Plasma proteomic signatures of cellular aging predict human disease.

Summary

Using >7,000 plasma proteins in 60,542 individuals, the authors built machine-learning models to estimate biological aging across >40 cell types. Cell-type aging signatures predicted incident disease and mortality, and in smokers, extreme aging of respiratory epithelial cells increased lung cancer risk by 58%. A polycellular aging risk score stratified mortality across cohorts and platforms.

Key Findings

• Built machine-learning models from >7,000 plasma proteins in 60,542 individuals to estimate biological age for >40 cell types.
• Cell-type aging signatures predicted incident disease and all-cause mortality over up to 15 years.
• In smokers, extreme respiratory epithelial aging conferred a 58% higher lung cancer risk versus smoking alone.
• APOE genotype showed divergent cell-type aging patterns (e.g., older astrocytes in APOE4 carriers) linked to disease risk.
• A polycellular aging risk score stratified mortality risk across cohorts and proteomics platforms.

Clinical Implications

Proteomic aging signatures could augment risk models for lung cancer (especially in smokers) and guide targeted prevention/screening strategies. Longitudinal monitoring may inform timing of interventions aimed at modulating biological aging.

Limitations

• Observational design precludes causal inference; interventions to modify aging signatures were not tested.
• Potential variability across proteomics platforms and cohorts requires further harmonization for clinical deployment.

Future Directions

Prospective clinical studies to integrate proteomic aging signatures into lung cancer screening algorithms, and interventional trials testing whether modifying aging biology reduces incident disease.