Introduction

Diabetes is characterized by hyperglycemia and is diagnosed using HbA1c and plasma glucose levels [1,2]. Diabetes represents a major risk factor for microvascular and macrovascular complications [3], making effective glycemic management essential to reduce these risks [4,5]. In 2024, the global prevalence of diabetes was estimated at 589 million individuals, accounting for 11.1% of adults aged 20–79 years [6]. In Japan, approximately 10.8 million people are affected [6]. With 29.3% of the Japanese population aged 65 years and older [7] and diabetes incidence increasing with age, the number of affected individuals is expected to rise further [8]. Older patients often present with multiple comorbidities, including hypertension and dyslipidemia, which increase the risk of renal failure and cardiovascular disease [9].

International guidelines, such as the American Diabetes Association Standards of Care, recommend therapies, including metformin, sodium–glucose cotransporter-2 (SGLT2) inhibitors, and glucagon-like peptide-1 (GLP-1) receptor agonists, tailored to a patient’s pathophysiology [10]. In the United Kingdom, treatment guidelines advocate biguanides, particularly metformin, as first-line therapy for type 2 diabetes [11]. In contrast, dipeptidyl peptidase-4 (DPP-4) inhibitors remain the most frequently prescribed glucose-lowering agents in Japan, followed by biguanides and SGLT2 inhibitors [12]. This pattern likely reflects their efficacy in lowering HbA1c, low hypoglycemia risk, and high treatment persistence among Japan’s predominantly older population with diabetes [12–14]. DPP-4 inhibitors are generally not associated with increased gastrointestinal symptoms [15,16] or digestive system infections [17]. Previous trials, including meta-analyses, have reported a low risk of hypoglycemia with sitagliptin, a widely used DPP-4 inhibitor [14,16].

Imeglimin is a novel therapeutic agent that lowers blood glucose by enhancing insulin secretion and improving insulin sensitivity [18]. It was first approved for clinical use in Japan, with further development and regulatory approvals ongoing in other Asian countries. According to the Japanese Risk Management Plan, hypoglycemia and gastrointestinal symptoms are identified as notable potential risks [19]. In line with this, randomized controlled trials (RCTs) have demonstrated that while imeglimin effectively reduces HbA1c by approximately 0.46% [20], it is also associated with adverse effects such as diarrhea (5.4%), gastritis (2.7%), abdominal discomfort (2.7%), and hypoglycemia (2.7%) [21].

Despite its clinical approval, comparative real-world data on hypoglycemia and gastrointestinal events between imeglimin and widely used DPP-4 inhibitors, especially in older populations, remain limited. Therefore, in this study, we aimed to evaluate these outcomes among new users of imeglimin and sitagliptin using a retrospective cohort design based on primary community pharmacy data. To distinguish clinical from statistical significance, a 10% absolute difference in the incidence of gastrointestinal symptoms was prespecified as the minimum clinically relevant threshold. In the absence of a universally established standard for this outcome [22], this threshold was chosen as a pragmatic benchmark to guide potential changes in prescribing practices, informed by the known safety profiles of oral hypoglycemic agents. By conducting this analysis, we sought to characterize the comparative safety of these therapies in a real-world setting and provide clinically relevant evidence to support treatment selection.

Materials and methods

Results

Patients’ characteristics

In September 2022, 2023, and 2024, a total of 3,621 patients were prescribed imeglimin and 24,046 patients were prescribed sitagliptin. During patient selection, 437 patients in the imeglimin group and 224 patients in the sitagliptin group who switched between imeglimin and sitagliptin during the observation period were excluded. Finally, 1,140 (31%) imeglimin users and 990 (4%) sitagliptin users met the criteria for “new users” and were included in the study. The mean age was 66.9 years in the imeglimin group and 69.9 years in the sitagliptin group (Table 1). Average daily doses were 1,619.4 mg for imeglimin and 46.2 mg for sitagliptin. Mean observation periods were similar, at 99 days for imeglimin and 101 days for sitagliptin. Notable baseline differences, reflected by absolute SMDs > 0.1, were observed for age, dyslipidemia, concomitant use of other antidiabetics or insulin, lipid-lowering drugs, and hospitalization history (Table 1).

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Baseline characteristics.

https://doi.org/10.1371/journal.pone.0359960.t001

Events

The incidence of at least one reported event was 31% (351 cases) for imeglimin and 22% (220 cases) for sitagliptin (crude RR: 1.39; 95% CI: 1.20–1.60). Among imeglimin users, the most frequently reported adverse events were gastrointestinal symptoms (12.8%), decreased therapeutic response (6.1%), and nausea (3.6%; Table 2). Notably, imeglimin use was associated with a significantly higher risk of nausea (adjusted HR [aHR]: 15.34; 95% CI: 3.50–67.23), diarrhea (aHR: 7.90; 95% CI: 2.54–24.62), elevated blood glucose (aHR: 4.57; 95% CI: 1.20–17.34), and abdominal discomfort (aHR: 6.02; 95% CI: 1.47–24.58). This association may reflect confounding by indication or temporary worsening of glycemic control during treatment discontinuation or medication switching rather than a direct pharmacological effect of imeglimin. Additionally, events such as vomiting (1.4%), soft stools (1.1%), and dyspepsia (0.4%) were reported exclusively in the imeglimin group. The Kaplan-Meier curves illustrating the cumulative incidence of gastrointestinal events during follow-up are presented in Fig 1.

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Incidence of common events among imeglimin and sitagliptin users.

https://doi.org/10.1371/journal.pone.0359960.t002

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Kaplan-Meier curves for the cumulative incidence of gastrointestinal events.

The curves show the cumulative incidence of gastrointestinal symptoms in patients newly prescribed imeglimin (dashed gray line) compared with that in patients prescribed sitagliptin (solid black line) during the follow-up period. The hazard ratio, 95% confidence interval, and p-value displayed in the figure were derived from the multivariable Cox proportional hazards model.

https://doi.org/10.1371/journal.pone.0359960.g001

Among sitagliptin users, the most common events were decreased therapeutic response (4.3%), improvement in pre-existing conditions (1.8%), and inadequate control of diabetes mellitus (1.4%). The rate of improvement of pre-existing conditions was higher in the sitagliptin group than in the imeglimin group (aHR: 0.25; 95% CI: 0.09–0.70; Table 2). To minimize potential confounding effects of concomitant antidiabetic medications, we performed sensitivity analyses restricted to patients receiving monotherapy (imeglimin, n = 114; sitagliptin, n = 561). In this monotherapy subpopulation, 19 gastrointestinal events occurred in the imeglimin group and 15 among sitagliptin users. The sensitivity analyses showed an elevated risk of gastrointestinal events among imeglimin users (aHR: 5.36; 95% CI: 2.63–10.91), with an absolute risk difference of 13.4% (95% CI: 7.6%–21.2%). Similarly, the aHR (5.20; 95% CI: 2.55–10.61) estimated using Firth’s penalized likelihood approach for Cox regression was comparable to the primary analysis. The results of the sensitivity analyses using Firth’s penalized likelihood approach for all events were also consistent with the primary analysis (S1 Table).

Add-on medications

Among imeglimin users, 25.3% (288 patients) had new medications added, compared with 31.8% (315 patients) of sitagliptin users (crude RR: 0.79; 95% CI: 0.69–0.91). Table 3 shows that the most common add-on medications among imeglimin users were non-insulin antidiabetic drugs (9.5%) and insulin or GLP-1 receptor agonists (2.6%). Imeglimin users showed a tendency toward higher use of antinauseants and non-insulin glucose-lowering agents, although this difference was not significant (Table 3). The results of sensitivity analyses using Firth’s penalized likelihood approach for all events were consistent with those of the primary analysis (S2 Table).

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Incidence of any add-on medication use among imeglimin and sitagliptin users.

https://doi.org/10.1371/journal.pone.0359960.t003

Discussion

In this retrospective cohort study, we evaluated the incidence of treatment-related events among new users of imeglimin and sitagliptin, with particular focus on patients aged ≥65 years. Cox proportional hazards analysis showed no significant difference in hypoglycemia risk between the two groups. Notably, no significant difference in hypoglycemia risk was observed between imeglimin and sitagliptin, despite a higher proportion of insulin use and greater clinical severity in the imeglimin group, which is consistent with the favorable safety profile reported in previous studies. However, imeglimin was associated with a markedly higher risk of gastrointestinal symptoms, including nausea, diarrhea, vomiting, and abdominal discomfort, than sitagliptin. The approximately 1.5-fold higher discontinuation rate observed with imeglimin likely reflects the increased incidence of gastrointestinal events and suboptimal glycemic control. Although some confidence intervals were wide owing to the limited sample size, these findings underscore the need for careful monitoring of gastrointestinal adverse events during the first 4 months of imeglimin therapy, particularly in older patients with type 2 diabetes.

This study is among the first real-world investigations to directly compare hypoglycemia risk between imeglimin and sitagliptin. Previous RCTs [21,23,30–36] evaluating each agent versus placebo reported no significant increase in hypoglycemia risk. The present findings are consistent with these reports, suggesting that imeglimin has a hypoglycemia safety profile comparable to that of sitagliptin.

Importantly, although the imeglimin group included a significantly higher proportion of insulin users (19.5% vs. 5.9%) than the sitagliptin group, hypoglycemia risk remained similar between groups. This observation further supports the favorable safety profile of imeglimin, even among patients with more advanced disease or those requiring complex treatment regimens.

Regarding gastrointestinal tolerability, prior studies [21,30] have reported a 1.3- to 1.5-fold increased risk of gastrointestinal symptoms with imeglimin compared with placebo, whereas meta-analyses [15] indicate that sitagliptin is associated with risks of nausea and diarrhea comparable to placebo. In contrast, the present real-world analysis demonstrated a five-fold higher risk of gastrointestinal symptoms among imeglimin users than among sitagliptin users. This finding was supported by sensitivity analyses restricted to monotherapy users. Although the multivariable-aHR for antinauseant use showed a relatively high point estimate (aHR: 2.35), the association was not significant, and the CI was wide, reflecting limited precision. However, the absolute incidence of antinauseant use was low in both the imeglimin (1.0%) and sitagliptin (0.8%) groups, corresponding to an absolute risk difference of only 0.2%. These findings indicate that while there may be a slight trend toward increased use of symptomatic medications, the absolute number of patients requiring antinauseant therapy was quite small in both groups, suggesting that these gastrointestinal events were generally manageable.

This study has several methodological strengths. Primary data collection by community pharmacists enabled detection of mild events, such as soft stools and abdominal distension, which are often underreported in large claims databases. In addition, the inclusion of detailed lifestyle data (e.g., smoking status, alcohol consumption, and OTC medication use) enhanced internal validity through more comprehensive adjustment. Generalizability is supported by the participation of approximately 10% of pharmacies affiliated with the JPA nationwide [37]. Furthermore, our sensitivity analyses using Firth’s penalized likelihood approach confirmed the robustness of the estimates despite the limited number of events.

This study has several limitations that warrant consideration. First, there are concerns regarding the potential for selection and attrition biases. Because new users were identified by pharmacists according to three predefined eligibility criteria, one of which required at least one follow-up visit, we could not determine the exact number of patients excluded owing to a single pharmacy visit (lost to follow-up). If patients in either group dropped out early owing to intolerable adverse events, this could introduce attrition bias. Additionally, the proportion of patients excluded because they were “switchers” between the study drugs was substantially higher in the imeglimin group (12%) than in the sitagliptin group (0.9%). This suggests that the imeglimin group may have included patients with more severe or refractory diabetes, and residual confounding by indication may have remained despite multivariable adjustments. Second, the non-significant difference in hypoglycemia risk, coupled with the wide CI (0.20–2.22), suggests limited precision, likely owing to the limited number of hypoglycemic events. Third, owing to the nature of our data collection approach, we only obtained information on the concomitant use of non-insulin antidiabetic medications or insulin rather than the specific names of the drugs. A nationwide Japanese study reported that while the use of GLP-1 receptor agonists is extremely low (<1%), biguanides (metformin) are prescribed in approximately 16% of patients [12]. Although our sensitivity analyses restricted to monotherapy users showed consistent results, we cannot completely rule out the possibility of residual confounding by specific concomitant medications in the overall cohort. In particular, since gastrointestinal events represent the most common side effects of metformin [38], the concomitant metformin use may have influenced the overall incidence observed for these gastrointestinal events. Fourth, the 4-month follow-up was shorter than that (9–13 months) typically observed in certain RCTs [31,34], potentially limiting the detection of long-term adverse outcomes. Fifth, the absence of longitudinal laboratory data (e.g., HbA1c) precluded comparative assessment of glycemic efficacy. Sixth, although a higher incidence of elevated blood glucose was observed in the imeglimin group, it is unclear whether this finding reflects confounding by indication, owing to more advanced or refractory disease, or clinical factors related to medication switching or treatment discontinuation. Seventh, incomplete event capture could occur if patients used multiple pharmacies; this risk was mitigated by restricting the analysis to “repeat users” of participating pharmacies. While awareness of imeglimin as a novel agent may have influenced reporting, the new-user design in both cohorts likely minimized such disparities. Finally, while sitagliptin was selected as a representative comparator owing to its highest prescription rate in Japan, it is difficult to attribute these findings to a class effect of DPP-4 inhibitors. Although DPP-4 inhibitors share a common pharmacological mechanism, individual agents may exhibit slightly different safety profiles, and further studies involving other DPP-4 inhibitors are required to confirm class-wide comparative safety against imeglimin.

Conclusion

This study revealed no significant difference in the risk of hypoglycemia between imeglimin and sitagliptin. However, imeglimin use was associated with a higher incidence of gastrointestinal adverse events, with the absolute risk difference closely approaching our prespecified threshold of 10% for clinical relevance. These findings suggest that patients initiating imeglimin therapy may benefit from careful monitoring for gastrointestinal symptoms, particularly during the first 4 months of therapy. Additional research is warranted to assess whether individuals with a history of gastrointestinal disorders are at an increased risk for these adverse events.

Supporting information

Acknowledgments

The authors wish to express their appreciation to the pharmacists of the participating pharmacies for providing data. We thank the staff of the Japan Pharmaceutical Association (Tokyo, Japan) for data collection and management.