FormalPara Key Summary Points

Why carry out this study?

As a result of advances in therapy, individuals with type 2 diabetes mellitus (T2D), including those in Asian countries, are living with their disease for longer.

Treatments such as glucagon-like peptide 1 (GLP-1) receptor agonists that target pancreatic β cells may show reduced efficacy over time because of diminishing β cell function.

The efficacy and safety of the GLP-1 receptor agonist lixisenatide plus basal insulin according to duration of T2D was investigated in a post hoc analysis using pooled data from Asian participants in three trials that were part of the GetGoal clinical trial program.

What was learnt from the study?

The duration of T2D at baseline did not significantly influence the change in glycated hemoglobin (HbA1c) achieved with add-on lixisenatide in Asian participants over 24 weeks.

Individuals with longer disease duration had a greater risk of symptomatic (but not severe) hypoglycemia.

Lixisenatide can be considered for use as add-on therapy in Asian people taking basal insulin, irrespective of time since diagnosis of T2D.

Introduction

Developments in type 2 diabetes mellitus (T2D) have led to earlier diagnosis, and people living longer with the disease [1]. These individuals frequently require more and more complex treatment regimens as agents are added over time to maintain glycemic control in the face of a progressive deterioration in pancreatic β cell function [1,2,3]. The choice of add-on therapy in T2D should take into account the mechanism of action of the existing therapies; agents with different but complementary mechanisms of action will augment glucose-lowering effects and counteract adverse events [1, 4].

When add-on therapy is required for individuals with T2D on basal insulin, the glucagon-like peptide 1 (GLP-1) receptor agonists are a recommended option [5]. These agents are associated with a low risk of hypoglycemia, and with the beneficial effect of body weight loss, which can help to counteract the body weight gain often seen with insulin [6, 7]. GLP-1 receptor agonists target a range of underlying pathogenic processes in addition to stimulating insulin secretion from β cells [4]. Short-acting agents such as lixisenatide slow gastric emptying and suppress glucagon secretion, leading to marked reductions in postprandial glucose (PPG) [6, 7]. GLP-1 receptor agonists also increase insulin sensitivity in peripheral tissues and reduce gluconeogenesis in the liver [4].

Because basal insulin lowers fasting plasma glucose (FPG) and GLP-1 receptor agonists reduce PPG, their use in combination has a rational therapeutic basis [1, 2]. In fact, in an indirect comparison of different antidiabetic treatments as add-on therapy to basal insulin, GLP-1 receptor agonists showed the most marked effect on glycated hemoglobin (HbA1c) [8].

Antidiabetic treatments that only target β cells might be expected to show limited efficacy in individuals with long-standing T2D because of diminishing β cell function over time, as has been seen with the sulfonylureas [3, 9, 10]. As the primary mechanism of action of GLP-1 receptor agonists (an increase in insulin secretion) is β cell dependent, investigating their use in individuals with long-standing diabetes is essential. Indeed, their use in clinical practice among patients with a T2D duration of > 10 years is limited [11], which may be due to the perception that the efficacy of these agents is reduced in patients with significant β cell function decline or patients with long T2D duration (> 10 years). However, there are limited data on the interaction between diabetes duration and the efficacy and safety of GLP-1 receptor agonists in Asian populations. Asia accounts for over half of the world’s diabetes population, and the prevalence of diabetes in Asia is projected to increase dramatically over the next two decades [12]. Thus, detailed knowledge of the effects of diabetes treatments in Asian populations is important.

The studies included in this post hoc analysis were part of the GetGoal clinical trial program. The GetGoal clinical trial program evaluated the efficacy and safety of lixisenatide as add-on therapy to basal insulin (± oral antidiabetic drugs [OADs]) in individuals with T2D [13,14,15,16]. These studies found that, compared with placebo + basal insulin, lixisenatide + basal insulin was consistently associated with a significantly greater reduction in HbA1c (mean reductions in HbA1c of − 0.62% to − 0.77% at week 24; all P < 0.001 vs placebo) and PPG (mean reductions in PPG of − 3.1 to − 8.0 mmol/L at week 24; all P < 0.001 vs placebo), irrespective of the concomitant OAD regimen [13,14,15,16].

In this post hoc analysis, the efficacy and safety of lixisenatide + basal insulin according to participants’ duration of diabetes was investigated using pooled data from Asian participants in three trials: GetGoal Duo-1 [14], GetGoal-L [13], and GetGoal-L-C [16].

Methods

GetGoal Study Design

The three GetGoal studies included in this post hoc analysis were randomized, double-blind, and placebo-controlled in design, and were conducted in individuals with T2D and an HbA1c ≥ 7% despite basal insulin and other OADs. In GetGoal-Duo 1 (ClinicalTrials.gov record NCT00975286), lixisenatide or placebo was added to treatment with insulin glargine and metformin ± a sulfonylurea, glinide or thiazolidinedione [14], and in GetGoal-L (ClinicalTrials.gov record NCT00715624), lixisenatide or placebo was added to treatment with insulin glargine ± metformin [13]. GetGoal-L-C (ClinicalTrials.gov record NCT01632163) had the same design as GetGoal-L, but in a predominantly Asian population [16].

In all three studies, lixisenatide was administered as a once-daily injection starting at a dose of 10 μg. In GetGoal-L and GetGoal-Duo 1, the lixisenatide dose was increased at 1-week intervals, first to 15 μg and then to 20 μg if tolerated [13, 14]. In GetGoal-L-C, the lixisenatide dose was increased to 20 μg after 2 weeks on the starting dose [16].

In this analysis, participant-level data were pooled from all individuals of Asian ethnicity in the intent-to-treat populations of these studies.

The protocol of each GetGoal clinical trial included in this post hoc analysis, GetGoal-Duo1 [14], GetGoal-L [13], and GetGoal-L-C [16], was approved by the relevant institutional review boards or ethics committees of the participating centers, and each was conducted in accordance with the principles of the Declaration of Helsinki and Good Clinical Practice guidelines. All participants provided written informed consent.

Efficacy and Safety Outcomes

The primary endpoint in all three studies was the change in HbA1c from baseline to week 24 [13, 14, 16]. Secondary endpoints included the proportion of individuals achieving an HbA1c < 7% at 24 weeks, as well as changes from baseline to week 24 in FPG, 2-h PPG, 2-h PPG excursion, body weight, body mass index (BMI), and basal insulin dose, as well as the basal insulin dose at study endpoint. The proportions of individuals achieving composite endpoints at 24 weeks were also assessed: HbA1c < 7% with no symptomatic confirmed hypoglycemia; HbA1c < 7% with no severe hypoglycemia; HbA1c < 7% without body weight gain; HbA1c < 7% with no symptomatic confirmed hypoglycemia and without body weight gain; and HbA1c < 7% with no severe hypoglycemia and without body weight gain.

Safety outcomes were the incidence and frequency of hypoglycemia (severe hypoglycemia, any symptomatic hypoglycemia, and symptomatic confirmed hypoglycemia [blood glucose < 3.3 mmol/L]), as well as the frequency of gastrointestinal adverse events (nausea, vomiting, and diarrhea).

Subgroup Analyses

The main objective of this post hoc analysis was to analyze efficacy and safety outcomes among Asian individuals with apparent decline of β cell function or long duration of T2D, to explore an alternative treatment to improve diabetes outcomes. Study investigators considered the use of tertiles to define subgroups as inappropriate for this analysis because β cell function declines progressively in Chinese individuals with T2D (decline of 2% annually) until about 10 years after diagnosis, whereupon a marked decline occurs [11]. Therefore, the following subgroups were defined on the basis of duration of T2D: < 10 years (group 1); 10 to < 15 years (group 2); and ≥ 15 years (group 3).

Statistical Analysis

Demographics, clinical characteristics, and efficacy and safety outcomes were summarized using descriptive statistics, including mean and standard deviation for continuous variables and number and proportion for categorical variables.

Differences between T2D duration subgroups were assessed using the chi-square test for categorical variables, and one-way analysis of variance (ANOVA) or Kruskal–Wallis test for continuous variables, where appropriate. Analysis of covariance (ANCOVA) models were used to examine differences between lixisenatide and placebo in changes from baseline to week 24 in HbA1c, FPG, 2-h PPG, 2-h PPG excursion, body weight, BMI, and basal insulin dose for each T2D-duration subgroup. Least squares mean (LSM) and standard error (SE) for each treatment group, and LSM difference (SE, 95% confidence intervals [CI], and P values) between treatment groups, were reported. Logistic regression was used to test treatment difference in each T2D-duration subgroup for HbA1c < 7% at week 24, and odds ratios (ORs), 95% CIs, and P values were calculated. A multiplicative interaction term of T2D-duration subgroup × treatment group was added to the model to assess potential inconsistency of the treatment effect of lixisenatide across T2D-duration subgroups, and a P value for interaction was obtained using the F test.

Covariates in all of the regression analyses included randomization strata of HbA1c (< 8%, ≥ 8%), study (GetGoal-Duo 1, GetGoal-L, GetGoal-L-C), age, sex, and baseline HbA1c, FPG, 2-h PPG, BMI, basal insulin dose, and duration of diabetes.

A two-sided P value ≤ 0.05 was considered to be statistically significant. All statistical analyses were performed using SAS® v9.3 (Cary, NC, USA).

Results

Participants

In total, 555 (40.0%) of the 1389 participants in the GetGoal-L, GetGoal-L-C, and GetGoal-Duo 1 studies were of Asian ethnicity and were included in the present analysis (Fig. 1). The majority of participants were from China (n = 251; 45.2%), India (n = 158; 28.5%), and Korea (n = 101; 18.2%), and the remainder were from Canada (n = 25; 4.5%) or the USA, Malaysia, Taiwan, France, and the UK (n = 20; 3.6%).

Fig. 1
figure 1

Data source and flow in this post hoc analysis

Full size image

Participant Characteristics

Over half of the study population (n = 284; 51.2%) were in group 1 (< 10 years), while 154 participants (27.7%) and 117 participants (21.1%) were in groups 2 (10 to < 15 years) and 3 (≥ 15 years), respectively (Table 1). The mean ± SD duration of diabetes at baseline was 5.3 ± 2.6 years in group 1, 12.1 ± 1.4 years in group 2, and 19.8 ± 4.0 years in group 3. The groups were not significantly different with respect to the male-to-female ratio or mean HbA1c. As expected, mean age, duration of basal insulin treatment and insulin dosage, as well as prevalence of diabetes complications (diabetic retinopathy, diabetic sensory or motor neuropathy, diabetic nephropathy) increased in line with the mean duration of diabetes, and thus differences between groups were significant for these baseline characteristics (P < 0.001 for age and duration of basal insulin; P = 0.01 for insulin dosage; P < 0.001 for diabetes complication prevalence).

Table 1 Baseline characteristics by baseline duration of type 2 diabetes
Full size table

Consistent with their higher mean insulin dosage, participants in group 3 had lower mean baseline FPG levels (6.5 ± 1.8 mmol/L) compared with those in group 2 (7.0 ± 2.0 mmol/L) or group 1 (7.2 ± 2.1 mmol/L; P = 0.01 for the comparison between groups). Mean body weight and BMI were lower in group 3 at baseline than in the other groups; the P value indicated significance for both variables.

Efficacy

There were no significant differences between the T2D-duration subgroups in terms of change from baseline in HbA1c, FPG, 2-h PPG, 2-h PPG excursion, body weight, or BMI (all P values for interaction > 0.05; Fig. 2). However, a significant difference between the subgroups was detected for the change in insulin dosage (U/day) during the study (P value for interaction =  0.038). This difference was most likely driven by the larger decrease in insulin dosage with lixisenatide versus placebo observed in group 3 compared with group 1 or 2, and may potentially have been influenced by the differences in basal insulin dosage between groups that were present at baseline (Table 1).

Fig. 2
figure 2

Difference in clinical outcomes in the change from baseline at 24 weeks between lixisenatide and placebo add-on treatment to basal insulin therapy ± oral antidiabetic drugs, by duration of diabetes at baseline. BMI body mass index, CI confidence interval, FPG fasting plasma glucose, h hour, HbA1c glycated hemoglobin, LSM least squares mean, PPG postprandial glucose, U units

Full size image

Compared with placebo, lixisenatide was consistently associated with significant reductions in HbA1c, 2-h PPG, 2-h PPG excursion, and body weight from baseline at 24 weeks in all T2D-duration subgroups (Fig. 2). LSM differences between lixisenatide and placebo in the changes in HbA1c were − 0.39% (95% CI − 0.63, − 0.14; P = 0.003), − 0.53% (95% CI − 0.87, − 0.19; P = 0.003), and − 0.85% (95% CI − 1.21, − 0.48; P < 0.001) in groups 1, 2, and 3, respectively. At 24 weeks, LSM differences in 2-h PPG, 2-h PPG excursion, and body weight between lixisenatide and placebo were respectively − 3.52 mmol/L (95% CI − 4.31, − 2.73; P < 0.001), − 3.09 mmol/L (95% CI − 3.81, − 2.38; P < 0.001), and − 1.00 kg (95% CI − 1.38, − 0.62; P < 0.001) in the overall cohort (all duration subgroups combined; Table 2). In addition, lixisenatide significantly reduced BMI relative to placebo in groups 1 (LSM difference − 0.41 kg/m2; 95% CI − 0.61, − 0.21; P < 0.001) and 2 (LSM difference − 0.29 kg/m2; 95% CI − 0.54, − 0.04; P = 0.022), but not in group 3 (LSM difference − 0.32 kg/m2; 95% CI − 0.71, 0.06; P = 0.096; Fig. 2). Conversely, change in basal insulin dose was significantly different between lixisenatide and placebo in group 3 (LSM difference − 5.03 U/day; 95% CI − 7.82, − 2.24; P = 0.001), but not in groups 1 (LSM difference − 1.34 U/day; 95% CI − 2.72, 0.05; P = 0.058) or 2 (LSM difference − 1.11 U/day; 95% CI − 2.96, 0.74; P = 0.238; Fig. 2).

Table 2 Clinical responses to lixisenatide or placebo add-on treatment to basal insulin therapy ± oral antidiabetic drugs in the overall pooled Asian population
Full size table

Overall, lixisenatide was associated with significantly greater probability of achieving an HbA1c < 7% at week 24 compared with placebo (Table 2). The proportions of participants who achieved an HbA1c < 7% ranged from 27.1% to 42.6% with lixisenatide, and from 12.0% and 14.8% with placebo across the subgroups. The probability of achieving HbA1c < 7% was not significantly different between the T2D-duration subgroups (P value for interaction = 0.622).

No interaction between diabetes duration and the probability of achieving any of the composite outcomes was observed in participants receiving lixisenatide (Supplementary Material Fig. S1; P values for interaction all > 0.05). However, the achievement of composite outcomes was more frequent in group 3 than in group 1. Composite outcomes that included hypoglycemia tended to favor group 2 over group 3, reflecting the greater likelihood of symptomatic confirmed hypoglycemia in group 3 (see “Safety”). More participants receiving lixisenatide (Supplementary Material Fig. S2a) achieved the composite endpoints compared with participants receiving placebo (Supplementary Material Fig. S2b).

Safety

No episodes of severe hypoglycemia were reported (Table 3). The proportion of participants with symptomatic hypoglycemia was higher in group 3 than in the other subgroups, for both lixisenatide and placebo, and a significant effect of T2D duration was found (P = 0.001). These differences across the T2D-duration subgroups were maintained when numbers of hypoglycemia events were analyzed per patient-year (P = 0.007).

Table 3 Hypoglycemia and gastrointestinal adverse events by duration of type 2 diabetes at baseline
Full size table

Gastrointestinal adverse events, particularly nausea and vomiting, occurred in a higher proportion of lixisenatide than placebo recipients in all T2D-duration subgroups (Table 3). Among lixisenatide recipients, gastrointestinal adverse events as a whole affected more participants in group 3 (49.2%) than in groups 1 or 2 (34.2% and 40.0%, respectively). This pattern was also seen with nausea and (to a lesser extent) diarrhea, but not with vomiting, which occurred in a greater percentage of participants in group 2 than in groups 1 or 3.

Multivariable Regression Analysis

After adjustment for baseline characteristics (age, sex, FPG, 2-h PPG, HbA1c, BMI, and basal insulin dose per kg body weight), no significant differences were found between group 1 and either group 2 or group 3 with respect to changes in HbA1c, FPG, or 2-h PPG among participants receiving lixisenatide (Table 4). However, the odds of achieving an HbA1c < 7% were significantly greater in group 2 versus group 1 (OR 1.77; 95% CI 1.01, 3.10; P = 0.047). Furthermore, changes in body weight (estimated difference − 0.71 kg; 95% CI − 1.27, − 0.14; P = 0.014) and basal insulin dose (estimated difference − 1.78 U/day; 95% CI − 3.39, − 0.17; P = 0.030) were found to be significantly different between groups 1 and 3.

Table 4 Summary of multivariable regression results of the association of duration subgroup with different clinical responses in Asian participants with type 2 diabetes
Full size table

Discussion

This post hoc analysis of data from the GetGoal clinical trial program investigated the effect of T2D duration on the efficacy and safety of lixisenatide in people of Asian ethnicity. Overall, duration of T2D at baseline did not significantly influence the change in HbA1c achieved with lixisenatide in Asian participants over 24 weeks. In fact, the mean change in HbA1c from baseline was larger in group 3 than in group 1 (− 0.99% vs − 0.22%), but no significant difference between the subgroups was found in the multivariable regression analysis. It is encouraging that, while the dose of insulin tended to increase with T2D duration, suggesting a reduced β cell function in participants with a longer duration of disease, HbA1c was consistently improved with lixisenatide in all T2D duration groups. This highlights the ability of lixisenatide to reduce hyperglycemia in individuals with long-standing T2D due to other underlying mechanisms independent of β cell function [17], for example by regulating α cell function and slowing gastric emptying on top of increasing insulin secretion [18, 19]. Similarly, no effect of T2D duration was found with respect to change in 2-h PPG, suggesting that there is no reduction in the glycemic efficacy of lixisenatide over time in the Asian population. Lixisenatide also significantly increased the probability of achieving an HbA1c < 7% at week 24, compared with placebo, in all T2D-duration subgroups. Longer duration of diabetes was also associated with a numerically higher probability of achieving HbA1c < 7% and no symptomatic confirmed hypoglycemia, no body weight gain, or neither of these outcomes. For composite endpoints that included hypoglycemia, the effect was curtailed in group 3 because there was a higher incidence of hypoglycemia in that group.

The results of this analysis are in line with those of the primary publications for the GetGoal-Duo 1, GetGoal-L, and GetGoal-L-C studies, which showed that lixisenatide as add-on therapy to insulin ± OADs was associated with significantly greater reductions in HbA1c and PPG compared with placebo in individuals with T2D [13, 14, 16]. The LSM difference between lixisenatide and placebo in HbA1c change at 24 weeks in the overall population (− 0.51%) and in groups 1, 2, and 3 (− 0.39%, − 0.53%, and − 0.85%, respectively) observed in this analysis were in the same range as in the primary publications, where the between-group difference ranged from − 0.3% to − 0.5%. Furthermore, the LSM difference between lixisenatide and placebo in PPG change at 24 weeks in the overall population (− 3.52 mmol/L) was comparable with that observed in the primary publications, where the between-group difference ranged from − 3.2 to − 3.8 mmol/L [13, 14, 16].

In common with other GLP-1 receptor agonists, lixisenatide promotes body weight loss in people with T2D [20]. Accordingly, this analysis found a modest mean reduction in body weight of 0.61 kg over 24 weeks in those receiving lixisenatide, against an increase of 0.39 kg in placebo recipients. The magnitude of the effect of lixisenatide on body weight loss did not differ between T2D-duration subgroups (P value for interaction = 0.878); however, after adjustment for baseline factors including body weight in the multivariate regression analysis, a significantly greater reduction in weight was seen in group 3 compared with group 1.

As expected, participants with a longer duration of diabetes were receiving a higher basal insulin dose, both at baseline and at the end of the study. This may explain why individuals with a longer T2D duration (≥ 15 years) had a greater reduction at 24 weeks in their insulin dose than those with a shorter T2D duration (< 10 years; estimated difference − 1.78 U/day; 95% CI − 3.39, − 0.17; P = 0.030).

With regard to safety, an interaction between T2D duration and the incidence of symptomatic (and symptomatic confirmed) hypoglycemia for both placebo and lixisenatide recipients was observed. This was not unexpected, because older age and longer diabetes duration both increase the risk of hypoglycemia in treated individuals with T2D [21]. Encouragingly, no severe hypoglycemic events were reported in any participant. These hypoglycemia results, taken together with the significant insulin dose reduction observed in participants with a longer T2D duration compared with the other duration groups, suggest that, in clinical practice when adding lixisenatide treatment to a basal insulin regimen in individuals with long-standing diabetes, physicians should consider reducing their basal dose.

The frequency of gastrointestinal adverse events was also found to increase with diabetes duration among lixisenatide recipients, and this was largely driven by an increase in nausea. Analyses of other GLP-1 receptor agonists have not found an association between the frequency of gastrointestinal adverse effects and the duration of T2D [22,23,24]. Furthermore, in pooled analyses of the GetGoal program, older age was not associated with an increased frequency of gastrointestinal adverse events among those receiving lixisenatide [25]. The findings of this analysis are therefore unexplained and may warrant further investigation.

The impact of diabetes duration on the efficacy of GLP-1 receptor agonists has been investigated in a number of post hoc studies and meta-analyses [22,23,24, 26,27,28,29,30,31]. Although the glycemic response to GLP-1 receptor agonists depends on residual β cell function, evidence from these studies has consistently shown that increasing duration of diabetes does not affect the ability of these drugs to lower HbA1c [17, 24, 26, 28,29,30,31]. Furthermore, the benefits of GLP-1 receptor agonists in lowering body weight [24, 29] and cardiorenal risk [27] are not attenuated by increased diabetes duration. These benefits are clinically important given that Asian individuals, who characteristically have β cell dysfunction that predisposes them to an increased risk of diabetes [32], also have an increased cardiovascular disease risk mediated by an interaction between glucose intolerance and insulin resistance [33]. The results of the current analysis provide further evidence that GLP-1 receptor agonists, such as lixisenatide, are a clinically relevant treatment option for Asian individuals with T2D, including those with a diabetes duration exceeding 10 years. The reduction in hyperglycemia in individuals with long-standing diabetes may be due to other underlying mechanisms of action of GLP receptor agonists that are independent of β cell function [17]. Further prospective research into the effectiveness of lixisenatide and other GLP-1 receptor agonists among Chinese patients with long-standing diabetes and effect on cardiorenal risk is warranted.

There are some limitations to this analysis, including its post hoc nature and the imbalance in participant numbers between subgroups. This analysis was conducted in a specific population (Asian ethnicity) and in individuals with uncontrolled diabetes despite treatment with basal insulin. Accordingly, the findings of this analysis cannot be generalized to other settings or patient groups, nor indeed can any possible differences between Asian and non-Asian individuals regarding the influence of T2D duration on treatment outcomes be determined. Lastly, the current analysis focused exclusively on T2D duration and a prior report examined the effect of baseline BMI on the effectiveness and safety of lixisenatide in Asian patients [34]; however, the effect on treatment outcomes based on other baseline patient or disease characteristics in Asian individuals may be worth investigating, including gender, age, baseline HbA1c, presence/absence of diabetes-related atherosclerotic cardiovascular disease, and the number of background oral antidiabetic agents.

Conclusions

The findings of this post hoc analysis indicate that lixisenatide significantly reduces HbA1c, 2-h PPG, and body weight among Asian individuals also receiving basal insulin, without an increased risk of hypoglycemia, regardless of the duration of T2D at baseline. Lixisenatide can therefore be considered for use as add-on therapy in Asian people taking basal insulin, irrespective of time since diagnosis of T2D. While individuals with longer disease duration had a greater risk of symptomatic (but not severe) hypoglycemia than those with shorter disease duration, no additional safety concerns were observed. A reduction in basal insulin dose, or the introduction of a GLP-1RA, should be considered for individuals with long-standing diabetes.