMIA INDUCED BY PHYSICAL
months. The patients were matched for Age and duration of
ACTIVITY IN TYPE 1 DIABETES (T1D) PATIENTS
Diabetes. [(MeanSD) age: 33.4 11.5 vs 34.2 11.4 years,
E. Patrakeeva1, K. Solovyova2, A. Mosikian1 p = NS; gender: 50 vs 53% males, p = NS; duration of diabetes:
1 20.5 11.8 vs 21.1 11.7 years, p = NS; BMI: 23.4 3.6 vs
Saint-Petersburg Medical University, Endocrinology, Saint-
24.5 3.4 kg/m2, p = NS; total daily insulin dose: 44.9 29.5 vs
Petersburg, Russia
2 46.5 28.5, p = NS)]. The effectiveness of these two sensor-
Federal Almazov North-West Medical Research Centre, Saint-
augmented pumps was evaluated using CareLink software data.
Petersburg, Russia
Hypoglycaemia was defined as an episode of interstitial glucose
54mg/dl for 20 minutes and expressed as episodes per week.
Background and Aims: Potential hypoglycemia is a sig- Results: Patients used the 640G system exhibited significantly
nificant barrier to physical activity (PA) in T1D patients. lower hypoglycaemia episodes/week (1.9 1.3 vs 3.6 1.9, p =
We investigate different strategies for prevention of exercise 0.0004) as well as better glycaemic control (HbA1c: 6.9 0.6 vs
induced hypoglycemia to find links between prevention 7.3 0.9%, p = 0.03). There was only one episode of severe hy-
methods and hypoglycemia frequency or quality of glycemic poglycaemia in the Veo group. 640G group had better hy-
control. poglycaemia awareness status (Gold Score: 1.8 0.8 vs 2.6 1.6,
Methods: We asked 53 regularly exercised T1D patients (age p = 0.01). The mean activation of the Suspend before low
28.5 7.2 years, HbA1c 7.08 1.15%) using anonymous ques- feature of the 640G pump was 2.2 1.6 times/day.
tionnaire in Google Forms spread by specialised social media Conclusions: Our results demonstrate that the use of the
group Diabet Connect. Patients described their main strategies MiniMed 640G sensor-augmented pump system can help to re-
using to prevent exercise induced hypoglycemia and gave in- duce the frequency of hypoglycaemia, in combination with better
formation about frequency of hypoglycemia during and after the glycaemic control.
exercise (including nighttime). The data were processed using
Fishers criteria.
Results: There were no difference in HbA1c between subjects 270
who had prepared for PA and who didnt use any strategy to
prevent hypoglycemia (p = 0.87). Also there were no links be- REAL-TIME BASAL INSULIN ATTENUATION BASED
tween hypoglycemia frequency (including late and nighttime ON CONTINUOUS GLUCOSE MONITORING (CGM):
episodes) and the method of insulin therapy (CSII p = 0.95; MDI ASSESSMENT OF STATE-OF-ART ALGORITHMS
p = 0.61). The frequency of late hypoglycemia didnt differ be- IN 14-DAY IN SILICO SCENARIO
tween subjects with low blood glucose level just after the exer- M. Vettoretti1, C. Fanton1, A. Facchinetti1, L. Meneghetti1,
cise comparing with subjects with normal or high blood glucose G. Sparacino1, C. Cobelli1
level at the same moment. Subjects used temporal basal rate 1
(including its combination with additional CHO intake or addi- University of Padova, Department of Information Engineering,
tional intake and bolus dose reduction) had no preferences Padova, Italy
in hypoglycemia prevention comparing with other strategies
(p = 0.29, p = 0.29, p = 0.30 respectively) Background and Aims: Automatic attenuation of insulin
Conclusions: There is still no clear dependence upon the pump basal delivery is employed as a safety measure against hy-
method of exercise induced hypoglycemia prevention and poglycemia in open- and closed-loop type 1 diabetes (T1D) ther-
frequency of hypoglycemic episodes (including late and apies. So far, three algorithms were proposed: brakes (B) and
nigthtime hypoglycemias). It can be the object for the future power brakes (PB), based on hypoglycemia risk index calculated
research. from original and predicted CGM data, respectively (Hughes et al.,
A-112 ATTD 2017 E-POSTER VIEWING ABSTRACTS

JDST 2010), and PB with insulin on board (IOB-PB) (Patek et al., 272
IEEE TBME, 2012). B and PB were tested on short scenarios in
HYPOGLYCEMIA AS A FUNCTION OF HBA1C
absence of meals. IOB-PB was assessed on a single-day scenario.
IN TYPE 2 DIABETES (T2DM): INSULIN GLARGINE
Here, we compare these algorithms on a 14-day scenario.
300 U/ML IN A PATIENT-LEVEL META-ANALYSIS
Methods: B, PB and IOB-PB algorithms are tested in silico in
OF EDITION 1, 2 AND 3
a 14-day scenario reproducing open-loop insulin pump therapy in
100 virtual subjects. The T1D patient decision-making model of P. Choudhary1, R. Bonadonna2, J.F. Yale3,
(Vettoretti et al., Proc IEEE EMBC, 2015) is used for simulation, C. Brulle-Wohlhueter4, E. Boelle-Le Corfec5, T. Bailey6
which includes the UVA/Padova T1D simulator (2013 release), 1
insulin sensitivity variability (Visentin et al., IEEE TBME, 2015), Department of Diabetes, Kings College London, London,
CGM error (Facchinetti et al., MBEC, 2015) and patients be- United Kingdom
2
haviour in making treatment decisions, also considering possible Department of Clinical and Experimental Medicine, University
patients mistakes, e.g. in carb-counting. of Parma, Parma, Italy
3
Results: Time in hypoglycemia (Thypo) is 24.4 min/day with B, McGill University, Department of Medicine, Montreal,
23.1 min/day with PB and 9.0 min/day with IOB-PB, reduced Canada
4
compared to no-attenuation scenario (Thypo = 26.9 min/day). As Sanofi, Diabetes Division, Paris, France
5
a drawback, time in hyperglycemia (Thyper) is increased with B Sanofi, Biostatistics and Programming, Paris, France
6
(Thyper = 9 h/day), PB (Thyper = 9 h/day) and IOB-PB (Thyper = 10.4 h/ AMCR Institute, Escondido, CA, USA
day) compared to no-attenuation scenario (Thyper = 8.5 h/day).
Conclusions: In a multiple-day in silico scenario IOB-PB Background and Aims: Basal insulin therapy can be a
shows the best performance in reducing Thypo, albeit Thyper in- compromise between achieving glycemic targets and avoiding
creases by almost 2 h/day. hypoglycemia, dependent on how intensively insulin is titrated.
In the phase 3a EDITION 1, 2 and 3 studies, insulin glargine
300 U/mL (Gla-300) provided equivalent glycemic control to
271 insulin glargine 100 U/mL (Gla-100) with less hypoglycemia in
people with T2DM. The objective of the current analysis was to
INSULIN PUMP SATISFACTION IN PEDIATRICS evaluate rates of confirmed (70 mg/dL) or severe hypoglycemia
M. Yafi1, M. Silva1, C. Bowden1, C. Hughes1, S. Ohland1, over 6 months of treatment with Gla-300 or Gla-100 in these
R. Yetman2 EDITION studies, as a function of HbA1c.
1
Methods: Meta-analysis was performed on patient-level
University of Texas at Houston Health Science Center, data, and annualized hypoglycemia rate as a function of HbA1c
Pediatric Endocrinology, Houston, USA at month 6 was fitted using a negative binomial regression
2
University of Texas at Houston Health Science Center, model.
Pediatrics, Houston, USA Results: Adding a treatment-by-HbA1c interaction term to the
model did not significantly improve the goodness of fit (inter-
Background and Aims: Insulin pump therapy or Continuous action p-value 0.937 and 0.829 for anytime [24 h] and nocturnal
Subcutaneous Insulin Infusion (CSII) offers many benefits for [00:0005:59 h] hypoglycemia, respectively). Therefore the
pediatric patients living with Type 1 Diabetes Mellitus (TIDM) model without interaction describes the data accurately: people
including potential improvement of Hemoglobin A1C, flexibility treated with Gla-300 experienced a consistently lower rate of
with physical activity and exercise, prevention of fluctuation of confirmed (70 mg/dL) or severe hypoglycemia vs those treated
blood glucose levels and improvement of quality of life. CSII use with Gla-100, regardless of HbA1c at month 6 (Figure).
can be effective in very young children, competitive athletes, Conclusions: These results suggest that treatment with Gla-
children with needle phobia and children with severe fluctuation 300 vs Gla-100 could allow people with T2DM to achieve
of blood glucose levels. equivalent glycemic control with less hypoglycemia.
Methods: We surveyed 30 families who have children with Studies sponsored by Sanofi (NCT01499082, NCT01499095,
TD1M utilizing CSII. The project was approved by the Institu- NCT01676220).
tional Review Board in our university.
Results: Sixty three percent of the children using CSII were
between the age of 13 and 18 years. Fifty six percent of the total
patients had the CSII for more than 5 years. All of the families
felt that CSII gave their children more flexibility in their lives,
made managing diabetes easier and reported that they would
recommend it to other families with children with diabetes. We
didnt evaluate HbA1C improvement but 87% of the families
believed that CSII improved the HbA1C.
The frequency of reported problems in 1 year using CSII was:
Diabetes ketoacidosis episodes 23%, pump failure 60% and se-
vere hypoglycemia 13%
Conclusions: CSII offers a great satisfaction rate among pe-
diatric patients with TDIM and their families. The best areas of
satisfaction are related to improving flexibility in lifestyle.
This small survey confirms the reported satisfaction in the
medical literature that patients and families get from CSII use.
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-113

273 tion with IDeg. HbA1c reduction was comparable with Gla-300
and Gla-100, whereas FPG reduction was significantly greater
CLINICAL PERSPECTIVES FROM THE BEGIN
with Gla-100 in the fixed but not random effect model. Risk of 1
AND EDITION PROGRAMS: TRIAL-LEVEL META-
confirmed (<56 mg/dL) or severe hypoglycemic event was lower
ANALYSES OUTCOMES WITH EITHER DEGLUDEC
with IDeg vs Gla-100 at night (00:0105:59 h) but comparable at
OR GLARGINE 300 U/ML VS GLARGINE 100 U/ML
any time (24 h). Risk of 1 confirmed (<54 mg/dL) or severe
IN T2DM
hypoglycemic event (nocturnal [00:0005:59 h] and anytime)
R. Roussel1, R. Ritzel2, S. Chevalier3, B. Balkau4, was lower with Gla-300 vs Gla-100. Risk of 1 documented
J. Rosenstock5 symptomatic (70 mg/dL) hypoglycemic event with IDeg or
1 Gla-300 vs Gla-100, nocturnal and anytime, closely reflected
Assistance Publique Hopitaux de Paris, Bichat Hospital, Paris, confirmed or severe events. Risk of 1 severe hypoglycemic
France event was comparable with IDeg or Gla-300 vs Gla-100.
2
Klinikum Schwabing, Stadtisches Klinikum Munchen GmbH, Conclusions: In trial-level meta-analyses in T2DM, Gla-100
Munich, Germany reduced HbA1c more than IDeg despite IDeg having a greater
3
Sanofi, Biostatistics and Programming, Paris, France FPG-lowering effect. Hypoglycemia risk was lower with IDeg vs
4
INSERM U1018, Center for Research in Epidemiology Gla-100 for nocturnal but not anytime events. Gla-300 provided
and Population Health, Villejuif, France comparable glycemic control to Gla-100 with lower risk of any-
5
Dallas Diabetes Research Center, Medical City, Dallas, USA time and nocturnal hypoglycemia. Head-to-head trials of IDeg vs
Gla-300 are warranted.
Background and Aims: The BEGIN and EDITION pro- Trial-level meta-analyses supported by Sanofi.
grams, including adults with T2DM on basal-bolus or basal-oral
therapy and those who were insulin nave, evaluated efficacy
and safety of insulin degludec (IDeg; BEGIN) and insulin glar-
gine 300 U/mL (Gla-300; EDITION) versus glargine 100 U/mL 274
(Gla-100).
Methods: HbA1c, fasting plasma glucose (FPG) and hypo- DRIVERS OF AND BARRIERS TO OPTIMAL BASAL
glycemia incidence with IDeg or Gla-300 vs Gla-100 were ex- INSULIN (BI) TITRATION: RESULTS OF A
plored in 2 trial-level meta-analyses. QUANTITATIVE SURVEY
Results: HbA1c reduction was significantly greater for Gla- L. Berard1, M. Bonnemaire2, M. Mical2, S. Edelman3,
100 vs IDeg despite significantly more pronounced FPG reduc- K. Khunti4
1
Winnipeg Regional Health Authority, Health Sciences Centre,
Winnipeg, Canada
2
Sanofi, Diabetes Division, Paris, France
3
University of California, San Diego, CA, USA
4
Diabetes Research Centre, University of Leicester, Leicester,
United Kingdom

Background and Aims: Evaluation of drivers of and barriers

to optimal BI titration in the USA, France, and Germany.
Methods: Online survey of 386 healthcare professionals
(HCPs) and 318 people with type 2 diabetes (T2DM) on long-
acting BI for 636 months, including 243 current BI users
(currently self-titrating n = 95), and 75 users who had dis-
continued BI for <12 months.
Results: Not all HCPs used fasting self-monitored plasma
glucose (SMPG) targets recommended in country-specific guide-
lines and preferred higher fasting SMPG targets. For current BI
users, mean start dose was 15 U and mean current dose was 25 U.
By 27 months, dose increases were <10 U for 62% of current BI
users, despite 49% reporting not reaching HbA1c target. Main
barriers to optimal titration for current BI users were concerns
over weight gain (52%), the time to reach goal (43%), perception
that dose increase meant worsening of disease (38%) and fear of
hypoglycemia (37%). HCPs perceived the main barriers to target
attainment in self-titrating patients to be fear of hypoglycemia
(74%) and low patient involvement/motivation (63%). Self-
titration was preferred by 26% of current BI users; 39% of cur-
rent BI users were self-titrating.
Conclusions: Generally, HCPs prefer a slower, safer ap-
proach to titration to higher glucose targets than recommended to
avoid hypoglycemia. However, patients not at target are frus-
trated about time taken to reach target and are less concerned
A-114 ATTD 2017 E-POSTER VIEWING ABSTRACTS

28.4% (EDITION) of patients, with a similar mean HbA1c (SD)

of 7.6% (0.9) at Week 12. Number of patients experiencing
nocturnal hypoglycemia (INSIGHT = 28.7%; EDITION =
27.5%), mean (SD) insulin dose at Week 12 (INSIGHT = 67.0 U
[37.8]; EDITION = 70.0 U [43.1]), and weight change from
baseline (INSIGHT = 0.41 kg [3.3]; EDITION = 0.15 kg [2.4])
were similar between groups. No between-group differences in
adverse events were noted.
Conclusions: Application of a 1 U/day self-titration algorithm
with Gla-300 is well-tolerated, effective and comparable to the
previously tested EDITION algorithm.
Study sponsored by Sanofi (NCT02401243).

276
WIDER WINDOWS FOR EVALUATING NOCTURNAL
HYPOGLYCEMIA CAPTURE MORE EVENTS
AND CONFIRM LOWER NOCTURNAL
about hypoglycemia than HCPs. Preference for self-titration HYPOGLYCEMIA RISK WITH INSULIN
needs improvement and patients need encouraging to self-titrate. GLARGINE 300 U/ML VERSUS 100 U/ML IN T2DM
Market survey project conducted by Hall & Partners US LLC,
funded by Sanofi. G. Bolli1, C. Wysham2, M. Fisher3, S. Chevalier4, A. Cali5,
B. Leroy5, M. Riddle6
1
University of Perugia, School of Medicine, Perugia, Italy
2
Diabetes and Endocrinology Center, Rockwood Clinic,
275 Spokane, USA
3
A PRAGMATIC SELF-TITRATION 1 UNIT/DAY Glasgow Royal Infirmary, University of Glasgow, Glasgow,
(INSIGHT) ALGORITHM FOR INSULIN GLARGINE United Kingdom
4
300 U/ML (GLA-300) IS WELL-TOLERATED Sanofi, Biostatistics and Programming, Paris, France
5
AND EFFECTIVE Sanofi, Diabetes Division, Paris, France
6
Oregon Health & Science University, Portland, OR, USA
J.F. Yale1, S. Harris2, L. Berard3, M. Groleau4, P. Javadi5,
J. Stewart4 Background and Aims: The EDITION clinical trials in
1
McGill University, Department of Medicine, Montreal, T2DM show comparable glycemic control with less nocturnal
Canada (00:0005:59 h) and anytime (24 h) hypoglycemia for insulin
2
Schulich School of Medicine & Dentistry, The University glargine 300 U/mL (Gla-300) versus insulin glargine 100 U/mL
of Western Ontario, London, Canada (Gla-100). However, the predefined nocturnal window may not
3
Winnipeg Regional Health Authority, Health Sciences Centre, capture all clinically relevant hypoglycemic events during the
Winnipeg, Canada true fasting period (i.e. late evening to prebreakfast).
4
Sanofi, Laval, QC, Canada Methods: Post hoc patient-level meta-analyses of pooled 6
5
Sanofi, Diabetes Division, Paris, France months data from EDITION 2, EDITION 3 and EDITION JP
2 (N = 1922; Fig) compared hypoglycemia using predefined
Background and Aims: Gla-300 provides comparable gly- (00:0005:59 h) and broader nocturnal windows (22:0005:59 h,
cemic control and less hypoglycemia versus insulin glargine
100 U/mL (Gla-100). In the phase 3 EDITION trials, insulin was
titrated once weekly (no more than every 3 days) by the HCP,
based on the median of the last three fasting prebreakfast SMPGs.
In Canada, the INSIGHT pragmatic 1 U/day self-titration protocol
is widely used with Gla-100.
Methods: This 12-week, randomized pilot study compared
safety and efficacy of the INSIGHT and EDITION titration al-
gorithms with Gla-300 in people with T2DM (insulin-nave or on
basal insulin OAD).
Results: 212 participants were randomized (37.0% insulin-
nave). Mean baseline characteristics: age 62.3 years, BMI
34.2 kg/m2, HbA1c 8.4%, prior basal insulin dose 57.2 U. Com-
parable numbers of patients in each titration group reached the
primary endpoint, fasting SMPG 5.6 mmol/L without nocturnal
(0:006:00 h) hypoglycemia (confirmed: SMPG 3.9 mmol/L or
symptomatic or severe) at 12 weeks (INSIGHT = 22.8%; EDI-
TION = 20.6%). No between-treatment differences in severe
hypoglycemic episodes were noted (INSIGHT = 1; EDITION =
3). HbA1c 7% was achieved by 28.7% (INSIGHT) versus
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-115

00:0007:59 h, 22:00 h to prebreakfast self-monitored plasma components of eA1c were computed separately, distinguishing
glucose [SMPG]). between contributions of fasting and postprandial SMBG values to
Results: Total numbers of events were consistently greatest overall eA1c estimates.
for 22:00 h to prebreakfast SMPG versus other windows, and Results: Accounting for baseline differences, Gla-300 low-
absolute differences (favoring Gla-300) were greater for windows ered fasting BG more than Gla-100, by -30mg/dl vs. -21.3mg/dl,
extending beyond 05:59 h (Fig). Risk of 1 confirmed (70 mg/ p < 0.001. As seen in Figure 1, the improvements of both the
dL) or severe event was lower for Gla-300 versus Gla-100 using fasting and the prandial components of eA1c were more pro-
the predefined and broader windows. Corresponding annualized nounced on Gla-300 than on Gla-100.
rates were also lower with Gla-300 for all windows (predefined Conclusions: In conclusion, while the EDITION 2 study has
41%, 22:0005:59 h 34%, 00:0007:59 h 30%, 22:00 h to pre- demonstrated that Gla-300 was similarly effective to Gla-100 in
breakfast SMPG 29% lower). A similar pattern of lower risk for terms of HbA1c lowering, Gla-300 achieved better overall eA1c
Gla-300 versus Gla-100 was seen with other hypoglycemia types. control than Gla-100 in completers, and for both fasting and
Conclusions: Nocturnal hypoglycemia risk was consistently prandial components.
lower for Gla-300 versus Gla-100 irrespective of hypoglycemia
definition or nocturnal window. 278
Studies sponsored by Sanofi (NCT01499095, NCT01676220,
NCT01689142). NEW INSULIN DEGLUDEC REDUCES
HYPOGLYCEMIA IN PATIENTS WITH TYPE 1
DIABETES AT HIGH RISK OF HYPOGLYCEMIA
M. Cokolic1
277
1
University Clinical Centre Maribor, Department of
ACTION PROFILES OF GLARGINE 300 AND Endocrinology and Diabetology- Internal Clinic, Maribor,
GLARGINE 100 IN TYPE 2 DIABETES: AN EA1C Slovenia
COMPARISON OF THE FASTING AND PRANDIAL
COMPONENTS OF AVERAGE GLYCEMIC CONTROL Background and Aims: Intensive Insulin Therapy (IIT) is a
M. Breton1, B. Kovatchev1, Z. Meng2, B. Leroy3, A. Cali3 therapeutic regimen for the treatment of type 1 diabetes (T1D). It is
1 a common notion that more frequent hypoglycemia is a disadvan-
UVA, Center for Diabetes Technology, Charlottesville, USA tage of intensive regimens. Many patients, who have had diabetes
2
Sanofi, Diabetes Division, Bridgewater, USA for a long time, do not notice the signs of hypoglycemia or they do
3
Sanofi, Diabetes Division, Paris, France not recognize the condition as it occurs without associated symp-
toms or during sleep. Severe hypoglycemia can be life-threatening.
Background and Aims: Insulin glargine 300U/mL (Gla-300) Methods: In 10 T1D patients suffering from frequent severe
is a long-acting insulin analog indicated for adults with diabetes hypoglycemia receiving treatment in the form of conventional
mellitus. Pivotal clinical trials compared Gla-300 vs. insulin IIT. We have replaced their conventional basal insulin with a
glargine 100U/mL (Gla-100), showing that Gla-300 and Gla-100 new ultra-long-acting basal insulin analogue degludec (IDeg).
achieve similar HbA1c outcomes. As HbA1c represents a sur- Results: This exploratory analysis included 10 T1D patients
rogate measure for average glycemia through hemoglobin gly- (30% female) with frequent low glucose concentrations and the
cation, it may deviate substantially from average glucose levels; following baseline characteristics: mean (SD) age 49.9 11.8
estimated A1c (eA1c) also indirectly measures average glycemia years, duration of T1D 17 8.5 years, HbA1c <8.52 1.5%, BMI
using a dynamical model with self-monitored blood glucose 25.5 4.6kg/m2, waist 94.7 10.1 cm, weight 77.3 12.1 kg.
(SMBG) as input. We propose to study differences in average HbA1c decreased from 8.52% to 8.07% after 10 months without
glycemia between Gla-300 and Gla-100 using eA1c, calibrated any changes in other parameters without hypoglycemia in the
with both reference HbA1c and SMBG day profiles. total treatment period with IDeg.
Methods: SMBG data from the EDITION-2 study (NCT0 Conclusions: Hypoglycemia represents a great challenge for
1499095) - a multicenter, open-label, clinical trial, which ran- both the patient and the healthcare professional; however, hy-
domized N = 811 participants to Gla-300 or Gla-100 once daily for poglycemia can be successfully managed. After replacing the
12 months (6m + 6m extension) were analyzed. N = 440 subjects conventional basal insulin with an ultra-long-acting basal insulin
with enough SMBG data for eA1c assessments at 0, 3, 6 and 12 analogue degludec this resulted in a consistent reduction in hy-
months (completers) were selected, and the fasting and prandial poglycemia with improved glycemic control in T1D on IIT pa-
tients at high risk of hypoglycemia.

279
SAFETY AND EFFICACY OF INSULIN DEGLUDEC
IN PATIENTS WITH TYPE 1 DIABETES MELLITUS
AFTER ONE YEAR OF TREATMENT
T. Didangelos1, K. Tziomalos1, A. Mourouglakis1, E. Karlafti1,
D. Stogianou1, S. Alkayiet1, A. Sofogianni1, A. Hatzitolios1
1
AHEPA University Hospital, 1st Propeudetic Department
of Internal Medicine, Thessaloniki, Greece

Background and Aims: To evaluate the safety and efficacy of

insulin degludec in patients with type 1 diabetes mellitus (T1DM).
A-116 ATTD 2017 E-POSTER VIEWING ABSTRACTS

Methods: We studied 31 patients with T1DM (61.3% males,

mean age 45.5 17.6 years), who were switched to insulin de-
gludec after at least two years of treatment with another basal
insulin. Patients were followed up another year after switching to
insulin degludec.
Results: Patients were previously treated with insulin glargine
(77.4%) or detemir (22.6%). During the 24 months prior to the
switch to insulin degludec, 29 patients (93.5%) experienced at
least 1 non-severe hypoglycemic episode (mean 12.7 3.7 epi-
sodes, among which 4.6 1.9 occurred during nighttime). During
the 12 months after the switch to insulin degludec, 4 patients
(12.9%) experienced a non-severe hypoglycemic episode (1, 2, 1
and 3 episodes, respectively, among which none occurred dur-
ing nighttime). At the time of switch to insulin degludec, HbA1c
and plasma glucose levels were 7.0 0.6% and 130 15 mg/dl,
respectively, and after 12 months decreased to 6.1 0.2% and
100 9 mg/dl, respectively (p < 0.001 for both comparisons). No
change observed on body weight, on dose of insulin degludec vs
previously administered insulin and on the prandial insulin,
during treatment with insulin degludec.
Conclusions: In patients with T1DM, switching from other
basal insulins to insulin degludec reduces the risk of hypogly- and Gla-100 regardless of renal function; no evidence of het-
cemia and improves glycemic control without body weight gain erogeneity of treatment effect was demonstrated across sub-
and without a need to increase insulin doses. groups (p = 0.46; Table). Risk of confirmed (70 mg/dL) or
severe hypoglycemia was significantly lower for nocturnal
(00:0005:59 h) events and comparable or lower for anytime
(24 h) events for Gla-300 versus Gla-100 across subgroups. Re-
280 nal function did not affect the lower rate of nocturnal or any time
hypoglycemia (no evidence of heterogeneity of treatment effect
GLYCEMIC CONTROL AND HYPOGLYCEMIA across subgroups: p = 0.73, p = 0.27).
BENEFITS WITH INSULIN GLARGINE 300 U/ML Conclusions: Gla-300 provided comparable glycemic control
(GLA-300) EXTEND TO PEOPLE WITH TYPE and consistently reduced nocturnal hypoglycemia versus Gla-
2 DIABETES (T2DM) AND MILD-TO-MODERATE 100 in participants with T2DM regardless of renal function, with
RENAL IMPAIRMENT no increase in any time hypoglycemia.
Studies sponsored by Sanofi EDITION 1 (NCT01499082),
J. Escalada1, S. Halimi2, P. Senior3, M. Bonnemaire4, A. Cali4,
EDITION 2 (NCT01499095), EDITION 3 (NCT01676220).
S. Chevalier5, J. Karalliedde6, R. Ritzel7
1
Clnic University of Navarra, Endocrinology and Nutrition, 281
Pamplona, Spain
2
Department of Diabetology, Endocrinology and Nutrition, INSULIN GLARGINE (IGLAR) AND INSULIN
Grenoble University Hospital Center, Grenoble, France DETEMIR (IDET) IN PATIENTS WITH TYPE 2
3
Division of Endocrinology, University of Alberta, Edmonton, DIABETES IN RUSSIA: A RETROSPECTIVE
Canada PRAGMATIC STUDY
4
Sanofi, Diabetes Division, Paris, France E. Patrakeeva1, O. Martyanova2, A. Mosikian1,
5
Sanofi, Biostatistics and Programming, Paris, France M. Zakharchenko1
6
Cardiovascular Division, Faculty of Life Sciences & Medicine,
1
Kings College London, London, United Kingdom Saint-Petersburg Medical University, Endocrinology,
7
Klinikum Schwabing, Stadtisches Klinikum Munchen GmbH, Saint-Petersburg, Russia
2
Munich, Germany Saint-Petersburg Outpatient Clinic No 27, Saint-Petersburg,
Russia
Background and Aims: EDITION 1, 2 and 3 showed that
over 6 months, Gla-300 provided comparable glycemic control Background and Aims: Meta-analyses of randomized clin-
to glargine 100 U/mL (Gla-100) and less hypoglycemia in people ical trials comparing IGlar and IDet show no difference in
with T2DM. As renal impairment increases the risk of hypo- their treatment efficacy, but the equality of its effectiveness
glycemia in people with T2DM, the aim was to investigate the in real-life clinical settings in Russia remains questionable.
impact of renal function on HbA1c reduction and risk of hypo- We aimed to provide this comparison in a retrospective prag-
glycemia in participants treated with Gla-300 and Gla-100. matic study.
Methods: The effects of Gla-300 versus Gla-100 on HbA1c Methods: We have analyzed 2000 existed medical records of
reduction and hypoglycemia were assessed in renal function patients with diabetes in an outpatient department for people older
subgroups (baseline eGFR 30 to <60, 60 to <90 and 90 mL/ than 18 y.o. All the records of patients who met the inclusion
min/1.73 m2) in a post hoc patient-level meta-analysis of data criteria (T2D for more than 6 month, IGlar of IDet treatment for
from EDITION 1, 2 and 3. more than 6 month) were included into analysis. Groups were
Results: Most participants (56%) had baseline eGFR 60 to compared with the use of Wilcoxon test or continuity-corrected
<90 mL/min/1.73 m2 (Table). Non-inferiority for HbA1c reduc- Chi-Squared test (were appropriate). Factors, which possibly in-
tion (defined in the EDITION studies) was shown for Gla-300 fluence on HbA1c level, were analyzed via two-way ANOVA.
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-117

Results: We included 133 medical records (70% IGlar, 53%

women) into analysis. HbA1c was (meanSE) 7.86 0.14% and
7.72 0.22% in IGlar and IDet groups, respectively (p = 0.54).
Average dose of basal insulin was 0.31 0.01 U/kg and 0.32
0.02 U/kg in IGlar and IDet groups, respectively (p = 0.31). There
was no between-group difference in percent of patients, reached
a target goal of HbA1c 7.5%. There was no between-group
difference for a target goal of HbA1c 7.0% neither.
Conclusions: No difference between IGlar and IDet effec-
tiveness for T2D treatment was shown in real-life clinical
settings in Russia. These results conform to ones shown in ran-
domized clinical trials meta-analyses. Thus, we propose to base a
choice mainly on the quality of patients life (frequency of in-
jections, hypoglycemic events and other adverse reactions).

282
SWITCH FROM GLARGINE TO DEGLUDEC:
EFFICACY AND SAFETY IN CHILDREN
AND ADOLESCENTS WITH TYPE 1 DIABETES
acting or regular mealtime (MT) (p = 0.033; Figure 1b) insulin.
L. Iughetti1, P. Bruzzi1, C. Toffoli1, S. Poluzzi1, F. Leo1, Despite the lack of statistical significance, during IDeg treatment
G. Maltoni2, T. Suprani3, S. Zucchini2, B. Predieri1 HbA1c values decreased by -0.35% (T1) and -0.20% (T2)
1
University of Modena and Reggio Emilia, Department points and FPG improved by 37.6%. z-score BMI did not change
of Medical and Surgical Sciences of the Mother, Children and no episode of severe hypoglycaemia was reported.
and Adults, Modena, Italy Conclusions: IDeg seems to improve the glycemic control
2
S. Orsola-Malpighi Hospital, Department of Pediatrics, than therapy with IGlar. Despite data from adults with T1D, our
Bologna, Italy results in childhood suggest that the dose of IDeg should be
3
Bufalini Hospital, Department of Pediatrics, Cesena, Italy reduced by median 5% and the meal-time insulin appropriate
replacement doses should be lowered by 10% for patients who
Background and Aims: Insulin degludec (IDeg) is a new previously received IGlar. IDeg might be considered a useful and
basal insulin that induces a flat and stable action profile with low well tolerated basal insulin also in childhood.
variability. Few data have been published on IDeg effects in
chilhood. Aim of this study was to assess the efficacy and the
safety of IDeg in children and adolescents with type 1 diabetes
(T1D) previously treated with IGlar. 283
Methods: Twenty patients (15.1 4.0 yrs; 9 males; T1D du- TO STUDY THE FREQUENCY OF NOCTURNAL
ration 7.2 3.7 yrs; IGlar treatment at least 1 year) were switched HYPOGLYCEMIA IN T2DM PATIENTS
to IDeg once daily. z-score BMI, HbA1c, fasting plasma glucose ON GLARGINE VS. DEGLUDEC USING
(FPG), severe hypoglycaemia rates, and insulin dose (IGlar or THE FLASH GLUCOSE MONITORING SYSTEM
IDeg plus short-acting or regular) were collected at baseline (T0, BY THE LIBRE PRO
during IGlar treatment), 3 months (T1), and 6 months (T2) after
IDeg was started. M. Saiyed1, B. Saboo1, H. Chandarana1,
Results: The switch from IGlar to IDeg allowed a significant D. Hasnani1, F. Patel1
reduction in doses of both basal (p = 0.018; Figure 1a) and short- 1
Dia Care Diabetes Care and Hormone Clinic, Diabetology,
Ahmedabad, India

Background and Aims: To find out the frequency of occur-

rence of nocturnal hypoglycemia in T2DM patients on Glargine
vs Degludec once daily using the Libre Pro sensor.
Methods: 30 adult T2DM Indian subjects who were on basal
insulin (Glargine) OHA (non SU) were studied.
Inclusion Criteria: HbA1c 7 - 8.5%.
Exclusion criteria: patients with symptomatic and documented
hypoglycemia and with renal dysfunction.
The Libre Pro flash monitoring sensor was applied to them
after taking consent. They were called on the 4th & 7th days after
application of sensor and the readings were taken and the diet and
doses of the basal insulin were optimised. The patients who had
nocturnal hypoglycemia after optimization of glargine doses to
achieve target fasting were shifted over to degludec.
On the 11th & 14th days, all the patients were again followed
up for optimization of therapy.
A-118 ATTD 2017 E-POSTER VIEWING ABSTRACTS

1
Results: Despite multiple optimization of glargine doses, 18 out IBILI - Institute for Biomedical Imaging and Life Sciences,
of 30 patients had nocturnal hypoglycemia and had to be shifted Faculty of Medicine, University of Coimbra, Laboratory
over to degludec. Also, up titration of glargine failed to achieve the of Physiology, Coimbra, Portugal
2
fasting glucose target. Insulin requirement of the patients on i3S - Instituto de Investigacao e Inovacao em Saude,
degludec reduced by 8% from the initial dose of glargine. Universidade do Porto, Biomaterials for Multistage Drug
Conclusions: We could observe that the patients on Degludec and Cell Delivery, Porto, Portugal
3
had less frequency of nocturnal hypoglycemia in comparison to INEB - Instituto de Engenharia Biomedica, Biocarrier, Porto,
glargine. Portugal
4
Faculty of Medicine, University of Coimbra, Laboratory
of Oncobiology and Hematology, University Clinic
of Hematology and Applied Molecular Biology Unit,
284
Coimbra, Portugal
5
MODELING SUBCUTANEOUS ABSORPTION CNC.IBILI - Center for Neuroscience and Cell Biology,
OF U100 AND U300 INSULIN GLARGINE University of Coimbra, Coimbra, Portugal
6
IN TYPE 1 DIABETES CIMAGO - Centre of Investigation in Environment Genetics
and Oncobiology, Faculty of Medicine, University of Coimbra,
M. Schiavon1, R. Visentin1, C. Dalla Man1, T. Klabunde2, Coimbra, Portugal
C. Cobelli1 7
FEUP - Faculdade de Engenharia, Universidade do Porto,
1 Porto, Portugal
University of Padova, Department of Information Engineering,
8
Padova, Italy ICBAS - Instituto de Ciencias Biomedicas Abel Salazar,
2 Universidade do Porto, Porto, Portugal
Sanofi-Aventis Deutschland GmbH, Drug Design, Science
& Medical Affairs, Frankfurt am Main, Germany
Background and Aims: The bioartificial pancreas is the most
Background and Aims: Subcutaneous administration of long- complex insulin-delivery system and presents a promising meth-
acting insulin analogs are often employed in multiple daily injec- odology to treat diabetes. To ameliorate the function of hydrogel-
tion (MDI) therapy of type 1 diabetes (T1D) to cover patients encapsulated beta cells, we describe a strategy aiming to mimic
basal insulin needs. Among these, U300 and U100 are formulations the in vivo cellular microenvironment by the establishment of
of insulin glargine indicated for once daily subcutaneous admin- interactions between cells and extracellular matrix (ECM)
istration in MDI therapy of type 1 diabetes (T1D). U300 is a new components.
formulation with a different absorption kinetics with respect to Methods: We chemically modified alginate hydrogels by
U100 resulting in less hypoglycemia in clinical trials. Some models using carbodiimide chemistry to incorporate ECM components,
have already been proposed but were not assessed under controlled namely the tripeptide Arg-Gly-Asp (RGD). After beta cell en-
experimental conditions for both formulations. The objective is to capsulation in RGD-alginate hydrogels, we analysed the effect of
develop a model of subcutaneous absorption of U100 and U300 these adhesive peptides on cells performance both in vitro and
glargine insulin formulations in T1D. in vivo.
Methods: The database consists of 24 T1D subjects who Results: The proposed RGD-alginate model demonstrated to
underwent a randomized, 4-sequence, cross-over, double-blind, be a good in vitro niche for supporting beta cells viability, pro-
dose-response euglycemic clamp study, receiving single subcu- liferation and activity, namely by improving the key feature of
taneous injections of 0.4, 0.6 and 0.9 U/kg U300 and 0.4 U/kg insulin secretion. The presence of RGD peptides promoted cell-
U100. Plasma insulin concentrations were measured for 36 hours matrix interactions, enhanced the expression of endogenous
using a validated radioimmunoassay. Model identification was ECM components and favoured the assembly of individual cells
performed on U100 and U300 data using a Bayesian Maximum a into multicellular spheroids, an essential configuration for proper
Posteriori technique. beta cell survival and functioning. In vivo, our pivotal model for
Results: The model well fits the data and provides precise diabetes treatment showed to improve glycaemic profile of type 2
parameter estimates for both insulin formulations. The model diabetic rats, where insulin secreted from encapsulated cells was
describes the gradual dissolution from the precipitate to soluble more efficiently used.
states and model parameters allow to characterize the different Conclusions: The RGD-alginate system stands out as a
rates of absorption between U100 and U300. promising approach to improve beta cell survival and function
Conclusions: The model will be incorporated into the UVA/ and increase the success of this therapeutic strategy that could
Padova T1D simulator together with the joint parameter distri- greatly improve the quality of life of an increasing number of
butions. This will open the door to perform in silico clinical trials diabetic patients worldwide.
for testing novel up-titration and insulin glargine switching rules.

286
285 PHYSIOLOGICAL EFFECTS OF INTRAPERITONEAL
VERSUS SUBCUTANEOUS INSULIN DELIVERY
THE EFFECT OF EXTRACELLULAR MATRIX IN PATIENTS WITH DIABETES MELLITUS TYPE 1:
COMPONENTS IN ALGINATE HYDROGELS A SYSTEMATIC REVIEW
FOR BIOARTIFICIAL PANCREAS: AN IDEAL
NICHE TO IMPROVE INSULIN SECRETION I. Dirnena-Fusini1,2, M.K. Am1,2,3, S.C. Christiansen1,2,4,
AND DIABETIC GLUCOSE PROFILE A.L. Fougner2,3,5, S.M. Carlsen1,2,4
J. Crisostomo1, A.M. Pereira1, S.J. Bidarra2,3, 1
Norwegian University of Science and Technology, Department
A.C. Goncalves4,5,6, P.L. Granja2,3,7, C.C. Barrias2,3,8, of Cancer Research and Molecular Medicine, Trondheim,
R. Seica1 Norway
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-119

2
Norwegian University of Science and Technology, Artificial center, single arm, treat to target study. The study comprised three
Pancreas Trondheim APT, Trondheim, Norway periods: baseline (conventional insulin injections), transition to
3
Central Norway Regional Health Authority, RHA, Trondheim, PAQ and PAQ treatment (12-weeks). Endpoints included: HbA1c,
Norway 7-point self-monitored blood glucose (SMBG), Barriers to Insulin
4
St. Olavs University Hospital, Department of Endocrinology, Treatment (BIT), Diabetes Treatment Satisfaction Questionnaire
Trondheim, Norway (DTSQ) and Short Form-36 (SF-36) survey.
5
Norwegian University of Science and Technology, Department Results: Mean age 63 7 years, 85% male, BMI 32.2 3.7 kg/
of Engineering Cygernetics, Trondheim, Norway m2, diabetes duration 15 7 years, HbA1c 70 12 mmol/mol.
Seventeen patients completed the study. After 12 weeks of PAQ
Background and Aims: Intraperitoneal (IP) insulin admin- treatment, significant reductions in HbA1c, -16 10 mmol/mol
istration in patients with diabetes mellitus type 1 (DM1) may (p 0001), and all 7-point SMBG values (p < 0.05) were seen. A
have physiological advantages but also carry potential risks moderate reduction in the total BIT score, -4.5 11.6 (p = 0.09) was
according to the invasive route of administration. It is impor- observed. The DTSQ total score significantly increased, 3.0 6.3
tant to weigh risks and advantages before initiating IP insulin (p<0.05), the hyperglycemia and hypoglycemia scores decreased,
infusion. -2.10 2.8 (p = 0.0041) and -0.50 2.5 (p = 0.3748), respectively.
Several studies have compared IP with subcutaneous (SC) in- The mental health score in the SF-36 increased 5.5 14.7 (p = 0.11)
sulin administration. However, no comparative summary of these while all other parameters were similar to baseline.
observations has been performed. Therefore, we initiated a sys- Conclusions: People with T2D who utilized PAQ achieved
tematic review to identify the possible different physiologic effects significantly better glycemic control, had a trend toward a re-
of IP versus SC insulin administration in patients with DM1. duction of BIT and had an increase in their DTS compared to
Methods: Specific search strategies in PubMed, Scopus, their baseline insulin therapy.
CENTRAL and Embase databases identified 1,517 publications
and 106 of these were recognized as important for the review.
After screening the references of the included publications
complementary materials were added to the research materials. 288
All results were analyzed, combined and calculated using sta-
SMARTCAP FOR TRACKING DISPOSABLE INSULIN
tistical software Rev Man 5.3.
INJECTION PENS
Results: The HbA1c levels of participants in the included
studies were compared at the beginning of IP insulin adminis- B. Peters1, A. Lepple-Wienhuese2, S. Rigante3, S. Joly3,
tration and after 3, 6, 12 and 24 months. R. El Hage Ali3
Our preliminary analyses of patients crossing over from SC to 1
IP treatment indicate a mean HbA1c decrease over time, with Valtronic Technologies Holding SA, Corp Business
some exceptions where mean HbA1c levels increased. Contrary, Development & Innovation, Les Charbonnieres, Switzerland
2
case-control studies indicate minimal superiority of IP delivery Valtronic Technologies Holding SA, Innovation, Lausanne,
evaluated by HbA1c changes. Switzerland
3
Conclusions: In patients switching from SC to IP insulin, Valtronic Technologies Holding SA, Innovation,
there was tendency towards improved blood glucose control Les Charbonnieres, Switzerland
based on changes of HbA1c. Metabolic improvements will be the
scope of ongoing analyses, as will results according to metabolic Background and Aims: 80% of diabetic patients use dis-
complications. posable insulin pens lacking data recording capability.
Methods: We present Smartcap, a novel intelligent cap re-
cording dose and time of injections automatically without user
input. Wireless data transfer helps caretakers and patients to
improve treatment. Dangerous dosing errors can be recognized
287 and alerted. Figure 1 shows Smartcap on a commercial insulin
PAQ INSULIN DELIVERY DEVICE IMPROVES pen. The device can be easily adapted to other pens. The insulin
GLYCEMIC CONTROL AND DIABETES
TREATEMENT SATISFACTION IN ADULTS
WITH TYPE 2 DIABETES
J. Mader1, N. Hermanns2, L. Lilly3, F. Aberer1, T. Poettler1,
C. Lanz1, J. Warner3, T. Pieber1
1
Medical University of Graz, Internal Medicine,
Graz, Austria
2
Diabetes Zentrum, Forschungsinstitut Diabetes-
Akademie Bad Mergentheim FIDAM GmbH, Bad
Mergentheim, Germany
3
CeQur, Clinical Research, Marlborough, USA

Background and Aims: PAQ is a patch on device which

provides set basal rates and bolus insulin on demand.
Methods: Twenty adults with type 2 diabetes (T2D) with
HbA1C 53 97 mmol/mol on an established regimen of basal-
bolus insulin therapy (2 injections per day) oral and/or in-
jectable glucose lowering agents were enrolled in this single
A-120 ATTD 2017 E-POSTER VIEWING ABSTRACTS

might be a good, less invasive, alternative way to implanted

pumps. Here we report the first Italian experience on patients
with the DiaPort system.
Methods: five patients (nos. 1-4 from Olbia and no. 5 from
Partinico) using the DIA, all patients were implanted after a
period in MDI and CSII
Results:

MDI MDI CSII CSII CIPII CIPII CIPII

PZ BM1 INSULIN HbA1c/HYP INSULIN HbA1c/HYP IMPL. INSULIN HbA1c/HYP

IU IU DATE IU
1 OLBIA 48.3 108 10.3% /NO 73 8.6% /NO SEPT 44 7.4% / NO
2015
2 OLBIA 26.6 82 8.4% /YES 54 6.9% / YES SEPT 40 6.5% /NO
2015
3 OLBIA 44.4 125 9.9% / NO 102 8.9% /NO SEPT 60 NA/ NO
2016
4 OLBIA 20.1 60 10.4%/ YES 45 8.6% /YES SEPT 35 NA/ NO
2016
5 PARTINICO 21.9 71 7.8% /YES 59 7.0% /YES JUNE 46 6.6% /NO
2016

Conclusions: A better metabolic control with less glycaemic

variability was evident. After 9 months implantation pz 2 ex-
perienced a reduction in catheter flow (obstruction for bolus of
dose measurements are possible through electrodes in the cap,
10 U or more) restored in the radiology unit. The diaport system
connected to a capacitance to digital converter circuit including a
is a good and easy to manage device in dedicated patients with
Bluetooth module for data transmission. The capacitance is a
improvement of clinical condition and excellent acceptability by
function of remaining volume in the drug container.
the patient. Rescue manoeuvres are less invasive than with the
Results: The precision is comparable to the analytical balance
implantable pumps.
determined weight for assessing the dispensed drug. An un-
equivocal identification of dialed-in dosing events is possible
solely based on capacity data, as demonstrated in Figure 2. In the
graph presented, a second order polynomial was fitted to the
290
capacitance data. The red boxes indicate the confidence interval
for C measurements with each dosing step (as if the pen was OBSERVATIONAL STUDY OF SAROGLITAZAR ON
emptied at 1 U steps). METABOLIC PARAMETERS IN INDIAN PATIENTS
The measured C values allow to accurately detect each real WITH DIABETIC DYSLIPIDEMIA: 43 WEEKS
dosing event, except 1 point at the full and 2 points at the empty OF CLINICAL EXPERIENCE
end where fringe effects of the field are not yet addressed in the
current prototype. S. Chatterjee1, A. Majumder2, S. Ray1
This is including and not corrected for factual inaccuracies of 1
Apollo Gleneagles Hospital, Diabetes and Endocrinology,
the mechanic-hydraulic delivery system and errors in the ca- Kolkata, India
pacity measurement. 2
KPC Medical College, Endocrinology, Kolkata, India
Conclusions: Capacitance changes can be recorded with good
linearity and precision down to less than 1 U. Background and Aims: Control of both glycemic and lipid
parameters are important for management of diabetic dyslipi-
demia (DD). Saroglitazar, the only dual peroxisome proliferator-
activated receptor (a/c) agonist available for clinical use, is
289 marketed in India for the treatment of DD since September 2013.
FIRST ITALIAN EXPERIENCE WITH CONTINUOUS Objective of this study was to evaluate the efficacy and safety of
INTRAPERITONEAL INSULIN INFUSION (CIPII) saroglitazar in Indian DD patients and also to compare our results
VIA DIAPORT SYSTEM with our previous findings.

I. Carta1, M.S. Ledda1, A. Carboni1, G. Giorgi1,

A. Scorsone2, D. Brancato2, S. Cherchi1, V. Provenzano2,
G. Tonolo1
1
SC Diabetologia aziendale, San Giovanni di Dio, Olbia, Italy
2
Struttura complessa diabetologia, Partinico Hospital,
Partinico, Italy

Background and Aims: CIPII is an alternative route of in-

sulin delivery when injection site issues make difficult subcuta-
neous insulin administration. In order to bypass the subcutaneous
tissue, the second generation Diaport system (DIA), allowing
intraperitoneal insulin infusion by an external insulin pump
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-121

Methods: This study is a retrospective analysis of 75 patients

with type 2 diabetes mellitus and dyslipidemia, who were re-
ceiving saroglitazar 4 mg once daily as per the approved indi-
cation and prescribing information of saroglitazar.
Results: Baseline demographics were: age 52.77 9.59
years (mean SD), body weight 68.24 10.50 kg; 64% par-
ticipants were males. At baseline, all patients were on anti-
hyperglycemic medication and 48 (64%) patients were on
statin therapy. Mean duration of follow-up was 42.36 weeks.
Compared to baseline values, there were statistically signifi-
cant changes observed in lipid and glycemic parameters
(table). Liver enzyme, alanine aminotransferase (ALT) was
significantly reduced from 42.61 24.03 U/L to 30.93 18.9 U/L
(mean SD; p < 0.0001). No serious adverse event was reported
during entire study follow-up. Adverse events reported were
mild and transient. No significant change was observed in se-
rum creatinine level. These results are consistent with our
previous reports with lesser number of patients and smaller
duration of study.
Conclusions: Our results showed that saroglitazar is effective
and safe for the treatment of type 2 diabetes with dyslipidemia.

291
EFFECTS ON METABOLIC PARAMETERS OF
ADDITION OF SGLT-2 INHIBITORS ON PATIENTS
WITH TYPE 2 DIABETES INADEQUATELY
CONTROLLED WITH DPP-IV INHIBITORS AND
METFORMIN.
S. Chatterjee1, S. Ray1, A. Mazumdar2,
K. Bhattacharjee3 Results: Results showed that addition of SGLT-2 inhibitors
produced favorable effects on all metabolic parameters studied
1
Apollo Gleneagles Hospital, Diabetology & Endocrinology, (Table 1 to Table 3).
Kolkata, India Conclusions: Our study shows that addition of SGLT-2 in-
2
K.P.C Medical College, Endocrinology Unit- Department of hibitors on existing therapy with DPP-IV inhibitors and metformin
Medicine, Kolkata, India produces favorable effects on metabolic parameters with the ad-
3
Assistant Manager, Biocon Ltd, Medical Services, Bangalore, vantage of weight loss and without producing major hypoglycemic
India events.

Background and Aims: Practice guidelines are open re-

garding choice of therapy after metformin. Insulin(INS), sul-
phonylureas(SU) and thiazolidinediones(TZD) may cause
hypoglycemia or weight gain. Dipeptidyl peptidase-IV (DPP-IV)
inhibitors are unlikely to produce that. Sodium-glucose
transporter-2 (SGLT-2) inhibitors are newer agents with the
advantage of weight loss. We have used SGLT-2 inhibitors
canagliflozin or, dapagliflozin, or, empagliflozin on T2D patients
inadequately controlled (HbA1c 7%) with DPP-IV inhibitors
plus Metformin and analyzed the results on glycemic and other
metabolic parameters. Since Indian data of such therapy is
scanty, this study is relevant because of the number of patients
studied, duration of study, study variables and effects of three
SGLT-2 inhibitors analyzed separately.
Methods: Data of 101 T2D patients inadequately controlled
(HbA1c 7%) with DPP-IV inhibitors plus Metformin who were
prescribed canagliflozin 100 mg or, dapagliflozin 5 or, 10 mg, or
empagliflozin 10 mg or, 25 mg once daily with mean follow-up
duration of 23 weeks were analyzed. Subjects receiving INS, SU
and TZD were excluded from analysis. Changes in weight, blood
pressure, glycemia, lipids, renal and hepatic parameters were
studied. Subgroup analyses were done to see effects of three
SGLT-2 inhibitors separately.
A-122 ATTD 2017 E-POSTER VIEWING ABSTRACTS

The mean plasma glucagon AUC observed during the control

study was not suppressed after treatment with liraglutide (control:
292 2352 997 vs. lira: 2705 1473 pg/ml*min, p = ns). Despite the
short duration of treatment, there was a reduction in both body
LACK OF GLUCAGON EFFECT OF GLP-1 RECEPTOR weight (4.8 2.4%, p = 0.001) and insulin dose (*25%, p = 0.001)
AGONISTS: ASSESSMENT DURING A MIXED MEAL with liraglutide.
TOLERANCE TEST Conclusions: Unlike the results of earlier acute infusion stud-
A. Galderisi1,2, J. Sherr1, N. Patel1, M. Van Name1, E. Cengiz1, ies, dysregulated plasma glucagon responses to mixed meal feed-
L. Carria1, K. Weyman1, M. Zgorski1, S. Weinzimer1, ings in T1D are not suppressed after 3-4-wks of treatment with
W. Tamborlane1 liraglutide. This suggests that the benefits of liraglutide as an ad-
1 junctive treatment for T1D are mediated by its effects on satiety,
Yale University, Pediatric Endocrinology, New Haven, USA leading to weight loss and reductions in insulin requirements.
2
University of Padova, Department of Womans and Childs
Health, Padova, Italy

Background and Aims: Postprandial hyperglycemia remains 293

challenging in type 1 diabetes (T1D) due, in part, to dysregulated
glucagon secretion. While acute GLP-1 receptor agonist infusion NASAL GLUCAGON (NG) FOR THE TREATMENT
studies suggested suppressed glucagon after mixed-meal feed- OF MODERATE TO SEVERE HYPOGLYCEMIA
ing in T1D, more recent GLP1 treatment studies demonstrate (HYPO) EPISODES IN CHILDREN AND
conflicting results. In this study, mixed-meal tolerance tests ADOLESCENTS WITH TYPE 1 DIABETES (T1D)
(MMTT) explored how the GLP1 receptor agonist, liraglutide, L. Deeb1, H. Dulude2, X.M. Zhang3, S. Zhang4, B. Reiner5,
effects meal-stimulated excursions in plasma glucose (PG) and C. Piche2, C.B. Guzman6
glucagon in patients with T1D.
1
Methods: Eleven participants (aged23 3 yrs, A1C7.6 0.9%) Florida State University College of Medicine, Department of
underwent two MMTTs, before and after treatment with liraglu- Pediatrics, Tallahassee, USA
2
tide. During the 3-4-wk outpatient period, liraglutide doses were Locemia Solutions, Clinical Development, Montreal, Canada
3
titrated to 1.8mg/day and insulin adjusted, as needed. The MMTTs Eli Lilly and Company, Medical Development- Diabetes
utilized Boost High Protein Energy Drink. PG and glucagon were Business Unit, Toronto, Canada
4
measured for 5-hrs. Normal fasting PG levels were achieved by Eli Lilly and Company, Statistics, Indianapolis, USA
5
overnight closed-loop insulin delivery. Pediatric Endocrine Office, Private Practice, Baltimore, USA
6
Results: Plasma glucose and glucagon were nearly identical in Eli Lilly and Company, Medical Development, Diabetes
both studies at baseline and following Boost ingestion (Figure). Business Unit, Indianapolis, USA
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-123

6
Background and Aims: This multi-center, open-label study Locemia Solutions, Pharmaceutical Product Development,
evaluated the effectiveness and ease of use of NG in treating Montreal, Canada
7
moderate or severe hypo episodes in patients (pts) with T1D, JSS Medical Research, Scientific Affairs, Montreal, Canada
8
aged 4 to <18 yrs. Eli Lilly and Company, Medical Development- Diabetes
Methods: Pts and caregivers (CGs) were taught how to use Business Unit, Toronto, Canada
9
NG. During naturally occurring symptomatic episodes of mod- Eli Lilly and Company, Medical Development- Diabetes
erate or severe hypo in real world settings, CGs administered Business Unit, Indianapolis, USA
3 mg NG and measured blood glucose (BG) levels over time.
Adverse events (AEs), recovery of symptoms, and ease of use Background and Aims: Nasal glucagon (NG), a needle-free
were solicited by questionnaires. glucagon that is absorbed through the nasal mucosa, is being
Results: Fourteen pts, who experienced 33 moderate hypo ep- developed to treat severe hypoglycemia. This single center, open-
isodes with neuroglycopenic symptoms and a BG level 70 mg/dL, label study evaluated the safety, pharmacokinetics (PK) and
were included in the efficacy and main safety analyses. Mean pharmacodynamics (PD) of NG in 36 otherwise healthy subjects
number of episodes per pt was 2.4 (range 1 to 4). In all episodes, pts with nasal congestion from common cold with or without con-
returned to normal status within 30 minutes after NG dose. No calls comitant nasal decongestant.
to 911 (emergency medical services) were needed. Mean baseline Methods: Cohort 1 (N = 18) received 2 NG doses: a 3 mg dose
BG level was 56 (range 42-70) mg/dL. Within 15 minutes after NG while suffering from common cold and a second 3 mg dose after
dose, mean BG level increased to 114 (range 79-173) mg/dL, and return to normal health. Cohort 2 (N = 18) received a single 3 mg
continued to increase (Figure). No serious AEs occurred. For most dose of NG 2 hours after receiving the nasal decongestant oxy-
episodes (61%), CGs administered NG in <30 seconds; in all cases, metazoline while suffering from common cold.
administration took <2 minutes. CGs were either satisfied or very Results: There were no serious adverse events (AE); the most
satisfied with NG after most episodes (91%). common AEs were transient lacrimation, nasal discomfort, rhi-
Conclusions: NG increased BG and resolved symptoms in all norrhea and nausea with reduced nasal symptoms and nausea in
reported episodes of hypo among children and adolescents with participants with normal health versus those with cold symp-
T1D. The majority of CGs were highly satisfied with NG. Data toms, with or without decongestant. Glucagon and glucose levels
suggest that NG is a viable alternative to currently available increased rapidly after treatment to peak glucagon levels at 20
injectable recombinant glucagons. minutes post dose and peak glucose levels at 30 to 40 minutes
post dose for all groups (Figure).
Conclusions: Nasal congestion, with or without concomitant
use of a nasal decongestant, did not significantly affect the PK
and PD of NG although transient AEs were more frequent in
participants with common cold than in healthy participants.
These data suggest that NG can be used to treat severe hypo-
glycemia in patients with nasal congestion.

294
EFFECTS OF NASAL CONGESTION FROM
COMMON COLD ON THE PHARMACOKINETICS
(PK) AND PHARMACODYNAMICS (PD)
OF NASAL GLUCAGON
H. Dulude1, E. Sicard2, M. Rufiange3, A. Sadoune4,
D. Carballo5, M. Triest6, M. Tafreshi7, E. Rampakakis7,
C. Piche1, X.M. Zhang8, C.B. Guzman9
1
Locemia Solutions, Clinical Development, Montreal, Canada
2
Altasciences/Algorithme Pharma, Principal Investigator,
Montreal, Canada
3
Altasciences/Algorithme Pharma, Scientific and Regulatory
Affairs, Montreal, Canada
4
Altasciences/Algorithme Pharma, Clinic Department,
Montreal, Canada
5
Locemia Solutions, Operations, Montreal, Canada
A-124 ATTD 2017 E-POSTER VIEWING ABSTRACTS

295 for non-medical people and use of a different route of ad-

ministration for glucagon may reduce the risk of accidental
NEEDLE-FREE NASAL DELIVERY
insulin injection.
OF GLUCAGON IS SUPERIOR TO INJECTABLE
DELIVERY IN SIMULATED HYPOGLYCEMIA
RESCUE
J.F. Yale1, M. Egeth2, C. Piche3, M. Lafontaine4, R. Margolies2, 296
E. Dissinger2, D. Carballo5, A. Shames2, N. Kaplowitz2,
NRF2 ACTIVATOR PROTECTS CORONARY BK
C.B. Guzman6, H. Dulude3
CHANNEL FUNCTION AND CORONARY
1
McGill University, Department of Medicine, Montreal, VASODILATION IN HIGH FAT DIET-INDUCED
Canada DIABETIC MICE
2
CORE Human Factors, Research, Bala Cynwyd, USA
3 T. Lu1, X.L. Wang1, X. Sun1, H.C. Lee1
Locemia Solutions, Clinical Development, Montreal, Canada
4 1
Locemia Solutions, Commercial, Montreal, Canada Mayo Clinic, Cardiovascular Diseases, Rochester- MN, USA
5
Locemia Solutions, Operations, Montreal, Canada
6
Eli Lilly and Company, Diabetes Business Unit. Development, Background and Aims: Vascular large conductance Ca2+-
Indianapolis, USA activated K+ (BK) channels, composed of four pore-forming
subunits (BK-a) and four regulatory subunits (BK-b1), are the
Background and Aims: This study evaluated the ease of use key determinant of coronary vascular tone. However, BK
of needle-free nasal glucagon (NG) devices versus injectable channel function is impaired in diabetic vessels. The nuclear
glucagon (Glucagon Emergency Kit GEK) in treating a sim- factor erythroid-2-related factor 2 (Nrf2) signaling is cytopro-
ulated severe hypoglycemia (SH) episode. tective against oxidative stress, but little is known about the role
Methods: The caregiver arm included 16 adult persons with of Nrf2 on regulating BK channel function.
diabetes (PWD) and their caregivers. The PWDs were taught Methods: Combining patch clamp and cellular biological
how to use one of the devices and they, in turn, instructed techniques, we examined the protective effects of Nrf2 on cor-
their caregivers. The acquaintance arm had 15 adults who onary BK channel function in diabetes.
were not trained on device use. Acquaintances were shown the Results: We found that BK currents were diminished in the
device before the simulation. During the simulation, each coronary smooth muscle cells (SMCs) of mice 6 months after
caregiver and acquaintance treated a manikin representing a feeding with high 60% fat-diet (HFD), compared to mice feeding
PWD with SH using both GEK and NG devices at different with 10% low fat-diet (LFD), which was accompanied by a re-
time-points. markable reduction of Nrf2, BK-a and BK-b1 protein expres-
Results: Fifteen of 16 caregivers and 14 of 15 acquaintances sions. Adenoviral delivery of Nrf2 gene (50 MOI, 48 h) produced
administered a full dose of NG (average time 0.27 and 0.44 min, a 2.9-fold and 3.4-fold upregulation of BK-a and BK-b1 protein
respectively). expressions in coronary SMCs respectively. The regulation of
Eight of 16 caregivers injected glucagon (average time BK-a protein expression by Nrf2 was attributed to increased BK-
1.89 min); of which 2 gave the full dose. Failures included in- a mRNA level, while that of BK-b1 expression by Nrf2 was
jection of diluent only (n = 4), insulin injection (n = 3), and bent associated with inhibited ubiquitin-dependent BK-b1 prote-
needle (n = 1). olysis. Oral administration of dimethyl fumarate (a FDA-
Three of 15 acquaintances injected glucagon (average time approved Nrf2 activator, 25 mg/kg/day, 10 days) enhanced Nrf2,
2.4 min); none gave the full dose. Failures included injection of BK-a and BK-b1 protein expressions, and preserved coronary
diluent only (n = 9), insulin injection (n = 1), refusal to inject BK channel function and coronary vasodilation in HFD mice.
(n = 1) and injection with empty syringe (n = 1). Conclusions: Our results indicate that Nrf2 signaling plays a
Conclusions: During simulated SH, caregivers and ac- critical role in coronary BK channelopathy and vasculopathy in
quaintances used GEK with low success rates and numerous HFD mice. Activation of Nrf2, therefore, may have a therapeutic
errors; however, they administered NG successfully and in potential for diabetic cardiovascular complications.
less time. This suggests that NG delivery is easier and faster

297
SAFETY OF SGLT2 INHIBITOR TREAMENT
IN TYPE 1 DIABETES: CHANGE IN THE PATTERN
OF EARLY METABOLIC DECOMPENSATION
FOLLOWING BASAL INSULIN SUSPENSION
N. Patel1, M. Van Name1, L. Carria1, E. Cengiz1,
S. Weinzimer1, W. Tamborlane1, J. Sherr1
1
Yale School of Medicine, Pediatric Endocrinology, New
Haven, USA

Background and Aims: Enthusiasm for the benefits of

SGLT2 inhibitors (SGLT2i) as adjunctive treatments in type 1
diabetes (T1D) has been offset by the possible increased risk of
diabetic ketoacidosis (DKA). Since pump-treated T1D patients
ATTD 2017 E-POSTER VIEWING ABSTRACTS A-125

are susceptible to DKA due to infusion site problems, this study add-on to insulin. The effects of CANA on daily mean glucose,
was undertaken to assess how treatment with SGLT2i affects glycemic variability, and time spent in glycemic ranges were
patterns of early metabolic decompensation following suspen- assessed.
sion of basal insulin. Methods: At baseline (week prior to randomization) and at
Methods: 10 participants with T1D [age 19-35y, duration the end of the study (Week 17-18), all patients (N = 351) were to
10 8y, A1c 7.4 0.8%) underwent an overnight pump sus- record 9-point self-monitoring blood glucose (SMBG) mea-
pension study before and after *3wks of SGLT2i treatment. surements; a subset of 89 patients also underwent continuous
On both study nights, basal insulin was suspended at 3AM and glucose monitoring (CGM).
plasma glucose (PG) and b-hydroxybutyrate (BHB) were Results: At Week 18, reductions in daily mean glucose and
measured every 30 minutes. Studies were terminated 6 hours standard deviation measured by 9-point SMBG were seen with
after suspension, or sooner if PG rose >350mg/dL or BHB CANA 100 and 300 mg vs placebo (PBO; Table). Consistent
>2.5mmol/L. with this, reductions in mean glucose and measures of gly-
Results: At the start of suspension, PG was similar in the two cemic variability assessed by CGM were seen with CANA 100
studies (Figure). Rate of rise in PG was markedly blunted by and 300 mg vs PBO (Table). Both CANA doses were asso-
SGLT2i treatment, with PG at end of suspension in control ciated with increases in the time patients spent within target
301 21mg/dL vs. SGLT2i 195 14mg/dL, (p = 0.002). In con- (glucose >70 to 180 mg/dL) and commensurate reductions in
trast, rise in BHB and BHB at end of suspension was similar time spent above target (glucose >180 mg/dL) vs PBO at the
(control 1.4 0.2mmol/L vs SGLT2i 1.8 0.3mmol/L, p = 0.06). end of the study. There were no meaningful changes in the
Differences in BHB were not considered clinically significant. time patients spent below target (glucose 70 mg/dL) across
Conclusions: In insulin-pump treated T1D patients receiving groups. CANA was generally well tolerated, with an increase
SGLT2i, difficulty recognizing early DKA following interrup- in ketone-related AEs.
tion of basal insulin is primarily due to the more gradual increase Conclusions: In summary, CANA improved indices of gly-
in PG rather than to more rapid increases in BHB. T1D patients cemic variability and increased the time spent in target without
using SGLT2i should test for ketones based on symptoms of increasing time spent below target vs PBO in T1DM patients.
illness rather than relying on persistent elevations of glucose to
identify impending DKA.
299
ENDOSCOPIC INSERTION OF A SOFT
298 POLYURETHANE TUBE INTO THE JEJUNUM
CANAGLIFLOZIN (CANA) IMPROVES GLYCEMIC TO DISSOCIATE MEAL FROM MUCOSA
CONTROL AND REDUCES GLYCEMIC VARIABILITY AS A SUBSTITUTE FOR BARIATRIC SURGERY
IN PATIENTS WITH TYPE 1 DIABETES MELLITUS K. Khodabakhshi Pirkalani1, Z. Talaeerad1
(T1DM) INADEQUATELY CONTROLLED 1
WITH INSULIN Mehr Medical Group, Internal Medicine/Gynecology, Tehran,
1 2 3 3
Iran
H.W. Rodbard , A.L. Peters , A. Cao , M. Alba
1
Endocrine and Metabolic Consultants, Rockville, MD, USA Background and Aims: To reduce side effects and cost of
2
Keck School of Medicine of the University of Southern bariatric surgery which is a complex, irreversible and hazardous
California, Los Angeles CA, USA procedure as many of the mechanism are still ill define we have
3
Janssen Research & Development LLC, Raritan, NJ, USA developed the following.
Methods: Based on the physiological function of the small
Background and Aims: CANA, a sodium glucose co- intestine we have designed a very soft polyurethane tube that
transporter 2 (SGLT2) inhibitor, provided glycemic improve- can be inserted via endoscopy as a small 3cmX3cmX3cm clump
ments in a randomized Phase 2 study in adults with T1DM as into the upper GI and inflated there so that it can cover about
90% of the jejunum mucosa and possibly 30% of the ileum. This
reduces the absorptive surface of the small intestine and causes
reduced absorption of the major nutrients just like a bariatric
surgery but without its drawbacks, side effects and irrevers-
ibility character. The 25mm diameter tube is double layer and
has a wall thickness of 0.3mm, completely impermeable to
water and is inflated via a second 2mm Archimedes Arc tube
filled with water during the procedure. The upper end of the
tube is fixed in the stomach via a larger ring in continuity with
the Archimedes Arc. The real beginning of the tube is at the
second part of the duodenum where bile and pancreatic juice are
drained into the tube as well.
Results: The design is completely successful and though
simple extremely effective. Preliminary results in dogs have
shown 88% reduction of different nutrient absorption.
Conclusions: To our knowledge this is the most effective
and simple substitute for bariatric surgery both in diabetic and
obesity setting with a cost of around 10% of it. Human trial is
underway.
A-126 ATTD 2017 E-POSTER VIEWING ABSTRACTS

300 Methods: In this 4-week feasibility study, volunteering adult

(n = 26) overweight (BMI >25 kg/m2) men and women were re-
LIPID STATUS IN PATIENTS WITH DIABETES
cruited from a diabetes outpatient clinic and from an occupational
MELLITUS TYPE 1 FROM VARIOUS ETHNIC GROUPS
health care unit. Participants used a smartphone application to
L.I. Kolesnikova1, S.I. Kolesnikov1, M.A. Darenskaya1, keep a visual food journal, share their meals and activity with peer
L.A. Grebenkina1, S.V. Gnusina1 group members and receive virtual coaching.
1 Outcomes were assessed via surveys both before and after the
Scientific Centre for Family Health and Human Reproduction intervention using a food frequency questionnaire (FFQ), group
Problems, Department of Pathophysiology, Irkutsk, Russia environment and social support scales. Height, weight and waist
circumference were measured. Also frequency of the application
Background and Aims: Diabetes mellitus type 1 (T1DM) is use was analysed.
the most severe form of Diabetes mellitus with clear clinical Results: After the intervention the daily use vegetables and
presentation and metabolic disturbances. Some studies reported fruits had increased by 2.15 (SD 2.64) portions (p < 0.001) com-
low T1DM morbidity rate among the aboriginal people in Arctic pared to baseline. Mean weight loss after the intervention was
and Siberian peoples resulted from the presence of the protective 1.5 kg (SD 1.82, p = 0.001). Mean reduction in waist circumference
alleles for this nosology. The aim of this study was to compare was 2.4 cm (SD 2.418, p = 0.001). 84% of participants agreed that
lipid status in Mongoloids and Caucasians with T1DM. they got support from other group members. The use of application
Methods: The study recruited the patients with T1DM: Mon- was high: average user uploaded 5.2 (SD 2.24) meals per day and
goloids (n = 38, mean age 34.4 1.9 years, disease duration 12.0 used the application total of 9.3 (SD 4.41) times per day.
1.5 years) and Caucasians (n = 27, mean age 32.7 2.3 years, Conclusions: Smartphone based virtual peer support can be
disease duration 12.9 2.0 years), which were the residents of used as a tool to promote healthy eating both in outpatient clinic
Siberian. The control group was comprised of practically healthy and occupational health settings.
persons that had no congenital diabetic history: Mongoloids
(n = 34, mean age 30.3 1.3 years) and Caucasians (n = 23, mean
age 27.1 1.9 years).
Results: Thus, T1DM Mongoloid patients had significantly
302
lower total lipids value in comparison with Caucasian patients. In
addition, T1DM Mongoloid patients were characterized with CONTINUOUS GLUCOSE MEASUREMENT,
significantly lower levels of triglycerides, total cholesterol and A TECHNICAL TOOL TO FACILITATE
low-density lipoproteins in comparison with Caucasian patients. OVERWEIGHT INDIVIDUALS TO ADHERE
It should be stressed that despite the absence of significant dif- TO A LOW CARB DIET PROGRAM
ferences in atherogenic coefficient between the patients in two
ethnic groups, atherogenic coefficient in Caucasian patients H.F. v. Lilienfeld-Toal1, D. Tews2
surpassed the normal value. Elevation of this parameter is an 1
Elte Sensoric GmbH, Gelnhausen, Germany
important prognostic factor for the course of diabetes, because 2
Diabetes Zentrum Gelnhausen, Outpatient Clinic,
dislipidemia promotes the development of morphological alter- Gelnhausen, Germany
ations in the basal vascular membrane and the hemodynamic
disturbances leading to generalized angio - and neuropathies. Background and Aims: It is difficult to keep motivation up
Conclusions: We can conclude a more favorable course of during attempts to reduce weight. After it became clear that dietary
the lipid metabolic processes in Mongoloids in comparison reduction of carbs - especially with a high glycemic index - is an
with Caucasians. essential part of a weight reduction project by avoiding insulin
peaks the measurement of glucose offers a technological tool for
instant monitoring of the correctness of the diet. Theoretically, this
301 enables the individual to collect experience with his favorite food
and allows for correction so that for him it becomes clear how his
VISUAL FOOD DIARY FOR SOCIAL SUPPORT, personal diet can be composed without losing palatability.
DIETARY CHANGES AND WEIGHT LOSS Methods: To test this possibility we have studied normal di-
H. Salmenius-Suominen1, M. Lehtovirta2, H. Vepsalainen1, etary situations while monitoring the subcutaneous glucose
H. Konttinen3, M. Erkkola1 concentration (FreeStyle Libre, Abbott). These situations in-
cluded comparison meals with different amounts of carb content
1
University of Helsinki, Department of Food and Environmental and glycemic index, fruits, nuts and the effects of different drinks
Sciences- Division of Nutrition, Helsinki, Finland (orange juice, apple juice, cola, beer, wine).
2
University of Helsinki, FiMM- Institute for Molecular Results: Parts of the diet induced high glucose peaks. Espe-
Medicine Finland, Helsinki, Finland cially in fruits with high glucose content, the effects of high in-
3
University of Helsinki, Department of Social Research, crease and fast decrease of tissue glucose combined with sudden
Helsinki, Finland attacks of hunger were impressive. The combination with fiber
rich parts of the meal showed that a total restriction of sweet
Background and Aims: Behaviour changes in diet may components in the diet was not necessary. It was possible to create
prevent or delay onset of chronic diseases. Social support is a a dietary composition which was well tolerated by the individual
known predictor of successful behaviour change. and resulted in glucose values constantly below 140 mg/dl.
To find out whether virtual peer support groups moderated by Conclusions: Through this approach the individual learns to
a nutrition professional using a mobile application can be used to take the right measures for palatable diet. This will probably lead
promote healthy eating. to better results in the project of weight reduction.
ATTD 2017 Read By Title

303 COMMERCIALIZING THE FIRST HYBRID CLOSED- S.C.D.U. Endocrinologia- Diabetologia e Metabolismo,
LOOP ARTIFICIAL PANCREAS SYSTEM Torino, Italy
L. Lintereur1, N. Kurtz2, A. Roy2, B. Grosman2, N. Parikh2, 2
Center for Diabetes Technology, Dept. of Psychiatry and
G. Voskanyan2 Neurobehavioral Science- University of Virginia, Charlottes-
1
Medtronic Diabetes, Systems Engineering, Northridge, USA ville, USA
2 3
Medtronic Diabetes, Advanced Research, Northridge, USA A.O.U. Citta della Salute e della Scienza di Torino, S.C.U.
Psicologia Clinica e Oncologica, Torino, Italy

304 SOME SUCCESSFUL EXPERIENCES IN MAGGOT

THERAPY IN BABOL, IRAN 308 THE EFFECT OF CSII ON GLYCEMIC
E. Shojaeizadeh1, B.D. Rahman2, S. Ali Asghar2, A. Kamran3, CONTROL INT2DM: A RETROSPECTIVE STUDY
N. Mahbobeh3, S. Davoud3 IN 60 PATIENTS WITH T2DM WITH A FOLLOW
1
School of Public Health, Tehran, Iran UP OF 24 MONTHS
2
Faculty of Medicine, University of Medical Sciences-Babol, R.K. Gonera1
Babol, Iran 1
3
, Rolde, The Netherlands
Tehran University of Medical Sciences, School of Public
Health, Tehran, Iran

309 REDUCED FORM OF GLUTATHIONE PROTECT

B-CELLS FROM DESTRUCTION CAUSED
305 THE INFLUENCE OF HEALTH BELIEF MODEL BY XANTHURENIC ACID
COMPONENT ON SELF EFFICACY AMONG G. Meyramov1, A. Shaybek1, E. Laryushina2, G. Kartbaeva1,
PATIENTS WITH TYPE 2 DIABETES O.N. Dupont3, A. Alina2, A. Abdraimova-Meyramova4,
D. Shojaeizadeh1, S. Esmaeil2, S. Maedeh3, G. Nahid4, A. Embergenova1, Z. Zhumagalieva1, A. Tulieva5
S. Motahareh5 1
Buketov Karaganda State University, Biology, Karaganda,
1
Tehran University of Medical Sciences, Tehran, Iran Kazakhstan
2
School of Public Health- Tehran University of Medical Sciences, 2
Karaganda State Medical University, Internal Diseases N2,
Department of Health Education & Promotion, Tehran, Iran Karaganda, Kazakhstan
3
Shahid Beheshti university of Medical Sciences, community 3
Hymnasium N3, Biology, Karaganda, Kazakhstan
health nursing education- Treatment Affairs, Tehran, Iran 4
Karaganda State Medical University, Physiology, Karaganda,
4
School of Public Health- Tehran University of Medical Sci- Kazakhstan
ences, Public Health, Tehran, Iran 5
Astana Medical University AG, Histology and Cytology,
5
School of Public Health, Tehran, Iran Astana, Kazakhstan

306 DIABETES MELLITUS IN STRUCTURE

OF PATIENTS WITH TERMINAL CHRONIC 310 FEAR OF SELF-INJECTING AND SELF-TESTING
RENAL FAILURE RECEIVING HEMODIALYSIS AND ITS ASSOCIATED RISK FACTORS AMONG
IN THE SAKHA REPUBLIC (YAKUTIA) ADOLESCENTS WITH TYPE 1 DIABETES : A CROSS
M. Petrova1, E. Unarova2 SECTIONAL STUDY
1
North-Eastern Federal University, Institute of Medicine, K. Almonaei1, A. Al Hayek1, A. Asirvatham1
1
Yakutsk, Russia PSMMC, Endocrinology and Diabetes, RIYADH,
2 Saudi Arabia
State Autonomous Institution of the Republic of Sakha Yakutia
Republican Hospital No 1 - National Medical Center, De-
partment of chronic hemodialysis n.a. Pavlov E.G., Yakutsk,
Russia
311 COMPARING AUTOMATED SEGMENTATION
FOR AUTOMATED CHC IN SELF-MANAGEMENT
307 NET GLYCEMIC VARIABILITY AND COPING OF DIABETES
STRATEGIES: A POSSIBLE RELATIONSHIP C. Baptista1, I. Cruz2, D. Martins1, P. Martins2, J. Cecilio2,
A. Bisio1, E. Campos-Nanez2, E. Gamarra1, P. Leombruni3, F. Carrilho1, P. Furtado2
G. Grassi1 1
CHUC, Endocrinology and Diabetes, Coimbra, Portugal
1 2
A.O.U. Citta della Salute e della Scienza di Torino, Universidade de Coimbra, DEI/CISUC, Coimbra, Portugal

A-127
A-128 ATTD 2017 READ BY TITLE

312 DENSITY SEGMENTATION IN DETECTION 317 PERSONAL VS PROFESSIONAL CONTINUOUS

OF DIABETIC RETINOPATHY GLUCOSE MONITORING IN TYPE 1 DIABETES: CAN
C. Baptista1, C. Travassos2, R. Monteiro2, S. Oliveira2, WE MAKE A DIFFERENCE?
D. Martins1, F. Carrilho1, P. Furtado2 G. Petrovski1, M. Zivkovic1
1 1
Centro Hospitalar e Universitario de Coimbra, University Clinic of Endocrinology, Center for Insulin Pump,
Endocrinology- Diabetes and Metabolism, Coimbra, Skopje, FYR Macedonia
Portugal
2
Universidade de Coimbra, DEI/CISUC, Coimbra, 318 CONTINUOUS SUBCUTANEOUS INSULIN
Portugal INFUSION (CSII) THERAPY: A COST EFFECTIVE
AND EFFICIENT ALTERNATIVE TO INTRAVENOUS
INSULIN INFUSION (IVII) THERAPY IN PATIENTS
REQUIRING HOSPITALIZATION OR ICU STAY
313 ARACHIDONIC ACID (AA) AS A POTENTIAL K. Singh1
ANTI-DIABETIC MOLECULE: AN IN VITRO 1
Fortis Hospital Mohali, Endocrinology, Chandigarh,
AND IN VIVO STUDY India
D. Undurti1
1
UND Life Sciences, R & D, Federal Way, USA
319 CONTINUOUS SUBCUTANEOUS INSULIN
INFUSION(CSII) PATCH PUMP IMPROVES QUALITY
OF LIFE (QOL) IN TYPE 1 PATIENTS USING FOOD
LIBRARY AND ACTIVITY TO CALCULATE
314 ACACIA NILOTICA (L.) DELILE COULD BE
INSULIN CARBOHYDRATE RATIO
A POTENTIAL DRUG COMBATING DIABETES:
P. Kelly1, S. Read1, P. English1
AN EVIDENCE-BASED AND IN-SILICO APPROACH 1
Plymouth Hospitals NHS Trust, Diabetes Centre, Plymouth,
M.R. Saha1, P. Dey2, I. Sarkar1, P. Kar1, D.D. Sarker3, S. Das4,
United Kingdom
B. Haldar5, T.K. Chaudhuri2, A. Sen1
1
University of North Bengal, Molecular Cytogenetics
Laboratory, Department of Botany, Siliguri, India 320 GLP-1 AGONIST IN COMBINATION WITH
2 INSULIN PUMP IN TYPE 2 DIABETES PATIENT
University of North Bengal, Cellular Immunology Laboratory,
Department of Zoology, Siliguri, India CAN IMPROVE GLUCOSE CONTROL:
3 A CASE REPORT
Raiganj University, Department of Botany, Raiganj, India
4
University of North Bengal, Department of Chemistry, Siliguri, M. Zivkovic1, G. Petrovski1, B. Todorova1
1
India University Clinic of Endocrinology, National Center
5
North Bengal Medical College, Department of Pathology, for Insulin Pump, Skopje, FYR Macedonia
Siliguri, India
321 CAN WIRELESS CARDIOVASCULAR FAST
TRACKING MONITORING SYSTEMS BE PREDICTIVE
315 ADDITION OF FASTER-ACTING INSULIN ASPART OF INPATIENT INTENSITY OF CARE IN DIABETES
TO BASAL INSULIN SIGNIFICANTLY IMPROVES SUBJECTS?
GLYCAEMIC CONTROL IN ADULTS WITH TYPE 2 A. Scorsone1, D. Brancato1, F. Gargano2, M. Fleres1,
DIABETES: THE ONSET 3 TRIAL L. Spano1, G. Picone2, V. Provenzano1
H. Rodbard1, D. Tripathy2,3, M. Vidrio-Velazquez4, 1
ASP 6 Palermo- Partinico Civic Hospital, Referral Centre for
M. Demissie5, S.C. Tamer5, M. Piletic6 Diabetes and Insulin Pumps Implantation, Palermo, Italy
1 2
Research Center, Endocrine and Metabolic Consultants, ASP 6 Palermo- Partinico Civic Hospital, Medicine Unit,
Rockville, USA Palermo, Italy
2
Texas Diabetes Institute, University of Texas Health Science
Center, San Antonio, USA
3 322 THE EFFECT OF THE BOLUS CALCULATOR USE
Audie L Murphy VA Hospital, South Texas Veterans Health
ON THE GLYCEMIC CONTROL AT THE INSULIN
Care System, San Antonio, USA
4 PUMP THERAPY
Endocrinology- Metabolism and Nutrition, Hospital General
B. Caric1, B. Vukovic1
Regional 110, Guadalajara, Mexico 1
5 University Clinical Center of Republic of Srpska, Department
Novo Nordisk A/S, Sborg, Denmark
6 for Endocrinology, Clinic for Internal Diseases, Banja Luka,
Nutritional Biochemistry, Nutrition and Dietetics, Diabetol-
Bosnia-Herzegovina
ogy, General Hospital, Novo Mesto, Slovenia

323 CLASSIFICATION, MODELING AND PREDICTION

OF GLYCEMIA IN DIABETES PATIENTS BY
316 EFFICACY AND SAFETY OF MINIMED 640G EVOLUTIONARY AND HEURISTIC COMPUTATION
INTEGRATED SENSOR-AUGMENTED PUMP FROM CONTINUOUS GLUCOSE MONITORING,
THERAPY (SAPT) SYSTEM WITH SMARTGUARD CARBOHYDRATES AND INSULIN RECORDS
IN PATIENTS WITH TYPE 1 DIABETES PREVIOUSLY J.M. Colmenar1, S.M. Winkler2, G. Kronberger2, E. Maqueda3,
TREATED WITH PARADIGM VEO M. Botella-Serrano4, A. Sanchez5, J.M. Velasco5, S. Contador5,
A.M. Gomez Medina1, D.C. Henao Carrillo1, L. Kattah1, O. Garnica5, J. Lanchares5, J.I. Hidalgo5
J.P. Llano1, A. Carol1, M. Rondon1 1
Universidad Rey Juan Carlos, Ciencias de la Computacion,
1
Pontificia Universidad Javeriana, Hospital San Ignacio, Mostoles, Spain
2
Unidad de Endocrinologa, Bogota, Colombia University of Applied Sciences Upper Austria, Heuristic and
ATTD 2017 READ BY TITLE A-129

Evolutionary Algorithm Laboratory, Hagenberg, Austria 327 SWITCHING TO INSULIN DEGLUDEC IN TYPE 1
3
Hospital Virgen de la Salud de Toledo, Servicio de En- DIABETES MELLITUS IN A REAL-LIFE SETTING
docrinologa y Nutricion, Toledo, Spain V. Bellido1, B. Gonzalez1, I. Goicolea1, F.J. Santamara1
4 1
Hospital Universitario Prncipe de Asturias, Servicio de En- Hospital Universitario Cruces, Endocrinologia, Barakaldo,
docrinologa y Nutricion, Alcala de Henares, Spain Spain
5
Universidad Complutense de Madrid, Adaptive and Bioin-
spired Systems Group, Madrid, Spain
328 MULTI CENTRE OPEN-LABEL CLINICAL STUDY
324 PREDICTIVE FACTORS OF THE ADHERENCE OF THE SUGARBEAT NON-INVASIVE CONTINUOUS
TO THE REAL-TIME CGM SENSORS: GLUCOSE MONITORING PATCH
A PROSPECTIVE OBSERVATIONAL STUDY T. Rahman1
1
(PARCS STUDY): STUDY DESIGN Nemaura Medical, Diagnostics, Loughborough, United King-
T. Murata1, M. Matsuhisa2, A. Kuroda2, M. Toyoda3, dom
Y. Hirota4, K. Kato5, H. Sawaki6, A. Tone7, S. Kawashima8,
A. Okada9, N. Sakane10
1
NHO Kyoto Medical Center, Diabetes Center, Kyoto, Japan
2
Tokushima University, Diabetes Therapeutics and Research 329 DIABETES (D) APPEARANCE IN ANOMALOUS
Center, Institute of Advanced Medical Sciences, Tokushima, MAGNETIC FIELD (AMF)
Japan N. Trifunovic1
1
3
Tokai University School of Medicine, Division of Nephrology- Belgrade, Serbia
Endocrinology and Metabolism, Department of Internal Medi-
cine, Isehara, Japan
4
Kobe University Hospital, Division of Diabetes and 330 BIOMETRIC DATA FROM HEALTH PATCH FOR
Endocrinology, Department of Internal Medicine, Kobe, GLYCEMIC MODELING
Japan D. Mul1, P. Funke2, R. Focia2, R. Vandenberg2, H.J. Aanstoot1,
5
NHO Osaka National Hospital, Diabetes Center, Osaka, R. Bruinsma3, H.J. Veeze1
Japan 1
Diabeter - Center for Pediatric and Adolescent Diabetes Care
6
Arisawa General Hospital, Diabetes Center, Hirakata, and Research, Pediatrics, Rotterdam, The Netherlands
Japan 2
Aspire Ventures, Technology and services, Lancaster PA, USA
7
Okayama University Hospital, Diabetes Center, Okayama, 3
Tempo Health, LLC, Rotterdam, The Netherlands
Japan
8
Kanda Naika Clinic, Department of Internal Medicine, Osaka,
Japan
9
Okada Clinic, Department of Internal Medicine, Fukuoka, 331 PROPOSAL FOR AN ITALIAN CONSENSUS FOR
Japan USE OF CSII IN TYPE 2 DIABETES:
10
NHO Kyoto Medical Center, Division of Preventive Medicine, IS IT FEASIBLE?
Clinical Research Institute, Kyoto, Japan G. Grassi1, G. Cristina2, C. Gauna2, P. Massucco3, L. Gentile4,
E. Ansaldi5, A. Girelli6, L. Sciangula7, A. Ciucci8, C. Tubili9,
C. Vitale10, L. De Candia11, A. Piccinni12, G. Tonolo13,
325 GLYCEMIC CONTROL WITH V-GO INSULIN E. Guastamacchia14
1
DELIVERY DEVICE COMPARED TO INSULIN PEN Citta della salute e della Scienza, Endocrinology, Torino, Italy
2
DEVICES IN PATIENTS REQUIRING BASAL- ASL TO2, Endocrinology and Metabolic Disease, Torino, Italy
3
PRANDIAL THERAPY San Luigi Gonzaga Faculty of Medicine of Turin University,
R. Lajara1, J. Davidson2, C. Nikkel3, T. Morris4 San Luigi Gonzaga Hospital, Medicine and Metabolic Disease
1 Unit, Torino, Italy
Diabetes America, Endocrinology, Plano, USA
2 4
University of Texas Southwestern Medical Center/Touchstone Cardinal Massaja Hospital, Endocrinology and Diabetes Unit,
Diabetes Center, Endocrinology, Dallas, USA Asti, Italy
3 5
Valeritas, Medical Affairs, Edmond, USA Santi Antonio e Biagio Hospital, Department of Endocrinology
4 and Metabolic Diseases, Alessandria, Italy
University of Central Oklahoma, Mathematics and Statistics,
6
Edmond, USA Spedali Civili of Brescia, Diabetologic Unit, Brescia, Italy
7
ASST Lariana- Presidio Polispecialistico, Diabetology Centre,
Como, Italy
8
326 EXAMINING THE EFFECTS OF INSULIN A. O. SantAnna di Como Ospedale di Mariano Comense,
PEN COMPLIANCE DEVICE ON HBA1C Diabetology Endocrinology, Como, Italy
9
AND LIFESTYLE ON INDIVIDUALS LIVING S. Camillo-Forlanini, Diabetic Unit, Roma, Italy
10
WITH TYPE 1 DIABETES MELLITUS ASREM Termoli/Larino, Italy, Diabetology Center, Termoli/
D. Minnock1, S. ODonnell2, G. Doyle2, R. Canavan3, Larino, Italy
D. OShea3 11
Ospedale M. Sarcone Terlizzi BA, Diabetology Center, Ter-
1
Health Sciences, School of Physiotherapy- Public Health and lizzi Bari, Italy
12
Sports science, Dublin 4, Ireland A. O. Card. G. Panico Tricase LE, UO Internal Medicine,
2
University College Dublin, UCD College Of Business, Dublin Lecce, Italy
13
4, Ireland Struttura sanitaria San Giovanni di Dio, Diabetology, Olbia,
3
St Vincents University Hospital, Endocrinology, Dublin, Ire- Italy
14
land Policlinico di Bari, Internal Medicine Department, Bari, Italy
A-130 ATTD 2017 READ BY TITLE

332 A SMART BASAL-BOLUS CALCULATOR FOR 338 COMPARISON THE DATA OF SMBG WITH CGMS
PEOPLE WITH DIABETES IN PREGNANT WOMEN WITH PRECONCEPTION
D. Boiroux1, T.B. Aradottir1, N.K. Poulsen1, H. Madsen1, DIABETES MELLITUS
J.B. Jorgensen1 T. Shestakova1, A. Dreval1, O. Dreval1, O. Medvedev2
1 1
Technical University of Denmark, Department of Applied Moscow Regional Research and Clinical Institute MONIKI,
Mathematics and Computer Science, Kgs. Lyngby, Denmark Endocrinology, Moscow, Russia
2
Lomonosov Moscow state university, Department of Basic
333 TELEMEDICINE CONSULTATION IN PATIENTS Medicine, Moscow, Russia
WITH TYPE 1 DIABETES MELLITUS TREATED WITH
CONTINUOUS SUBCUTANEOUS INSULIN INFUSION
(CSII): ANALISYS OF EFFICIENCY AND EFFICACY
339 EVALUATING THE ROLE OF MHEALTH
AFTER TWO YEARS
INTERVENTIONS IN PATIENT EMPOWERMENT
V.M. Andia1, M. Picallo1, A. Lopez Guerra1, M. Moyano1,
FOR CHRONIC DISEASE MANAGEMENT SUCH
M.J. Rodrguez Calero1, M.L. Ramrez Rodrguez1,
AS DIABETES
M. Requena1, Y. Olmedilla1, M. Arnoriaga1, J. Agreda1,
O. Saadatfard1, D. Wiljer2, A.G. Ekeland3, M. Bradway1,
M.A. Velez1, J. Atencia1, B. Weber1, C. Gonzalez Antiguedad1,
E. Arsand3
M. Sambo1, S. Monereo1 1
1 UNN - University Hospital of North-Norway, Norwegian
Hospital General Universitario Gregorio Maranon, En-
Centre for eHealth Research, Troms, Norway
docrinology, Madrid, Spain 2
University Hospital Network, Education Technology Innova-
tion, Toronto, Canada
3
334 SELF-REPORTED HYPOGLYCEMIA REDUCTION UNN - University Hospital of North-Norway, Norwegian
IN TANDEM PUMP USE COMPARED TO PREVIOUS Centre for E-health Research, Troms, Norway
METHODS OF DIABETES THERAPY
G. Marin1
1
Tandem Diabetes, Marketing, San Diego, USA
340 AN INTEGRATED MOBILE DIABETES
335 TELEMEDICINE IN GESTATIONAL DIABETES APPLICATION CAN BENEFIT GLUCOSE MANAGE-
(GDM): FEASIBILITY, DIABETOLOGICAL AND MENT AND DECREASE DISEASE BURDEN IN TYPE 1
OBSTETRICS-NEONATAL OUTCOMES, COSTS DIABETES: A MIXED METHOD STUDY
AND PATIENTS SATISFACTION ANALYSIS T.H. van de Belt1, J. Lancee1, B.C. Heeren1, L.J. Engelen1,
E. Gamarra1, V. Rovei2, C. Monzeglio2, D. Ciochetto2, M.M. Van Gelder2, C.J. Tack3
1
F. Leone3, S. Rizzo2, A. Valla3, G. Maletto4, A. Bisio5, Radboud University Medical Center, Radboudumc REshape
F. Broglio1, E. Ghigo1, G. Grassi1 Innovation Center, Nijmegen, The Netherlands
2
1
AO Citta della Salute e della Scienza di Torino, Radboud University Medical Center, Department for Health
S.C.D.U. Endocrinologia, Diabetologia e Metabolismo, Evidence- Radboud Institute for Health Sciences, Nijmegen,
Torino, Italy The Netherlands
3
2
AO Citta della Salute e della Scienza di Torino, Radboud University Medical Center, Department of Internal
P.O. SantAnna - S.C.D.U. Ginecologia e Ostetricia, Medicine, Nijmegen, The Netherlands
Torino, Italy
3
AO Citta della Salute e della Scienza di Torino, P.O.
SantAnna - Servizio di Dietetica, Torino, Italy
4
AO Citta della Salute e della Scienza di Torino, 341 ASSESSMENT OF INSULIN DISTRIBUTION
Dipartimento di Scienze Mediche, Torino, Italy IN SUBCUTANEOUS ADIPOSE TISSUE WHEN
5
Universita degli Studi di Torino, Facolta di Medicina e USING NOVEL INSULIN INFUSION CATHETERS
Chirurgia, Torino, Italy T. Altendorfer-Kroath1, P.K. Schondorff2, R. Juliussen2,
M. Heschel2, F. Sinner1, T. Birngruber1
1
JOANNEUM RESEARCH Forschungsgesellschaft mbH,
336 THE BURDEN OF SEVERE HYPOGLICAEMIAS HEALTH, Graz, Austria
AND DIABETES KETOACIDOSIS IN T1DM PATIENTS 2
Unomedical a/s, A ConvaTec Company, Lejre, Denmark
R. Ciampichini1, P. Cozzolino2, P. Cortesi1, C. Fornari1,
F. Madotto1, V. Chiodini1, L. Mantovani1, G. Cesana1
1
University of Milano-Bicocca, Research Centre on Public
Health, Monza, Italy 342 NOVELTY IN TREATMENT OF DIABETIC
2
CHARTA Foundation, Public Health, Monza, Italy ULCER WITH ABU-KHALSA (ARNEBIA EUCHROMA)
OINTMENT
M. Amiri1, Z. Abdi2, R. Alipour3, M. Alipour4
1
337 MORTALITY OF HYPO DM PATIENTS: A COM- Faculty of Medicine, Islamic Azad University, Kazerun
PARISON WITH AMI AND PAD FOR FURTHER Branch, Milad Medical Center, Shiraz, Iran
2
DISCUSSION Shiraz High School of Medical Sciences, Milad Medical
R. Ciampichini1, P. Cozzolino2, P. Cortesi1, C. Fornari1, Center, Shiraz, Iran
F. Madotto1, V. Chiodini1, L. Mantovani1, G. Cesana1 3
Faculty of Nursing, Islamic Azad University, Tehran Branch,
1
University of Milano-Bicocca, Research Centre on Public Milad Medical Center, Shiraz, Iran
4
Health, Monza, Italy Islamic Azad University, Kazerun Branch, Milad Medical
2
CHARTA Foundation, Public Health, Monza, Italy Center, Shiraz, Iran
ATTD 2017 READ BY TITLE A-131

343 AUTOMATIC PROCESSING OF THE SELF- 349 WHAT MEDICINES ARE USED FOR DIABETES
MONITORING OF BLOOD GLUCOSE (SMBG) DATA IN BRAZIL: DATA FROM AN NATIONAL SURVEY
WITH GLUCOPRINT DEVICE: SMBG FREQUENCY M.M. Milward de Azevedo Meiners1, N. Urruth Leao Tavares1,
AND HBA1C TRENDS S. Serrate Mengue2, P.F. Couto Vieira1, A. Turmina Fonta-
A. Mayorov1, E. Surkova1, G. Galstyan1 nella2, E. Merchan-Hamann1
1 1
Endocrinology Research Centre, Diabetes Institution, Moscow, Health Sciences Facultat, University of Braslia, Post Gradu-
Russia ation Program on Public Health, Brasilia, Brazil
2
University of Rio Grande do Sul, Post Graduation Program on
344 OPTIMISING EDIARY DESIGN FOR DIABETES Epidemiology, Porto Alegre, Brazil
CLINICAL TRIALS USING REAL WORLD USAGE
DATA
J. Thilaganathan1, A. Heikkila1, P. ODonohoe1, O. Kotiranta1 350 IMPROVING GLYCEMIC CONTROL WITH
1
CRF Health, Product & Services Management, London, INSULIN PUMP THERAPY: THE IMPACT
United Kingdom OF DIABETES DURATION
S. Pejakovic1
1
Clinical Center of Vojvodina, Clinic for Endocrinology,
345 ESTIMATION OF HBA1C BASED ON 30-DAY Diabetes and Metabolic Disorders, Novi Sad, Serbia
SELF-MONITORING OF BLOOD GLUCOSE
N. Hosszufalusi1, A.G. Tabak2, G. Jermendy3, G. Vandorfi4,
A. Fulcz5, G. Bibok6, L. Koranyi7
1 351 IMPACT OF INSULIN FORMULATIONS
Semmelweis University, 3rd Department of Internal Medicine,
ON SUCCESSFUL CONTINUOUS SUBCUTANEOUS
Budapest, Hungary
2 INSULIN INFUSION (CSII): ROLE OF DILUENTS
Semmelweis University, 1st Department of Internal Medicine,
ON CSII INDUCED TISSUE REACTIONS
Budapest, Hungary
3 U. Klueh1, Y. Qiao2, I. Ludzinska2, C. Czajkowski2,
Bajcsy-Zsilinszky Hospital, 3rd Department of Internal Medi-
D. Kreutzer2
cine, Budapest, Hungary 1
4 Wayne State University, Biomedical Engineering, Detroit,
Vanderlich, Health Center, Veszprem, Hungary
5 Michigan, USA
Csolnoky Ferenc Hospital, Endocrine and diabetes center, 2
University of Connecticut School of Medicine, Surgery,
Veszprem, Hungary
6 Farmington, USA
77 Elektronika, Medical Department, Budapest, Hungary
7
DRC, Research Center, Balatonfured, Hungary

352 CONTINUOUS GLUCOSE MONITORING

346 THE METHOD OF INSULIN ADMINISTRATION IN GLYCOGEN STORAGE DISEASE TYPE IA
AND PERINATAL OUTCOMES IN PREGNANT A MAJOR IMPROVEMENT FOR PATIENTS
WOMEN WITH TYPE 1 DIABETES D. Martins1, D. Oliveira1, C. Baptista1, S. Paiva1, H. Esperto2,
E. Deryabina1 S. Moreira2, H. Clemente2, N. Vicente1, L. Cardoso1, A. Lages1,
1
Ural Research Institute of Maternity and Infancy Welfare, M. Ventura1, N. Cunha1, F. Carrilho1
Yekaterinburg, Russia 1
Centro Hospitalar e Universitario de Coimbra, Servico de
Endocrinologia, Diabetes e Metabolismo, Coimbra, Portugal
2
Centro Hospitalar e Universitario de Coimbra, Servico de
347 THE ROLE OF CONTINUOUS GLUCOSE Medicina Interna, Coimbra, Portugal
MONITORING SYSTEM ON THERAPY OF CYSTIC
FIBROSIS RELATED DIABETES IN YOUTH - PILOT
PROJECT
353 MODELING TREATMENT WITH METFORMIN,
S. Todorovic1, T. Milenkovic1, K. Mitrovic1, R. Vukovic1,
WITH AND WITHOUT INSULIN IN A VIRTUAL T2DM
L. Plavsic1
1 POPULATION IN AN ENVIRONMENT WITH THE
Mother and Chiled Health Care Institute, Endocrinology,
VARIABILITY OF COMMON SELF-MANAGEMENT
Belgrade, Serbia
BEHAVIORS
Y.Y. Chang1, M. Weinheimer1, S. Riddle1, G. Kongable1
1
The Epsilon Group, Simulation and Modeling, Charlottesville,
348 HELPING ON DIABETES MANAGEMENT: USA
SHOWING PATTERNS ON RECORDS
AND PROVIDING ADVICES ON A SMARTPHONE APP
D. Machado1, T. Paiva1, I. Dutra2, P. Brandao1, C. Neves3,
S. Oliveira3, C. Esteves3, C. Arteiro4, D. Carvalho3 354 TIME SPIRALS: NEW APPROACH TO VISUALIZE
1
Faculdade de Ciencias da Univ. Porto/Inst. Telecomunicacoes, LARGE DATASETS OBTAINED FROM CGM
Computer Science, Porto, Portugal MEASUREMENTS OF PATIENTS WITH DIABETES
2
Faculdade de Ciencias da Univ. Porto/CRACS, Computer D. Akotia1, A. Holubova1,2, J. Muzik1,2
1
Science, Porto, Portugal Charles University, Spin-off Company and Research Results
3
Faculdade de Medicina, Universidade do Porto, Servico de Commercialization Center of the First Faculty of Medicine,
Endocrinologia, Porto, Portugal Prague, Czech Republic
4 2
Faculdade de Ciencias da Nutricao e Alimentacao, Uni- Czech Technical University in Prague, Faculty of Biomedical
versidade do Porto, Servico de Nutricao, Porto, Portugal Engineering, Prague, Czech Republic
A-132 ATTD 2017 READ BY TITLE

355 AUTOMATIC DETECTION OF MEALTIME 360 THE EFFICACY AND SAFETY OF INSULIN
SITUATIONS IN DAILY REGIMEN OF PATIENTS DEGLUDEC AS BASAL-BOLUS INSULIN THERAPY
WITH TYPE 1 DIABETES WHO USE MHEALTH IN CHILDREN AND ADOLESCENTS WITH TYPE 1
TECHNOLOGIES DIABETES IN ROUTINE CLINICAL PRACTICE
A. Holubova 1,2, J. Muzk1,2, M. Polacek2, D. Fiala1, A. Kiiaev1, O. Koksharova1, L. Chernikh2, O. Pollyak2,
E. Arsand3,4, M. Bradway3,4, J. Kaspar1,2, K. Hana2, A. Yusupova2, I. Korneva3, M. Slovak3
P. Smrcka2, M. Kvapil5, J. Broz5 1
Ural State Medical University, Department of Outpatient Pe-
1
Charles University, Spin-off Company and Research Results diatrics, Ekaterinburg, Russia
2
Commercialization Center of the First Faculty of Medicine, Regional Pediatrics University Hospital No 1, Department of
Prague, Czech Republic Endocrinology, Ekaterinburg, Russia
2 3
Czech Technical University in Prague, Faculty of Biomedical University Hospital No 40, Department of Endocrinology,
Engineering, Prague, Czech Republic Ekaterinburg, Russia
3
Norwegian Centre for E-Health Research, University Hospital
of North Norway, Troms, Norway
4
UiT The Arctic University of Norway, Department of Clinical 361 FLASH GLUCOSE MONITORING IN CLINICAL
Medicine, Troms, Norway PRACTICE: COMPARISON BETWEEN BASIC
5
Second Faculty of Medicine, Department of Internal Medicine, AND PROFESSIONAL APPROACH
Prague, Czech Republic A. Tumminia1, V. Rapisarda2, C. Egiziano2, L. Frittitta1,
L. Sciacca1, L. Tomaselli2
1
Garibaldi-Nesima Hospital, Endocrinology Section, Clinical
and Experimental Medicine, Catania, Italy
2
356 COMMON BARRIERS TO USING NEW SITES Garibaldi-Nesima Hospital, Endocrinology Unit,
FOR INSULIN INJECTION/INFUSION IN ADULTS Catania, Italy
WITH TYPE 1 DIABETES (T1D)
E. Patrakeeva1, E. Mokhova1, T. Zagorovskaya2,
A. Mosikian1 362 COMPARATIVE PHARMACOECONOMIC
1
Saint-Petersburg Medical University, Endocrinology, Saint- ASSESSMENT OF THE TYPE 2 DIABETES
Petersburg, Russia TREATMENT WITH INSULIN DEGLUDEC
2
Saint-Petersburg State University, Saint-Petersburg, Russia AND INSULIN GLARGINE IN BASAL-BOLUS
INSULIN THERAPY
M. Shestakova1, A. Kolbin2, G. Galstyan3, A. Kurilev2,
Z. Wenlong2, J. Balykina4, M. Proskurin4, Y. Alekseeva5,
357 CONNECTING DIABETES TELEMEDICINE E. Startseva6
1
SYSTEM TO A NONSTOP EMERGENCY Endocrinology Research Centre, Diabetes Institute, Moscow,
HELPDESK Russia
2
J. Muzik1, A. Holubova1, M. Vlasakova2, D. Fiala1, Pavlov First Saint Petersburg State Medical University, De-
M. Polacek1, K. Hana1, K. Jan1, M. Muzny2, J. Broz3 partment of Clinical Pharmacology and Evidence-based Med-
1
Czech Technical University in Prague, Faculty of Biomedical icine, Saint Petersburg, Russia
3
Engineering, Prague 2, Czech Republic Endocrinology Research Centre, Diabetic Foot Department,
2
Charles University, Spin-off Company and Research Results Moscow, Russia
4
Commercialization Center, First Faculty of Medicine, Prague, Saint Petersburg State University, Department of Mathema-
Czech Republic tical Modelling of Energy Systems, Saint Petersburg, Russia
5
3
Charles University, Department of Internal Medicine- 2nd Novo Nordisk, Medical Affairs, Moscow, Russia
6
Faculty of Medicine, Prague, Czech Republic Novo Nordisk, Clinical, Medical and Regulatory Department,
Moscow, Russia

363 USE OF MOBILE-BASED TECHNOLOGIES

358 TREATMENT OF AUTOIMMUNE DIABETES TO IMPROVE DIABETES SELF-MANAGEMENT
AND INSULITIS IN NOD MICE WITH A SYNTHETIC BEHAVIOR AND POSTPRANDIAL GLUCOSE
PEPTIDE VACCINE BASED ON EPITOPE OF DPP4 CONTROL
AND P277 E. Toschi- MD1,2, S. Edwards1, H. Wolpert- MD1,2
T. Li1, X. Wei1, Z. Li1, J. Fang1 1
Joslin Diabetes Center, Clinical- Behavioral- and Outcomes
1
China Pharmaceutical University, Life Science and Technology Research, Boston, USA
College, Nanjing, China 2
Harvard Medical School, Boston, USA

364 MENTAL HEALTH AND DIABETES: CAN TECH-

359 INSULIN DEGLUDEC AND PREGNANCY: NOLOGY BE A GAME CHANGER AND TRANSFORM
CASES REPORT IN TYPE 1 DIABETIC WOMEN CARE IN PATIENTS WITH BOTH DIABETES
E. Gamarra1, C. Monzeglio2, G. Grassi1 AND MENTAL HEALTH PROBLEMS?
1
AO Citta della Salute e della Scienza G. Mlawa1,2
1
di Torino, S.C.D.U. Endocrinologia, Diabetologia e Metabo- Queens Hospital-London, Acute Medicine/Diabetes & Endo-
lismo, Torino, Italy crinology, London, United Kingdom
2 2
AO Citta della Salute e della Scienza di Torino, Ginecologia e King George Hospital, Acute Medicine/Diabetes & Endo-
Ostetricia, Torino, Italy crinology, London, United Kingdom
ATTD 2017 READ BY TITLE A-133

365 GLYCEMIC CONTROL DURING PREGNANCY 367 THE COURSE OF DIABETES IN MONGOLOIDS
IN WOMEN WITH TYPE 1 DIABETES MELLITUS L. Kolesnikova1, S.I. Kolesnikov1, M.A. Darenskaya1,
TREATED THE PREDICTIVE LOW-GLUCOSE L.A. Grebenkina1, S.V. Gnusina2
1
SUSPEND INSULIN PUMP (640G) A CASE Scientific Centre for Family Health and Human Reproduction
REPORT Problems, Department of Pathophysiology, Irkutsk, Russia
K. Cyganek1, T. Klupa1,2, M. Malecki1,2 2
Scientific Centre for Family Health and Human Reproduction
1
Hospital University, Department of Metabolic Diseases, Kra- Problems, Department of Pathophysiology, Irkutsk, Russia
kow, Poland
2
Jagiellonian University, Medical College, Department
of Metabolic Diseases, Krakow, Poland
368 HEALING OF DIABETIC FOOT ULCER USING
AUTOLOGOUS PERIPHERAL BLOOD
MONONUCLEAR CELLS
B. Supartono1,2, P. Kusumaningsih3, M. Sakiinah4
1
366 EFFECTIVENESS OF TRANSITION BETWEEN University of Pembangunan Nasional Veteran Jakarta,
MULTIPLE DAILY INJECTIONS TO CONTINUOUS Stem Cell Research and Tissue Engineering Center, Jakarta,
SUBCUTANEOUS INSULIN INFUSION IN PATIENTS Indonesia
2
WITH TYPE 1 DIABETES National Sport Hospital, Orthopaedic Department, Jakarta,
C. Neves1, S. Oliveira1, J.S. Neves1, M. Pereira1, C. Esteves1, Indonesia
C. Arteiro1, C. Redondo1, A. Costa1, D. Carvalho1 3
RSU Al-Fauzan, Obstetrics and gynecology, Jakarta, In-
1
Hospital de Sao Joao, Department of Endocrinology, Diabetes donesia
4
and Metabolism, Porto, Portugal RSU Al-Fauzan, Emergency Room, Jakarta, Indonesia