Scientists Develop Blood Protein Score That Predicts Years of Healthy Life

Global life expectancy has risen dramatically, but healthy life expectancy has not kept pace, leaving millions of older adults burdened by chronic disease. Geroscience proposes that targeting the underlying hallmarks of aging—rather than individual diseases—could simultaneously delay multiple age-related conditions. A key obstacle to testing this hypothesis is the lack of validated surrogate biomarkers that can measure healthspan in clinical trials without requiring decades of follow-up.

To address this, researchers from the University of Connecticut, University of Exeter, and University of Helsinki developed the Healthspan Proteomic Score (HPS) using data from the UK Biobank Pharma Proteomics Project (UKB PPP). Healthspan was defined as years lived free from eight major conditions: cancer (excluding nonmelanoma skin cancer), type I and II diabetes, heart failure, myocardial infarction, stroke, COPD, dementia, and death. Of 53,018 participants with proteomic data from the Olink Explore 3072 panel (2,920 proteins), 43,119 were free from these conditions at baseline and were split 70/30 into training and test sets.

Using LASSO-penalized Cox regression followed by a Gompertz survival model, the team identified 86 proteins and chronological age that together predicted 10-year risk of developing a first healthspan-defining condition. HPS was computed as one minus that risk; lower scores indicate higher risk. The median HPS was 0.84 in disease-free individuals and dropped progressively to 0.07 in those with five concurrent conditions. The model achieved a C-statistic of 0.718 in the test set, with proteins alone contributing nearly all predictive power (C-statistic 0.715), indicating that biological information in the proteome largely supersedes chronological age.

A lower HPS was significantly associated with higher all-cause mortality and each individual disease outcome. Crucially, HPS demonstrated superior predictive accuracy compared to established biological age measures including proteomic clocks (PhenoAge, aging.ai) and epigenetic clocks (Horvath, GrimAge, DunedinPACE) when tested in both the internal UKB test cohort and the external Essential Hypertension Epigenetics (EH-Epi) Finnish twin cohort. Gene set enrichment analysis of HPS-associated proteins highlighted hallmark biological pathways including innate and adaptive immune response, inflammatory signaling, cellular metabolism, and extracellular matrix remodeling—consistent with known mechanisms of biological aging.

External validation in the EH-Epi study confirmed that HPS correlated meaningfully with epigenetic aging clocks and health outcomes in an independent population. The authors propose HPS as a practical surrogate endpoint for geroscience clinical trials, enabling shorter study durations and smaller sample sizes by substituting hard clinical endpoints with a validated proteomic proxy of overall health trajectory.