Proteomic Risk Score for Prediction of Incident Hypertension

Proteomic signatures may enhance the prediction of cardiometabolic diseases for targeted prevention. We evaluated whether a proteomic risk score (ProtRS) improves the prediction of incident hypertension.

Within the UK Biobank Proteomics data set, participants at risk for incident hypertension were randomly split into derivation (n=25 158) and test (n=10 781) sets. A ProtRS was trained in the derivation cohort using least absolute shrinkage and selection operator penalized Cox regression and evaluated in the test set with sequential adjustment for demographics, clinical risk factors, and a polygenic risk score (PRS). External replication was performed in the Asklepios study (n=793).

In UK Biobank Proteomics (2312 events), each 1-SD higher ProtRS was associated with incident hypertension after covariate adjustment (hazard ratio, 1.68 [95% CI, 1.59-1.78]; P<0.001). Adding the protein score to demographics, clinical variables, and PRS improved model fit (global χ² from 1733 to 2048, P<0.001), discrimination (C-index, 0.745-0.765), net reclassification improvement (19.0% [95% CI, 15.9-22.0]), and integrated discrimination improvement (2.9% [95% CI, 2.4-3.8]). In Asklepios (221 events), the ProtRS was associated with an increased risk of hypertension (hazard ratio, 1.32 [95% CI, 1.13-1.55]; P<0.001) after covariate adjustment, and improved global χ² (from 120.4 to 132.4; P<0.001), C-index (from 0.723 to 0.736) and discrimination (net reclassification improvement, 13.1% [95% CI, 1.2%-24.1%]). Pathway analysis highlighted neutrophil degranulation, insulin-like growth factor, PI3K signaling, and extracellular matrix remodeling.

A ProtRS improves prediction of incident hypertension beyond demographics, clinical, and genetic information in UK Biobank Proteomics, with supportive external replication. Proteomic information may enhance individualized hypertension risk stratification and provide biologic insights into disease development.