Genetic clusters of BMI reveal symptom-specific causal effects on depression

While BMI is phenotypically and genetically associated with depression, the extent to which BMI causes depression and the potential causal mechanisms underlying this effect remain unclear. We used PheWAS-based Clustering of Mendelian Randomization instruments (PWC-MR) on 324 BMI-associated SNPs across 407 UK Biobank traits, which identified six genetic clusters. We focused on two well characterised clusters: one enriched by body mass-related traits (e.g. fat-free mass) and another by socioeconomic traits (e.g. income). We estimated the causal effects of these two clusters on nine symptoms of depression, individually measured in the latest genome-wide meta-analysis of depression symptoms (N = 224,535–308,421), including core psychological symptoms such as depressed mood and anhedonia, as well as somatic symptoms such as appetite changes and tiredness. BMI exhibited differential effects across clusters and symptoms not captured in univariate Mendelian randomization. Using instruments in the mass cluster, BMI had the greatest effect on appetite changes (β = 0.27, 95% CI [0.24, 0.30], p = 1.28×10⁻⁶⁵), with some effects on tiredness (β = 0.11, 95% CI [0.07, 0.15], p = 1.69×10⁻⁷) and anhedonia (β = 0.10, 95% CI [0.07, 0.14], p = 5.47×10⁻⁸). We found no evidence for an effect on suicidal ideation (β = 0.02, 95% CI [0.00, 0.04], p = 5.01×10⁻¹). This pattern of effects was broadly similar in the socioeconomic cluster. In sensitivity analyses, the effect on appetite was maintained while effects on other symptoms (depressed mood, psychomotor changes, sleep problems) disappeared. Differential effects of BMI across depression symptoms illustrate that the effect typically detected using Mendelian randomization appears to be driven by the expected effect of BMI on appetite, with smaller effects on core psychological symptoms.