Melancholic Symptoms in Bipolar Depression and Responsiveness to Lamotrigine

Patients suffering from depression with melancholic symptoms (i.e., anhedonia, flat affect, diurnal mood variation, terminal insomnia, psychomotor disturbances, decreased weight/appetite, and excessive guilt) respond better to certain antidepressants. Melancholic symptoms also occur in bipolar depression, although they have received less research. Lamotrigine has been shown to alter some of the biological processes that are known to occur in melancholic depression. The purpose of this study is to determine if patients with melancholic bipolar II depression are more responsive to lamotrigine than patients with non-melancholic bipolar II depression. This study will re-analyze data from a previous 8-week, randomized, placebo-controlled trial that evaluated lamotrigine as a treatment for bipolar II depression (GSK-SCA100223; NCT00274677). The original study data was made available by GlaxoSmithKline as part of an initiative to make clinical trials data available for research use. Access was applied for via https://www.clinicalstudydatarequest.com. The analysis strategy will be comparable to the original study, although the investigators will first classify participants as suffering from either melancholic or non-melancholic depression. The diagnosis of melancholic depression was established according to baseline responses to the Hamilton Depression Rating Scale (HAMD-17) and the Montgomery-Åsberg Depression Rating Scale (MADRS), according to the DSM-IV-TR diagnostic criteria. HAMD-17 and MADRS change scores will be compared between the treatment and placebo groups using Analysis of Variance (ANOVA). Both ANOVA models will include a test for an interaction between treatment group (lamotrigine vs. placebo) and melancholic depression (melancholic depression vs. non-melancholic depression). To handle missing data, each ANOVA model will be computed with only complete-case data first and subsequently using inverse probability weights that account for the probability of drop out. Inverse probability weights will be created based on covariates that predict missing responses. HAMD-17 and MADRS response rates between the treatment and placebo groups will be evaluated with a Cox proportional hazard regression analysis. There will be two separate analyses, one including participants with melancholic depression, and one including participants with non-melancholic depression. Statistical models will also adjust for baseline depression severity, if participants with melancholic depression are found to have more severe depressive symptoms at baseline. Given the delay between antidepressant initiation and response, trial-and-error prescribing is an inevitably lengthy process. The investigators hope the results of this study will enable more timely and effective treatment for patients with bipolar depression.