Dipeptidyl peptidase-4 (DPP-4) inhibitors stimulate glucose-dependent insulin secretion by increasing the activity of endogenous glucagon-like peptide (GLP-1) and glucose-dependent insulinotropic polypeptides (GIP), which are hormones released from intestinal cells in response to carbohydrate absorption [1] . Thus, DPP-4 inhibitors can improve both fasting and postprandial glucose levels without hypoglycemic episodes, which provides a definite benefit as compared with conventional insulin secretagogues, i.e. sulfonylureas (SU) [2] [3] . At present, seven DPP-4 inhibitors are available in Japan: sitagliptin, vildagliptin, alogliptin, linagliptin, teneligliptin, anagliptin and saxagliptin [4] -[9] . Sitagliptin, the first of these DPP-4 inhibitors, was approved at the end of 2010, and is now widely available in Japan. Despite being the most commonly used DPP-4 inhibitor in Japan; predictive factors of the HbA1c-lowering effects of sitagliptin have not yet been elucidated. Herein, we investigated the efficacy of sitagliptin and also patient characteristics related to a good response to sitagliptin at multiple medical centers. Six months after the initiation of 50 mg sitagliptin treatment, glycemic control was significantly improved. The only baseline characteristics significantly associated with a good response to sitagliptin in the multivariable regression analysis were high HbA1c and combination therapy with an α-glucosidase inhibitor (α-GI). Our results provide guidance for the optimal use of sitagliptin for managing patients with type 2 diabetes.

Changes in the main parameters in Studies 1 and 2 are shown in Table 1. In total, 25 patients dropped out because of drug side effects or failure to return for hospital visits. Glycemic control was markedly improved, in terms of both HbA1c and FPG, in both Study 1 and Study 2. Particularly, HbA1c was reduced by 0.7%, despite the glimepiride dose being decreased from 2.6 ± 0.1 mg to 1.2 ± 0.1 mg in Study 2. In addition to the improvements in both HOMA-R and HOMA-β, the PI/I ratio also improved, suggesting that the recovery of β cell function was not due to forced overloading, and that there had actually been a load reduction. The 1,5-AG level was also markedly increased, indicating glycemic fluctuation to be attenuated. Study 3, which examined the effects of switching agents, is summarized, with average doses and the number of patients, in Table 2. The average doses of each of the OHA classes were calculated as follows; 40 mg of gliclazide was equivalent to 1mg of glimepiride, 0.6 mg of voglibose was equivalent to 50 mg of miglitol, and 90 mg of nateglinide was equivalent to 10 mg of mitiglinide. As a result, small doses of SU and biguanides (BG), and standard doses of α-GI, glinides and thiazolidinediones were switched to 50 mg of sitagliptin. Table 3 shows the results of study 3. Glycemic control was slightly but significantly improved, whereas HOMA-β was decreased.

Table 4 shows single or multiple correlations between each parameter at baseline and effectiveness, defined as a more than 0.9% HbA1c reduction at the end of the study. Among single correlations, high HbA1c, low HOMA-β, high PI/I ratio, high 1,5-AG and combination therapy with SU, α-GI, or BG were associated with the effectiveness of sitagliptin. For multivariable regression analysis, we adopted two models (Models 1 and 2), containing HOMA-β or the PI/I ratio, respectively, because these two parameters are interdependent. In both Model 1 and Model 2, high HbA1c and combination therapy with an α-GI were still positively associated with the effectiveness of sitagliptin, which provides information on the optimal use of DPP-4 inhibitors in type 2 diabetic patients.

In the present study, we investigated the effect of administering 50 mg of sitagliptin, which is a standard starting dosage for this medication in Japan. However, outside of Japan, double this dose, i.e. 100 mg of sitagliptin, is generally used [10] . The dose-ranging efficacy of sitagliptin was tested in a previous study [11] , in which no differences in HbA1c were observed among sitagliptin doses of 50 mg, 100 mg, and 200 mg in Japanese patients with type 2 diabetes mellitus. According to these results, the 50 mg dose of sitagliptin was adopted as the standard for Japanese patients. Thus, Japanese patients are considered to be more responsive to sitagliptin than non-Japanese patients. A meta-analysis revealed that, despite the smaller sitagliptin dose of only 50 mg, a greater HbA1c reduction (−0.99% versus placebo) was observed in Japanese than in non-Japanese patients [12] . A marked HbA1c reduction (−1.0% versus baseline) was also observed in this study with add-on therapy, which is consistent with the previous results. Despite GLP-1 levels not being decreased in Japanese patients [13] , a good response to sitagliptin was still demonstrated. One possible explanation is that the body mass index (BMI) is generally lower in Japanese than in non-Japanese patients [12] . DPP-4 was previously reported to be one of the adipokines, playing a role in the induction of insulin resistance and providing a link between obesity and type 2 diabetes [14] . In addition, non-esterified fatty acids (NEFA) were demonstrated to be involved in reducing incretin signaling [15] or incretin secretion from the small intestine [16] . Thus, in obese patients, whose serum levels of DPP-4 or NEFA are predicted to be higher than those of normal weight subjects, the glucose-lowering

effect of DPP-4 inhibitors may be attenuated.

In addition to the prominent HbA1c reduction, marked recovery of the PI/I ratio was also observed in the present study (study 1). These results reflect the protective effect of GLP-1 on beta cells via alleviation of endoplasmic reticulum stress [17] . Though a similar effect was previously reported in non-Japanese patients [18] , our report is, to our knowledge, the first to document a PI/I ratio reduction in response to taking a DPP-4 inhibitor in Japanese patients with type 2 diabetes. The PI/I ratio reduction was greater in Japanese than in non-Japa- nese patients (19% and 12% reductions, respectively). These observations may explain the beneficial effects of DPP-4 inhibitors in Japanese patients whose beta cell function is exhausted judging from their low of HOMA-β levels. Taking into consideration that a greater PI/I ratio reduction (22%) was obtained when the SU dosage was reduced in Study 2, add-on therapy with 50mg of sitagliptin with a reduced SU dosage may confer a major protective effect on beta cells in Japanese diabetic patients.

Several studies have reported factors contributing to the efficacy of sitagliptin in Japanese diabetic patients [19] -[21] . In these studies, low BMI and high HbA1c were common predictive factors for HbA1c reduction. Though high baseline HbA1c is consistent with our present data, why low BMI was not associated with the efficacy of sitagliptin is unknown. The method of sitagliptin treatment, i.e. add-on or switching therapy, the study scale and the study duration may have influenced the relevance of obesity to the HbA1c reduction. Instead of BMI, our study indicated diminished beta cell function, i.e. low HOMA-β and a high PI/I ratio, to predict HbA1c reduction. Though DPP-4 inhibitors ameliorate insulin resistance (HOMA-R) probably by inhibiting glucagon secretion [11] , DPP-4 inhibitors belong to a class of insulin secretagogues. Our results are considered to essentially reflect the original characteristics of DPP-4 inhibitors, which are known to improve glycemic control by enhancing insulin secretion, in diabetic patients with decreased beta cell function.

It has yet to be determined which OHA would most effectively reduce HbA1c when combined with DPP-4 inhibitors. Among possible candidates, combination therapy with metformin was reported to be beneficial, because DPP-4 inhibitors and metformin have complementary mechanisms of action and additive effects with respect to increasing the concentrations of active GLP-1 in plasma [22] . Inhibition of DPP-4 activity by metformin is speculated to be the mechanism by which metformin enhances GLP-1 [23] . Though our data also support compatibility between DPP-4 inhibitors and metformin, combination therapy with an α-GI showed a stronger relationship with HbA1c reduction in our multivariable regression analysis. Therapy combining sitagliptin with miglitol, in which the ability of α-GI to enhance GLP-1 secretion was discovered, was previously reported [24] . Our study is the first to clinically demonstrate the compatibility between a DPP-4 inhibitor, sitagliptin, and an α-GI in Japanese patients with type 2 diabetes.

In conclusion, the addition of sitagliptin to an already administered regimen or switching from another OHA resulted in reduced HbA1c due to recovery of β-cell function. In addition, we suggest that a good response to sitagliptin can be predicted when this agent is used in the combination with an α-GI. These results provide useful insights for managing diabetic patients by using DPP-4 inhibitors.

None of the authors has any potential conflicts of interest associated with this research.