Duloxetine treatment and glycemic controls in patients with diagnoses other than diabetic peripheral neuropathic pain: a meta-analysis

Objective: Mood disorders are often associated with poor glycemic control, and antidepressant treatments for mood and pain disorders can alter plasma glucose levels in patients with diabetes. A previous meta-analysis from three studies showed that duloxetine modestly increased fasting plasma glucose (FPG) and HbA(1c) levels in patients with diabetic peripheral neuropathic pain (DPNP). This meta-analysis examined whether there were any short- and long-term effects of duloxetine (20-120 mg/day) on glycemic control in patients with diagnoses other than DPNP. Research design and methods: Short-term data (9-27 weeks): seven studies of duloxetine in general anxiety disorder, fibromyalgia, and chronic lower back pain (CLBP). Long-term data: 41-week, uncontrolled extension of the short-term CLBP study and 52-week study in patients with recurrence of major depressive disorder. Main outcome measures: Baseline-to-endpoint changes in FPG and HbA(1c) levels. Results: In short-term studies, patients were randomly assigned to placebo (n = 1098) or duloxetine (n = 1563). Mean baseline-to-endpoint changes in FPG and HbA(1c) did not significantly differ in duloxetine-treated patients compared with placebo-treated patients. In the 41-week study (n = 181), duloxetine-treated patients experienced a small but significant within-group baseline-to-endpoint increase in HbA(1c) (mean change = 0.1%; p < 0.001). This result was in contrast to absence of effect on mean baseline-to-endpoint within-group changes in FPG (p = 0.326) in that study, and to absence of between-treatment changes in FPG (p = 0.744) and HbA(1c) (p = 0.180) in the 52-week placebo-controlled study. Conclusion: Duloxetine treatment did not significantly alter FPG and HbA(1c) levels compared with placebo treatment in the short-term studies. A small but statistically significant within-group increase in HbA(1c) was found in the 41-week study, but not in between-treatment group differences in the 52-week study. Neither of the long-term studies showed significant changes in the FPG levels. The small, non-reproducible HbA(1c) increase in one study of patients without DPNP may have resulted from patients with unrecognized diabetes in these trials.