Proteomics risk scores and mortality in heart failure: Generalizability across populations.
JJ, S., J, J., CG, D., KO, K., M, H., E, L., SJ, B., & VL, R. (2026). Proteomics risk scores and mortality in heart failure: Generalizability across populations.. PloS one. https://doi.org/10.1371/journal.pone.0350697
JJ S, J J, CG D, KO K, M H, E L, et al. Proteomics risk scores and mortality in heart failure: Generalizability across populations.. PloS one. 2026; doi: 10.1371/journal.pone.0350697
JJ S, J J, CG D, et al. Proteomics risk scores and mortality in heart failure: Generalizability across populations.[J]. PloS one. 2026. DOI: 10.1371/journal.pone.0350697.
@article{jj2026,
author = {Shearer JJ and Joo J and Downie CG and Kuku KO and Hashemian M and Leifer E and Bielinski SJ and Roger VL},
title = {Proteomics risk scores and mortality in heart failure: Generalizability across populations.},
journal = {PloS one},
year = {2026},
doi = {10.1371/journal.pone.0350697},
note = {PMID: 42335150},
}
TY - JOUR AU - Shearer JJ AU - Joo J AU - Downie CG AU - Kuku KO AU - Hashemian M AU - Leifer E AU - Bielinski SJ AU - Roger VL TI - Proteomics risk scores and mortality in heart failure: Generalizability across populations. T2 - PloS one PY - 2026 DO - 10.1371/journal.pone.0350697 AN - PMID:42335150 ER -
BACKGROUND: Heart failure (HF) is a complex syndrome with high mortality. Proteomics risk scores have shown promise in predicting mortality beyond guideline-recommended clinical tools. It is crucial to understand how risk scores generated by different methods and populations perform, and whether they highlight the same protein targets relevant to outcomes. METHODS: To examine whether the study design impacts proteomics scores designed to predict mortality in HF, we evaluated three published risk scores that used the SomaScan assay to measure plasma proteins in a community cohort, clinical trial, and registry. Each score was assessed in the aforementioned community cohort and Cox models examined the association of a 1-standard deviation increase in score with mortality, with and without adjustment for clinical covariates. Performance of each risk score to predict 5-year mortality risk was assessed using calibration plots and time-dependent area under the curve and compared with a clinical model. RESULTS: Risk scores were similarly distributed and moderately correlated (Pearson correlation coefficient = 0.59-0.76). A 1-standard deviation increase in each risk score was associated with an increased risk of all-cause mortality: community cohort (HR = 2.70, 95% CI: 2.50-2.91); clinical trial (HR = 1.76, 95% CI: 1.65-1.88); registry (HR = 1.70, 95% CI: 1.6-1.81). Risk remained after adjustment for clinical covariates, although slightly attenuated, and similar across different ejection fraction categories. All risk scores showed strong calibration across the risk levels, alone, with an average expected over observed ratio ranging between 0.96-1.56. Seven proteins were included in at least two risk scores, with renin being included in all three. CONCLUSIONS: All three proteomics risk scores improved risk stratification in HF patients beyond guideline recommended clinical tools, independent of study design and ejection fraction. These results demonstrate that proteomics risk scores can enhance risk stratification across the HF syndrome, even when derived from different methods and populations.