DIABETES TECHNOLOGY & THERAPEUTICS

Volume 25, Number 9, 2023

ª Mary Ann Liebert, Inc.
DOI: 10.1089/dia.2023.0178

Open camera or QR reader and

scan code to access this article
and other resources online.

ORIGINAL ARTICLE

A Comparison of Faster Insulin Aspart

with Standard Insulin Aspart Using Hybrid
Automated Insulin Delivery System in Active Children
and Adolescents with Type 1 Diabetes:
A Randomized Double-Blind Crossover Trial
Klemen Dovc, MD,1,2 Simon Bergford, MS,3 Elke Fröhlich-Reiterer, MD,4
Dessi P. Zaharieva, PhD,5 Nejka Potocnik, MD,6 Alexander Müller, MS,7 Ziva Lenarcic, MD,1,2
Peter Calhoun, PhD,3 Maria Fritsch, MD,4 Harald Sourij, MD,5 Natasa Bratina, MD,1,2
Craig Kollman, PhD,3 and Tadej Battelino, MD1,2

Abstract
Objective: To evaluate the use of faster acting (FIA) and standard insulin aspart (SIA) with hybrid automated
insulin delivery (AID) in active youth with type 1 diabetes.
Research Design and Methods: In this double-blind multinational randomized crossover trial, 30 children and
adolescents with type 1 diabetes (16 females; aged 15.0 – 1.7 years; baseline HbA1c 7.5% – 0.9%
[58 – 9.8 mmol/mol]) underwent two unrestricted 4-week periods using hybrid AID with either FIA or SIA in
random order. During both interventions, participants were using the hybrid AID (investigational version of
MiniMed 780G; Medtronic). Participants were encouraged to exercise as frequently as possible, capturing
physical activity with an activity monitor. The primary outcome was the percentage of sensor glucose time
above range (180 mg/dL [10.0 mmol/L]) measured by continuous glucose monitoring.
Results: In an intention-to-treat analysis, mean time above range was 31% – 15% at baseline, 19% – 6% during
FIA use, and 20% – 6% during SIA use with no difference between treatments: mean difference = -0.9%; 95%
CI: -2.4% to 0.6%; P = 0.23. Similarly, there was no difference in mean time in range (TIR) (78% and 77%) or
median time below range (2.5% and 2.8%). Glycemic outcomes during exercise or postprandial periods were
comparable for the two treatment arms. No severe hypoglycemia or diabetic ketoacidosis events occurred.

1
Department of Endocrinology, Diabetes and Metabolism, University Children’s Hospital, Ljubljana, Slovenia.
2
Faculty of Medicine, University of Ljubljana, Ljubljana, Slovenia.
3
Jaeb Center for Health Research Foundation, Inc., Tampa, Florida, USA.
4
Department of Pediatrics and Adolescent Medicine, Medical University Graz, Graz, Austria.
5
Division of Endocrinology, Department of Pediatrics, Stanford University, Stanford, California, USA.
6
Faculty of Medicine, Institute of Physiology, University of Ljubljana, Ljubljana, Slovenia.
7
Interdisciplinary Metabolic Medicine Trials Unit, Medical University of Graz, Graz, Austria.
8
Division of Endocrinology and Diabetology, Medical University of Graz, Graz, Austria.
Parts of this study were presented at the 15th International Conference on Advanced Technologies & Treatments for Diabetes (ATTD
2022), Barcelona, Spain.

612
FACT STUDY 613

Conclusions: FIA was not superior to SIA with hybrid AID system use in physically active children and
adolescents with type 1 diabetes. Nonetheless, both insulin formulations enabled high overall TIR and low time
above and below ranges, even during and after documented exercise.
Trial Registration Clinicaltrials.gov: NCT04853030.

Keywords: Exercise, Faster acting insulin, Children, Adolescents, Automated insulin delivery, Type 1 diabetes.

Introduction Methods
Study design
A utomated insulin delivery (AID), with its glucose-
responsive approach and mimicking the physiological
response of a healthy beta cell, has revolutionized the
This double-blind, multinational, two-period randomized
crossover trial was conducted at two locations: University
management of type 1 diabetes. Numerous clinical studies Children’s Hospital Ljubljana in Slovenia and Medical
evaluating various AID systems have unequivocally dem- University of Graz in Austria. Written informed consent was
onstrated safe improvements in glycemic outcomes in indi- obtained from all participants before any study-related ac-
viduals with type 1 diabetes of different ages, genders, and tivities. The protocol and informed consent or assent forms
diabetes durations.1–6 As such, current clinical guidelines were approved by the appropriate institutional review boards
recommend offering AID systems to all youth and adults with and ethics committees, and regulatory approval to conduct the
type 1 diabetes.7,8 study was obtained in both countries. The study is listed on
All current commercially available AID systems adopt the clinicaltrials.gov under registration number NCT04853030.
hybrid approach, requiring prior meal and exercise an- During the two unrestricted 4-week periods, each partici-
nouncements by the user to achieve recommended glycemic pant used one of the two insulin formulations with the
targets.9,10 Part of the challenge with current AID technology MiniMed 670G 4.0 (investigational version of MiniMed
is due to the pharmacokinetic and pharmacodynamic delay, 780G with equivalent algorithm but without Bluetooth con-
and comparatively slow absorption time from the subcuta- nectivity and with two glucose set points 100 and 120 mg/dL)
neous administration of insulin.11 Faster insulin formulations Medtronic AID system: FIA or SIA, assigned in random or-
are continuously being developed that have the potential to der. Randomization was done using a computer-generated
further advance the efficacy and safety of AID systems, with sequence with a permuted block size of four. Study partici-
the goal of becoming fully AID technology in the future.12 pants and investigators were blinded to the treatment
Faster acting insulin aspart (FIA) is a recent aspart (Novo allocation.
Nordisk, Denmark) formulation, to which two excipients Eligible participants were children and adolescents aged
have been added to increase the early absorption (nicotin- 10–18 years, diagnosed with type 1 diabetes for at least 6
amide) and to optimize the stability (l-arginine) of the for- months, using an insulin pump for at least 3 months, and had a
mulation.13 Studies have shown that administration of FIA screening HbA1c <11% (97 mmol/mol). Key exclusion cri-
bolus, either with subcutaneous injections or with an insulin teria were untreated celiac or thyroid disease, current treat-
pump, are associated with earlier insulin exposure and action, ment with drugs known to interfere with glucose metabolism,
and earlier offset of exposure than with standard insulin as- and diabetes management using an ultrarapid acting insulin
part (SIA).14–16 analogue. A full list of the inclusion and exclusion criteria is
In a first double-blind randomized clinical trial evaluating given in Supplementary Table S1.
glycemic outcomes using AID with either FIA or SIA in Before randomization, participants completed a run-in
adults with type 1 diabetes, time in range (TIR; 70– period of at least 1 week using the study insulin pump
180 mg/dL) was comparable between the two arms.17 Note- (without AID mode) and continuous glucose monitoring
worthy, this study was conducted in a supervised inpatient (CGM; Guardian Sensor 3). Capillary fingerstick blood
setting, with a fully AID approach involving unannounced/ glucose testing was performed using a Contour Next Link 2.4
uncovered meals and an unannounced afternoon exercise blood glucose meter (Ascensia Diabetes Care), and blood
protocol.17 Recently, randomized clinical trials including ketone testing was performed using the Abbott Precision Xtra
adults with type 1 diabetes have demonstrated a modest im- meter (Abbott Laboratories). Successful completion of the
provement in glycemic outcomes with FIA compared with run-in period required at least 80% CGM wear time during
SIA using various hybrid AID systems over several the prior 7 days and an average of at least three blood glucose
weeks.18–20 However, there is currently limited evidence meter tests per day.
from randomized controlled trials enrolling children and Randomization occurred on the same day that the run-in
adolescents with type 1 diabetes.21 period was completed. Participants received AID system
We aimed to evaluate the use of FIA in a second- training at the beginning of each crossover period. Partici-
generation hybrid AID system in youth with type 1 diabetes pants were contacted 6–10 days into each study period to
over a longer period that included frequent physical activity. initialize AID mode, and then continued to use the AID
Based on available data, we hypothesized that AID with FIA system for 4 weeks. All participants started with an AID
would improve glucose time above range (>180 mg/dL) mode target glucose set point of 100 mg/dL and active insulin
due to improved postprandial glycemic outcomes compared time set at 2 h. At the AID mode initiation contact, partici-
with SIA. pants were reminded to obtain an overnight fingerstick blood
614 DOVC ET AL.

glucose measurement (between 2 and 3 AM) for one night distribution. A post hoc analysis was conducted examining
after AID initiation, and if blood glucose was <70 mg/dL, to postprandial excursions by time of day: breakfast (4 AM–<11
treat with fast-acting carbohydrates, discontinue AID mode, AM), lunch (11 AM–<4 PM), and dinner (4 PM–<9 PM).
and notify the investigators the next day for advice to increase Participants completed various quality-of-life surveys in-
the glucose set point and/or active insulin time. cluding the Diabetes Distress Scale, the Glucose Monitoring
During both periods after AID mode was initialized, par- Satisfaction Survey, Hypoglycemia Confidence Survey, the
ticipants had telephone contacts at 24 h, 3 days, and 2 weeks, Diabetes Technology Attitudes Survey, and Insulin Dosing
and clinic visits at 1, 3, and 4 weeks. Participants were asked Systems: Perceptions, Ideas, Reflections, and Expectations
to upload device data for study staff review before each (INSPIRE) Survey.
contact, and device data were collected during each follow- A sample size of 30 participants was selected to provide
up visit. Participants were encouraged to engage in exercise 80% power with a two-sided type 1 error rate of 5% to reject
frequently to meet current exercise guidelines of at least the null hypothesis of no between-group difference in per-
60 min of moderate-to-vigorous exercise per day.22 centage time >180 mg/dL, under the assumption that the
In line with current consensus recommendations, partici- mean percentage time >180 mg/dL in the FIA group would be
pants were also advised to set a higher (exercise) glucose 6.7% lower than that in the SIA group, with a standard de-
target (150 mg/dL) at least 1 h before exercise11,23 and to viation of paired differences of 12.7%.
capture the exercise on a commercially available physical Statistical analyses were performed on an intention-to-
activity monitor (Garmin Venu Sq, Garmin). In the second treat basis, and all randomized participants were included in
half of each study period, participants were invited to par- the primary and secondary analyses. A per-protocol analysis
ticipate in a 1-day supervised sports camp (*5 h of exercise was conducted including participants who used the CGM and
per day). AID feature at least 80% of the time in both periods. Sub-
Adverse events were recorded throughout the trial. An group analyses were conducted to examine the relationship
adverse event was defined as any untoward medical occur- between the primary outcome and several baseline factors,
rence in a study participant, irrespective of the relationship including age, baseline HbA1c, type 1 diabetes duration,
between the adverse event and the treatments under investi- gender, and random C-peptide level. For the primary out-
gation. Reportable adverse events were any serious adverse come, the treatments were compared using a repeated mea-
event, an adverse device effect, an adverse event occurring in sures least squares regression model with an unstructured
association with a study procedure, an adverse event that covariance structure adjusting for study period and HbA1c at
leads to discontinuation of a study device for 2 or more hours, randomization as fixed effects.
severe hypoglycemia, or diabetic ketoacidosis.24,25 Missing data were handled by means of direct likelihood.
Analyses for the secondary outcomes, per-protocol analysis,
and subgroup analyses paralleled the primary analysis. For
Statistical methods
secondary analyses, the false discovery rate was controlled
The primary outcome of the trial was time above range using the adaptive two-stage Benjamini–Hochberg proce-
(>180 mg/dL) as measured by CGM tested for superiority. dure. All P values are two tailed. Analyses were performed
Secondary outcomes included mean glucose concentration, using SAS software, version 9.4 (SAS Institute).
glucose coefficient of variation, TIR (70–180 mg/dL), and the
percentage of time >250 mg/dL, time below range
Results
(<70 mg/dL), and time <54 mg/dL. CGM and insulin metrics
were calculated over a 24-h period, and separately for daytime Between April 21, 2021, and June 28, 2021, 30 participants
(6 AM–11:59 PM), nighttime (12 AM–5:59 AM), postmeal were screened and enrolled. University Children’s Hospital
periods (3 h after carbohydrates were entered), and exercise Ljubljana enrolled 24 participants and the Medical Uni-
periods (from start of exercise to 2 h after exercise cessation). versity of Graz enrolled 6 participants. Participants ranged in
Postprandial periods began when carbohydrates were first age from 11 to 18 years, and HbA1c at screening ranged from
entered and stopped either 3 h after a meal, when additional 5.9% to 9.9% (Table 1). All participants enrolled were non-
carbohydrates were entered, or exercise began. Postexercise Hispanic White.
periods began when the exercise session was initiated and All participants who were randomized completed both
stopped either 2 h after the exercise session ended or when periods using each treatment arm and visit and phone com-
carbohydrates were entered or another exercise session be- pletion rates were 100% (Fig. 1). CGM and AID use were
gan. Data from the sports camp days were analyzed sepa- high with a median percentage time spent using CGM of 91%
rately and were not included in the main analysis. CGM and during the FIA arm and 89% during the SIA arm, and a
insulin outcomes were calculated in each respective period if median percentage time spent using AID mode of 89% during
a participant had at least 3 days of CGM data for the 24-h the FIA arm and 87% during the SIA arm (Supplementary
period, 2 days for daytime period, 1 day for nighttime and Table S2).
postmeal periods, and 4 h for exercise periods. Mean percentage of time above range (>180 mg/dL) was
Additional secondary outcomes for postmeal periods in- 31% – 15% at baseline, 19% – 6% during the FIA arm, and
cluded peak glucose concentration, glucose excursion, time 20% – 6% during the SIA arm (mean difference [FIA-
to peak glucose concentration, and postprandial area under SIA] = -0.9%; 95% CI -2.4% to 0.6%; P = 0.23; Table 2 and
the curve. These additional outcomes were only calculated if Fig. 2). There was no evidence of a treatment effect by period
each individual postmeal period lasted at least 2 h and then carryover effect (P = 0.89; Supplementary Fig. S1). The per-
were averaged over all meals for each participant. All CGM protocol analysis showed results similar to the primary
and insulin outcomes were summarized appropriately to their analysis (Supplementary Table S3).
FACT STUDY 615

Table 1. Participant Characteristics at Enrollment

Overall N = 30 FIA first N = 15 SIA first N = 15
Age at randomization (years)
10–13 6 (20%) 2 (13%) 4 (27%)
14–18 24 (80%) 13 (87%) 11 (73%)
Mean – SD 15.0 – 1.7 15.1 – 1.6 15.0 – 1.7
Range 11.8–18.3 12.2–18.3 11.8–17.9
Gender—Male 14 (47%) 7 (47%) 7 (47%)
Race—White non-Hispanic 30 (100%) 15 (100%) 15 (100%)
Diabetes duration at randomization (years)
Mean – SD 7.8 – 3.8 8.8 – 4.2 6.8 – 3.3
Range 1.3–14.8 1.3–14.8 2.9–12.6
HbA1c at screening (%)
£8.5 24 (80%) 12 (80%) 12 (80%)
‡8.6 6 (20%) 3 (20%) 3 (20%)
Mean – SD 7.5 – 0.9 7.3 – 0.9 7.7 – 0.9
Range 5.9 to 9.9 5.9 to 8.9 6.8 to 9.9
C-Peptide at screening (ng/mL)
Mean – SD 0.10 – 0.06 0.10 – 0.05 0.11 – 0.06
<0.1 13 (43%) 8 (53%) 5 (33%)
Baseline % time in range (70–180 mg/dL)
Mean – SD 66% – 14% 68% – 10% 63% – 17%
FIA, faster acting insulin aspart; SIA, standard insulin aspart.

Mean glucose was 158 – 24 mg/dL at baseline, CI -0.08 to 0.05 units/kg per day; P = 0.80; Supplementary
140 – 10 mg/dL during the FIA arm, and 141 – 8 mg/dL Table S6). Total, basal, and bolus daily insulin were similar
during the SIA arm (mean difference [FIA-SIA] = -1.2 mg/ for the two treatment arms during daytime, nighttime, and
dL; 95% CI -4.9 to 2.5 mg/dL; P = 0.45; Table 2). Mean postmeal periods.
percentage of TIR was 66% – 14% during baseline, Treatment group differences between FIA and SIA on
78% – 6% during the FIA arm, and 77% – 7% during the SIA percentage time >180 mg/dL did not differ by age, HbA1c,
arm (mean difference [FIA-SIA] = 1.4%; 95% CI -0.8% to diabetes duration, gender, or C-peptide level (Supplementary
3.6; P = 0.18). Median percentage of time with glucose levels Table S7). The treatment effect by age group and by baseline
<54 mg/dL was 0.36% (0.06%, 0.96%) during baseline, HbA1c group is shown in Supplementary Tables S8 and S9.
0.41% (0.14%, 1.05%) during the FIA arm, and 0.42% Exploratory outcomes for the sports camp days and meal
(0.24%, 1.27%) during the SIA arm (mean difference [FIA- challenge test are given in Supplementary Tables S10 and
SIA] = -0.11%; 95% CI -0.33% to 0.08%; P = 0.23). S11. TIR and time >180 mg/dL for the sports camp days were
During postmeal periods, mean percentage time similar to the 24-h outcomes. Insulin use for the sports camp
>180 mg/dL was 37% – 16% during baseline, 28% – 9% days was similar to that of the rest of the study. TIR (70–
during the FIA arm, and 29% – 9% during the SIA arm (mean 180 mg/dL) was 41% – 21% for FIA and 55% – 28% for SIA
difference [FIA-SIA] = -1.8%; 95% CI -5.8% to 2.2%; for the missed bolus test and 51% – 29% for FIA and
P = 0.48; Table 2). Other postprandial CGM metrics includ- 54% – 29% for SIA for the delayed bolus test. Questionnaire
ing peak glucose and postprandial area under the curve were results are reported in Supplementary Table S12.
not significantly different for the two treatment arms. Post- One adverse event was reported during baseline, two ad-
prandial excursion for breakfast meals was 42 – 38 mg/dL verse events during the FIA period, and two adverse events
during baseline, 64 – 21 mg/dL during the FIA arm, and during the SIA period. One of the adverse events in the FIA
77 – 21 mg/dL during the SIA arm (mean difference [FIA- period was ketonemia, and the rest were viral infections.
SIA] = -12 mg/dL; 95% CI -20 to -3 mg/dL; P = 0.004; None of these adverse events were serious, and there were no
Supplementary Table S13). There was no significant differ- cases of severe hypoglycemia nor diabetic ketoacidosis
ence in postprandial excursion during the lunch or dinner during the study.
meals.
The mean glucose over the 24 h was largely similar for the
Discussion
two treatment arms (Fig. 3). Other CGM outcomes by time of
day were also similar for the two treatment arms (Supple- In this randomized controlled trial investigating the use of a
mentary Table S4). There were also no treatment group dif- faster acting insulin analogue with hybrid AID in youth with
ferences for CGM outcomes during exercise (Supplementary type 1 diabetes, we demonstrate comparable glycemic out-
Table S5). comes between FIA and SIA. The primary outcome of this
Mean total daily insulin dose was 1.19 – 0.63 units/kg per study, the superiority of FIA compared with SIA in time above
day during baseline, 0.75 – 0.12 units/kg per day during the range (>180 mg/dL), was not met. However, given that the
FIA arm, and 0.77 – 0.15 units/kg per day during the SIA arm achieved glycemic outcomes in both study arms, which ex-
(mean difference [FIA-SIA] = -0.02 units/kg per day; 95% ceeded the current consensus clinical recommendations (time
616 DOVC ET AL.

FIG. 1. Visit completion flowchart.

above range 19% and 20% and TIR 78% and 77%, respec- observed in adults using different insulin delivery modali-
tively), an additional significant improvement in glycemic ties,17,26–29 including three recent studies evaluating FIA
outcomes with FIA could be challenging to achieve. with different AID systems in an unsupervised environ-
The performance of the hybrid AID system during the ment.18–20 Lee et al.20 reported small (-1.4%) improvements
study was noteworthy, with a reduction in time above range in time above range and in TIR (+1.9%) with FIA use over 6
from 31% during the baseline period to *19% during the weeks, and Beck et al.18 reported similar improvements in
observational period, and a concomitant improvement in TIR time above range (-2%) and TIR (2%) with FIA use over 13
from 66% during the baseline to *78% during the obser- weeks. Importantly, both the above-mentioned studies were
vational period, irrespective of insulin formulation. open label, hence participants’ knowledge of the insulin
Results of this study observed in active children and ad- formulation used could have influenced their decision mak-
olescents with type 1 diabetes are complementing results ing when administering insulin.
 Table 2. Continuous Glucose Monitoring Outcomes over 24-Hour Period and After Meals
Adjusted difference FIA minus SIA
Baseline (N = 30) FIA (N = 30) SIA (N = 30) (95% CI) [P]a
Primary outcome: 24-h period
% Time >180 mg/dL, mean – SD 31% – 15% 19% – 6% 20% – 6% -0.9% (-2.4% to 0.6%) [0.23]
Secondary outcomes: 24-h period
Hours of data, median (quartiles) 148 (140, 161) 609 (577, 622) 599 (571, 612)
Mean glucose (mg/dL), mean – SD 158 – 24 140 – 10 141 – 8 -1.2 (-4.9 to 2.5) [0.45]
% TIR 70–180 mg/dL, mean – SD 66% – 14% 78% – 6% 77% – 7% 1.4% (-0.8% to 3.6%) [0.18]
Glucose CV (%), median (quartiles) 36% (33%, 39%) 35% (33%, 37%) 35% (33%, 39%) -0.8% (-2.2% to 0.5%) [0.18]
% Time >250 mg/dL, median (quartiles) 5.0% (2.2%, 15.8%) 3.0% (1.7%, 5.2%) 3.1% (1.8%, 6.0%) -0.1% (-0.8% to 0.5%) [0.63]
% Time <70 mg/dL, median (quartiles) 2.1% (1.4%, 4.2%) 2.5% (1.2%, 4.0%) 2.8% (1.8%, 4.1%) -0.5% (-1.1% to 0.1%) [0.11]
% Time <54 mg/dL, median (quartiles) 0.36% (0.06%, 0.96%) 0.41% (0.14%, 1.05%) 0.42% (0.24%, 1.27%) -0.11% (-0.33% to 0.08%) [0.23]
Secondary outcomes: postmealb
Hours of CGM data, median (quartiles) 93 (78, 101) 303 (253, 343) 289 (269, 344)
Mean glucose (mg/dL), mean – SD 168 – 25 153 – 15 155 – 13 -2.3 (-8.1 to 3.5) [0.53]

617
% TIR 70–180 mg/dL, mean – SD 60% – 15% 69% – 9% 67% – 9% 2.0% (-1.9% to 6.0%) [0.44]
Peak glucose (mg/dL), mean – SDc 210 – 31 199 – 19 202 – 16 -4.0 (-10.1 to 2.1) [0.34]
Time to peak glucose (min), mean – SDc 59 – 11 74 – 12 70 – 12 3.5 (-3.3 to 10.2) [0.44]
Excursion (mg/dL), mean – SDc 46 – 23 60 – 12 65 – 14 -4.7 (-11.9 to 2.6) [0.34]
Postprandial AUC (mg/dL), mean – SDc 23 – 14 30 – 8 31 – 9 -0.7 (-6.1 to 4.6) [0.72]
Glucose CV (%), median (quartiles) 34% (31%, 39%) 35% (33%, 37%) 35% (32%, 39%) -0.4% (-2.3% to 1.6%) [0.69]
% Time >180 mg/dL, mean – SD 37% – 16% 28% – 9% 29% – 9% -1.8% (-5.8% to 2.2%) [0.48]
% Time >250 mg/dL, median (quartiles) 7.0% (3.8%, 19.1%) 4.4% (2.7%, 7.3%) 4.9% (3.0%, 10.0%) -0.3% (-2.0% to 1.1%) [0.69]
% Time <70 mg/dL, median (quartiles) 2.1% (1.1%, 3.7%) 2.9% (1.4%, 4.4%) 2.9% (1.9%, 4.3%) -0.2% (-1.1% to 0.8%) [0.69]
% Time <54 mg/dL, median (quartiles) 0.22% (0.00%, 0.86%) 0.68% (0.18%, 1.24%) 0.50% (0.38%, 0.97%) -0.07% (-0.42% to 0.30%) [0.69]
a
P values and 95% CIs are from a repeated measures least squares regression model with an unstructured covariance structure adjusting for period and HbA1c at randomization as fixed
effects. Owing to a skewed distribution, % time >250, <70, and <54 mg/dL were transformed using a rank normal transformation. Multiple comparisons for secondary outcomes were adjusted
using the two-stage Benjamini–Hochberg adaptive false discovery rate procedure. The carryover effect for % time >180 mg/dL was tested by adding a period by treatment interaction term to the
model. The carryover effect P value was 0.89.
b
Postmeal periods begin when carbohydrates are first entered, lasting up to 3 h after a meal or until another carbohydrate is entered or exercise begins.
c
Metric is calculated for each meal, and then averaged across all meals for each participant.
AUC, area under the curve; CGM, continuous glucose monitoring; CV, coefficient of variation; TIR, time in range (70–180 mg/dL).
618 DOVC ET AL.

maintain near-normoglycemia (TIR 78% and 81%, respec-

tively) regardless of insulin formulation used. Although time
below range was moderately higher than overall results
(7.2% with FIA and 5.9% with SIA), time in clinically rele-
vant hypoglycemia <54 mg/dL was low in both study arms
(0.00% with FIA and 0.12% with SIA).
A previous free-living study estimated glycemic outcomes
associated with exercise indirectly through setting a higher
(exercise) target (150 mg/dL) in two different AID systems.30
The study demonstrated that the use of a higher target, re-
commended to be initiated before exercise, maintained the
same TIR and time in clinically relevant hypoglycemia as in
matched periods when a higher target was not used. Notably,
without activity monitor data, no temporal association be-
tween exercise intensity, duration, and timing of higher
glucose target initiation could be made.30
FIG. 2. Boxplots for % time >180 mg/dL by treatment
group and period. Box plots of the % time >180 mg/dL Ekhlaspour et al.31 demonstrated improved TIR with hy-
during baseline, period 1, and period 2 are shown. Baseline brid AID use during a 48-h ski camp with prolonged physical
is represented by the grey box, FIA period is represented by activity in children and adolescents with type 1 diabetes.
red boxes, and SIA period is represented by blue boxes. Under real-time remote monitoring, daytime (62.4%) and
Black dots indicate the mean values, horizontal bars in the overall (66.4%) TIR achieved with hybrid AID use were
boxes indicate the medians, and the bottom and top of each comparable with our observations. Recently, Morrison
box represent the 25th and 75th percentiles. FIA, faster et al.32 showed no difference in TIR between FIA and SIA
acting insulin aspart; SIA, standard insulin aspart. (81% for both insulin formulations) with AID use over 24 h
after exercise in adults with type 1 diabetes.
On the contrary, Boughton et al.19 reported unchanged time We evaluated postprandial glycemic outcomes, and TIR
above range and TIR with double-blinded FIA use in another approached the recommended clinical consensus target of
AID system over 8 weeks and reporting a slight improvement >70% (i.e., 69% with FIA and 67% with SIA) in this period
in time below range (-0.3%) with FIA use. In our study, there with no difference between the arms. However, a post hoc
was no difference between study interventions in hypogly- analysis assessing postprandial excursion by type of meal
cemia metrics, and both time <70 mg/dL and time <54 g/dL found FIA has a lower excursion than SIA during breakfast.
values were within the consensus clinical recommendations.9 Previous studies have demonstrated that FIA provides mod-
To our knowledge, this is the first free-living study to est improvements in postprandial glycemic outcomes20,28 or
evaluate glycemic outcomes in children and adolescents with no differences at all27 when used within AID.
type 1 diabetes using AID during unsupervised exercise. It might be possible that due to the highly adaptable AID
During exercise and recovery periods, AID was able to control algorithm that readjusts insulin delivery based on

FIG. 3. Mean glucose by 15-min period over 24-h day. The dots and solid curve represent the median value, the shaded
region represents the 25th and 75th percentiles, and the dashed curves represent the 10th and 90th percentiles.
FACT STUDY 619

glycemic profiles, differences in pharmacokinetics and article. All contributing authors approved the final version of
pharmacodynamics of currently approved insulin analogues the article. T.B. is the guarantor of this study and, as such, had
are not of sufficient extent to provide a clinically meaningful full access to all the data in the study and take responsibility
advantage.11 To further improve postprandial glycemic out- for the integrity of the data and the accuracy of the data
comes, significantly faster insulin analogues33 or adjunctive analysis.
therapies are likely needed.34
The second-generation AID systems have been adapted to Author Disclosure Statement
further improve glycemic management and usability, with
K.D. served on advisory boards of Novo Nordisk, Pfizer,
several algorithmic advancements, including adjustable glu-
and Sanofi. K.D. received honoraria for participation in the
cose targets, automated correction boluses, factory-calibrated
speaker’s bureau of Abbott, Eli Lilly, Medtronic, Novo
CGM technology, and updated controllers that ensure a more
Nordisk, and Pfizer. E.F.-R. reports having received speaker
robust personalization of the therapy and increased time in
honoraria from Eli Lilly, Novo Nordisk, and Merck. E.F.-R.
AID mode.1 Certain AID systems also have a meal detection
served on advisory boards for Eli Lilly and Sanofi. D.P.Z. has
feature that, if prompted, can alert the system to deliver more
received honoraria for speaking engagements from Ascensia
aggressive automated-correction doses.29,35,36
Diabetes, Insulet Canada, and Medtronic Diabetes. D.P.Z.
Importantly, during the entire study duration, participants
served on an advisory board for Dexcom and has received an
maintained a high level of AID use (*90%), reflecting the
ISPAD-JDRF Research Fellowship and research support
usability of the system, a key element to fulfilling the gly-
from Insulet and the Leona M. and Harry B. Helmsley
cemic benefits of AID use also in children and adolescents.
Charitable Trust.
The strengths of our study include the multinational,
H.S. served on advisory boards of Amarin, Amgen,
double-blind, crossover design and including children and
Boehringer Ingelheim, Eli Lilly, Novartis, Novo Nordisk, and
adolescents. In addition, there was no remote monitoring or
Sanofi. P.C. reports no personal financial disclosures but re-
close supervision, hence the assessment of glycemic out-
ports that his current employer has received consulting pay-
comes with AID use, including during free-living exercise
ments on his behalf from vTv Therapeutics, Beta Bionics,
captured with activity monitors, supports generalizability of
Dexcom, and Diasome. H.S. received honoraria for partici-
these findings.
pation in the speaker’s bureau of Amarin, Amgen, Bayer,
Our study also has limitations worth noting. Our study
Boehringer Ingelheim, Eli Lilly, Daiichi Sankyo, Novo Nor-
cohort included only participants who were already using an
disk, and Sanofi. H.S.’s institution received investigator-
insulin pump and/or CGM (including first-generation AID
initiated grant support from Boehringer Ingelheim, Eli Lilly,
system), which does not necessarily represent the broader
Novo Nordisk, MSD, and Sanofi. N.B. received honoraria for
type 1 diabetes population. Furthermore, despite the rela-
participation in the speaker’s bureau of Abbott and Medtronic.
tively favorable baseline HbA1c levels (mean HbA1c 7.5%,
T.B. served on advisory boards of Novo Nordisk, Sanofi,
range 5.9%–9.9%), the participants consistently achieved
Eli Lilly, Boehringer, Medtronic, Indigo, and DreaMed
recommended glycemic outcomes throughout the study pe-
Diabetes. T.B. received honoraria for participating in the
riod, irrespective of their diverse socioeconomic back-
speaker’s bureaux of Eli Lilly, Novo Nordisk, Medtronic,
grounds and baseline glycemic status. Finally, each study
Abbott, Sanofi, Aventis, Astra Zeneca, and Roche. T.B. owns
period was only 4 weeks, so the change in HbA1c could not
stocks of DreamMed Diabetes. T.B.’s institution received
be compared between the two treatment arms.
research grant support from Abbott, Medtronic, Novo Nor-
However, CGM metrics may provide a more comprehen-
disk, GluSense, Sanofi, Novartis, Sandoz, and Zealand
sive assessment of glycemic control.37 In addition, CGM
Pharma. No other financial disclosures were reported.
metrics calculated over 4 weeks of CGM use have been
shown to be strongly correlated with their 3-month CGM
metric, so we expect these results to be similar over a longer Funding Information
study duration.38 This was an investigator-initiated study, sponsored by the
In conclusion, this randomized, double-blind clinical trial Faculty of Medicine, University of Ljubljana. The study was
did not demonstrate superiority of FIA over SIA in physically funded in part by University Medical Centre Ljubljana Re-
active children and adolescents with type 1 diabetes using an search and Development Grant 20210205 and by Medtronic
advanced hybrid AID system. Both insulin formulations en- Research Grant ERP-2019-11958. Medtronic provided study
abled high TIR and low time above range and time below materials. K.D., N.B., and T.B. were funded in part by Slo-
range, including during periods with documented physical venian National Research Agency Grants J3-2536, J7-1820,
activity. and P3-0343.
Acknowledgments
Supplementary Material
The authors thank all the participants, nurses, and nurse
Supplementary Figure S1
educators for their dedication and commitment to this study.
Supplementary Table S1
Supplementary Table S2
Authors’ Contributions
Supplementary Table S3
K.D., S.B., D.P.Z., and T.B. drafted the article. S.B., P.C., Supplementary Table S4
and C.K. did the statistical analysis and verified the under- Supplementary Table S5
lying data. K.D., S.B., E.F.-R, D.P.Z., N.P., A.M., Z.L., P.C., Supplementary Table S6
M.F., H.S., N.B., C.K., and T.B. reviewed and edited the Supplementary Table S7
620 DOVC ET AL.

Supplementary Table S8 Diabetes Care 2019;42(7):1255–1262; doi: 10.2337/dc19-

Supplementary Table S9 0009
Supplementary Table S10 15. Heise T, Hövelmann U, Zijlstra E, et al. A comparison of
Supplementary Table S11 pharmacokinetic and pharmacodynamic properties between
Supplementary Table S12 faster-acting insulin aspart and insulin aspart in elderly
Supplementary Table S13 subjects with type 1 diabetes mellitus. Drugs Aging 2017;
34(1):29–38; doi: 10.1007/s40266-016-0418-6
16. Heise T, Stender-Petersen K, Hövelmann U, et al. Phar-
References
macokinetic and pharmacodynamic properties of faster-
1. Bergenstal RM, Nimri R, Beck RW, et al. A comparison of acting insulin aspart versus insulin aspart across a clinically
two hybrid closed-loop systems in adolescents and young relevant dose range in subjects with type 1 diabetes mel-
adults with type 1 diabetes (FLAIR): A multicentre ran- litus. Clin Pharmacokinet 2017;56(6):649–660; doi: 10
domised crossover trial. Lancet 2021;397(10270):208–219; .1007/s40262-016-0473-5
doi: 10.1016/s0140-6736(20)32514-9 17. Dovc K, Piona C, Yesxiltepe Mutlu G, et al. Faster compared
2. Boughton CK, Allen JM, Ware J, et al. Closed-loop therapy with standard insulin aspart during day-and-night fully
and preservation of C-peptide secretion in type 1 diabetes. closed-loop insulin therapy in type 1 diabetes: A double-
N Engl J Med 2022;387(10):882–893; doi: 10.1056/ blind randomized crossover trial. Diabetes Care 2020;
NEJMoa2203496 43(1):29–36; doi: 10.2337/dc19-0895
3. Brown SA, Kovatchev BP, Raghinaru D, et al. Six-month 18. Beck RW, Russell SJ, Damiano ER, et al. A multicenter
randomized, multicenter trial of closed-loop control in type randomized trial evaluating fast-acting insulin aspart in the
1 diabetes. N Engl J Med 2019;381(18):1707–1717; doi: 10 bionic pancreas in adults with type 1 diabetes. Diabetes
.1056/NEJMoa1907863 Technol Ther 2022;24(10):681–696; doi: 10.1089/dia.2022
4. Burnside MJ, Lewis DM, Crocket HR, et al. Open-source .0167
automated insulin delivery in type 1 diabetes. N Engl J Med 19. Boughton CK, Hartnell S, Thabit H, et al. Hybrid closed-
2022;387(10):869–881; doi: 10.1056/NEJMoa2203913 loop glucose control with faster insulin aspart compared
5. Russell SJ, Beck RW, Damiano ER, et al. Multicenter, with standard insulin aspart in adults with type 1 diabetes:
randomized trial of a bionic pancreas in type 1 diabetes. A double-blind, multicentre, multinational, randomized,
N Engl J Med 2022;387(13):1161–1172; doi: 10.1056/ crossover study. Diabetes Obes Metab 2021;23(6):1389–
NEJMoa2205225 1396; doi: 10.1111/dom.14355
6. Ware J, Allen JM, Boughton CK, et al. Randomized trial of 20. Lee MH, Paldus B, Vogrin S, et al. Fast-acting insulin
closed-loop control in very young children with type 1 aspart versus insulin aspart using a second-generation hy-
diabetes. N Engl J Med 2022;386(3):209–219; doi: 10 brid closed-loop system in adults with type 1 diabetes: A
.1056/NEJMoa2111673 randomized, open-label, crossover trial. Diabetes Care
7. ElSayed NA, Aleppo G, Aroda VR, et al. 7. Diabetes 2021; doi: 10.2337/dc21-0814
technology: Standards of care in diabetes-2023. Diabetes 21. Ware J, Allen JM, Boughton CK, et al. Hybrid closed-loop
Care 2023;46(Suppl 1):S111–s127; doi: 10.2337/dc23-S007 with faster insulin aspart compared with standard insulin
8. Sherr JL, Schoelwer M, Dos Santos TJ, et al. ISPAD aspart in very young children with type 1 diabetes: A
clinical practice consensus guidelines 2022: Diabetes double-blind, multicenter, randomized, crossover study.
technologies: Insulin delivery. Pediatr Diabetes 2022;23(8): Diabetes Technol Ther. 2023;25(6):431–436. doi: 10.1089/
1406–1431; doi: 10.1111/pedi.13421 dia.2023.0042
9. Battelino T, Danne T, Bergenstal RM, et al. Clinical targets 22. Adolfsson P, Taplin CE, Zaharieva DP, et al. ISPAD
for continuous glucose monitoring data interpretation: Re- Clinical Practice Consensus Guidelines 2022: Exercise in
commendations from the international consensus on time in children and adolescents with diabetes. Pediatr Diabetes
range. Diabetes care 2019;42(8):1593–1603; doi: 10.2337/ 2022;23(8):1341–1372; doi: 10.1111/pedi.13452
dci19-0028 23. Riddell MC, Peters AL. Exercise in adults with type 1 di-
10. Ceriello A, Prattichizzo F, Phillip M, et al. Glycaemic abetes mellitus. Nat Rev Endocrinol 2023;19(2):98–111;
management in diabetes: Old and new approaches. Lancet doi: 10.1038/s41574-022-00756-6
Diabetes Endocrinol 2022;10(1):75–84; doi: 10.1016/ 24. Abraham MB, Karges B, Dovc K, et al. ISPAD clinical
s2213-8587(21)00245-x practice consensus guidelines 2022: Assessment and man-
11. Phillip M, Nimri R, Bergenstal RM, et al. Consensus rec- agement of hypoglycemia in children and adolescents with
ommendations for the use of automated insulin delivery diabetes. Pediatr Diabetes 2022;23(8):1322–1340; doi: 10
(AID) technologies in clinical practice. Endocr Rev 2022; .1111/pedi.13443
doi: 10.1210/endrev/bnac022 25. Glaser N, Fritsch M, Priyambada L, et al. ISPAD clinical
12. Kowalski A. Pathway to artificial pancreas systems re- practice consensus guidelines 2022: Diabetic ketoacidosis
visited: Moving downstream. Diabetes Care 2015;38(6): and hyperglycemic hyperosmolar state. Pediatr Diabetes
1036–1043; doi: 10.2337/dc15-0364 2022;23(7):835–856; doi: 10.1111/pedi.13406
13. Kildegaard J, Buckley ST, Nielsen RH, et al. Elucidating 26. Avgerinos I, Papanastasiou G, Karagiannis T, et al. Ultra-
the mechanism of absorption of fast-acting insulin aspart: rapid-acting insulins for adults with diabetes: A systematic
The role of niacinamide. Pharm Res 2019;36(3):49; doi: 10 review and meta-analysis. Diabetes Obes Metab 2021;
.1007/s11095-019-2578-7 23(10):2395–2401; doi: 10.1111/dom.14461
14. Bode BW, Iotova V, Kovarenko M, et al. Efficacy and 27. Hsu L, Buckingham B, Basina M, et al. Fast-acting insu-
safety of fast-acting insulin aspart compared with insulin lin aspart use with the MiniMed(TM) 670G system. Dia-
aspart, both in combination with insulin degludec, in chil- betes Technol Ther 2021;23(1):1–7; doi: 10.1089/dia.2020
dren and adolescents with type 1 diabetes: The onset 7 trial. .0083
FACT STUDY 621

28. Ozer K, Cooper AM, Ahn LP, et al. Fast acting insulin betes: A single-centre, open-label, randomised controlled,
aspart compared with insulin aspart in the Medtronic 670G crossover, non-inferiority trial. Lancet Digit Health 2021;
hybrid closed loop system in type 1 diabetes: An open label 3(11):e723–e732; doi: 10.1016/s2589-7500(21)00139-4
crossover study. Diabetes Technol Ther 2021;23(4):286– 35. Phillip M, Battelino T, Atlas E, et al. Nocturnal glucose
292; doi: 10.1089/dia.2020.0500 control with an artificial pancreas at a diabetes camp.
29. Stamati A, Karagiannis T, Tsapas A, et al. Efficacy and N Engl J Med 2013;368(9):824–833; doi: 10.1056/
safety of ultra-rapid insulin analogues in insulin pumps in NEJMoa1206881
patients with type 1 diabetes mellitus: A systematic review 36. Weinzimer SA, Bailey RJ, Bergenstal RM, et al. A compari-
and meta-analysis. Diabetes Res Clin Pract 2022; son of postprandial glucose control in the medtronic advanced
193(110144; doi: 10.1016/j.diabres.2022.110144 hybrid closed-loop system versus 670G. Diabetes Technol
30. Dovc K, Battelino T, Beck RW, et al. Impact of temporary Ther 2022;24(8):573–582; doi: 10.1089/dia.2021.0568
glycemic target use in the hybrid and advanced hybrid 37. Rodbard D. Metrics to evaluate quality of glycemic control:
closed-loop systems. Diabetes Technol Ther 2022;24(11): comparison of time in target, hypoglycemic, and hyper-
848–852; doi: 10.1089/dia.2022.0153 glycemic ranges with ‘‘risk indices’’. Diabetes Technol
31. Ekhlaspour L, Forlenza GP, Chernavvsky D, et al. Closed Ther 2018;20(5):325–334; doi: 10.1089/dia.2017.0416
loop control in adolescents and children during winter 38. Riddlesworth TD, Beck RW, Gal RL, et al. Optimal sam-
sports: Use of the tandem control-IQ AP system. Pediatr pling duration for continuous glucose monitoring to deter-
Diabetes 2019;20(6):759–768; doi: 10.1111/pedi.12867 mine long-term glycemic control. Diabetes Technol Ther
32. Morrison D, Zaharieva DP, Lee MH, et al. Comparable 2018;20(4):314–316; doi: 10.1089/dia.2017.0455
glucose control with fast-acting insulin aspart versus insulin
aspart using a second-generation hybrid closed-loop system Address correspondence to:
during exercise. Diabetes Technol Ther 2022;24(2):93– Tadej Battelino, MD
101; doi: 10.1089/dia.2021.0221 Department of Endocrinology
33. Svehlikova E, Mursic I, Augustin T, et al. Pharmacoki- Diabetes and Metabolism
netics and pharmacodynamics of three different formula- University Children’s Hospital
tions of insulin aspart: A randomized, double-blind, Bohoriceva 20
crossover study in men with type 1 diabetes. Diabetes Care Ljubljana 1000
2021;44(2):448–455; doi: 10.2337/dc20-1017 Slovenia
34. Tsoukas MA, Majdpour D, Yale JF, et al. A fully artificial
pancreas versus a hybrid artificial pancreas for type 1 dia- E-mail: [email protected]