Dynamic protein trajectories and inflammaging landscapes throughout human aging revealed by large-scale proteomics

Abstract Background Aging is a heterogeneous process characterized by systemic molecular remodeling and chronic inflammation ("inflammaging(definition)"), which collectively drive cardiovascular multi-morbidity. While chronological age is the dominant risk factor for disease, it fails to capture the individualized biological pathways and cellular drivers of physiological decline. Purpose To delineate age-dependent protein trajectories, identify the modulation of these by cardiovascular risk factors (CVRFs), and construct a proteomic "Healthy Aging Clock" to evaluate biological age acceleration, chronic inflammation status and associations with all-cause mortality and 23 age-related morbidities. Methods High-throughput untargeted proteomics by data independent acquisition LC-MS/MS with extracellular vesicle enrichment by Mag-Net (610 proteins) was performed on plasma samples from 3,113 participants in the MyoVasc cohort. Proteins were clustered into age-dependent trajectories, and cellular origins were inferred using bulk transcriptomics-integrated ensemble machine learning. Generalized Additive Models (GAMs) and linear interaction models assessed risk factor-modified aging kinetics and sex-specific divergences. To quantify biological aging, a "Healthy Aging Clock" was trained using elastic net regression exclusively on a strictly defined reference sub-cohort of "optimal agers" (N=172). Biological age acceleration (AgeAccel) and C-reactive protein (CRP) levels were used to stratify participants into distinct risk phenotypes. Results Four distinct age-dependent protein trajectories were identified. Cluster 1 was enriched in secreted proteins derived from platelet granules and myeloid lineages, whereas Clusters 2-4 reflected hemostasis, lipid metabolism, and extracellular matrix remodeling. Systematic interaction analysis revealed divergent aging kinetics between sexes for key immunomodulatory and lipid metabolism-related proteins, including A2M, ATRN, COL18A1 and LDLR (all FDR-adjusted interaction p < 0.05). Crucially, GAMs demonstrated that major CVRFs selectively hijack the platelet-immune axis, attenuating its physiological downregulation and maintaining a maladaptive pro-thrombotic state. The 92-protein healthy aging clock accurately predicted biological age (training set R=0.96; cross-validation test set R=0.82). Stratification based on AgeAccel and CRP identified a "double-hit" phenotype (accelerated aging with chronic inflammation) that was robustly associated with the highest risk of all-cause mortality and prevalent multi-morbidity. Conclusions Pathological aging is actively driven by the remodeling of the platelet-immune axis, a pathway selectively hijacked and accelerated by major CVRFs. A proteomic clock trained strictly on healthy individuals, when combined with systemic inflammatory status, effectively captured low and high risk proteomic aging endotypes, offering a robust framework for precision risk stratification.