Transdiagnostic Polygenic Risk Models for Psychopathology and Comorbidity: Cross-Ancestry Analysis in the All of Us Research Program

Abstract

Psychiatric disorders share substantial genetic liability, but the predictive utility of transdiagnostic polygenic risk scores (PRSs) remains unclear. In 102,091 All of Us Research Program participants, we compared a common psychiatric genetic (CPG) factor PRS with standard single-disorder and CPG-residualized disorder-specific PRSs across 11 psychiatric conditions and comorbidity burden. The CPG PRS showed stronger and more consistent associations with psychiatric diagnoses than most single-disorder-based scores, explaining up to 24.6-fold more phenotypic variance, and was strongly associated with comorbidity burden. CPG-residualized disorder-specific PRSs retained smaller independent contributions for several disorders, supporting complementary shared and disorder-specific genetic risk. Cross-ancestry analyses showed more consistent portability for comorbidity burden than for individual disorder prediction, while still underscoring limitations of Eurocentric discovery GWASs. These findings support the value of transdiagnostic PRSs as pragmatic indices of broad psychiatric liability and comorbidity burden, while emphasizing the limited utility of current PRS models as disorder-specific diagnostic tools.