Adiposity-modified proteomic signatures characterize heterogeneity in heart failure risk and cardiometabolic trajectories among metabolically unhealthy individuals
4 affiliations
- 1Research Center of Clinical Epidemiology, Peking University Third Hospital
- 2Healthcare Big Data Research Institute, School of Public Health, Cheeloo College of Medicine, Shandong University
- 3Department of Medical Dataology, School of Public Health, Cheeloo College of Medicine, Shandong University
- 4Department of Epidemiology and Biostatistics, School of Public Health, Peking University
Abstract
Background: Obesity increases the risk of heart failure (HF), but substantial risk heterogeneity exists among metabolically unhealthy individuals. We aimed to identify adiposity-modified proteomic signatures associated with incident HF and cardiometabolic disease progression.
Methods: We included 42,324 metabolically unhealthy UK Biobank participants without baseline HF. Cox models evaluated associations of adiposity measures with incident HF and identified HF-related proteins modified by adiposity. Unsupervised weighted gene co-expression network analysis identified protein modules, followed by stability-based elastic-net Cox regression to prioritize module-specific core signatures. Multi-state models evaluated cardiometabolic trajectories preceding HF. Incremental prediction beyond clinical factors, continuous WHtR, and NT-proBNP was assessed using measures of discrimination, reclassification, calibration, and clinical utility with bootstrap internal validation.
Results: During a median follow-up of 13.65 years, 2,178 incident HF events occurred. Among 2,923 proteins, 1,406 were associated with incident HF after false discovery rate (FDR) correction. Interaction analyses identified 117 significant protein–adiposity pairs involving 76 unique proteins. Network analysis organized these proteins into six co-expression modules with distinct associations with HF, adiposity, and cardiometabolic diseases, from which five core signatures were prioritized. Among participants with central obesity, the turquoise top4 signature comprising WFDC2, GDF15, OCLN, and SHISA5 showed the strongest association with incident HF (HR per SD, 1.93; 1.82–2.04), while its simplified top3 signature comprising SHISA5, WFDC2, and GDF15 retained a comparable association (HR, 1.90; 1.79–2.02). PRAP1 captured a complementary liver-metabolic pattern, whereas the brown NCAN–PON3–ADGRG2 signature was inversely associated with HF (HR, 0.72; 0.68–0.76). The signatures showed differential associations across cardiometabolic disease development and subsequent progression to HF, and their associations persisted after adjustment for continuous waist-to-height ratio. Beyond clinical factors and NT-proBNP, combining the turquoise top3 signature with PRAP1 increased the C-index by 0.021 and the 10-year AUC by 0.017, with event and nonevent net reclassification improvements of 0.095 and 0.258, respectively.
Conclusions: Adiposity-modified plasma proteins characterized molecular heterogeneity in incident HF and its cardiometabolic antecedents. Network-based analyses highlighted biological processes involving metabolic stress, immune activation, tissue remodeling, lipid metabolism, and inter-organ interactions. A parsimonious four-protein panel comprising SHISA5, WFDC2, GDF15, and PRAP1 provided incremental predictive information, although external validation is required.
Related
- Divergent ectopic fat distribution and adipokine/hepatokine levels characterize the metabolically healthy overweight/obesity phenotype
- Statistical models to characterize colon tumor stiffness heterogeneity through representative atomic force microscopy maps
- Common Carotid Artery Extra-Media Thickness as a Marker of Perivascular Obesity and Cardiometabolic Risk in Individuals Without Cardiovascular Disease
- Parental obesity and offspring cardiometabolic trajectories from childhood to adulthood
- Cardiometabolic Risk Trajectories in Young Adults Across 16 Middle East and North African Countries, 1990-2018: A Sex-Asymmetric Divergence
BibTeX
@article{lv2026adipositymodified,
title = {{Adiposity-modified proteomic signatures characterize heterogeneity in heart failure risk and cardiometabolic trajectories among metabolically unhealthy individuals}},
author = {Jiayu Lv and Shucheng Si and Zhaoyu Wang and Zhaoji Li and Jue Liu and Liyuan Tao},
journal = {Frontiers in Endocrinology},
year = {2026},
month = nov,
doi = {10.3389/fendo.2026.1955231},
url = {https://www.frontiersin.org/articles/10.3389/fendo.2026.1955231/full},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Adiposity-modified proteomic signatures characterize heterogeneity in heart failure risk and cardiometabolic trajectories among metabolically unhealthy individuals
AU - Jiayu Lv
AU - Shucheng Si
AU - Zhaoyu Wang
AU - Zhaoji Li
AU - Jue Liu
AU - Liyuan Tao
PY - 2026
DA - 2026/11/09
JO - Frontiers in Endocrinology
DO - 10.3389/fendo.2026.1955231
UR - https://www.frontiersin.org/articles/10.3389/fendo.2026.1955231/full
AB - Background: Obesity increases the risk of heart failure (HF), but substantial risk heterogeneity exists among metabolically unhealthy individuals. We aimed to identify adiposity-modified proteomic signatures associated with incident HF and cardiometabolic disease progression. Methods: We included 42,324 metabolically unhealthy UK Biobank participants without baseline HF. Cox models evaluated associations of adiposity measures with incident HF and identified HF-related proteins modified by adiposity. Unsupervised weighted gene co-expression network analysis identified protein modules, followed by stability-based elastic-net Cox regression to prioritize module-specific core signatures. Multi-state models evaluated cardiometabolic trajectories preceding HF. Incremental prediction beyond clinical factors, continuous WHtR, and NT-proBNP was assessed using measures of discrimination, reclassification, calibration, and clinical utility with bootstrap internal validation. Results: During a median follow-up of 13.65 years, 2,178 incident HF events occurred. Among 2,923 proteins, 1,406 were associated with incident HF after false discovery rate (FDR) correction. Interaction analyses identified 117 significant protein–adiposity pairs involving 76 unique proteins. Network analysis organized these proteins into six co-expression modules with distinct associations with HF, adiposity, and cardiometabolic diseases, from which five core signatures were prioritized. Among participants with central obesity, the turquoise top4 signature comprising WFDC2, GDF15, OCLN, and SHISA5 showed the strongest association with incident HF (HR per SD, 1.93; 1.82–2.04), while its simplified top3 signature comprising SHISA5, WFDC2, and GDF15 retained a comparable association (HR, 1.90; 1.79–2.02). PRAP1 captured a complementary liver-metabolic pattern, whereas the brown NCAN–PON3–ADGRG2 signature was inversely associated with HF (HR, 0.72; 0.68–0.76). The signatures showed differential associations across cardiometabolic disease development and subsequent progression to HF, and their associations persisted after adjustment for continuous waist-to-height ratio. Beyond clinical factors and NT-proBNP, combining the turquoise top3 signature with PRAP1 increased the C-index by 0.021 and the 10-year AUC by 0.017, with event and nonevent net reclassification improvements of 0.095 and 0.258, respectively. Conclusions: Adiposity-modified plasma proteins characterized molecular heterogeneity in incident HF and its cardiometabolic antecedents. Network-based analyses highlighted biological processes involving metabolic stress, immune activation, tissue remodeling, lipid metabolism, and inter-organ interactions. A parsimonious four-protein panel comprising SHISA5, WFDC2, GDF15, and PRAP1 provided incremental predictive information, although external validation is required.
SN - 1664-2392
ER -