Evaluation of Large-Scale Plasma Proteomics for Prediction of Heart Failure in Individuals with A Full Range of Glucose Metabolism Profiles
European Journal of Preventive Cardiology, 2025
Zhang Y., Hou R., Sun W., Guo J., Chen Z., Li H., Li C., Wu L., Ji J., Zheng D.
| Disease area | Application area | Sample type | Products |
|---|---|---|---|
CVD | Patient Stratification | Plasma |  Olink Explore 3072/384 |
Abstract
Aims
Individuals with abnormal glucose metabolism are at a significantly higher risk of developing heart failure (HF). However, strategies for early identification of HF in this high-risk population remain inadequate. This study aimed to identify plasma protein biomarkers associated with HF development and construct predictive models to identify at-risk individuals.
Methods
We analyzed HF development in abnormal glucose metabolism population using data from 6,517 participants in discovery cohort and 2,783 in validation cohort, all from the UK Biobank, with no prior history of HF. Proteomic profiling was performed, and Lasso-Cox regression was used to identify protein associations, followed by Cox regression to develop predictive models. The model incorporated four proteins (NTproBNP, LTBP2, REN, GDF15) and clinical factors to create a protein-panel-clinical-factors (PPCF) model. For comparison, the model’s performance was also evaluated in individuals with normal glucose metabolism.
Results
Over a median follow-up of 13.90 years, 555 incident HF cases were recorded in discovery cohort. The PPCF model achieved an AUC of 0.823 (95% CI: 0.785 – 0.860) in validation cohort, improving predictive performance by 0.05 (P < 0.001) compared to clinical factors-only model. In general population of 23,107 individuals, PPCF model obtained an AUC of 0.807 (95% CI: 0.786 – 0.829). Both protein panel model and PPCF model demonstrated superior net benefits over clinical factors model in abnormal glucose metabolism population.
Conclusion
This study identified plasma protein biomarkers linked to HF development in abnormal glucose metabolism population and established the predictive models. These findings support early identification in high-risk populations.