Efficacy and Safety of Sitagliptin Added to Insulin in Japanese Patients with Type 2 Diabetes: The EDIT Randomized Trial

Aims To clarify the efficacy and safety of adding sitagliptin to insulin therapy in Japanese patients with suboptimally controlled type 2 diabetes (T2DM). Study Design and Methods This was a 24-week, prospective, randomized, open-labeled, controlled trial. Patients with T2DM who were suboptimally controlled despite receiving at least twice daily injection of insulin were enrolled in the study. The patients were randomized to continuation of insulin treatment (Insulin group) or addition of sitagliptin 50 to 100 mg daily to insulin treatment (Ins+Sita group). The primary outcome was change in HbA1c at week 24. Results Adding sitagliptin to insulin significantly reduced HbA1c from 7.9 ± 1.0% at baseline to 7.0 ± 0.8% at week 24 (P <0.0001), while there was no significant change in HbA1c in the Insulin group (7.8 ± 0.7% vs. 7.8 ± 1.1%, P = 0.32). The difference in HbA1c reduction between the groups was 0.9% (95% confidence interval, 0.4 to 1.5, P = 0.01). There was no significant weight gain in either group. Incidence of hypoglycemia was significantly reduced in the Ins+Sita group compared with the Insulin group. Treatment satisfaction was improved in the Ins+Sita group. Baseline HbA1c level and beta cell function were associated with the magnitude of reduction in HbA1c in the Ins+Sita group. Conclusion Adding sitagliptin to insulin reduced HbA1c without weight gain or increase in hypoglycemia, and improved treatment satisfaction in Japanese patients with T2DM who were suboptimally controlled despite at least twice daily injection of insulin. Trial Registration The University Hospital Medical Information Network (UMIN) Clinical Trials Registry UMIN000004678


Introduction
Type 2 diabetes (T2DM) is characterized by beta cell dysfunction and insulin resistance [1].It is a progressive disease, and most patients with T2DM eventually require insulin therapy to achieve optimal glycemic control [2].Insulin is the most effective glucose-lowering agent; however, since increased risk of hypoglycemia, weight gain, and fear or unwillingness to inject limits optimization of the dose and number of insulin injections, many patients treated with insulin still do not achieve their glycemic goal [2][3][4].
Dipeptidyl peptidase-4 (DPP-4) inhibitors slow the degradation of incretin hormones, glucagon-like peptide-1 (GLP-1) and glucose-dependent insulinotropic polypeptide (GIP), and thereby enhance the action of endogenous incretin [5].Since incretin hormones stimulate insulin secretion in a glucose-dependent manner, DPP-4 inhibitors improve hyperglycemia without an increase in risk of hypoglycemia and weight gain [6].DPP-4 inhibitors have also been shown to improve glucagon dynamics [7].Furthermore, animal studies suggest that chronic exposure to GLP-1 may increase beta cell mass by promoting proliferation and differentiation and inhibiting apoptosis of beta cells [8].
Based on these characteristics of DPP-4 inhibitors, adding a DPP-4 inhibitor to insulin is expected to improve glycemic control without an increase in risk of hypoglycemia and weight gain.Previous studies conducted in USA and Europe have shown that adding a DPP-4 inhibitor to insulin in patients with T2DM reduced HbA1c [9][10][11], but the incidence of hypoglycemia was increased in one study in which sitagliptin was used [9].Moreover, since the glucose-lowering effect of DPP-4 inhibitors appears to be greater in Asians compared with Caucasians [12], the efficacy and safety of DPP-4 inhibitors added to insulin need to be clarified in the Asian population as well as in other ethnic groups.
In Japan, the first DPP-4 inhibitor, sitagliptin, has been marketed since 2009.Therefore, in this study we aimed to evaluate the efficacy and safety of adding sitagliptin in Japanese patients with T2DM whose glycemic control is suboptimal despite insulin therapy.

Research Design and Methods Subjects
The protocol for this trial and supporting CONSORT checklist are available as supporting information; see S1 CONSORT Checklist, S1 and S2 Protocol.We enrolled outpatients with T2DM who had been treated with at least twice daily injections of insulin for at least 2 months, and who had a suboptimal level of glycated hemoglobin (HbA1c) (i.e., !7.0%, according to the guidelines of the Japan Diabetes Society (JDS) [13]) between January 2011 and March 2013.A total of 54 patients were enrolled in the study, but four patients withdrew their consent before randomization (Fig. 1).No patients had been treated with sitagliptin or other DPP-4 inhibitors prior to study entry, since combination therapy with sitagliptin, the first approved DPP-4 inhibitor in Japan, and insulin was not approved in Japan until September 2011.Patients with type 1 diabetes, patients who had changed oral hypoglycemic agents within 2 months prior to study entry, and patients with severe liver disease or renal failure were excluded from the study.Written informed consent was obtained from all patients.This study was conducted according to the principles expressed in the Declaration of Helsinki and was approved by the ethical review committee of Keio University School of Medicine, Tokyo, Japan.The study was registered in the University Hospital Medical Information Network (UMIN) Clinical Trials Registry (http://www.umin.ac.jp/ctr/) as "Efficacy and safety of DPP-4 inhibitor added to Insulin treatment in patients with type 2 diabetes (EDIT) study; UMIN000004678".

Study design
This was a 24-week, prospective, randomized, open-labeled, controlled trial.A total of 50 subjects were randomized to the insulin (Insulin) group or the insulin plus sitagliptin (Ins+Sita) group based on a table of random numbers (simple randomization), and the treatment allocation was blinded to both attending physicians and patients until day 0 of the study.Subjects in the Insulin group continued their previous treatment during the study, and the insulin dose was adjusted by the attending physicians at each visit according to the guidelines of the Japan Diabetes Society (JDS) to achieve HbA1c below 7.0% as well as fasting plasma glucose levels below 130 mg/dL and 2 h postprandial plasma glucose level below 180 mg/dL [13].In subjects in the Ins+Sita group, sitagliptin 50 mg once daily was added to the previous treatment at week 0, and if the patient randomized to the Ins+Sita group was treated with a sulfonylurea or glinide, these drugs were discontinued at week 0. Further, insulin dose, regardless of the type of insulin, was decreased by 20% at week 0 if the patient's HbA1c was less than 8.0% at week 0.Then, the insulin dose was adjusted by the attending physicians at each visit during the study according to the JDS guidelines [13].Adherence to sitagliptin treatment (75% or greater administration) was confirmed at each visit by assessing the medication possession ratio (% MPR; 99.1 ± 2.7%).If HbA1c did not achieve a level below 7.0% at week 12, the attending physicians were able to uptitrate the dose of sitagliptin to 100 mg daily.Other treatment including lifestyle modification was continued in both groups during the study.Patients in both groups were asked to visit the clinic every 4 weeks and received standard care similarly, except that patients allocated to the Ins+Sita group were also asked to attend for an additional visit at week 2, when the attending physicians judged whether further reduction in insulin was needed.The primary outcome of the study was change in HbA1c at week 24.
Blood and urine sampling was conducted at weeks 0, 4, 8, 12, 16, 20 and 24, together with body weight and blood pressure measurement.The patients were allowed to continue their self-monitoring of blood glucose (SMBG) during the study and instructed to measure glucose level if they felt hypoglycemic.In addition, the patients were asked to conduct 7-point SMBG at week 0 and 24 to assess the daily glycemic profile.Seven-point SMBG was conducted seven times a day, before and 1 h after three meals and at bedtime, for two consecutive days, and mean values were used for analysis.The incidence and severity of hypoglycemia were assessed by a questionnaire in the patients at each visit.Hypoglycemia was defined as having hypoglycemic symptoms and/or blood glucose less than 70 mg/dL.Severe hypoglycemia was defined as hypoglycemia requiring the assistance of a third party for recovery.The patients were also asked to answer the Diabetes Treatment Satisfaction Questionnaire (DTSQ) [14] and the Diabetes Medication Satisfaction Questionnaire (DiabMedSat) [15] to assess treatment satisfaction at weeks 0 and 24.

Meal tolerance test
A mixed meal tolerance test (MTT) was administered at weeks 0 and 24.On the day of MTT, patients were asked to attend the hospital after an overnight fast and instructed to consume a meal within 15 min.The meal consisted of crackers, pudding and bisque soup (460 kcal; 56.5 g carbohydrate, 18 g fat and 18 g protein, Janefu E460F18, Kewpie Corporation, Tokyo, Japan).During MTT, the patients were asked to take medication as usual, except that the insulin dose was reduced by 50% to prevent hypoglycemia.Blood sampling was conducted at 0, 30, 60, 90 and 120 min after ingestion of the meal.

Statistical analysis
The study was powered to show superiority in the primary endpoint, HbA1c, at week 24.In order to detect a difference in HbA1c of 0.6% [9] with SD of 0.7% between the two treatment groups, 23 subjects per group would yield a power of 80% with a 5% two-sided significance level.Assuming a withdrawal rate of 10%, enrollment of 25 subjects per group was planned.
Data are presented as mean ± SD in the text and tables, and as mean ± 1.96 x standard error (SE) (i.e., 95% confidence interval; CI) in the figures.Non-normally distributed data are expressed as median (interquartile range; IQR).Statistical analyses were performed using SPSS version 22 (IBM, Chicago, IL, USA).Mann-Whitney U test or Fisher's exact test was used to analyze the difference between the groups.The change in each parameter from baseline was analyzed by Wilcoxon signed-rank test.Spearman's rank correlation coefficient was used to test the relationship between change in HbA1c and clinical parameters.For analysis of sensitivity, the longitudinal profile was also analyzed by mixed model repeated measures (MMRM).A Pvalue <0.05 was considered statistically significant.

Patients' characteristics
One patient in the Insulin group was excluded because of protocol violation (Fig. 1).Eventually, 49 patients (24 in Insulin group, 25 in Ins+Sita group) completed the study and were included in the analysis (Table 1).There was no significant difference in sex, age, duration of diabetes, BMI and insulin dose between the two groups (all P >0.05).

Body weight, hypoglycemia and other adverse events
There was no significant change in body weight during the study in either group (S1 Fig. ), and the change in body weight at week 24 was not significantly different between the groups (0.4 ± 1.6 vs. -0.2± 2.0 kg, difference between the groups; -0.5 kg, 95% CI, -1.6 to 0.5, P = 0.16, Fig. 3a).There was also no significant difference in change in waist circumference at week 24 between the groups (1.13 ± 5.3 vs. 0.88 ± 5.4 cm, difference between the groups; -0.2 cm, 95% CI, -3.3 to 2.8, P = 0.94, Fig. 3b).

Meal tolerance test (MTT)
Plasma glucose, CPR, and glucagon levels during MTT at weeks 0 and 24 are shown in Fig. 4.There was no significant change in plasma glucose and CPR levels in the Insulin group at week 24.In the Ins+Sita group, plasma glucose levels at 30, 60, 90 and 120 min after meal ingestion were significantly decreased at week 24 (all P <0.05).CPR levels at 90 and 120 min were significantly increased at week 24 in the Ins+Sita group (median (IQR) 1.70 (1.35-2.55) vs. 2.30 (1.50-2.80)ng/mL, P = 0.02 and 1.80 (1.55-2.45) vs. 2.60 (1.55-3.20)ng/mL, P = 0.03, respectively).There was no significant change in plasma glucagon level during MTT at week 24 in either group.

Predictors of change in HbA1c after sitagliptin add-on treatment
Correlations between change in HbA1c during the study (ΔHbA1c) and other parameters are shown in Table 2.There were significant correlations between ΔHbA1c and baseline HbA1c (r = -0.49,P = 0.01, S3  AUCCPR/AUCglucose ratio during the study (ΔAUCCPR/AUCglucose ratio) was significantly correlated with ΔHbA1c in the Ins+Sita group (r = -0.49,P = 0.01).A significant correlation between ΔHbA1c and baseline GA or 1,5-AG was also observed in the Insulin group (S3 Fig. ) as well as in total subjects (Table 2).

Discussion
In this study, adding sitagliptin effectively reduced HbA1c by 0.9% without an increased risk of hypoglycemia or weight gain in Japanese patients with T2DM whose glycemic control was suboptimal despite at least twice daily injections of insulin.On the other hand, there was no significant improvement in HbA1c in the Insulin group, in which physicians tried to achieve the glycemic goal by adjustment of insulin dose.Improvement of glycemic control by adding sitagliptin was also confirmed by a significant reduction in GA and a concomitant increase in 1,5-AG during the study.The reduction in HbA1c in this study was comparable or greater than that in previous studies conducted in USA or Europe (0.5-0.7%) despite the comparable baseline HbA1c level [9][10][11], which was consistent with the previous report showing a greater glucose-lowering effect of DPP-4 inhibitors in Asians compared with Caucasians [12].Moreover, in this study, we also examined the daily glycemic profile by assessing 7-point SMBG, and showed that adding sitagliptin to insulin significantly reduced the postprandial glucose level.
Vilsboll et al. have reported that adding sitagliptin to insulin increased the incidence of hypoglycemia in patients with T2DM [9].The improvement of glycemic control by adding sitagliptin without an increased incidence of hypoglycemia observed in this study might be due to difference in the study protocol.In this study, insulin dose was reduced if initial HbA1c level was less than 8%, and the concomitant use of insulin secretagogues was discontinued during the study.Although the attending physician was able to adjust insulin dose during the study, the daily insulin dose remained significantly lower in the Ins+Sita group compared with the Insulin group.Thus, our results suggest that reducing the insulin dose by up to 20% and discontinuing insulin secretagogues may prevent hypoglycemia when a DPP-4 inhibitor is added to insulin therapy.Hong et al. have also recently reported the efficacy and safety of adding sitagliptin in patients with T2DM whose HbA1c was poorly controlled with insulin [21].Adding sitagliptin resulted in a 0.6% reduction in HbA1c without an increase in incidence of hypoglycemia compared with the control group in which insulin dose was up-titrated by 25% during the study.
Treatment satisfaction assessed by DTSQ was also improved in the Ins+Sita group, which was mainly derived from the improvement of patients' satisfaction with controlling hyper-and hypoglycemia.On the other hand, there was no change in DiabMedSat.The reason for this inconsistency is unclear, but the results of the latter might be affected by ceiling effects [22].Since there was no serious adverse event during the study, consistent with prior reports [23][24][25], our results proved that adding sitagliptin to insulin effectively and safely improves glycemic control and patients' satisfaction in Japanese patients with T2DM.In addition, in this study, sitagliptin treatment showed some reduction in oxidative stress markers, which may also be favorable to prevent the development or progression of vascular complications, although a favorable effect of DPP-4 inhibitors on cardiovascular outcome remains to be established [26,27].
To explore the mechanism by which sitagliptin improved glycemic control, MTT was also performed.As a result, MTT revealed that adding sitagliptin decreased postprandial glycemic excursion through enhancing postprandial insulin secretion.This change was accompanied by an increase in intact GLP-1 level after adding sitagliptin, confirming enhancement of incretin action.On the other hand, we did not observe a significant change in glucagon level after adding sitagliptin.It has been reported that DPP-4 inhibitors suppress glucagon level [28][29][30].This inconsistency may be in part due to the inaccuracy of current methods of glucagon measurement [31].It has been reported that GLP-1 suppresses glucagon secretion through enhancing insulin secretion [32].Thus, the insufficient glucagon suppression observed in this study may have been due to marked beta cell dysfunction in our patients with diabetes duration of about 20 years, although it has also been reported that GLP-1 suppresses glucagon secretion through stimulating somatostatin secretion from delta cells [33].
Finally, we tried to explore the predictors of HbA1c reduction by adding sitagliptin.As a result, baseline HbA1c level was a strong predictor of HbA1c reduction, consistent with previous reports [34,35].However, baseline GA and 1,5-AG levels were more strongly correlated with HbA1c reduction by adding sitagliptin than with baseline HbA1c, and were also associated with HbA1c reduction in the Insulin group as well as in all patients, suggesting that baseline GA and 1,5-AG levels similarly or even more sensitively predict a reduction in HbA1c after intensification of treatment than does baseline HbA1c.Our study also revealed that baseline beta cell function and improvement of beta cell function during the study were also predictors of HbA1c reduction after adding sitagliptin.It has been reported that baseline beta cell function and its progressive decline are important predictors of future glycemic control in patients with T2DM [36][37][38][39].Thus, these findings suggest that beta cell function as well as baseline HbA1c level is a major contributor to the glucose-lowering effect of sitagliptin added to insulin in Japanese patients with T2DM.
The small sample size and open-labeled design are limitations of this study.However, this study was conducted at a single institution and all attending physicians were specialists in diabetes, reflecting the "real world" clinical practice setting.Treatment allocation was blinded to both the physicians and patients until day 0 to avoid selection bias, and both groups similarly received standard care during the study.In addition, half of the patients were also randomized to a control (Insulin) group; therefore, we believe that we could at least partly eliminate the effects of study enrollment by comparison to the control group.In this study, we did not use a strict algorithm for insulin titration during the study and the insulin dose was mainly adjusted by the attending physicians at each visit, also reflecting the "real world" usual care setting.This may be the reason that the change in insulin dose in this study was relatively small.Recent studies have suggested that self-titration of insulin dose by patients based on their SMBG values resulted in better glycemic control compared with titration by physicians [40,41].Thus, introducing self-titration of insulin dose by patients might have resulted in better glycemic control in this study, although this would have been the case in both groups.On the other hand, comprehensive assessments including MTT and treatment satisfaction in addition to other clinical outcomes were a strength of this study, which allowed us to better understand the efficacy and safety of sitagliptin added to insulin therapy.
In conclusion, adding sitagliptin to insulin reduced HbA1c level by 0.9% without weight gain or increase in hypoglycemia, and improved treatment satisfaction in Japanese patients with suboptimally controlled T2DM.Adding a DPP-4 inhibitor may be an option when the glycemic goal is not achieved by at least twice daily injection of insulin.Longer-term efficacy and safety should be tested in future studies.

Table 1 .
Baseline characteristics of patients.

Table 2 .
Correlation between change in HbA1c during the study (ΔHbA1c) and clinical parameters.