Multi-omics models for hypertension prediction using blood pressure PRS and omics scores

Background: Hypertension is a major modifiable risk factor for cardiovascular disease, yet the relative contributions of proteomic (ProtS), metabolomic (MetS), and polygenic risk scores (PRS) to incident hypertension risk remain incompletely characterized. Methods: We developed ProtS and MetS for systolic (SBP) and diastolic blood pressure (DBP), and hypertension using Elastic Net regression in the Multi-Ethnic Study of Atherosclerosis (MESA). We incorporated these scores with previously developed BP PRSs into 5 incremental XGBoost models, with external validation in the Jackson Heart Study (JHS). We further applied transfer learning to adapt the MESA-derived models to JHS. Elastic Net selected proteins were prioritized using Mendelian randomization, colocalization, and functional analyses. Results: In MESA, model discrimination improved with the addition of omics scores, with the combined model achieving an area under the curve (AUC) of 0.72 compared with 0.63 for the clinical baseline model. In JHS, ProtS showed the highest AUC of 0.68 and the smallest cross-cohort performance decline, compared to MetS and PRS. Incorporating measured SBP and DBP substantially improved discrimination, with the PRS+BP model achieving an AUC of 0.80 in MESA and the baseline+BP model achieving an AUC of 0.74 in JHS. Among JHS participants with normal baseline BP, ProtS provided the strongest discrimination among models without measured BP (AUC = 0.62), while PRS+BP model achieved an AUC of 0.69. Leaf-weight refreshing consistently improved calibration in JHS while preserving discrimination. Mendelian randomization identified 77 proteins associated with BP traits or hypertension, involved in vascular, inflammatory, renal, and natriuretic pathways. Conclusions: ProtS demonstrated superior predictive performance and cross-cohort transferability among the omics modalities evaluated, while measured BP remained the strongest predictors of incident hypertension. Proteomic and genetic information may provide complementary biological signals for identifying individuals at elevated hypertension risk, particularly before clinically elevated BP is detectable.

Authors

Institutions

Publication Details

Journal
medRxiv
Published
2026-11-16
DOI
https://doi.org/10.64898/2026.11.13.26368055
Primary Topic
Blood Pressure and Hypertension Studies
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Multi-omics models for hypertension prediction using blood pressure PRS and omics scores

Brian Spitzer, Usman A. Tahir, Yana Hrytsenko, Tamar Sofer et al.
medRxiv
Blood Pressure and Hypertension Studies
preprint

Multi-omics models for hypertension prediction using blood pressure PRS and omics scores

Brian Spitzer, Usman A. Tahir, Yana Hrytsenko, Tamar Sofer, Laura M. Raffield, Yogesh Purushotham, Ziqing Wang, Zsu Zsu Chen, Jerome Rotter, Sivan Almogy, Kent Taylor, Henry J. J. Lin, Steve Rich
preprint en

Abstract

Background: Hypertension is a major modifiable risk factor for cardiovascular disease, yet the relative contributions of proteomic (ProtS), metabolomic (MetS), and polygenic risk scores (PRS) to incident hypertension risk remain incompletely characterized. Methods: We developed ProtS and MetS for systolic (SBP) and diastolic blood pressure (DBP), and hypertension using Elastic Net regression in the Multi-Ethnic Study of Atherosclerosis (MESA). We incorporated these scores with previously developed BP PRSs into 5 incremental XGBoost models, with external validation in the Jackson Heart Study (JHS). We further applied transfer learning to adapt the MESA-derived models to JHS. Elastic Net selected proteins were prioritized using Mendelian randomization, colocalization, and functional analyses. Results: In MESA, model discrimination improved with the addition of omics scores, with the combined model achieving an area under the curve (AUC) of 0.72 compared with 0.63 for the clinical baseline model. In JHS, ProtS showed the highest AUC of 0.68 and the smallest cross-cohort performance decline, compared to MetS and PRS. Incorporating measured SBP and DBP substantially improved discrimination, with the PRS+BP model achieving an AUC of 0.80 in MESA and the baseline+BP model achieving an AUC of 0.74 in JHS. Among JHS participants with normal baseline BP, ProtS provided the strongest discrimination among models without measured BP (AUC = 0.62), while PRS+BP model achieved an AUC of 0.69. Leaf-weight refreshing consistently improved calibration in JHS while preserving discrimination. Mendelian randomization identified 77 proteins associated with BP traits or hypertension, involved in vascular, inflammatory, renal, and natriuretic pathways. Conclusions: ProtS demonstrated superior predictive performance and cross-cohort transferability among the omics modalities evaluated, while measured BP remained the strongest predictors of incident hypertension. Proteomic and genetic information may provide complementary biological signals for identifying individuals at elevated hypertension risk, particularly before clinically elevated BP is detectable.

medRxiv
Beth Israel Deaconess Medical Center (US), The Lundquist Institute (US), Beth Israel Deaconess Hospital (US), Yale University (US)
Blood Pressure and Hypertension Studies
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon