Background
The high prevalence of type 2 diabetes mellitus (T2DM) is a worldwide public health concern. Asian countries are currently facing a greater burden of T2DM; more than 60 % of the world’s T2DM patients are in Asia [
1].
Efficacy evaluation of treatment with diabetes commonly uses hemoglobin A1c (HbA1c); however, the level of this index does not allow for evaluation of treatment effects on insulin secretion or insulin resistance [
2‐
4]. β-cell function in patients with T2DM is approximately 50 % that of healthy individuals at the time of diagnosis, and decreases yearly thereafter [
5]. Hence, evaluation of β-cell function as well as HbA1c is desirable for assessment of drug efficacy in treatment of diabetes. An important characteristic of T2DM is elevation of the fasting glucose level as well as that of postprandial glucose level, which are mainly affected by postprandial insulin secretion. Insulinogenic index is commonly used to assess early-phase insulin secretion in response to glucose [
6‐
8]. It is reported that a reduced insulinogenic index represents the main abnormality in the transition from normal glucose tolerance (NGT) to T2DM, resulting in the elevation of postprandial glucose levels in Asian subjects [
7‐
12]. In addition, maintenance of an appropriate insulinogenic index decreases incidence of microalbuminuria in T2DM [
13]. Considered together, these findings suggest that the insulinogenic index may be a critical factor in progression to T2DM, maintenance of postprandial glucose levels, and prevention of the complications of diabetes.
The American Diabetes Association and European Association for the Study of Diabetes consensus for treating T2DM recommend biguanides as first-line therapy [
14]. This treatment, however, has not been established in Asian countries, including Japan [
15,
16]. Unlike the insulin resistance seen in Caucasians, Asian patients with T2DM have a relatively low BMI and a predominant insulin secretory defect [
3,
7‐
12,
17‐
23]. Therefore, insulin secretagogues, particularly sulfonylureas and dipeptidyl peptidase-4 (DPP-4) inhibitors are widely used in Japan [
24]. Meta-analysis has shown that DPP-4 inhibitors are more effective in Asian compared to non-Asian patients [
23]. Effectiveness of DPP-4 inhibitors on insulin secretion stimulated by glucose for 12-week has shown in Korean patients with T2DM [
25]. However, little is known the effects of DPP-4 inhibitors on the insulinogenic index as the primary endpoint compared to sulfonylureas monotherapy.
We conducted a multicenter, randomized controlled trial to compare the effect of glimepiride and sitagliptin on the insulinogenic index after 52-week treatment in Japanese patients with T2DM.
Subjects and methods
Trial design and participants
A randomized, open-label, parallel-group trial was conducted over a period of 52 weeks from February 10, 2011 to March 31, 2013 at 18 centers across Japan, including clinics and general and university hospitals. Eligibility criteria were outpatients with T2DM aged < 80 years with an HbA1c level < 8.4 % who had received no pharmacological treatment for diabetes for at least 1 month prior to participation in this trial. Exclusion criteria were renal or liver dysfunction, pancreatic or hematological operation, severe complications of diabetes, being pregnant or possibly pregnant, malignancy under treatment and medications known to affect glucose metabolism.
Ethics
The protocol was approved by the University hospital Medical Information Network (UMIN) (Clinical Trial Registry No. 000004791), the Ethics Committee of Kyoto University Graduate School and Faculty of Medicine, as well as the Ethics Committee of each study center. The trial was performed in accordance with the Declaration of Helsinki upon obtaining written informed consent from all participants, and was reported in accordance with the Consolidated Standards of Reporting Trials (CONSORT) Statement [
26].
Intervention and maintenance
Each participant was administered glimepiride (titrated upward to 1.0 mg) or sitagliptin (titrated upward to 100 mg) once daily in the morning for 52 weeks. The starting dose was decided by the respective physicians based on the baseline condition of each participant. When HbA1c levels exceeded 6.9 % after 6 months or later, glimepiride and sitagliptin doses were increased to each titrated dose. Physicians were allowed to decrease the doses at any point to prevent the occurrence of a hypoglycemic event. On the other hand, if participants did not meet the specified glycemic control criteria with the setup dose, physicians were allowed to add or switch medications and the participants were discontinued from the trial.
Outcome measurements
The primary outcome measurement was the difference in post-treatment insulinogenic index between the two groups. Secondary outcome measurements were the levels of plasma glucose (PG) (mmol/l), immunoreactive insulin (IRI) (pmol/l), C-peptide (CPR) (nmol/l), glucagon (ng/l) (Millipore Corporation, Bilerica, MA), and insulin sensitivity index (ISI; an index of insulin resistance) during 75 g oral glucose tolerance tests (OGTTs) before and after 52-week treatment [
27]. In addition, HbA1c (%), glycated albumin (GA) (%), and BMI (kg/m
2) after treatment also were evaluated as secondary outcome measurements. Each outcome was calculated as follows: HbA1c was expressed as a National Glycohemoglobin Standardization Program (NGSP) equivalent value calculated by the following formula: HbA1c (NGSP value) (%) = 1.02 × HbA1c (Japan Diabetes Society value) (%) + 0.25 [
28]. The estimated glomerular filtration rate (eGFR) (ml/min/1.73 m
2) was calculated by 194 × serum Cr
−1.094 × age
−0.287 for men, 194 × serum Cr
−1.094 × age
−0.287 × 0.739 for women [
29]. The insulinogenic index was calculated by the following equation: (IRI at 30 min - fasting IRI)/(PG at 30 min—fasting PG) [
6‐
8]. ISI composite was calculated by 10,000/{[fasting PG (mmol/l) × fasting IRI (pmol/l) × mean 75 g OGTT PG (mmol/l) × mean 75 g OGTT IRI (pmol/l)]
0.5}1/3 [
27].
Sample size
Given the lack of differences and variance in the insulinogenic index between two similar groups in a previous study, an effect size of 0.6, which is conventionally accepted as a medium effect, was used to calculate an appropriate sample size. We estimated that 100 participants would provide at least 80 % power to detect a statistically significant difference (α = 0.05, two-sided test, and withdrawal rate of 10 % per year) between the two groups.
Randomization
We used the UMIN system, a computer-generated random sequence, to assign participants to either glimepiride or sitagliptin in a 1:1 ratio by minimization, based on sex, center, age, and HbA1c. Collaborating physicians enrolled the participants, and during the follow-up period, this trial was performed without blinding. That is, both physicians and participants were aware of which drug was allocated.
Procedures
Upon obtaining informed consent, OGTTs were performed before (0 week) and after 52-week treatment. The levels of PG, IRI, CPR, and glucagon were measured at 0, 15, 30, 60, and 120 min (min) during OGTTs. After treatment, OGTTs were performed with a 24-h washout period. GA was measured at 0 and 52 weeks, and glutamic acid decarboxylase (GAD) antibody was measured at 0 week. HbA1c, PG, body weight, and clinical biochemical tests were measured (0, 4, 12, 24, 36, and 52 weeks). Safety monitoring for hypoglycemia was performed during treatment. All samples were labeled with a code assigned to each participant and routinely analyzed at a laboratory of the SRL Corporation (Tokyo, Japan).
Statistical analyses
All statistical analyses were performed with a blind procedure by an independent third party, Statcom Company Limited. For the primary outcome measurement, the main analysis was performed on the full analysis set (FAS) of all randomized participants who took medications as allocated and underwent the OGTTs before and after treatment, excluding those with a hemolyzed sample, those who were added or changed therapy, and those who were withdrawn from the trial before treatment following consent acquisition. Subgroup analysis was performed on the per-protocol set (PPS), which excluded positive result of GAD antibody, protocol violations, or poor compliance from the FAS. Analysis of covariance (ANCOVA) was used to evaluate the primary outcome measurement based on baseline log-transformed insulinogenic index, and allocation variables, including age, sex, and HbA1c as covariates. For secondary outcome measurements, repeated measures analysis using a mixed model with terms for visit, treatment, and interaction was performed for OGTT, HbA1c, and BMI, including baseline values as covariates. Least-squares means (lsmeans) with 95 % confidence intervals (CIs) were obtained from the model, estimating from the mixed model. Other secondary outcome measurements were compared between the two groups using Analysis of variance (ANOVA) for the evaluation of GA and ISI. As an exploratory analysis, the change of insulinogenic index in each group was analyzed by Paired t test. The achievement rate of HbA1c < 7.0 % between the two groups were compared by Fisher’s Exact test. All statistical analyses were performed using SAS version 9.2 (SAS Institute Inc, Cary, NC) and JMP® 9 (SAS Institute Inc., Cary, NC, USA). No interim analysis was performed.
Discussion
In this trial, sitagliptin monotherapy resulted in significantly higher insulinogenic index compared with that of glimepiride monotherapy in Japanese patients with T2DM after 52-week treatment (Fig.
2a). It is reported that the DPP-4. Inhibitors conserved β-cell function in patients with T2DM and autoimmune diabetes [
30,
31]. Our result is possibly due to conserving β-cell function by DPP-4 inhibitor. In addition, the insulinogenic index was negatively correlated with glucose levels at 60 min during OGTT after treatment (Fig.
2b). This result suggests that maintenance of the insulingenic index is important to preserve postprandial glucose levels in early T2DM and indicates that sitagliptin may have a better effect on lowering the postprandial plasma glucose levels than glimepride in Asian patients with T2DM. It was reported that DPP-4 inhibitors including sitagliptin suppress glucagon secretion [
32,
33]. However, there were no effects of sitagliptin on glucagon secretion in this trial. This result suggests that the effect of sitagliptin on insulin secretion but not on glucagon secretion might continue after a drug washout period of 24 h after 52-week treatment.
Insulin secretion stimulated by sulfonylureas is independent of the glucose concentration, while DPP-4 inhibitors increase an active form of incretin peptide by DPP-4 inhibition and potentiate glucose-dependent insulin secretion [
34‐
36]. Accordingly, sulfonylureas are associated with risk of hypoglycemia and weight gain; DPP-4 inhibitors are associated with lower frequency of hypoglycemia and are weight neutral [
34‐
36]. In this trial, there was no hypoglycemia or weight gain in the sitagliptin group (Tables
1 and
2). However, it should be noted that glimepiride treatment also did not induce weight gain or incidence of severe hypoglycemia (Tables
1 and
2). This might be attributed to the low-dose of glimepiride used. Based on the Japanese claims database, 72 % of patients with single sulfonylurea treatment used 1.0 mg/day for 1 year, and 70 % of the patients treated with 1.0 mg/day of glimepiride achieved 1.0 % reduction of HbA1c [
37]. Thus, the dose of glimepiride was set to 1.0 mg/day in this trial. Consequently, a similar reduction in HbA1c levels was observed during 52-week treatment and no significant differences were found in HbA1c and GA levels between the two groups (Table
2, Additional file
2: Figure S2). These results made it possible to evaluate insulin secretion under the same glucose control conditions. Accordingly, the dose selection of less than 1.0 mg/day of glimepiride, the effect of which on HbA1c reduction is comparable to that of less than 100 mg/day of sitagliptin, seemed reasonable in this trial. In addition, our results also suggest that in regard to HbA1c-lowering efficacy, a low-dose of glimepiride is similarly effective as sitagliptin without weight gain or severe hypoglycemia at the early stage of T2DM with low insulin secretion (Table
2 and Additional file
2: Figure S2). Although a similar reduction in HbA1c and GA was achieved, post-challenge plasma glucose levels were significantly lower with sitagliptin than with glimepiride (Table
2, Fig.
3a, and Additional file
2: Figure S2). This result indicates that sitagliptin has better effects on insulinogenic index after 24-h wash out period at 52-week treatment.
This trial has several limitations. First, low-dose glimepiride was used in Japanese patients with T2DM who have low BMI and insulin secretion [
3,
7‐
9,
12,
17,
22‐
25]. The maximum dose of glimepiride was set at 1.0 mg/day to achieve the same improvement of HbA1c between the two groups. Although a significantly better level of insulinogenic index was shown by glimepiride treatment (p < 0.05, data not shown), it is unknown to what extent high-dose glimepiride improves the insulinogenic index or HbA1c in Japanese patients with T2DM. Second, meal tolerance tests (MTTs) were not performed in this trial. The primary endpoint is to evaluate early-phase insulin secretion in response to glucose. The insulinogenic index has already been established as an index of early-phase insulin secretion during OGTT [
6‐
8]. However, the indices of the meal-stimulated insulin secretion by MMTs have not established, mainly because the total calories and contents of the meal differ among the previous studies [
32,
33]. Third, the follow up rate of 78 % might be relatively low. The main reason is hemolysis of samples, which are important for the calculation of insulinogenic index. If we had planned to use multiple imputation method for missing data in the protocol, we might obtain higher follow up rate. The strength of this trial lies in its design, i.e., a multicenter, randomized, controlled trial involving clinics and a university hospital, which enhances the generalizability of our results. Furthermore, an active-controlled trial comparing the most widely used insulin secretagogues is practical for a daily clinical setting. This trial focused on the pathophysiology and treatment efficacy of T2DM in Asia. Our finding of a better effect on early-phase insulin secretion is of clinical importance for Asian patients with T2DM.
In conclusion, sitagliptin showed better effects on insulinogenic index after 52-week treatment compared to glimepiride in Japanese patients with T2DM. Further research is required to assess early-phase insulin secretion in patients treated with these drugs for longer period.
Authors’ contributions
Author contributions were as follows: YK designed the study, wrote the research protocol, collected and managed the research data, interpreted the data, and drafted the initial manuscript; NH designed the study, wrote the research protocol, collected and managed the research data, interpreted the data, and drafted, reviewed, and edited the manuscript; AH, SK, KY, EO, SH, HY, YF, NK, and YN designed the study and collected data; FM and MS planned and performed the statistical analysis; SH managed the research data, interpreted the data, and drafted, reviewed, and edited the manuscript; TN and NI designed the study, drafted the research protocol, managed the research data, interpreted the data, and drafted, reviewed, and edited the manuscript. Nobuya Inagaki is the guarantor of this trial. All authors read and approved the final manuscript.
Acknowledgements
This trial was supported by the Japan Diabetes Foundation. Nobuya Inagaki received research grant from the Japan Diabetes Foundation for sponsoring this trial; the Japan Diabetes Foundation was not involved in the study design, data collection, analysis or interpretation of the results, or preparation of the manuscript.
The authors cordially thank the participants and collaborators of the MAIKO study for their contributions to this project. The complete list of collaborators in the MAIKO Study is as follows: Nobuya Inagaki, Kazuaki Nagashima, Shin-ichi Harashima, Kentaro Toyoda, Akihiro Hamasaki, Norio Harada, Masahito Ogura, Yoshihito Fujita, Shunsuke Yamane, Kazuataka Nagai, Yuichi Sato, Kaori Ikeda, Atsushi Muraoka, Ayano Shiho, Daniela Nasteska, Yaeko Kondo, Takanari Harada, Kimitaka Shibue, Erina Joo, Kazuki Sasaki, Fusako Uwamori, Ayako Tsutano, Michiko Kontani, Ayako Yasui, Noriko Shinozuka, Yuko Nakahigashi (Department of Diabetes, Endocrinology and Clinical Nutrition, Kyoto University); Shizuka Kaneko (Takatsuki Red Cross Hospital); Koichiro Yasuda, Motoya Koie, Tomohiro Santo, Hijiri Morita (Osaka Saiseikai Noe Hospital); Kyohei Maruyama, Masaya Hosokawa, Yumiko Tahara, Yaeko Kondo (Kyoto Kizugawa Hospital); Nobuhisa Mizuno, Eiichi Ogawa (Shiga Medical Center for Adults); Hiroko Yoneda (The Japan Baptist Hospital); Norikazu Kitano, Makiko Numa, Ayako Tazoh (Hyogo Prefectural Tsukaguchi Hospital); Yoshio Nakamura (Hyogo Prefectural Amagasaki Hospital); Gen Inoue, Shogo Funakoshi, Tomonao Hirobata (Wakayama Red Cross Hospital); Hideki Yano, Akira Kuroe (Hikone Municipal Hospital); Takashi Matsuoka, Atsushi Yoshida (Kurashiki Central Hospital); Motozumi Okamoto, Takao Taniguchi, Miki Araki (Otsu Red Cross Hospital); Shibue Kimitaka, Yasushi Kondo (Tanabe Central Hospital); Norio Harada (Nagisa Clinic); Fujita Yoshihito (Nishio Clinic); Mikiko Okamoto, Kazuyo Suzuki (Kyoto Minami Hospital); Tadasumi Nakano (Mitsubishi Kyoto Hospital); Akira Takahashi (Nakamura Hospital); Fujio Matsuo, Megumi Shinji (Statcom Company Limited); Shimpei Fujimoto (Department of Endocrinology, Metabolism and Nephrology, Kochi Medical School.), Shiro Hinotsu (Center for Innovative Clinical Medicine, Okayama University Hospital); and Takeo Nakayama (Department of Health Informatics, School of Public Health, Kyoto University).