Introduction

India has the second highest number of people with diabetes worldwide, with approximately 8.6% of the population reported to have this disease [1]. Glycemic control is poor in Indian diabetic patients, with a mean glycated hemoglobin (HbA1c) of 8.9 ± 2.1% (74 ± 23 mmol/mol) and only 19.7% of the Indian diabetic population achieving the American Diabetes Association/European Association for the Study of Diabetes HbA1c target of <7% (<53 mmol/mol) [2, 3]. In addition to the increased prevalence of diabetes, higher rates of diabetes-related complications and mortality are also observed in South Asian people [4]. The factors contributing to this situation are multifactorial and include a genetic predisposition to diabetes, increased visceral adiposity and insulin resistance in this population, which are compounded by lifestyle factors [5]. Despite these numbers, South Asian participants in trials of glucose-lowering therapies are under-represented [6]; hence, it is important to report trial data from diverse populations.

IDegLira is a novel, once-daily, titratable, fixed-ratio combination of insulin degludec (IDeg) and the glucagon-like peptide-1 receptor agonist (GLP-1RA) liraglutide (Lira) that has been developed for the treatment of people with type 2 diabetes [7–9]. Importantly, GLP-1RAs act in a glucose-dependent manner and are also commonly associated with a low risk of hypoglycemia and weight loss [10]. The efficacy and safety of IDegLira and the benefits of its complementary mode of action have been examined in the large global DUAL clinical trial program [7–9, 11]. Here we describe a post hoc analysis of the efficacy and safety of IDegLira in participants from Indian trial sites included in the DUAL program. Our hypothesis was that the safety and efficacy of IDegLira in the Indian subpopulation would be similar to that demonstrated for the global population.

Methods

This post hoc analysis used data from the IDegLira (Novo Nordisk, Bagsværd, Denmark) phase 3 trials which included participants with type 2 diabetes from Indian trial sites, namely, the DUAL I extension (52 weeks of data), DUAL IV and DUAL II trials, which included Indian sites [8, 9, 11].

Study Designs

The study designs of DUAL I extension, DUAL II and DUAL IV have been described previously [8, 9, 11] and include participants uncontrolled on oral antidiabetic drugs (OADs; DUAL I extension and DUAL IV) and participants uncontrolled on basal insulin + OADs (DUAL II). The aim of DUAL I extension was to investigate the efficacy and safety of IDegLira versus its individual components IDeg or Lira alone in participants inadequately controlled by metformin ± pioglitazone over 52 weeks [8]. In DUAL IV IDegLira was compared with placebo, both added on to a therapeutic regimen of sulphonylurea ± metformin [9], and in DUAL II, IDegLira was compared to IDeg (maximum dose of 50 U) in participants inadequately controlled on 20–40 U of basal insulin + one to two OADs [11]. Owing to differences in trial design, individual trial data were analyzed separately, rather than pooling the IDegLira data.

Endpoints and Statistical Analyses

The primary endpoint in each study was change from baseline in HbA1c. Secondary efficacy endpoints included participants achieving a HbA1c of <7% (<53 mmol/mol), change in body weight, insulin dose and laboratory-measured fasting plasma glucose (FPG). Safety endpoints included adverse events (AEs) and confirmed hypoglycemia [severe (unable to self-treat) and/or plasma glucose level of <3.1 1 mmol/L]. Change from baseline in HbA1c, FPG and body weight were analyzed using an analysis of covariance (ANCOVA) model with treatment and pre-trial medication as fixed factors, and baseline HbA1c stratum (DUAL I only), sub-study (DUAL I only) and baseline value as covariates—all performed on the full analysis set using last observation carried forward (LOCF) method to impute missing values. Attainment of a HbA1c level of <7% was analyzed using a logistic regression model (with LOCF), with the same explanatory variables as used for the primary endpoint. The results of the analyses were also summarized descriptively for other endpoints (insulin dose, rates of hypoglycemia and other safety endpoints).

As a post hoc analysis, the original individual trials were not powered to perform statistical analyses in the Indian subpopulations. The study reported here is based on a post hoc analysis of previously conducted studies and does not involve any new studies of human or animal subjects performed by any of the authors.

Results

The Indian subpopulations comprised 15, 16 and 15% of the global trial populations of the DUAL I extension, DUAL II and DUAL IV trials, corresponding to 251, 64 and 64 participants, respectively. Baseline characteristics of the global trial populations and Indian subpopulations are shown in Table 1. Overall, the baseline characteristics of the Indian subpopulations were largely comparable to those of the global trial populations, although the Indian participants had lower FPG and body mass index (BMI).

Table 1 Baseline characteristics of overall and Indian subpopulations in the DUAL I extension, DUAL II and DUAL IV trials (full analysis set)
Full size table

Glucose Control

The change in HbA1c from baseline and end-of-trial HbA1c in the global study populations and the Indian subpopulations from the three studies are shown by patient population in Fig. 1. Considering patients previously uncontrolled on OAD(s), in the Indian subpopulation of DUAL I extension, IDegLira resulted in a statistically significantly greater HbA1c reduction versus IDeg [estimated treatment difference (ETD) −0.91%; 95% confidence interval (CI) −1.21 to −0.61 (−10 mmol/mol; 95% CI −13 to −7); p < 0.0001] and versus Lira [ETD −0.77%; 95% CI −1.08 to −0.46 (−8 mmol/mol; 95% CI −12 to −5); p < 0.0001]. In DUAL IV, a significantly greater reduction was observed with IDegLira versus placebo [ETD: −0.78%; 95% CI −1.25 to −0.31 (−9 mmol/mol; 95% CI −14 to −3); p = 0.002] in the Indian subpopulation. In patients previously uncontrolled on basal insulin (DUAL II), IDegLira also resulted in a greater HbA1c reduction versus IDeg [maximum dose 50 U; ETD −1.32%; 95% CI −1.80 to −0.85 (−14 mmol/mol; 95% CI −20 to −9); p < 0.0001] in the Indian subpopulation.

Fig. 1
figure 1

Mean change in glycated hemoglobin (HbA1c) and end-of-trial HbA1c in the global study population and the Indian subpopulations of the DUAL I extension, DUAL IV and DUAL II trials (maximum 50 U) [8, 9, 11, 12]. Mean observed values are based on full analysis set and data inputted using the last observation carried forward (LOCF) method. Treatment difference is estimated from an analysis of covariance (ANCOVA) analysis. CI Confidence interval EOT end of trial, ETD estimated treatment difference, IDeg insulin degludec, IDegLira insulin degludec/liraglutide, Lira liraglutide, OAD oral antidiabetic drug

Full size image

The proportion of participants in the global study populations and the Indian subpopulations achieving HbA1c <7% (<53 mmol/mol) are shown in Fig. 2. In the Indian subpopulation of the DUAL I extension trial, participants were significantly more likely to achieve an HbA1c of <7% (<53 mmol/mol) with IDegLira than with IDeg [84.0 vs. 43.1%; odds ratio (OR) 6.77; 95% CI 3.35–13.65; p < 0.0001] or with Lira (84.0 vs. 47.5%; OR 5.84; 95% CI 2.86–11.90; p < 0.0001). Participants were also more likely to achieve an HbA1c of <7% (<53 mmol/mol) with IDegLira than with placebo in the DUAL IV trial (69.6 vs. 16.7%; OR 11.47; 95% CI 2.79–47.17; p < 0.001). In DUAL II, participants were also more likely to achieve an HbA1c of <7% (53 mmol/mol) with IDegLira versus IDeg (65.6 vs. 21.9%; OR 7.36; 95% CI 2.31–23.50; p <0.0007).

Fig. 2
figure 2

Percentages of participants achieving the HbA1c target of <7% in the global study population and the Indian subpopulations of the DUAL I extension, DUAL IV and DUAL II trials (maximum 50 U) [8, 9, 11, 12]. Mean observed values are based on the full analysis set and LOCF inputted data. The odds ratio (OR) is estimated from a logistic regression analysis

Full size image

In the DUAL I extension trial, IDegLira resulted in a greater FPG reduction versus Lira (ETD −29.80 mg/dL; 95% CI −39.62 to −19.98; p < 0.0001); differences versus IDeg were not statistically significant [TD −7.04 mg/dL; 95% CI −16.57 to 2.50; p = not significant (ns)]. Differences versus placebo in the DUAL IV trial were also not statistically significant (ETD −18.65 mg/dL; 95% CI −38.80 to 1.50; p = ns). IDegLira resulted in a greater FPG reduction versus IDeg (maximum dose 50 U) in DUAL II (ETD −34.26 mg/dL; 95% CI −53.28 to −15.24; p = 0.0006).

Body Weight

In the Indian subpopulation of the DUAL I extension trial, IDegLira was weight neutral (−0.3 kg), whereas body weight increased with IDeg (1.8 kg; ETD −1.61 kg; 95% CI −2.45 to −0.78; p = 0.0002) and decreased with Lira (−2.3 kg; ETD 1.72 kg; 95% CI 0.87–2.58; p < 0.0001) (Fig. 3). In DUAL IV, the mean weight change was 0.6 kg with IDegLira and −0.9 kg with placebo (ETD 1.37 kg; 95% CI 0.31–2.43; p = 0.0124). IDegLira resulted in a greater weight loss versus IDeg (ETD −1.91 kg; 95% CI −3.39 to −0.43; p = 0.01223) in DUAL II. Across the studies, baseline body weights were lower in the Indian subpopulations.

Fig. 3
figure 3

Mean change in body weight in the global study population and Indian subpopulations of the DUAL I extension, DUAL II and DUAL IV trials [8, 9, 11]. Maximum dose of IDeg was 50 U in DUAL II. Mean observed values are based on full analysis set and LOCF inputted data. Treatment difference is estimated from an ANCOVA analysis

Full size image

Insulin Dose

End-of-trial insulin dose was 35 U with IDegLira and 59 U with IDeg in the DUAL I extension trial, 25 U with IDegLira in the DUAL IV trial and 42 and 46 U with IDegLira and IDeg in the DUAL II trial.

Safety Endpoints

Adverse events for the global populations and Indian subpopulations are summarized in Table 2. Rates of gastrointestinal AEs (combined diarrhea, nausea and vomiting) were 25.3 events per 100 patient-years of exposure (PYE) in the IDegLira group, 8.5 per 100 PYE in the IDeg group and 75.3 per 100 PYE in the Lira-treated group in the DUAL I extension trial, and 9.4 (IDegLira) and 36.2 (placebo) events/100 PYE in DUAL IV. In DUAL II, the rates were 82.1 (IDegLira) and 20.7 (IDeg) events/100 PYE.

Table 2 Adverse events reported in the DUAL I extension, DUAL II and DUAL IV trials (full analysis set)
Full size table

There were no severe hypoglycemic episodes in the Indian subpopulations of the three trials. Confirmed hypoglycemia rates were 1.98, 1.08 and 0.37 events/PYE in the IDegLira, IDeg and Lira groups, respectively, in the DUAL I extension trial (compared with 1.77, 2.79 and 0.19 events/PYE in the global trial population) and 4.06 and 0.36 events/PYE with IDegLira and placebo, respectively, in DUAL IV (compared with 3.52 and 1.35 events/PYE in the global trial population). In DUAL II, confirmed hypoglycemia rates were 1.16 and 0.83 events/PYE with IDegLira and IDeg, respectively (compared with 1.53 and 2.63 events/PYE in the global trial population).

Discussion

In the Indian subpopulations, IDegLira resulted in substantial HbA1c reductions that were significantly greater than those achieved by all comparators. Mean end-of-trial HbA1c with IDegLira was <7% (<53 mmol/mol) in all trials and more participants achieved a HbA1c target of <7% (<53 mmol/mol) versus all comparators, consistent with the global trial populations [8, 9, 11].

IDegLira was generally weight-neutral when used as an add-on to oral therapy (DUAL I extension/DUAL IV) and resulted in weight loss when used after basal insulin (DUAL II).

Considering IDegLira alone, rates of confirmed hypoglycemia were consistent with those observed with IDegLira in the global trial populations. However, the rate of confirmed hypoglycemia with IDeg was substantially lower in the Indian subpopulations (by around 1.7/1.8 events/PYE) versus the global trial populations of DUAL I and II. As a result, the rate of confirmed hypoglycemia was numerically lower with IDeg versus IDegLira in the Indian subpopulations of DUAL I and II, while in the global trial population IDegLira resulted in a lower rate of hypoglycemia versus IDeg, reaching statistical significance in the DUAL I trial [8, 9]. This difference might be explained by the small number of participants—and therefore hypoglycemic events—in the Indian subpopulations and by baseline differences. The hypoglycemia rates should also be considered in the context of the significantly lower mean end-of-trial HbA1c achieved with IDegLira compared to comparators.

IDegLira was generally well tolerated, with a lower rate of gastrointestinal AEs in the Indian subpopulation versus the global trial populations in DUAL I and IV, and a higher rate in DUAL II. Across all trials, the proportion of participants experiencing AEs was higher in the Indian subpopulation than in the global trial population, but the pattern of AEs did not differ [8, 9, 11].

An inherent limitation of this study is that it is a post hoc analysis of three randomized controlled DUAL trials that were not designed to compare the safety and efficacy of IDegLira in the Indian subpopulation versus the global population. As such, there were differences at baseline in terms of FPG and BMI with the Indian subpopulation compared with the global population. Furthermore, owing to trial design differences, individual trial data were analyzed separately rather than the IDegLira data being pooled, and therefore sample sizes were small. Differences between Indian and non-Indian populations were not analyzed as the Indian population constituted a relatively low proportion of the global population.

Conclusion

In conclusion, results from the Indian subpopulations largely reflected those of the global study populations. These results provide evidence to support the efficacy and safety of IDegLira as a treatment modality for participants with type 2 diabetes uncontrolled on OADs and/or basal insulin of the South Asian population, offering better glucose control than either component alone, improved weight profile versus the basal insulin component, and with fewer gastrointestinal side effects than a GLP-1 analogue.