This was a double-blind, double-dummy, parallel group, randomized study carried out over a period of 24 weeks. Patients were recruited from 24 centres across five countries, as listed in the Acknowledgements. The study was carried out according to Good Clinical Practice guidelines and the Declaration of Helsinki. The protocol was approved by the appropriate ethical boards and all patients provided written informed consent. Patients were eligible if they had Type 2 diabetes for a minimum of 6 months and had received metformin monotherapy for at least 3 months; the patients also had to be on a minimum metformin dose of 1000 mg per day continuously for at least 2 months prior to study entry, but remain inadequately controlled by medication, diet and physical exercise. Other inclusion criteria were a baseline HbA1c 6.8–9.0%, and a body mass index (BMI) between 20 and 35 kg/m2. In the 8 weeks preceding the study, and throughout the study, patients remained on their individual maximally tolerated dose of metformin.
Eligible patients received either nateglinide (Starlix, Novartis Pharma, Basel, Switzerland) or gliclazide in combination with metformin during the 24-week period. Randomization to treatment was by a computer-generated schedule via an interactive voice-responding system that assigned randomization on a study-centre basis with a block size of 4. A double-dummy technique, using identical-looking nateglinide and placebo tablets and identical-looking gliclazide and placebo capsules, was used to blind study medication assignment. Treatment regimens of nateglinide and gliclazide were started at the lowest levels (60 mg three times a day before meals and 80 mg once per day, respectively) and were titrated to the next dose level on a monthly basis up to a maximum of 180 mg before meals and 240 mg per day, respectively, during the first 3 months. Dose levels of study medication were increased if the fasting plasma glucose (FPG) level was > 7 mmol/l, if the patient had not experienced any confirmed hypoglycaemic events (symptomatic and/or asymptomatic events with plasma glucose concentration ≤ 4.0 mmol/l) and if the patient had not experienced more than three hypoglycaemic events in the past month.
Efficacy was determined from the change in HbA1c levels from baseline to end point. Treatments were also assessed for the percentage of patients reaching a treatment target (end point HbA1c < 7% and/or a decrease ≥ 0.5% HbA1c), FPG and body weight after 24 weeks of double-blind treatment. At baseline and after 24 weeks, patients attended the clinic after an overnight fast and received a standard breakfast consisting of 180 ml orange juice (grapefruit juice was not allowed), 60 g of bread (white, wheat or rye), 20 g jam or preserves, 10 g butter or margarine, 120 ml whole milk (3–4% fat) or the equivalent amount of cheese plus 120 ml water, and decaffeinated coffee or tea if desired. The meal was consumed within 15 min and blood samples were taken at 0, 15, 30, 60, 90, 120, 180 and 240 min. Glucose and insulin concentrations were measured and excursions were calculated from the difference between fasting and postprandial levels; the area under the glucose curve from 0 to 4 h (AUC0−4 h), adjusted for pre-meal values, was calculated for patients who had at least six measurements. In addition, insulin secretion index was calculated as: homeostatis model assessment (HOMA)-B = [20 × fasting insulin (mU/l)]/[fasting glucose (mmol/l) – 3.5].
Determinations of HbA1c and glucose and insulin from the meal tests were carried out at two central laboratories, one in Europe and one in the USA. Additionally, all patients were provided with glucose-monitoring devices and supplies, and were instructed on their use. Hypoglycaemic events were analysed according to whether they were asymptomatic, symptomatic or confirmed events; number of hypoglycaemic clinical symptoms and incidence rate (number of symptoms per 100 patients per month) were also determined. Symptoms suggestive of hypoglycaemia were tachycardia, palpitations, shakiness, sweating and dizziness, hunger, blurred vision, impairment of motor function, confusion or inappropriate behaviour. An event was considered as confirmed if the self-monitored plasma glucose value obtained at the time of the event was ≤ 4.0 mmol/l (corresponding to blood glucose value of ≤ 3.6 mmol/l) or if hypoglycaemia was classified as grade 2 (i.e. an episode with sufficient neurological impairment that the patient was unable to initiate self-treatment and required assistance or hospitalization). The incidence rate of hypoglycaemic events per 100 patients per month was calculated as [(total number of events across all patients)/(total duration on treatment in days)] × 30 × 100. All adverse events were recorded; they were defined as serious if they were fatal, life-threatening, required prolonged hospitalization, resulted in persistent or significant disability/incapacity, constituted a congenital anomaly/birth defect or were considered medically significant. For each adverse event, the relationship to study drug treatment was classified by the investigator as suspected or not suspected.
A planned sample size of 120 patients per treatment was considered sufficient to detect an HbA1c difference of 0.5% with 90% power, assuming a dropout rate of 15% and an sd of 1.1. An ancova model was used to test the null hypothesis that nateglinide plus metformin combination therapy was as effective as gliclazide plus metformin combination therapy. The primary ancova model included effects for treatment, study centre, baseline HbA1c and treatment by baseline HbA1c interaction. Efficacy analyses used the intent-to-treat population which included all randomized patients with at least one post-baseline efficacy evaluation, and safety was assessed for all randomized patients with a post-baseline safety assessment. Evaluations from the meal test were assessed for all patients with available data; during storage and transport to the central laboratories a small number of samples were lost and thus numbers of patients with available data were reduced slightly, although analysis indicated that the population with data was equivalent to the total population. The statistical tests were conducted at the two-sided significance level of 0.05.