Pharmacological treatments for atypical depression: A systematic review and network meta-analysis of randomized controlled trials

Introduction: Atypical depression is a highly prevalent subtype that includes mood reactivity, hypersomnia, and leaden paralysis, necessitating different therapeutic approaches than melancholic depression. No network metaanalysis has been conducted on pharmacological treatments for atypical depression. Methods: We performed a PRISMA-compliant systematic review and network meta-analysis searching PubMed/ Central, Clinicaltrials.gov, Embase, PsycINFO, Scopus, WebOfScience for randomized controlled trials (RCTs) testing pharmacological interventions for atypical depression until 04/24/24 (PROSPERO: CRD42024540262). Depressive symptom change (standardized mean difference/SMD), response, and all-cause discontinuation (acceptability) (risk ratio/RR) were co-primary outcomes; tolerability was the secondary outcome. Risk-of-bias and global/local inconsistencies were measured, and Confidence in Network Meta-Analysis (CINeMA) was used to assess the confidence in the evidence. Results: Out of 2214 hits, we included 21 eligible RCTs, 20 entering the NMA. For efficacy (k = 16, N = 903, treatments=12), only phenelzine outperformed placebo (SMD=-1.31, 95 %C.I.=[-2.14;-0.49]). Phenelzine, moclobemide, isocarboxazid, imipramine, selegiline, sertraline, and fluoxetine all outperformed nortriptyline (from SMD=-4.54, 95 %C.I.=[-8.02;-1.07] to SMD=-3.08, 95 %C.I.=[-5.42; -0.75]). Regarding response (k = 13, N = 1442, treatments=7), phenelzine (RR=2.58, 95 %C.I.=[2.02-3.31]), sertraline (RR=2.25, 95 %C.I.= [1.01-4.99]), moclobemide (RR=2.16, 95 %C.I.=[1.12-4.19]), fluoxetine (RR=1.89, 95 %C.I.=[1.30-2.76]) and imipramine (RR=1.76, 95 %C.I.=[1.35-2.28]) outperformed placebo, and phenelzine also outperformed imipramine (RR=1.56, 95 %C.I.=[1.25-1.96]). No treatment was significantly different from placebo for acceptability. No intervention outperformed placebo on any outcome in sensitivity analyses upon exclusion of high-risk-of-bias and intention-to-treat trials, likely due to a loss in power of the analysis, and overall CINeMA ratings were low/very low. Conclusions: Phenelzine might perform better than other compounds, but several drugs outperformed placebo in response. Nortriptyline performed worse than other treatments. High-quality studies are needed.