Genomic profile of pharmacogenetic  markers and fitting prediction models for maintenance doses of warfarin

• Ano de defesa: 2024
• Autor(a) principal: Neyra, Jennifer Eliana Montoya
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Biblioteca Digitais de Teses e Dissertações da USP
• Palavras-chave em Português: admixed population; ancestralidade local; Farmacogenética; local ancestry; modelos de predição; Pharmacogenetics; população mista; prediction models; varfarina.; warfarin.
• Link de acesso: https://www.teses.usp.br/teses/disponiveis/95/95131/tde-23082024-202433/

The admixed Brazilian population is highly heterogeneous, which has implications in health, disease, and also susceptibility to therapeutic regimens due to the underlying genetic variability from ancestral populations. Genomic data were composed of independent individuals selected from the ISA-Nutrição projects (N = 681), Projeto Coração Baependi (N = 548) and the Trios do project (N = 136). These data were subjected to genetic quality control per individual and polymorphism, identification of the haplotypic phase, imputation of variants not covered by the genotyping, and, filtering of variants with low imputation quality. Then, haplotypes were derived for global and local ancestry and the filtered post-imputation data were used to characterize the distribution of the pharmacogenetic markers of interest. Global ancestry was analyzed using Principal Component Analysis, and the average distribution of each population inferred by clustering with FastStructure was also evaluated. Local ancestry was calculated using RFMix and was explored by the average distribution at each locus, in addition, the polygenic risk score was calculated to compare regions of lower and higher ancestral variability between the studied cohorts. Also, we look for evidence of positive selection in the genome based on the average contribution of local ancestry. Six SNPs (rs10916661, rs10871454, rs749671, rs34707540, rs6897106, and rs2108622) differed from the global population panel. The overall average ancestral composition was 18.3% (SD = 18.5%) African, 5.2% (SD = 6%) Native American, 3.1% (SD = 3%) Asian, and 75.4% (SD = 19.6%) European compared to local ancestry with 22.2% (SD = 1.2), 5.1% (SD = 1.5), 2.1% (SD = 1.0) and 70.4% (SD = 2,3) respectively. The inferred local genetic profile presents heterogeneity between individuals, we have those with homogeneous ancestry (> 99%) and others in different stages of miscegenation, including individuals with a heterogeneous profile with a predominance of a single ancestor of around 30%. It was found that 48.2% of the samples had a polygenic risk greater than zero. Furthermore, no areas with signs of positive selection were found throughout the genome in our data. The variable ancestral profile was identified between our Brazilian admixed samples and genetic variants exhibiting varying frequencies in homogeneous global populations. However, the regression model with variables selected by the Stepwise method (AIC) did not perform well for individuals who required warfarin doses greater than 50 mg/week, possibly due to the absence of a factor capable of controlling dose variability in the model.