Factors Predicting Therapeutic Efficacy of Combination Treatment With Sitagliptin and Insulin in Type 2 Diabetic Patients: The ASSIST-K Study

Background It is unclear whether dipeptidyl peptidase-4 inhibitors decrease hemoglobin A1c (HbA1c) in a glucose-dependent manner in patients on insulin therapy who have impaired insulin secretion. This study investigated factors influencing the efficacy of sitagliptin when used concomitantly with insulin to treat type 2 diabetes mellitus (T2DM) in the real-world setting. Methods A retrospective study was conducted of 1,004 T2DM patients at 36 Japanese clinics associated with the Diabetes Task Force of the Kanagawa Physicians Association. Eligible patients had been on insulin for at least 6 months, with a baseline HbA1c of 7.0% (53 mmol/mol) or higher. Baseline characteristics and laboratory data from 495 patients were subjected to multiple regression analysis to identify factors influencing the change of HbA1c. Results Most patients (n = 809) received sitagliptin at a dose of 50 mg. In the 1,004 patients, HbA1c decreased by 0.74% (6 mmol/mol) and body weight increased by 0.1 kg after 6 months of combination therapy. Multiple regression analysis showed that a higher baseline HbA1c, older age, and lower body mass index influenced the change of HbA1c after 6 months. Hypoglycemic symptoms occurred in 7.4%, but none were severe. Conclusions These results emphasize the importance of a higher HbA1c at the commencement of sitagliptin therapy in patients on insulin. Glucose-dependent suppression of glucagon secretion by sitagliptin may be useful in patients with impaired insulin secretion. Sitagliptin can be used concomitantly with insulin irrespective of the insulin regimen, duration of insulin treatment, and concomitant medications.


Introduction
Among the dipeptidyl peptidase-4 (DPP-4) inhibitors that are widely used, sitagliptin is the first to be approved for patients with type 2 diabetes mellitus (T2DM) on insulin therapy. Despite its established position, relatively few clinical studies of sitagliptin have been performed. These include a US study of patients who were predominantly on once-daily long-acting, insulin therapy [1], a Japanese study of patients on mixed insulin therapy [2], and observational studies with small populations [3,4].
T2DM patients receiving insulin often have a longer disease duration and a reduced insulin secretion capacity. Because their baseline characteristics and insulin doses vary, clinical trials alone do not provide enough information about these patients [5]. Moreover, the effectiveness of DPP-4 inhibitors at stimulating an insulin response in patients with compromised insulin secretion on insulin therapy has not been examined. A small-scale randomized controlled trial (RCT) has shown that sitagliptin improves hemoglobin A 1c (HbA 1c ) and postprandial glucose levels in patients with type 1 diabetes [6]. The effect of sitagliptin in increasing glucagon-like peptide 1 (GLP-1), which results in inhibition of glucagon secretion rather than stimulation of insulin secretion, has attracted attention [7,8].
RCTs play a significant role in evidence-based medicine. However, such trials exclude patients who do not meet the specified inclusion criteria, even though these patients are often important in the real-world clinical setting. This is particu- In the present study, we investigated the efficacy and safety of insulin-sitagliptin combination therapy in a multicenter study to verify the usefulness of this regimen for T2DM patients on various types of insulin therapy. Our findings add some useful information to the evidence already established by RCTs.

Methods
This retrospective study included patients receiving sitagliptin in addition to insulin from November 2011 to March 2013 at 36 diabetes clinics in Kanagawa Prefecture, Japan. Sitagliptin was started if the attending physician considered that insulin was not achieving adequate glycemic control. Glycemic control, the insulin dose, concomitant drugs, blood pressure (BP), body weight, and laboratory data were analyzed, as well as the occurrence of adverse events. Eligible patients had been on insulin for 6 months or longer, with an HbA 1c of 7.0% (53 mmol/mol) or higher, and had undergone follow-up for at least 6 months.
The primary endpoint was the change of HbA 1c . To explore factors contributing to the improvement of HbA 1c by insulin-sitagliptin combination therapy, multiple regression analysis was performed with the following variables: age, sex, baseline body mass index (BMI), duration of insulin therapy, baseline HbA 1c , daily insulin dose, presence/absence of diabetic neuropathy, smoking, alcohol consumption, and concomitant use of sulfonylureas (SUs), biguanides (BGs), alpha-glucosidase inhibitors (α-GIs), and thiazolidinediones (TZDs). Patients who had previously been treated with a DPP-4 inhibitor were excluded.
Results are reported as mean ± standard deviation. All analyses were carried out using SPSS version 19 software (SPSS Inc., Chicago, IL, USA). The effect of sitagliptin treatment was assessed by one-way analysis of variance and P < 0.05 was accepted as indicating statistical significance.
This study was registered with the Clinical Trials Registry (http://clinicaltrials.gov; NCT01855087) and was undertaken in accordance with the Ethical Guidelines for Clinical Studies of the Japanese Ministry of Health, Labor, and Welfare. Informed consent was not required because this was a retrospective analysis.

Results
A total of 1,169 case record forms were collected. Then patients with sitagliptin treatment for < 6 months (n = 54), baseline HbA 1c < 7.0% (53 mmol/mol) (n = 81), and incomplete data (n = 30) were excluded. The remaining 1,004 patients were available for analysis and their characteristics are shown in Table 1. Based on the results of Japanese clinical studies employing sitagliptin, the drug was up-titrated in some patients from a starting dose of 50 mg to 100 mg if the former dose was not effective, while a low dose of 25 mg was selected for 65 patients with impaired renal function [9]. Among the 1,004 patients, medications other than insulin and sitagliptin were altered in 185 patients, including 144 who discontinued α-GIs or TZDs and 41 who received add-on biguanides.
Hypoglycemic symptoms were noted in 7.4% of the subjects, but were not severe. Other adverse events included abdominal distension (0.9%) and constipation (0.4%). With regard to changes in laboratory values, there was a significant decrease of systolic BP, fasting blood glucose, glycated albumin, alkaline phosphatase, total cholesterol, high-density lipoprotein (HDL) cholesterol, and triglycerides, while there was a significant increase of uric acid and serum creatinine ( Table 2).
Multiple regression analysis was performed to examine factors related to the change of HbA 1c within 6 months of starting sitagliptin. Complete baseline data on the age, sex, BMI, duration of insulin treatment, HbA 1c , daily insulin dose, presence/absence of diabetic neuropathy, smoking, alcohol consumption, and concomitant use of SUs, BGs, α-GIs and TZDs were available for 495 patients. These characteristics of the 495 patients were similar to those of the main study population (data not shown).
A high HbA 1c at baseline was the strongest contributor to reduction of HbA 1c at 6 months after starting treatment with insulin-sitagliptin combination therapy (Table 3). In addition, BMI and age made slight, but significant, contributions to the reduction of HbA 1c (Table 3).
We also performed multiple regression analysis on a cohort of 114 patients who had C-peptide data before commencement of sitagliptin therapy. This showed that C-peptide had no influence on the effectiveness of sitagliptin (data not shown).

Discussion
The Diabetes Task Force of the Kanagawa Physicians Association previously conducted a large-scale trial involving more than 1,000 patients (the ASSET-K study), which established that a higher baseline HbA 1c , shorter disease duration, and lower baseline BMI were factors contributing to reduction of HbA 1c in patients receiving sitagliptin combined with another oral antidiabetic drug [10][11][12][13][14]. However, the factors influencing the HbA 1c -lowering effect of sitagliptin when it is used concomitantly with insulin were not fully elucidated. Therefore, we conducted the present study to investigate this issue.
We confirmed the effectiveness and safety of combining sitagliptin with insulin in a large patient population (n = 1,004) treated in the real-world setting. HbA 1c decreased by 0.74% (6 mmol/mol) and body weight increased by 0.1 kg, while there were no appreciable changes in the doses of insulin or other oral antidiabetic agents. In addition, there were no episodes of severe hypoglycemia, although hypoglycemic events were reported in 7.4% of the patients. In general, the formulation, dose, and frequency of insulin administration are altered if an HbA 1c < 7.0% (53 mmol/mol) is not achieved after a certain period. However, increasing the insulin dose may lead to adverse events such as weight gain and hypoglycemia. The present study demonstrated the effectiveness of insulinsitagliptin combination therapy in the real-world setting. As well as improving HbA 1c , there were significant reductions of systolic BP, fasting blood glucose, glycated albumin, alkaline phosphatase, total cholesterol, HDL cholesterol, and triglycerides, which suggest that sitagliptin may also exert a favorable effect on BP and the lipid profile, as reported in the ASSET-K study [14]. The significant increase of uric acid and serum creatinine were also in agreement with the results of the ASSET-K study, and can be explained by the fact that sitagliptin increases GLP-1, an incretin hormone with diuretic properties. In fact, a correlation between the reduction of BP and an increase of creatinine and uric acid levels has been reported [14]. We performed multiple regression analysis to identify patients who were likely to respond to addition of sitagliptin to insulin therapy. We analyzed various factors that had a potential influence on HbA 1c including diabetic neuropathy, smoking, and alcohol, because diabetic neuropathy may inhibit the effect of GLP-1 and patients who smoke or drink regularly are more likely to lead an unhealthy lifestyle with suboptimal diet and exercise. We identified a higher baseline HbA 1c as the strongest contributing factor, while a lower baseline BMI and older age were also significant. This is in agreement with the fact that older patients tended to have a lower BMI in our study. There was a positive correlation between the duration of diabetes and the duration of insulin therapy (Pearson's correlation coefficient = 0.40), as well as between the frequency of insulin administration and the daily insulin dose (Pearson's correlation coefficient = 0.51). Therefore, the duration of insulin therapy and the daily insulin dose were included in the multiple regression analysis as background factors, revealing that the duration of insulin therapy was not associated with the effectiveness of sitagliptin. These results support the addition of sitagliptin to insulin therapy, irrespective of the duration of insulin treatment, the daily insulin dose, concomitant medications, and the C-peptide profile. Even in patients with compromised insulin secretion, sitagliptin can still effectively stimulate insulin secretion and it also has other potential benefits. Inhibition of glucagon secretion may be involved in the efficacy of DPP-4 inhibitors for insulin-treated patients [6] and inhibition of glucagon secretion as a GLP-1 receptor agonist may be important in type 1 diabetes [15]. It has been reported that T2DM patients in East Asian countries (including Japan) have lower endogenous insulin secretion and that their BMI tends to be lower [16]. Kim et al reported that sitagliptin was particularly effective in the Japanese population [17]. In addition to the ASSET-K study, several observational studies have demonstrated the effectiveness of sitagliptin in patients with a lower BMI [11,13,18]. However, more research is needed to determine whether our results are related to ethnic factors, especially studies in non-Japanese populations.
There were some limitations of this study. First, it was retrospective. Second, all of the patients received sitagliptin and there was no control group without sitagliptin treatment. Third, markers for the assessment of beta-cell function (HOMA-beta) and the basal level of active GLP-1, which are factors that could influence the reduction of HbA 1c by sitagliptin therapy, were not measured in all patients. In fact, only 495 out of 1,004 patients were included in the detailed analysis, but these patients were considered to be reasonably representative of the entire population since their characteristics did not differ from those of the whole population (data not shown).
In conclusion, this study provided useful new information about the efficacy and safety of sitagliptin that can be added to the findings of relevant RCTs. We demonstrated that a higher baseline HbA 1c was associated with greater improvement of HbA 1c in patients given sitagliptin as add-on therapy to insulin, while other factors were not clinically relevant. This suggests that patients might be candidates for add-on sitagliptin therapy irrespective of their characteristics, insulin dose, and concomitant medications. Because a significant increase of body weight was not observed, sitagliptin may be added whenever the HbA 1c is not well controlled by insulin therapy.