Clustering of major depressive disorder genetic instruments identifies distinct and directionally opposing effects on cardiometabolic risk

BackgroundMajor depressive disorder (MDD) is a highly heterogeneous condition that is frequently co-morbid with type 2 diabetes (T2D), and yet the biological mechanisms linking these diseases remain unclear. We aim to identify distinct biological pathways in depression that may modify T2D risk. MethodsUsing Clustered Mendelian randomisation (MR-Clust), we analysed 621 genome-wide significant MDD variants to identify clusters of variants with similar causal effects on T2D. These clusters were validated, and their causal effects were comprehensively tested against glycaemic traits, depression subtypes, T2D risk factors, and cardiometabolic biomarkers using external GWAS data and the UK Biobank. Functional annotation of these clusters was performed using FUMA. ResultsMR-Clust identified three distinct clusters of MDD-associated variants. Two clusters (MDD1 and MDD2) were causal for higher T2D and its related risk factors, adverse glycaemic and cardiometabolic profiles. Functional annotation implicated brain expression that overlapped strongly with depression-related traits such as smoking and neuroticism. By contrast, MDD3 was causal for lower T2D risk, more favourable glycaemic and cardiometabolic biomarker profiles, and was enriched for gene sets linked to fatty acid metabolism and steroid biosynthesis. MDD1 and MDD2 clusters were associated with atypical-like depression symptoms, whereas MDD3 was associated with melancholic depression symptoms. ConclusionOur findings demonstrate a heterogenous genetic architecture for depression, with distinct biological pathways conferring opposing effects on cardiometabolic health. Understanding this heterogeneity could help tailor prevention and treatment strategies for people with depression at greatest metabolic risk.