Antipsychotic-associated metabolic toxicity remains one of the most persistent clinical problems in psychopharmacology. Clozapine and olanzapine are especially effective for psychosis but carry high liability for weight gain, dyslipidemia, insulin resistance, and type 2 diabetes. Current monitoring recommendations recognize this risk, yet they remain largely uniform across patients and do not incorporate ancestry-specific genetic risk or mechanistic drug-gene information. We performed a transcriptome-wide association study-informed drug-gene prioritization analysis to examine whether approved antipsychotic target genes overlap with genes whose genetically predicted expression is associated with type 2 diabetes. The analysis used ancestry-specific type 2 diabetes transcriptome-wide association study (TWAS) results derived from a multi-ancestry genome-wide association study (GWAS) and approved antipsychotic drug-gene interactions from the Drug-Gene Interaction Database (DGIdb). For each drug, target genes were matched to TWAS genes across six metabolic tissues, and a weighted risk score was calculated as the sum of the absolute TWAS z-score multiplied by the drug-gene interaction score for significant targets. Follow-up analyses decomposed signals into targets with positive and negative TWAS directions, operationally interpreted as aggravating and compensatory, while also examining curated metabolic axis genes including GLP1R, GIPR, PPARG, and SLC2A4. The analysis identified recurrent exploratory target-overlap signals for clozapine and olanzapine. Clozapine showed the most consistent cross-ancestry aggravating profile, with recurrent target overlap involving GLP1R and immune-related genes. Olanzapine showed strong mechanistic-axis overlap involving GLP1R, GIPR, and PPARG, although its simple TWAS directionality was often classified as compensatory. Trifluoperazine emerged as a notable candidate, with significant target enrichment in the European ancestry analysis and top ranking in the Hispanic analysis. Fluspirilene also met the combined enrichment false discovery rate threshold in the European ancestry analysis, although its clinical metabolic interpretation was less direct. Haloperidol decanoate showed a high burden driven partly by SLC2A4, but its directionality was frequently mixed or compensatory. These findings nominate the incretin axis as a plausible translational bridge between antipsychotic metabolic liability and existing interventions such as GLP-1 receptor agonists. They also identify a key methodological gap. Future models must incorporate the pharmacologic mode of action to distinguish receptor blockade from activation. Except for the enrichment-positive European ancestry findings for trifluoperazine and fluspirilene, the results are exploratory prioritization signals and do not establish drug-specific metabolic effects, causal mechanisms, or ancestry-specific treatment effects. Overall, this TWAS-informed analysis provides a hypothesis-generating framework for ancestry-aware metabolic monitoring and targeted validation studies.
