516 Diabetes Care Volume 37, February 2014

Annett Stahn,1 Frank Pistrosch,1,2

Relationship Between Xenia Ganz,1,2 Madlen Teige,2
Carsta Koehler,3 Stefan Bornstein,1
Hypoglycemic Episodes and and Markolf Hanefeld1,2

Ventricular Arrhythmias in
Patients With Type 2 Diabetes and
Cardiovascular Diseases: Silent
Hypoglycemias and Silent
Arrhythmias

OBJECTIVE
In patients with type 2 diabetes and cardiovascular diseases (CVDs), intensive
treatment with insulin and/or sulfonylurea (SU) may be associated with excessive
increased risk of hypoglycemic episodes. To evaluate the risk of critical arrhythmias
related to glycemic variability, we carried out an observational study in type 2
diabetes patients with CVD.

RESEARCH DESIGN AND METHODS

Thirty patients with type 2 diabetes and documented CVD who had been treated
CARDIOVASCULAR AND METABOLIC RISK

with insulin and/or SU underwent 5 days of monitoring with a continuous glucose

measurement system along with parallel electrocardiogram recording for moni-
toring of ventricular arrhythmias. Twelve age-matched patients with documented
CVD who received treatment with metformin and/or dipeptidyl peptidase-4 in-
hibitor served as the control group. Patients were receiving stable treatment,
and were instructed to notice symptoms of arrhythmias and hypoglycemia,
respectively. 1
Universitätsklinikum Carl Gustav Carus, MK III,
RESULTS Dresden, Germany
2
GWT-TUD GmbH, Study Center Professor
We observed a high incidence of asymptomatic severe episodes of hypoglycemia Hanefeld, Dresden, Germany
3
(<3.1 mmol/L) in patients receiving treatment with insulin and/or SU, whereas GWT-TUD GmbH, Medical Consulting, Dresden,
severe hypoglycemia did not develop in any of the control subjects. Patients with Germany
severe hypoglycemia (n = 12) had a higher number of severe ventricular Corresponding author: Markolf Hanefeld,
[email protected].
arrhythmias (patients with versus without severe hypoglycemia, respectively:
ventricular couplets 41.7 6 81.8 vs. 5.5 6 16.7; ventricular tachycardia 1.0 6 1.9 Received 24 March 2013 and accepted 5
September 2013.
vs. 0.1 6 0.3). No direct correlation could be found among different variables of
This article contains Supplementary Data online
glucose profile, corrected QT interval, and ventricular arrhythmias. at http://care.diabetesjournals.org/lookup/
suppl/doi:10.2337/dc13-0600/-/DC1.
CONCLUSIONS
A.S. and F.P. contributed equally to this study.
Our results suggest that severe episodes of hypoglycemia are associated with an
© 2014 by the American Diabetes Association.
increased risk of severe ventricular arrhythmias. See http://creativecommons.org/licenses/by-
Diabetes Care 2014;37:516–520 | DOI: 10.2337/dc13-0600 nc-nd/3.0/ for details.
care.diabetesjournals.org Stahn and Associates 517

Patients with type 2 diabetes and relationship between these events in treated with sulfonylurea, and 9
cardiovascular diseases (CVDs) may type 2 diabetes patients with CVD. patients (30%) were treated with a
represent a vulnerable high-risk group Therefore, we used continuous glucose combination of insulin and sulfonylurea
for hypoglycemic episodes (HEs) measurement and continuous (comedications: anti-hypertensives
associated with arrhythmias and sudden electrocardiogram (ECG) monitoring in 90%, b-blockers 70%, statins 93%). The
death. parallel for ;5 days in patients with following cardiovascular comorbidities
In patients with type 1 diabetes, sudden type 2 diabetes, who had been treated were diagnosed: coronary heart disease
nocturnal death is described as “dead- with insulin and/or sulfonylurea, and 19 patients; combination of coronary
in-bed” syndrome. A cause for this could documented CVD to investigate this in a heart disease and stroke or peripheral
be QTc prolongation leading to fatal cohort of frail patients. arterial disease 4 patients;
ventricular arrhythmias due to long cerebrovascular disease 4 patients;
RESEARCH DESIGN AND METHODS stroke 2 patients; and peripheral arterial
episodes of severe nocturnal
hypoglycemia (1). Methods disease 1 patients. Twelve age-matched
Inclusion criteria were as follows: type 2 patients with type 2 diabetes and
Diverse case reports about diabetes; age 50–80 years; proven CVD, confirmed CVD served as the control
hypoglycemia and cardiac events are including coronary heart disease, group. These patients were treated with
described in the literature (2,3), but it is stroke, peripheral arterial disease, or anti-hyperglycemic agents that do not
still difficult to document these events cerebrovascular disease; and insulin increase the risk of hypoglycemia: five
in parallel. Moreover, many episodes and/or sulfonylurea treatment. patients received metformin, and seven
with critical low glucose levels may be patients received metformin in
Exclusion criteria were as follows: type 1
asymptomatic, and arrhythmias will be combination with a dipeptidyl
diabetes; ECG changes like AV-Block II8
unrecognized during sleep. peptidase-4 inhibitor. Eight patients
and III8, or atrial fibrillation; implanted
Pathophysiological studies, partly with pacemaker or defibrillator; and (66%) used a b-blocker, and all of them
clamps, showed correlations between treatment with antiarrhythmic drugs received treatment with statins. The
hypoglycemia and an increase of besides b-blockers and calcium channel following cardiovascular comorbidities
hormones like epinephrine and blockers. Patients were excluded from were diagnosed: coronary heart disease,
norepinephrine, which induce the study if a blood sample, obtained at 8 patients; combination of coronary
vasoconstriction, platelet aggregation, baseline, showed hyponatremia/ heart disease and stroke or peripheral
and thereby ischemia (4,5). Therefore, hypernatremia, hypokalemia/ artery disease, 1; cerebrovascular
the question arose of whether strict hyperkalemia, or hypothyroidism/ disease, 2; and peripheral artery
glucose control with insulin and/or hyperthyroidism. disease, 1.
sulfonylurea, which is associated with The baseline characteristics of both
We recorded in parallel interstitial
increased risk of hypoglycemia and groups are shown in Table 1.
glucose (IG) concentrations with
excessive glucose fluctuations, may
continuous glucose measurement
have harmful effects on electric stability Target Parameters
system (CGMS) and long-term ECG for
and myocardial blood flow. We calculated the following parameters
5 days.
In the ACCORD (Action to Control with CGMS: average IG and SD; mean
IG levels were measured every 5 min amplitude of glucose excursions
Cardiovascular Risk in Diabetes) trial,
with CGMS Medtronic MiniMed Gold, (MAGE); area under the curve (AUC);
excessive mortality in the group
and at day 2 patients ingested a minimum and maximum IG; frequency
receiving intensified glucose-lowering
standardized test meal. For ECG and time of HE. We calculated the
treatment has been discussed with
analysis, we used the Amedtec ECGpro following parameters with ECG: heart
respect to hypoglycemia being the
system. IG levels between 3.1 and 3.9 rate (in beats per minute [bpm]); QTc
cause for cardiovascular events (6).
mmol/L were defined as mild HEs, and time; ventricular extrasystoles (VESs);
Severe, but not mild, hypoglycemia was
IG levels ,3.1 mmol/L IG were defined couplets; triplets; and ventricular
associated with an odds ratio of 3.4 in a
as severe episodes. Patients were tachycardias (VTs). Heart rate variability
recently published retrospective
instructed to notice all symptoms of was assessed as the SD of normal RR
observational study after adjustment
hypoglycemia and arrhythmias with intervals (SDNNs).
for major risk factors and Charlson
date and time in a predefined protocol
comorbidity index (7). A recently
during CGMS and ECG measurement. RESULTS
published retrospective register study in
veterans reported a hazard ratio of 2.00 Patient Characteristics As shown in Table 1, both groups were
for cardiovascular events, and 1.76 for Thirty patients with type 2 diabetes and well-balanced for HbA1c, blood
microvascular events for patients with confirmed CVD, HbA1c levels ,9% pressure, parameters of lipid
HEs (8). (75 mmol/mol), and age 56–80 years, metabolism, and renal function.
Little is known about the frequency of who had been treated with insulin Average IG as well as maximum IG, SD of
asymptomatic hypoglycemia, in and/or sulfonylurea, were considered. IG values, MAGE, or the 24 h IG AUC
particular nocturnal hypoglycemia; Seventeen patients (57%) were treated were not different between the group at
frequency of silent arrhythmias; and the with insulin, 4 patients (13%) were risk and the control group (Table 1);
518 Silent Hypoglycemias and Arrhythmias Diabetes Care Volume 37, February 2014

Table 1—Baseline characteristics and parameters of glycemic control during P , 0.05). During the whole recording
5 days of CGMS recording time, patients in the group at risk
Characteristics Group Mean SD P value Range spent 39.5 6 70.5 min with an IG level
,3.1 mmol/L, and 156 6 255 min with
Age (years) 1 67.6 6.4 0.540 56–80
2 66.2 7.0 51–77 an IG level ,3.9 mmol/L (Table 1).
HbA1c (%) 1 7.3 0.8 0.349 5.9–8.9 However, only 14 mild symptomatic
2 7.6 0.7 6.6–9.0 hypoglycemias and 1 serious
HbA1c (mmol/mol) 1 56 6.1 0.349 41–74 hypoglycemia were noticed by the
2 60 5.5 51–75 patients. Silent HEs occurred more often
Systolic BP (mmHg) 1 146.2 20.7 0.508 105–183 at night (26 episodes at night vs. 11 during
2 142.0 10.1 125–155 the daytime).
Diastolic BP (mmHg) 1 78.8 9.2 0.528 61–98
In the group at risk, 28 patients had
2 80.6 4.6 73–90
VESs, but 11 control subjects also had
Sodium (mmol/L) 1 138.1 2.9 0.684 132.4–144
2 137.7 3.2 132.5–143.2
VESs (Table 2). We observed, however,
Potassium (mmol/L) 1 4.42 0.36 0.727 3.6–4.9
multiple ventricular arrhythmias
2 4.38 0.39 3.77–5.06 (couplets 17 patients [56.7%]; triplets 10
Cholesterol (mmol/L) 1 4.7 0.9 0.940 3.2–7.0 patients [33.3%]; VTs 5 patients [16.7%])
2 4.7 1.3 2.7–6.1 in the high-risk group (Table 2). The
HDL cholesterol (mmol/L) 1 1.2 0.3 0.180 0.7–1.8 longest VT continued for about 9 beats.
2 1.4 0.4 0.7–2.1 We found a maximum of 258 couplets
LDL cholesterol (mmol/L) 1 2.6 0.7 0.942 1.0–4.0 per patient, 32 triplets per patient, and
2 2.5 0.9 1.3–3.6 6 VTs per patient. However, only one
Triglycerides (mmol/L) 1 2.4 1.7 0.052 0.4–7.9 patient noticed cardiac arrhythmias. No
2 1.4 0.3 0.6–1.7 VT was observed among control
Creatinine (mmol/L) 1 82.3 21.3 0.280 68–124 subjects, and triplets were distinctly
2 76.4 14.0 49–98
rarer among controls. SDNN as a marker
Mean IG 1 8.20 1.67 0.422 5.6–11.4
of cardiac autonomic neuropathy was
2 8.63 1.15 5.9–10.3
not significantly different between
Maximal IG (mmol/L) 1 15.7 3.3 0.578 9.5–22.2
2 15.1 2.2 11.7–17.9
patients in the high-risk group versus
Minimal IG (mmol/L) 1 3.28 0.9 0.000 2.2–6.0
subjects in the control group (Table 2).
2 5.01 1.0 3.7–6.7 Furthermore, we did not find a
SD of IG (mmol/L) 1 2.33 0.78 0.148 1.3–4.3 significant correlation between SDNNs
2 1.97 0.55 1.2–3.0 and ventricular arrhythmias.
AUC_D2 (mmol/L-1 h) 1 2,368 613 0.129 1,630–3,869 We divided the whole study population
2 2,670 438 1,714–3,479 into patients in whom severe HEs
MAGE (mmol/L) 1 4.9 1.5 0.649 2.78–9.42 developed (n = 12) and patients in
2 4.5 1.7 2.71–6.55
whom severe HEs did not develop
HEs per patient with IG ,3.1 mmol/L 1 1.0 1.6 0.049 0–6
(n = 30), independent from their
2 0 0 0
glucose-lowering medication (Table 3).
HEs per patient with IG ,3.9 mmol/L 1 2.6 3.1 0.010 0–12
2 0.2 0.4 0–1 Again, we found a significantly higher
Time spent with IG ,3.1 mmol/L (min) 1 39.5 70.5 0.008 0–240
frequency of severe cardiac arrhythmias
2 0 0 0 such as couplets or VTs in patients with
Time spent with IG ,3.9 mmol/L (min) 1 156 255 0.047 0–996 severe HEs.
2 2.9 7.5 0–25 Furthermore, we analyzed the
Distribution of HEs* relationship between markers of
Daytime (6:00 A.M.–10:00 P.M.): 11
glucose fluctuations and ventricular
Nocturnal (10:00 P.M.–6:00 A.M.): 26
arrhythmias in the high-risk group. The
Group 1, high-risk group (29 men, 1 woman); group 2, control group (9 men, 3 women). SD of IG levels was not related to VESs or
AUC_D2, AUC for glucose at recording day 2. x2 test for sex: P = 0.063. *,3.1 mmol/L for
group 1. more severe arrhythmias. For MAGE, we
calculated the following tertiles: ,4.02
mmol/L, 4.02–5.61 mmol/L, and .5.61
however, the minimum IG was per patient in the group at risk. We did mmol/L. The duration of severe HEs
significantly lower in the group at risk not detect any severe HEs in the control (IG , 3.1 mmol/L) was classified into 0,
for hypoglycemia (3.28 6 0.9 vs. 5.01 6 group (Table 1). Furthermore, we 1–35, and .35 min. We found a
1.0 mmol/L, P , 0.01). found a higher frequency of mild HEs in significantly higher incidence of VESs in
Over 5 days, the average number of HEs patients at risk compared with the patients with a MAGE .5.61 mmol/L
with an IG ,3.1 mmol/L was 1.0 6 1.6 control group (2.6 6 3.1 vs. 0.2 6 0.4 HEs, and severe HEs of .35 min (P = 0.008)
care.diabetesjournals.org Stahn and Associates 519

Table 2—ECG parameters during 5 days of parallel recording

Parameters Group Mean SD P value Range Comment
Mean heart rate (bpm) 1 69.4 8.6 0.096 55–89
2 78.0 15.8 64–96
Minimal heart rate (bpm) 1 53.1 8.9 0.603 43–78
2 54.5 3.9 46–60
Maximal heart rate (bpm) 1 110.5 17.7 0.293 79–159
2 117.6 23.9 72–154
Mean QTc (ms) 1 376.9 48.7 0.531 300–475
2 388.0 50.2 307–445
SDNNs (ms) 1 121.2 36.2 0.448 55–225
2 112.6 22 84–184
SVESs per patient (n) 1 633 1,345 0.458 6–6,938
2 322 764 0–2,723
VESs per patient (n) 1 3,607 7,977 0.027 1–35,328 28 patients in group 1
2 144 217 0–732 11 patients in group 2
Couplets per patient (n) 1 20.5 53.3 0.054 0–258 17 patients (56.7%) in group 1
2 0.9 1.2 0–4 6 patients (50%) in group 2
Triplets per patient (n) 1 2.2 6.3 0.080 0–32 10 patients (33.3%) in group 1
2 0.1 0.3 0–1 1 patient (8.3%) in group 2
VTs per patient (n) 1 0.5 1.3 0.05 0–5 5 patients (16.7%) in group 1
2 0 0 0 None in group 2
Group 1, high-risk group; group 2, control group; SVES, supra-VES.

(Supplementary Table 1). More severe silent severe ventricular arrhythmias in but only one patient reported
ventricular arrhythmias were, however, CVD patients with type 2 diabetes who arrhythmias. Ten patients (33.3%) had
not different between tertiles of MAGE were treated with insulin substitution triplets, and 5 patients (16.7%) had VTs.
(data not shown). Correlation analysis therapy and/or sulfonylurea. Although These silent ventricular arrhythmias also
did not show a direct relationship only 1 serious HE and 14 mild occurred more at night than at daytime.
between parameters of glycemic symptomatic HEs were reported from In the ACCORD and ADVANCE studies,
variability, quality of diabetes control, 30 participants, we found a high severe hypoglycemia did not show a
and risk of severe ventricular incidence of asymptomatic episodes direct relationship to cardiovascular
arrhythmias in this heterogeneous with IG levels ,3.1 mmol/L (one HE per events and mortality (6,9). Thus, the
population. Case reports are patient per 5 days). Twice as many HEs question was whether hypoglycemia
represented in Supplementary Figs. 1 were recognized at night as during the labels a vulnerable group of patients,
and 2. daytime. This points to the fact that rather than the cause of death. Some
nocturnal HEs are under-reported but investigators even suggest that mild
CONCLUSIONS may be associated with serious hypoglycemia has no harmful effect on
Our investigations of parallel recording cardiovascular complications. Even cardiovascular outcome (ischemic
of CGMS and ECG reveal a high more critical, we detected a high conditioning). ECG recording in the
incidence of both asymptomatic incidence of asymptomatic serious dead-in-bed syndrome in patients with
episodes of critically low IG levels and ventricular arrhythmias in our patients, type 1 diabetes without CVD revealed
an increase in QTc time and rhythm
disturbances only for long-lasting HEs of
Table 3—Relationship between severe HEs and cardiac arrhythmia during .30 to 150 min (1). Accordingly, we
5 days of parallel recording
found a significantly higher number of
Outcomes HE Mean SD P value
VTs in patients with HEs ,3.1 mmol/L
VESs per patient (n) Yes 3,377 7,219 0.688 (Table 3). Of notice, the first data in
No 2,371 6,416 patients with heart failure and CGMS
Couplets per patient (n) Yes 41.7 81.8 0.024 recording have shown that rapid
No 5.5 16.7 glycemic fluctuations calculated as
Triplets per patient (n) Yes 2.36 4.3 0.597 MAGE are a significant risk factor for
No 1.33 5.8
arrhythmias and all-cause mortality if
VTs per patient (n) Yes 1.0 1.9 0.017
MAGE exceeds 5 mmol/L (10). In our
No 0.1 0.3
high-risk group who had been treated
HE “Yes,” n = 12; HE “No,” n = 30. with insulin and/or sulfonylurea, we
found significantly more VESs in
520 Silent Hypoglycemias and Arrhythmias Diabetes Care Volume 37, February 2014

patients with MAGE .5.61, patient-centered treatment decision. counter-regulatory responses to

representing the highest tertile of Prospective trials with parallel recording hypoglycaemia. Clin Sci (Lond) 2000;99:
351–362
MAGE (Supplementary Table 1). For this of CGMS and ECG are urgently needed
high range of fluctuations, Monnier to evaluate the predictors of harmful 5. Heller SR. Abnormalities of the
electrocardiogram during hypoglycaemia:
et al. (11) could show increased cardiovascular events of intensive the cause of the dead in bed syndrome? Int
oxidative stress. Oxidative stress is likely glucose control. J Clin Pract Suppl 2002;129:27–32
to predispose patients to atrial 6. Gerstein HC, Miller ME, Byington RP, et al.;
fibrillation, a frequent finding in patients Action to Control Cardiovascular Risk in
with type 2 diabetes (12). High MAGE Acknowledgments. The authors thank Diabetes Study Group. Effects of intensive
was, however, not associated with a Medtronic for supplying the continuous glucose glucose lowering in type 2 diabetes. N Engl J
significantly increased number of monitoring system equipment. Med 2008;358:2545–2559
triplets and VTs in our patients with CVD Funding. This study was funded by the 7. McCoy RG, Van Houten HK, Ziegenfuss JY,
Shah ND, Wermers RA, Smith SA. Increased
but without symptoms of severe heart researchers.
mortality of patients with diabetes
failure. Mechanistic studies of Duality of Interest. No potential conflicts of reporting severe hypoglycemia. Diabetes
hyperinsulinemic hypoglycemia in interest relevant to this study were reported. Care 2012;35:1897–1901
healthy subjects have shown a Author Contributions. A.S. and F.P. designed 8. Zhao Y, Campbell CR, Fonseca V, Shi L.
prolongation of QTc time and the and coordinated the project, collected and Impact of hypoglycemia associated with
production of proarrhythmogenic analyzed data, and wrote and reviewed the antihyperglycemic medications on vascular
manuscript. X.G. and M.T. collected and risks in veterans with type 2 diabetes.
catecholamines. Thus, sympathetic analyzed data. C.K. performed statistical Diabetes Care 2012;35:1126–1132
activation by rapid fluctuations and analysis. S.B. reviewed the manuscript. M.H.
critically low glucose levels may be an designed the project and wrote and reviewed 9. Patel A, MacMahon S, Chalmers J, et al.;
the manuscript. M.H. is the guarantor of this ADVANCE Collaborative Group. Intensive
important link in provoking severe blood glucose control and vascular
arrhythmias (13). However, we did not work and, as such, had full access to all the data
in the study and takes responsibility for the outcomes in patients with type 2 diabetes.
find a correlation between QTc and HEs integrity of the data and the accuracy of the N Engl J Med 2008;358:2560–2572
in our study population. Another data analysis. 10. Pochinka I, Strongin L, Struchkova J.
important mechanism in the Prior Presentation. Parts of this study were Glycaemic variability and ventricular
pathogenesis of critical arrhythmias not presented in poster form at the World Diabetes cardiac arrhythmias in type 2 diabetic
considered in our study is the Congress 2011 of the International Diabetes patients with chronic heart failure.
Federation, Dubai, United Arab Emirates, 4–8 Diabetologia 2012;55(Suppl. 1):S102
association of HEs and cardiac ischemia.
December 2011. 11. Monnier L, Mas E, Ginet C, et al. Activation
As described by Desouza et al. (14),
of oxidative stress by acute glucose
cardiac ischemia is particularly common References fluctuations compared with sustained
in patients with high fluctuations 1. Gill GV, Woodward A, Casson IF, Weston PJ. chronic hyperglycemia in patients with
(MAGE) in blood glucose who are Cardiac arrhythmia and nocturnal type 2 diabetes. JAMA 2006;295:1681–
receiving treatment with insulin. This hypoglycaemia in type 1 diabetesdthe 1687
“dead in bed” syndrome revisited.
may explain why in our patients with 12. Neuman RB, Bloom HL, Shukrullah I, et al.
Diabetologia 2009;52:42–45 Oxidative stress markers are associated
pre-existing CVD a high MAGE during
2. Frier BM, Barr StCG, Walker JD. Fatal with persistent atrial fibrillation. Clin Chem
longer periods of critically low glucose
cardiac arrest following acute 2007;53:1652–1657
levels was associated with a higher rate hypoglycaemia in a diabetic patient. Pract 13. Laitinen T, Lyyra-Laitinen T, Huopio H, et al.
of VESs. Diabetes Int 1995;12:284 Electrocardiographic alterations during
In conclusion, patients with type 2 3. Tanenberg RJ, Newton CA, Drake AJ. hyperinsulinemic hypoglycemia in healthy
diabetes and CVD who receive Confirmation of hypoglycemia in the subjects. Ann Noninvasive Electrocardiol
“dead-in-bed” syndrome, as captured by a 2008;13:97–105
treatment with insulin and/or retrospective continuous glucose
sulfonylurea exhibit a high incidence of 14. Desouza C, Salazar H, Cheong B, Murgo J,
monitoring system. Endocr Pract 2010;16: Fonseca V. Association of hypoglycemia
both asymptomatic HEs and silent 244–248 and cardiac ischemia: a study based on
severe arrhythmias, particularly at 4. Galassetti P, Davis SN. Effects of insulin per continuous monitoring. Diabetes Care
bedtime. This has to be considered in a se on neuroendocrine and metabolic 2003;26:1485–1489