Impact of using PRS-CSx and pruning and thresholding for polygenic partitioning of apparent treatment resistant hypertension

Apparent treatment-resistant hypertension (aTRH) is a clinically challenging condition with heterogeneous etiologies. Understanding the biological pathways underlying resistance to antihypertensive treatment could inform targeted therapeutic strategies. To evaluate how methodological choices in SNP selection influence biological inference, we applied two approaches to select aTRH-associated variants for clustering: PRS-CSx and pruning and thresholding (P&T). Using k-means clustering, we grouped aTRH-associated variants based on their association profiles across 91 cardiometabolic-related phenotypes. We then performed pathway and tissue enrichment analyses to evaluate the biological processes represented by each cluster. Both methods identified multiple genetic clusters, but the distribution of variants and biological signals differed. Clustering based on PRS-CSx produced unequally distributed clusters of SNPs and yielded limited tissue enrichment, while P&T-based clustering captured more uniform trends across cardiometabolic traits and broader tissue and pathway enrichment. These results demonstrate that methodological choices in SNP selection influence downstream clustering and biological interpretation. Despite some overlap in identified pathways and tissue enrichment, each approach identified unique biological signals, highlighting the potential of pairing polygenic methods and k-means clustering to elucidate the biological heterogeneity of aTRH and guide future mechanistic studies.