Population pharmacogenomics of CYP1A2, CYP2C19, CYP2D6 and CYP3A4 in a Uruguayan cohort with schizophrenia

Latin American populations remain underrepresented in pharmacogenomic research, limiting the applicability of genomic findings to these groups. Uruguay, with its unique admixed demographic history, constitutes a relevant case study for characterizing clinically significant pharmacogene variation. In this work, we characterized haplotypic diversity across four CYP450 genes, CYP1A2 , CYP2C19 , CYP2D6 , and CYP3A4 , in a cohort of 47 Uruguayan patients with treatment-resistant schizophrenia (TRS) receiving clozapine. Star-allele frequencies in the TRS cohort were contextualized against the Urugenomes Phase 2 dataset (U2; n = 30), used as the primary Uruguayan population reference, alongside two ancestry-enriched datasets comprising Indigenous-admixed (n = 10) and Afro-Uruguayan (n = 10) individuals. Star-allele and diplotype inference was performed using Aldy v4.8.3, and population-level frequencies were contextualized against global biogeographic references from PharmFreq and gnomAD. CYP1A2*30 predominated across all Uruguayan datasets, while CYP3A4 showed near-complete dominance of the *1 allele. CYP2C19 and CYP2D6 displayed greater inter-dataset heterogeneity, with metabolizer phenotype distributions differing from global reference populations. To assess the clinical relevance of this pharmacogenomic variation, multivariable linear regression models were constructed to predict plasma clozapine concentrations, incorporating clinical, behavioral, and genetic predictors. A model combining HDL, triglycerides, glycemia, cigarette consumption, and CYP1A2*30 allele count explained 41% of the variance in clozapine levels (adjusted R² = 0.41), representing a statistically significant improvement over the clinical-only model (ANOVA p = 0.034). Bootstrap resampling confirmed the stability of all estimated effects, and leave-one-out cross-validation yielded a corrected R² of 0.31. These findings contribute to characterizing the pharmacogenomic landscape of an underrepresented Latin American population and suggest that locally generated CYP haplotype data may complement clinical factors in explaining interindividual variability in clozapine exposure in this population.