Adjunctive Low-dose Metformin in Patients With Schizophrenia and Metabolic Abnormalities

Methods In this 12-week, randomized, double-blind, placebo-controlled trial, metformin 500 mg/d or 1000 mg/d or placebo was prescribed to clozapine-treated patients with schizophrenia having pre-existing metabolic abnormalities. The study was approved by an institutional review board and was conducted at Taipei Medical University-Wan Fang Hospital and Taipei City Psychiatric Center from May 2013 to January 2015. All clinical investigation had been conducted according to the principles expressed in the Declaration of Helsinki. The investigators screened clozapine-treated patients in the first phase and enrolled eligible patients in this clinical trial. Written informed consent was obtained from all patients before the screening.

Patients in the first-phase screening Patients diagnosed with schizophrenia or schizoaffective disorder according to the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition; aged 20-65 years; and who had taken clozapine for at least 3 months were invited to the first-phase screening. Clinical interviews were conducted, and medical records were evaluated for collecting patient demographics and clinical information, namely diagnosis, age of onset, and dosage of clozapine use.

Measurements The height, BW, waist circumference (WC), and blood pressure (BP) of all patients were measured. The body mass index (BMI) was calculated as the BW in kilograms divided by the square of the height in meters. After overnight fasting, blood was collected for analyzing the fasting plasma glucose (FPG), TG, and high-density lipoprotein cholesterol (HDL-C) levels. Under manufacture's guide, fasting TG, FPG, and HDL-C levels were measured using an automated system (Roche Cobas C501).

Patients in the metformin trial Patients in the first-phase screening were enrolled in the present trial if they had at least one of the following metabolic abnormalities: BMI ≥ 24; WC > 90 cm (men) or 80 cm (women); fasting serum TG level ≥ 150 mg/dL; fasting serum HDL-C level ≤ 40 mg/dL (men) or 50 mg/dL (women); systolic BP ≥ 130 or diastolic BP ≥ 85 mm Hg; current use of antihypertensive agents; and FPG level = 100-126 mg/dL. The exclusion criteria were the following: history of diabetes mellitus (DM); current use of hypoglycemic or hypolipidemic agents; pregnancy; allergy to metformin; a creatinine level > 1.4 ng/dL; an abnormal liver function test result; and chronic cardiopulmonary insufficiency.

Trial procedures Patients recruited in the trial were randomized to three groups, metformin 500 mg/d, metformin 1000 mg/d, or placebo. The randomization was conducted by a research assistant, who was blinded to the patient's status. To ensure the concealment of the randomization, the metformin and placebo were provided in coded containers. The appearance of the placebo was identical to that of metformin tablets. All patients, caregivers, and investigators were masked to the randomization.

In the first week, 500 mg of metformin was administered in the morning to the groups of metformin 500 mg/d and 1000 mg/d, and placebo was administered to the placebo group. In the second week, the dosage was revised to 500 mg of metformin in the morning and the placebo in the evening for the group of metformin 500 mg/d, 500 mg of metformin twice a day for the group of metformin 1000 mg/d, and placebo twice daily for the placebo group. For all patients, the clozapine dosage remained unchanged throughout the intervention. The recruited patients underwent physical and laboratory evaluations at week-4, week-8, and week-12.

Statistical analyses The investigators used descriptive statistics for summarizing the baseline clinical characteristics of patients and analysis of variance for examining the differences in these characteristics among all groups. Patients who continued the intervention for at least 4 weeks were included in the analyses. The investigators adopted the last observation carried forward (LOCF) approach for replacing missing data, assuming no change in the missing values of metabolic indices after an event of dropout. Furthermore, for investigating whether repeated measures collected in a longitudinal manner change over time, the investigators first used the paired t test for examining differences between the baseline and follow-up measures. Repeated measure analyses were conducted using 2-way within-subjects analysis of variance for examining group (df = 2, between groups), time (df = 3, within subjects), and interaction (df = 6, interaction between time and groups) effects on the changes in metabolic profiles over time. The analyses were conducted using SPSS Version 20.0 (IBM, Armonk, NY). P < 0.05 was considered to indicate statistical significance.