Correlation of QT interval and heart rate variability in type 2 diabetic patients: Possible evidence for autonomic regulation of QT interval

Corresponding Author:

Toru Maruyama
Campus Life Health Center
Kyushu University
Fukuoka, Japan
E-mail: maruyama@artsci.kyushu-u.ac.jp

Abstract

Objective: QT interval length in Electrocardiogram (ECG) has prognostic impact on the patients with diabetes. This interval is influenced by autonomic nervous function, which is, however, impaired in diabetic patients. Since exact mechanisms of autonomic regulation of QT interval in diabetics remains unclear, this study aimed to investigate the relationship between heartrate- corrected QT interval (QTc interval) and Heart Rate Variability (HRV) in diabetic patients. Methods: Coefficient of variance of R-R intervals (CVRR:%) in digital ECG, as a representative time-domain measure of HRV, was estimated immediately after recording 12-lead ECG in type 2 diabetic (n=60) and nondiabetic control (n=62) groups. QTc interval was obtained by Bazett’s formula where QTc=QT/RR1/2. Demographic and laboratory data were extracted from medical records. Significant contributors of multiple clinical factors to the QTc interval length were analyzed by multivariate analysis. Results: Body mass index (p<0.001) and CVRR (p=0.015) contributed significantly to the QTc interval regulation in diabetic group, whereas CVRR (p=0.037) alone contributed significantly to this interval in controls. However, CVRR was correlated with the QTc interval negatively in diabetic group (standardized ÃŽ²=-0.306), whereas the opposite was found in the case of control group (standardized ÃŽ²=0.266). Conclusion: Although this allows a cross-sectional case-control study design, positive correlation of BMI and QTc interval in diabetic patients indicates cardiometabolic linkage (r=0.339, p=0.008). CVRR was considered to regulate QTc intervals. However, the details of this autonomic regulation in diabetics differed from those in controls. The mechanisms of autonomic QTc regulation in diabetics require future investigation.

Keywords

■ autonomic nerve ■diabetes ■heart rate variability ■QTc interval

Introduction

Type 2 diabetes mellitus is a global healthcare burden, because the prevalence of this common disease is increasing significantly. Urbanization and economic trend are the two major factors influencing the prevalence of diabetes, which differs among several population groups. There is a close link between diabetes and cardiovascular diseases, and Electrocardiogram (ECG) is a routine cardiac examination [1,2]. QT interval in ECG is measured from the beginning of QRS complex to the end of the T wave, as a measure of ‘electrical’ ventricular contraction time, and is highly dependent on heart rate. Therefore, QT interval length corrected by heart rate (QTc interval) is used in clinical practice. This interval is easily accessible by digital ECG as a simple, low-cost measure to predict all-cause and cardiovascular mortality in general population by a meta-analysis [3]. In patients with type 2 diabetes, this ECG parameter has prognostic impact with respect to all-cause mortality and future adverse events such as myocardial infarction, heart failure and stroke [4,5]. Therefore, the QTc interval is used to predict the future cardiovascular risk in diabetic patients [6]. However, this interval is regulated strictly by many factors such as electrolyte balance, sex hormone and autonomic nervous function that is impaired in diabetics. Autonomic QTc interval regulation is still a challenging topic, because digital ECG evaluates QTc interval automatically, and current methodologies estimating autonomic nerve activity are advanced. Therefore, the present study aimed to investigate the correlation between the Heart Rate Variability (HRV) reflecting autonomic nerve function and QTc intervals in diabetics and nondiabetics separately, because this regulatory system may be different each other.

Methods

■ Selection of subjects

The present study was performed from August 2016 to November 2018. Sixty Japanese patients with type 2 diabetes mellitus and 62 non-diabetic control subjects were enrolled into the present study. Diabetic patients were under the monthly visit to the outpatient division of BOOCS Clinic (Fukuoka, Japan). Control subjects were non diabetics undergoing annual health-check program in this clinic, although some of them had life-style-related diseases other than type 2 diabetes. Type 2 diabetes was defined as fasting serum glucose >126 mg/dl, casual serum glucose >200 mg/dl, hemoglobin A1c (%) estimated according to the National Glycohemoglobin Standardization Program (NGSP) >6.5% and/ or current antidiabetic medication. Diabetic patients were treated under the discretion of the treating physicians in this clinic. Diabetic complications of retinopathy and nephropathy were evaluated by experienced diabetologists by means of ophthalmoscopy and urine albumin excretion, respectively. Diabetic neuropathy was estimated according to the diagnostic criteria proposed by Toronto Diabetic Neuropathy Expert Group [7]. Demographic variables of the enrolled participants were extracted from personal medical records. Routine laboratory tests were performed in diabetic group under the monthly visit and in control group as annual health-check program. Body Mass Index (BMI) was calculated by BW (kg) divided by square of height (m). Blood pressure was measured by sphygmomanometer in sitting position after taking a few minutes rest. Hypertension was defined as casual blood pressure >140/90 mmHg and/or treatment with antihypertensive drugs. Dyslipidemia was also defined as serum LDL cholesterol >140 mg/dl, serum HDL cholesterol <40 mg/dl and/or prescription of lipid-lowering drugs [8]. Exclusion criteria for diabetic patients included those with unstable diabetic condition (i.e., ketoacidosis, diabetic coma and frequent hypoglycemic episodes), those with advanced diabetic nephropathy requiring hemodialysis, and those under the insulin therapy including insulin analogues such as insulin degludec and insulin glargine or Basal-supported Oral Therapy (BOT). Patients under the treatment with drugs affecting on QTc interval such as probucol, antiarrhythmic drugs, antipsychotic drugs, antifungal and antibiotic agents were also excluded.

■ ECG recordings

All patients underwent resting 12-lead ECG in the morning of regular visit or health-check program to minimize the potential bias due to circadian rhythm of HRV. ECG was recorded in supine position after several minutes’ rest using digital surface ECG recorder (CARDISUNY C310, FUKUDA ME, Tokyo, Japan), and was printed at a paper speed of 25 mm/sec and amplitude of 10 mm/mV or 5 mm/mV as appropriate. Absolute QT interval (msec) was measured from the beginning of the QRS complex to the end of the T wave. The end of the T wave was defined as the intercept between the isoelectric line and the regression line derived from the descending slope of the T wave. U wave was carefully distinguished from T wave. Strictly, absolute QT interval is different in different leads. In the present study, QT interval in limb lead II was adopted, because this interval is longest in lead II, and T wave is easily distinguished from U wave in this lead [9]. QT interval was corrected by the mostly used Bazett’s formula (QTc: msec), where QTc=QT/RR^1/2. Coefficient of variance of R-R intervals (CVRR:%) is a representative time-domain measure of HRV. This parameter was calculated automatically by sampling of 100 beats during sinus rhythm. ECG data were diagnosed based on Minnesota code, transferred using A/D converter, and stored automatically to a personal computer (VAIO, model SVP132A16N, SONY®, Tokyo, Japan). Thereafter, ECG recordings were reviewed by experienced cardiologists in a blind manner. ECG exclusion criteria were documented atrial fibrillation and other sustained arrhythmias, manifest WPW syndrome, right or left bundle branch block, atrioventricular block with its degree of second or more and intraventricular conduction disturbance showing QRS width >120 msec.

■ Data analyses

All data were expressed as means ± Standard Deviation (SD). Data sets were examined by Kolmogorov-Smirnov test and Shapiro-Wilks test for normality. None of the variables with missing data qualified. Comparison of continuous variables between the two groups was conducted with unpaired Student’s t-test or Mann-Whitney U-test which depended on the normality. Discrete variables were analyzed as a contingency table using Fisher’s exact test or Pearson’s χ2 test. Pearson’s or Spearman’s correlation analysis was performed depending on the normality. Multiple regression analysis was applied to determine the significant contributors to the objective variables. The criteria to enter the variables into the regression model was statistical significance or otherwise clinical importance. Explanatory variables were checked for confounding factor and Variance Inflation Factor (VIF) >10 was defined as multicollinearity. These analyses were performed using Bell Curve for Excels version 2.12 (Social Survey Research Information Co., Ltd., Tokyo, Japan). Differences with two-sided p<0.05 were considered significant.

■ Ethical considerations

This study was conducted according to the updated Declaration of Helsinki in 2008 and the Ethical Guidelines for Epidemiological Studies published by the Japanese Ministry of Health, Labor and Welfare and Ministry of Education, Culture, Sports, Science and Technology [10]. The study design was approved by the internal ethics committee of The Institute of Rheological Function of Foods Co. Ltd (Hisayama, Fukuoka, Japan). Signed informed consent was obtained from each subject prior to the enrollment into the study.

Results

■ Baseline characteristics of diabetic and control groups

TABLE 1 demonstrated baseline characteristics in diabetic and control groups. There were no significant differences in the mean age between the two groups. With respect to categorical variables, there were no significant differences between the two groups in the gender balance and the prevalence of hypertension, dyslipidemia and obesity. BMI in diabetic group did not differ from that in control group, and heart rate estimated by ECG recording in diabetic group was greater than that in control group (p=0.012). Systolic and diastolic blood pressures in the former group showed no difference from the corresponding blood pressures in the latter group. In laboratory data, total and LDL cholesterol levels in the diabetic group were greater than the corresponding cholesterol levels in the controls. As expected, HbA1c (NGSP) in the diabetic group was far higher than that in the control group (p<0.001), whereas serum K concentration (mEq/L) showed no intergroup difference. Digital ECG recording demonstrated significantly longer QTc interval (p=0.010) and reduced CVRR (p<0.001) in the diabetic group compared with control group.

Table 1: Baseline characteristics in diabetic and control groups

■ Prescriptions in diabetic and control groups

Diabetic patients (n=60) were prescribed with sulfonylurea (SU)-type oral hypoglycemic agents (n=6), dipeptidyl peptidase-4 (DPP4) inhibitors (n=8), sodium glucose cotransporter-2 (SGLT2) inhibitors (n=10), and metformin (n=4). However, the combined prescription was the most frequent (n=32). For hypertensive patients in the diabetic group (n=10), angiotensinreceptor blockers (ARB; 30%), calcium channel blockers (CCB; 20%) and β blockers (BB; 10%) were prescribed, although the combination of these antihypertensive agents was the most prevalent (40%). Antihypertensive prescription for control group (n=14) included ARB (21%), CCB (21%), BB (7%) and the combination of these was (50%). Intergroup comparison of the drug distribution was not significant (p=0.948). With respect to the lipid-lowering agents, 3-hydroxyl-3-methylglutaryl coenzymeA (HMGCoA) inhibitors (statin; 63%), ezetimibe (25%) and the combination of these two agents (12%) were prescribed in the dyslipidemic patients (n=16) of diabetic group, and the trend was the same (p=0.979) as in the dyslipidemic patients (9) of control group showing HMGCoA inhibitors (67%), ezetimibe (22%) and these combination (11%).

■ QTc intervals in diabetic and control groups

As QTc interval is influenced by many factors, this index was analyzed by multivariate analysis. Multiple regression analysis was applied to the respective diabetic and control groups to find covariates contributing significantly to the QTc interval. As shown in TABLE 2, statistical significance was not obtained in the control group by regression model with forced injection of variables (R2=0.150, F=1.912 and p=0.107). Then, stepwise multiple regression analysis demonstrated that CVRR (p=0.037) was the sole contributor to the QTc interval regulation (TABLE 2) in the significant regression model (R2=0.121, F=3.906 and p=0.026). QTc interval estimated in diabetic group was also fitted to the multiple regression model, and the significant regression model (R2=0.345, F=4.294 and p=0.002) demonstrated that BMI (p<0.001) and CVRR (p=0.008) were the significant contributors to the QTc intervals (TABLE 3). Of note is that CVRR was correlated to the QTc intervals positively in controls but negatively in diabetics. Partial regression coefficient (β) was 0.004 (95% Confidence Interval (CI) of 0.000– 0.007, p=0.037), and the standardized β was 0.266 in the control group (TABLE 2), whereas β was -0.005 (95% CI of -0.009– -0.001, p=0.015), and the standardized β was -0.306 in the diabetic group (TABLE 3).

Table 2: Multiple regression analysis predicting contributors to QTc intervals in control subjects

Table 3:  Multiple regression analysis predicting contributors to QTc intervals in diabetic patients

Discussion

The present study demonstrated the contribution of multiple clinical factors to the QTc interval estimated in diabetic and control groups. However, this contribution in diabetic patients differed from that in the control subjects. BMI and CVRR were the main contributors to the QTc interval regulation in type 2 diabetic patients, whereas CVRR contributed mainly to the QTc interval in controls. CVRR was found to be associated with QTc interval regulation in both groups. However, the details of this association in diabetics differ from those in nondiabetics. CVRR regulated QTc interval negatively in diabetics but positively in nondiabetics (TABLES 2 & 3). QTc interval regulation is mediated mainly by autonomic influences, electrolytes balance and sex hormone actions and so on. Mean QTc interval in females is longer than this interval in males, and this interval in diabetics with autonomic neuropathy is longer than that interval in diabetics without neuropathy [11]. Such ECG consensus was confirmed in this study. Mean QTc interval in males (n=74) was 406.5 ± 23.8 msec, whereas this interval in females (n=48) was 418.8 ± 31.0 msec (p=0.006). This interval in diabetic patients with neuropathy was 417.2 ± 24.6 msec (n=41), whereas this interval in those without neuropathy was 413.3 ± 35.2 msec (n=19), although this difference was not significant (p=0.635). However, gender was not a main regulator of the QTc intervals by multivariate analysis (TABLES 2 & 3). The reason is not clear, but gender difference in QTc interval decreases with aging due to decline of testosterone level [12]. In our study, QTc intervals in diabetics were explained mostly by the combination of CVRR and BMI (TABLE 3). No significant correlation between HbA1c and CVRR was obtained in diabetic group (r=0.086, p=0.524, n=60), indicating that HbA1c is a reliable marker of diabetic control status, and that CVRR reflects an autonomic diabetic complication irrespective of the current diabetic control. Reportedly, individuals with higher BMI show longer QTc intervals in ECG, and higher BMI or longer QTc are associated with increases of all-cause mortality and cardiovascular mortality in the third US National Health and Nutrition Examination Survey [13]. The positive linear correlation between QTc interval and BMI was confirmed in this study (FIGURE 1), and optimal regression equation between QTc and BMI is reported to be fourthorder degree polynomial [14]. These indicate cardiometabolic abnormality in patients with type 2 diabetes.

Although there are many confounding factors, CVRR is an established time-domain index of HRV reflecting sympathovagal interaction [15]. QTc interval prolongation is associated with autonomic neuropathy in diabetic patients [11], but Fiorentini et al. [16], reported that QTc interval is lengthened in pre-diabetic subjects showing insulin resistance and impaired glucose tolerance. They also found the correlation between the QTc interval and various parameters of HRV, i.e., standard deviation of normalto- normal RR intervals during sinus rhythm (SDNN) as an overall time-domain HRV index, Low-Frequency component (LF) and the ratio of low frequency to high frequency component (L/H) in the power spectral analysis of HRV [16]. Contrarily, Takahashi et al. [17], investigated autonomic neuropathy as it relates to frequencydomain HRV, baroreflex sensitivity and washout rate of ¹²³I-metaiodobenzylguanidine in diabetic patients. They concluded that none of the parameters of HRV correlated with QTc intervals. There are many differences among their and our studies, i.e., baseline characteristics of enrolled patients such as control level of diabetes are different (mean HbA1c of 9.1 ± 2.4% in [17] vs. 7.7 ± 1.7% in our study). Methodologies of HRV analysis are different, i.e., time-domain analysis in our study, frequency-domain analysis in [17], and both analyses in [16]. Such differences may affect the discrepant outcomes, and baroreflex sensitivity is reported to correlate negatively with QTc intervals [17]. This and our findings commonly suggest that increased vagal activity abbreviates the QTc intervals in diabetics (FIGURE 2), although baroreflex sensitivity reflects phasic, whereas HRV indicates static vagal tone.

QTc interval dynamicity under the vagal predominance in healthy subjects remains controversial. In the last century, autonomic influences on the uncorrected and corrected QT intervals in healthy volunteers have been investigated by atrial pacing under the pharmacological autonomic blockade using propranolol and atropine. Cappato et al. [18], observed that baseline QTc interval (418 ± 26 msec) was shortened significantly (p<0.01) by sympathetic blockade alone (400 ± 22 msec) but not changed by sympathovagal blockade (423 ± 12 msec) during sinus rhythm. However, these relationships disappeared by removing autonomic influences under the constant atrial pacing at its rate ranging from 100 to 150 ppm. They supposed integrated direct and indirect (heart-rate mediated) autonomic regulation of QTc intervals and concluded that vagal activity lengthens the QTc interval which is not mediate by rate-dependent mechanisms. Likewise, Ahnve and Vallin demonstrated the findings similar to that of Cappato et al. [18], under the same protocol applied to the healthy subjects [19]. Although baseline autonomic balance and intrinsic ventricular properties differ among individuals at rest, these findings commonly indicate vagally-mediated lengthening of QTc interval length in healthy subjects. In this century, Xhaet et al. [20], reported the QTc interval prolongation under the reflex vagal activation induced by phenirephrine in healthy subjects, which is compatible with the results of pharmacological autonomic blocking studies. Our results of positive correlation between CVRR and QTc interval in nondiabetics is not contradictory to this line of evidence (TABLE 2), because CVRR is an overall index of HRV regulated predominantly by vagal activity at rest. In this sense, negative linearity among QTc and CVRR in diabetics (FIGURE 2) demonstrates pathological correlation, indicating longer QTc interval under the reduced HRV as observed frequently in diabetic patients with neuropathy. This study allows several limitations. First limitation is a research protocol of a cross-sectional case-control study design with relatively small sample size. Longitudinal study may provide more distinct information concerning autonomic QTc interval regulation in the progression of diabetes. The second one is the treatment of diabetic patients. They were treated under the discretion of the treating physicians, although our antidiabetic medication complied with the guideline of the Japanese Society of Diabetes [21]. The third one is the limited selection of the explanatory variables among laboratory data and baseline characteristics. Cutting edge autonomic function tests may provide more sophisticated regression model. Finally, Bazett’s formula is characterized by excessive heart-rate correction of measured QT interval showing proportionally longer QTc interval as heart rate increases [19,22]. Since basic heart rates in our two groups ranged from 60 to 70 bpm (TABLE 1), overcorrection by heart rate was unlikely. With respect to prescription, statin is reported to abbreviate QTc interval and improve HRV in patients with Heart Failure (HF) reflecting pleiotropic effect [23,24], although the incidence of statin prescription in diabetic group was equivalent to that in control group. Further, recently available DPP4 and SGLT2 inhibitors benefit diabetic patients with HF. These agents are reported to have no meaningful effects on QT and QTc intervals in literature [25-27] including thorough QT/QTc studies [28,29]. With such limitations in mind, the complicated correlation between QTc interval and HRV should be validated in a future large cohort.

Conclusion

This study demonstrated the multifactorial regulation of the QTc interval estimated in diabetic and control groups. BMI was correlated positively with QTc interval in diabetic group, indicating cardiometabolic linkage. CVRR was also associated with QTc interval regulation. However, CVRR was correlated negatively with QTc interval in diabetic group, and opposite correlation was found in the control group. This may reflect pathological autonomic QT interval regulation, abnormal intrinsic myocardial properties or both in diabetic patients especially with autonomic neuropathy.

Acknowledgement

The authors thank staffs of BOOCS Clinic for clinical assistance.

Disclosure

The authors declare that there are no competing interests in relation to this manuscript.

Funding Source

This work was supported in part by an Academic Support from SOUSEIKAI Global Clinical Research Center (LTA Medical Corporation, Japan), and a Grant-In-Aid for Supporting Industry Program (so-called ‘Suppoin’) from the Japanese Ministry of Economy, Trade and Industry, Japan (20180830-52).

References

1. Leon B, Maddox T. Diabetes and cardiovascular disease: Epidemiology, biological mechanisms, treatment recommendations and future research. World. J. Diabetes. 6(13), 1246-1258 (2015).
2. Glovaci D, Fan W, Wong N. Epidemiology of diabetes mellitus and cardiovascular disease. Current. Cardiol. Rep. 21(4), 21(2019).
3. Zhang Y, Post W, Blasco-Colmenares E et al. Electrocardiographic QT interval and mortality: A meta-analysis. Epidemiology. 22(5), 660-670 (2011).
4. Fagher K, Nilsson A, Londahl M. Heart rate-corrected QT interval prolongation as a prognostic marker for 3-year survival in people with type 2 diabetes undergoing above-ankle amputation. Diabet. Med. 32(5), 679-685 (2015).
5. Pfister R, Cairns R, Erdmann E et al. Prognostic impact of electrocardiographic signs in patients with type 2 diabetes and cardiovascular disease: Results from the PROactive study. Diabet. Med. 28(10), 1206-1212 (2011).
6. Veglio M, Chinaglia A, Cavallo Perin P. The clinical utility of QT interval assessment in diabetes. Diabetes. Nutr. Metab. 13(6), 356-365 (2000).
7. Tesfaye S, Boulton A, Dyck P et al. Diabetic neuropathies: update on defintions, diagnostic criteria, estimation of severity, and treatments. Diabetes. Care. 33(10), 2285-2293 (2010).
8. Ohta Y, Tsuchihashi T, Onaka U et al. Clustering of cardiovascular risk factors and blood pressure control status in hypertensive patients. Intern. Med. 49(15), 1483-1487 (2010).
9. Postema P, Wilde A. The measurement of the QT interval. Curr. Cardiol. Rev. 10(3), 287-294 (2014).
10. https://www.mhlw.go.jp/general/seido/kousei/i-kenkyu/ekigaku/0504sisin.html
11. Kobayashi S, Nagao M, Asai A et al. Severity and multiplicity of microvascular complications are associated with QT interval prolongation in patients with type 2 diabetes. J. Diabetes. Investig. 9(4), 946-51 (2018).
12. Vicente J, Johannesen L, Galeotti L et al. Mechanisms of sex and age differences in ventricular repolarization in humans. Am. Heart. J. 168(5), 749-756 (2014).
13. Waheed S, Dawn B, Gupta K. Association of corrected QT interval with body mass index, and the impact of this association on mortality: Results from the Third National Health and Nutrition Examination Survey. Obes. Res. Clin. Pract . 11(4), 426-434 (2017).
14. Curione M, Tego A, Capoccia D et al. Prediction of QTc length as function of BMI: A clinical tool to establish arrhythmias risk in obesity. Clin. Ter. 162(6), e155-9 (2011).
15. Murata K, Araki S. Assessment of autonomic neurotoxicity in occupational and environmental health as determined by ECG R-R interval variability: A review. Am. J. Ind. Med. 30(2), 155-163 (1996).
16. Fiorentini A, Perciaccante A, Valente R et al. The correlation among QTc interval, hyperglycemia and the impaired autonomic activity. Auton. Neurosci. 154, 94-98 (2010).
17. Takahashi N, Nakagawa M, Saikawa T et al. Regulation of QT indices mediated by autonomic nervous function in patients with type 2 diabetes. Int. J. Cardiol. 96(3), 375-379 (2004).
18. Cappato R, Alboni P, Pedroni P et al. Sympathetic and vagal influences on rate-dependent changes of QT interval in healthy subjects. Am. J. Cardiol. 68(11), 1188-1193 (1991).
19. Ahnve S, Vallin H. Influence of heart rate and inhibition of autonomic tone on the QT interval. Circulation. 65(3), 435-439 (1982).
20. Xhaet O, Argacha J, Pathak A et al. Sympathoexcitation increases the QT/RR slope in healthy men: differential effects of hypoxia, dobutamine, and phenylephrine. J .Cardiovasc. Electrophysiol. 19(2), 178-184 (2008).
21. Haneda M, Noda M, Origasa H et al. Japanese clinical practice guideline for diabetes 2016. Diabetol. Intern. 9: 1-45 (2018).
22. Arai K, Nakagawa Y, Iwata T et al. Relationships between QT interval and heart rate variability at rest and the covariates in healthy young adults. Auton. Neurosci. 173(1-2), 53-57 (2013).
23. Xie R, Cui W, Liu F et al. Statin therapy shortens QTc, QTcd, and improves cardiac function in patients with chronic heart failure. Int. J. Cardiol. 140(22), 255-257 (2010).
24. Vrtovec B, Okrajsek R, Golicnik A et al. Atorvastatin therapy increases heart rate variability, decreases QT variability, and shortens QTc interval duration in patients with advanced chronic heart failure. J. Card. Fail. 11(9), 684-690 (2005).
25. Bloomfield DM, Krishna R, Hreniuk D et al. A thorough QTc study to assess the effect of sitagliptin, a DPP4 inhibitor, on ventricular repolarization in healthy subjects. J. Clin. Pharmacol. 49(8), 937-946 (2009).
26. Sato T, Miki T, Ohnishi H et al. Effect of sodium-glucose co-transporter-2 inhibitors on impaired ventricular repolarization in people with Type 2 diabetes. Diabet. Med. 34(10), 1367-1371 (2017).
27. Zhang W, Smulders R, Abeyratne A et al. Ipragliflozin does not prolong QTc interval in healthy male and female subjects: a phase I study. Clin. Ther. 35(8), 1150-1161(2013).
28. Carlson G, Tou C, Parikh S et al. Evaluation of the effect of dapagliflozin on cardiac repolarization: A thorough QT/QTc study. Diabetes. Ther. 2(3), 123-32 (2011).
29. Kumagai Y, Hasunuma T, Sakai S et al. Randomized, controlled, thorough QT/QTc study shows absence of QT Prolongation with Luseogliflozin in healthy Japanese subjects. PLoS. One. 10, e0139873 (2015).