BMI-driven multi-omics analysis profiles the molecular mechanism of psycho-metabolic nexus and contributes to risk stratification

Obesity and psychiatric disorders frequently co-occur, yet the heterogeneous genetic mechanisms underlying this comorbidity remain elusive. Here, we stratify the polygenic architecture of body mass index (BMI) into five biologically distinct clusters based on metabolic signatures and construct partitioned polygenic scores (pPGSs) for each. Using individual-level data from the UK Biobank (N = 310,708), we identify cluster-specific psychiatric risk profiles, with amino acid-related clusters showing the strongest associations. These effects are markedly modulated by sex and BMI: underweight individuals and females show amplified psychiatric vulnerability, while obesity attenuates certain associations. Integration of proteomic and metabolomic data reveals key molecular mediators, including TNN, LEP, and ANGPT2, that influence psychiatric outcomes in a BMI- and sex-dependent manner. Mediation analyses demonstrate that pPGSs primarily exert indirect effects on psychiatric disorders through immune and metabolic molecular networks. These findings highlight the functional heterogeneity of BMI genetics, providing a framework for precision psychiatry. Metabolic clustering of BMI-associated genetic variants reveals sex- and BMI-dependent psychiatric risk heterogeneity, mediated by coordinated immune and metabolic molecular networks.