Proteomics Reveal Clusters of Hypertension Cases Associated with Differing Prevalence of Cardiovascular and Renal Complications

BackgroundHypertension affects over 30% of adults and is the leading risk factor for cardiovascular disease. It often presents without obvious symptoms, meaning that, although effective therapies exist, hypertension remains widely undiagnosed and insufficiently treated. Genomics-based prediction methods have shown only modest benefits for these disorders, but proteomic markers have demonstrated potential for greater predictive and clinical value. MethodsWe applied a novel machine-learning based patient stratification analysis pipeline to proteomics data for 7,086 hypertension patients from UK Biobanks Pharma Proteomics Project cohort (2,911 proteins). We evaluated the contribution of each protein to the output of a tree-based risk model to explore the combinations of protein expression values that naturally separate hypertension cases into clusters and assessed the prevalence of cardiovascular and renal complications within each obtained cluster. ResultsWe identified 10 clusters of hypertension patients segregated by differential expression of HAVCR1, PLAT, PTPRB, REN and RTN4R. Four of these clusters showed statistically significant enrichment for cardiovascular and renal complications, and three of them had significantly lower prevalence of complications than expected among hypertension patients. ConclusionWe hypothesize that the hypertension clusters identified may represent distinct mechanistic subtypes. With further study this could help focus studies on subgroups of hypertension patients with a shared disease etiology, identify more personalized precision medicine treatment options for each subgroup, and develop mechanism-based biomarker tests to support enriched clinical trial recruitment.