Key result
A 10-protein model plus traditional risk factors predicts incident hypertension with ~0.75 AUC.
Why the study?
Longitudinal proteomic studies evaluating the epidemiological associations of circulating proteins with hypertension remain scarce.
Are specific plasma proteins associated with the development of hypertension in individuals without baseline hypertension?
Cohort (n=22,788)
Are specific plasma proteins associated with the development of hypertension in individuals without baseline hypertension?
Effect estimate: AUC 0.751 (95% CI 0.727-0.773)
Specific plasma proteomic profiles, particularly those related to immunity and lipid metabolism, are significantly associated with incident hypertension and can improve predictive modeling.
May identify protein biomarkers for incident hypertension risk; leaves open whether targeting these alters clinical outcomes.
Background: Longitudinal proteomic studies on hypertension remain scarce. We evaluated the epidemiological associations of circulating proteins with hypertension on the basis of the UK Biobank (UKB) and public GWAS-summary data in this study. Methods: This study included 22 788 participants without hypertension on baseline and 2910 plasma proteins from the UKB. The competing risk regression and Mendelian randomization analysis were used to examine the association between plasma proteins and hypertension. Machine learning was used to identify proteins with strong predictive power for hypertension, and WGCNA was employed to characterize their patterns. Results: A total of 462 proteins exhibited significant associations with hypertension: 328 proteins of them had significant associations with hypertension, SBP, and DBP. Both the Fine-Gray competing risk regression model and the Mendelian randomization analysis revealed eight proteins (ADM, ELN, NCAN, PRSS8, ZBTB17, FURIN, TINAGL1, and WARS) with consistent directional effects. We identified potential 13 drugs targeting using the DGIdb database. Further, we created a predictive model using the 10 proteins and traditional risk factors by random forest and light gradient boosting machine model. The ROC curve had an AUC of 0.751 (95% confidence interval: 0.727–0.773). Immunity and disordered lipid metabolism-related protein modules identified by WGCNA were associated with an increased risk of hypertension. Conclusion: The 462 proteins are significantly associated with hypertension. The eight proteins have causal associations with hypertension. The characteristic patterns of immunity and disordered lipid metabolism are capable of contributing to the pathogenesis of hypertension.
No takes yet. Share an insight, caveat, or question.
Zhang et al. (2026) conducted a cohort in Hypertension (n=22,788). Circulating plasma proteins was evaluated on Incident hypertension (AUC 0.751, 95% CI 0.727-0.773). A predictive model using 10 plasma proteins and traditional risk factors yielded an AUC of 0.751 (95% CI: 0.727-0.773) for hypertension, with 8 proteins showing causal associations.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: