This was a randomized (1:1), multicenter, parallel-group trial (outpatient setting) comparing saxagliptin 2.5 mg BID versus placebo BID as add-on therapy to metformin IR BID (ClinicalTrials.gov registration: NCT00885378. The study took place at 25 sites in the United States, 9 sites in Germany, 5 sites in Hungary, and 4 sites in Puerto Rico. The protocol, amendments, and patient informed consent were approved by the institutional review board(IRB)/independent ethics committee (IEC) at each site before study initiation, and the study was performed in accordance with the Declaration of Helsinki and the International Conference on Harmonisation/Good Clinical Practice. Patients provided informed consent before study participation. Each IRB/IEC was composed of a review panel that was responsible for ensuring the protection of the rights, safety, and well-being of human subjects involved in the clinical investigation and was adequately constituted to provide assurance of that protection.
Patients were men and nonpregnant nonlactating women aged 18 to 78 years receiving stable metformin IR monotherapy (daily dose ≥1500 mg, given BID) for ≥8 weeks before enrollment and having inadequate glycemic control (HbA1c level 7.0%–10.0% with diet and exercise). Additional inclusion criteria were a fasting C-peptide value ≥0.8 ng/mL and body mass index (BMI) ≤45.0 kg/m. Patients were excluded if they had marked polydipsia and polyuria with >10% weight loss <3 months before screening, history of diabetic ketoacidosis or hyperosmolar nonketotic coma, or insulin therapy (except during short-term hospitalization or gestational diabetes) within 1 year of screening. Additional exclusion criteria included a cardiovascular event within 3 months of screening, congestive heart failure New York Heart Association class III/IV, known ejection fraction ≤40%, or history of hemoglobinopathies.
Patients were enrolled by investigators at each study site. At the screening visit, each patient was assigned a unique sequential subject number by an Interactive Voice Response System (IVRS), which was used for identification throughout the study. After a 2-week single-blind lead-in period with placebo BID added to metformin IR BID (with diet and exercise), patients were randomized 1:1 to 12 weeks of double-blind treatment with saxagliptin 2.5 mg BID or matching placebo BID added on to metformin IR using a blocked randomization schedule. The computer-generated randomization scheme was developed and kept by the study sponsor. Randomization was performed by calling the IVRS. Placebo tablets were identical in appearance to the saxagliptin tablets, and medication was dispensed using bottle numbers assigned by the IVRS. Titration or adjustment of blinded saxagliptin or metformin was not allowed during the study.
The primary end point was the change in HbA1c from baseline to week 12 (or last observation carried forward [LOCF], if no week-12 value was available). Key secondary end points were FPG change from baseline to week 12, proportion of patients achieving HbA1c <7.0% at week 12, and the proportion of patients achieving HbA1c ≤6.5% at week 12. Other efficacy variables included the proportions of patients with 1) reductions in HbA1c ≥0.5% or 2) ≥0.7%, 3) with FPG <110 mg/dL or 4) FPG <126 mg/dL, and 5) proportions requiring discontinuation for lack of glycemic control (FPG ≥270 mg/dL at 4 weeks or ≥240 mg/dL at 8 weeks). Body weight was evaluated post hoc as an exploratory end point.
Safety assessments included AEs, electrocardiograms, vital signs, physical exams, and clinical laboratory tests (ie, hematology, serum chemistry [liver function, kidney function, creatinine kinase, electrolytes, protein], and urine testing).
Baseline and change from baseline efficacy assessments were analyzed in the Randomized Patients Population (those who received randomized study drug with ≥1 postbaseline assessment). The primary efficacy analysis was an analysis of covariance (ANCOVA) of the adjusted mean change in HbA1c (least-squares mean adjusted for baseline HbA1c value) from baseline to week 12 (or LOCF) during the double-blind period with treatment as a fixed effect and baseline HbA1c as a covariate. A sample size of 152 (n = 76 per treatment group) was calculated to provide 90% power to detect an estimated 0.6% difference in HbA1c with SD of 1.1%, assuming 5% of patients were unevaluable and a 2-sided α of 0.05. In addition, subgroups of HbA1c, gender, age and ethnicity were also analyzed by ANCOVA similar to the primary analysis.
Change from baseline to week 12 in FPG was analyzed in the same way as the primary end point. The number and proportion of patients achieving a therapeutic glycemic response (HbA1c <7.0%) at week 12 LOCF was compared between groups using the methodology of Zhang et al. and Tsiatis et al. A similar analysis was conducted for the proportion of patients achieving HbA1c ≤6.5% at week 12. As a post hoc analysis, mean (95% CI) changes in body weight (LOCF) from baseline to week 12 were summarized by treatment group. Other efficacy end points were analyzed using descriptive statistics (n [%]).
The protocol specified the sequential testing approach to control the type I error rate at the 0.05 level. Once an end point failed to achieve statistical significance, subsequent end points in the sequence were assessed using only summary statistics, including the 95% CI. The order of testing was 1) primary end point, 2) change from baseline in FPG, 3) proportion of patients with HbA1c <7.0% at week 12, and 4) proportion of patients with HbA1c ≤6.5% at week 12. Statistical analyses were performed using SAS Version 8.2 or higher (SAS Institute Inc., Cary, NC).
Safety analyses were conducted using the Treated Patients Population that comprised all patients who received study drug during the double-blind treatment period. Clinical AEs were summarized using Medical Dictionary for Regulatory Activities (MedDRA) version 12.1. Events of special interest were based on a list of predefined MedDRA terms prior to database lock and unblinding of the data.