Frequently Asked Questions

Metformin was associated with better outcomes in treatment-resistant bipolar depression when it successfully reversed insulin resistance. At the 14-week primary endpoint, 10 of 20 metformin-treated patients (50%) no longer met insulin resistance criteria compared with 1 of 25 placebo-assigned patients (4%) (Fisher exact P = .0009). Patients who converted from insulin-resistant to insulin-sensitive had significantly greater improvement in depressive symptoms beginning at week 6 and maintained through week 26, with large effect sizes for MADRS scores at week 14 (d = 1.17) and week 26 (d = 1.04).

At week 14, 50% of metformin-treated patients reversed insulin resistance, compared with 4% of patients assigned to placebo. Specifically, 10 of 20 patients in the metformin group became converters versus 1 of 25 in the placebo group, a significant difference (Fisher exact P = .0009).

The study found that the antidepressant signal tracked more closely with reversal of insulin resistance than with simple assignment to metformin. Converters had significantly greater reductions in MADRS depression scores than non-converters beginning at week 6 and continuing through week 26. Among non-converters, 4 of 10 metformin-treated patients (40%) and 7 of 24 placebo-treated patients (29.2%) still met the study's 30% MADRS response criterion, indicating that metformin assignment alone did not explain the full clinical effect.

Clinical improvement emerged by week 6 among patients whose insulin resistance reversed. In the mixed-effects analyses, converters showed significantly greater improvement in both MADRS depression scores and GAF functioning scores starting at week 6, and those gains were sustained through week 26.

At week 14, 81.8% of converters met the study's treatment response threshold, compared with 39.3% of non-converters. This was defined as at least a 30% reduction from baseline MADRS score in this treatment-resistant population, and the between-group difference was significant (9/11 vs 11/28; Fisher exact P = .031).

Yes. Patients whose insulin resistance reversed had significantly better anxiety and functioning outcomes than non-converters. GAF scores improved significantly beginning at week 6 and remained better through week 26, with large effect sizes at week 14 (d = 1.47) and week 26 (d = 1.58). HAM-A anxiety scores also improved significantly in favor of converters at week 14 (d = 0.61) and week 26 (d = 1.0).

The study did not find evidence that metformin destabilized mood into mania. YMRS mean scores were low at baseline and remained low in converters and non-converters throughout the 26-week trial. Suicidal ideation scores did not worsen in most patients (40 patients); in 5 patients, scores fluctuated but did not lead to study withdrawal.

The most common adverse effects with metformin were gastrointestinal, especially loose stool or diarrhea (40%), nausea (35%), and vomiting (10%). These were expected with metformin, and the study reported no significant between-treatment differences for adverse events occurring in at least 5% of either group. No serious adverse events occurred, and no subjects withdrew because of side effects or noncompliance.

At week 14, metformin-treated patients lost weight while placebo-assigned patients gained weight. The metformin group lost 1.67 kg and the placebo group gained 1.35 kg, for a mean difference of 3.02 ± 1.31 kg (t122.99 = 2.29, P = .023). Weight differences still favored metformin at week 26, but they were no longer statistically significant.

The trial enrolled adults aged 18 years or older with DSM-5 bipolar I or bipolar II disorder who had unremitting depressive symptoms despite optimal mood-stabilizing treatment and who also had insulin resistance. Depressive symptoms required a MADRS score of at least 15 for at least 4 weeks, and insulin resistance was defined by a HOMA-IR value of at least 1.8. Patients with type 2 diabetes mellitus were excluded.

The article describes treatment-resistant bipolar depression as failure to reach sustained remission after two 8-week trials of recommended medications, including combination therapy, given at therapeutic doses. For trial entry, patients also had to remain depressed despite stable optimal treatment based on 2013 CANMAT guideline-supported mood-stabilizing monotherapy or combinations for at least 4 weeks.

Insulin resistance was measured using the Homeostatic Model Assessment-Insulin Resistance (HOMA-IR) equation. The study used concurrent fasting plasma glucose and fasting serum insulin values to calculate HOMA-IR, and a cutoff of at least 1.8 was used to define insulin resistance for study eligibility.

Participants started metformin at 500 mg twice daily for 1 week, for a total of 1,000 mg/day, and then were titrated to 1,000 mg twice daily, or 2,000 mg/day, if tolerated, for 25 more weeks. Slower titration was allowed for tolerability, and all subjects were maintained on at least 1,500 mg/day. Among the 10 metformin-treated converters at week 14, 3 were taking 1,500 mg/day and 7 were taking 2,000 mg/day.

This was a highly refractory cohort. Patients had a mean of 8.6 lifetime failed medication trials for bipolar disorder, and 91.2% had failed drugs from at least 3 psychotropic classes, while 55.6% had failed drugs from all 4 classes studied: lithium, antiepileptics, antipsychotics, and antidepressants. Mean baseline MADRS scores were above 28, mean HAM-A scores were above 16, mean GAF scores were below 50, and 89% had a chronic or interepisode symptomatic course.

The study suggests that insulin resistance may be a clinically relevant treatment target in treatment-resistant bipolar depression. The trial specifically enrolled patients with insulin resistance and found that reversal of insulin resistance was associated with significantly better depression, anxiety, and functioning outcomes. The authors conclude that treatment of modifiable clinical risk factors for insulin resistance in treatment-resistant bipolar depression, such as obesity or long-term antipsychotic use, should be considered.

The main limitations were the small sample size and the fact that only 50% of metformin-treated patients reversed insulin resistance. The small sample prevented deeper analysis of which clinical or metabolic features predicted benefit. The authors also note that it is unknown whether higher metformin doses or extended-release formulations would improve outcomes, and they caution that metformin may not be an adequate long-term treatment for all patients.