Outlive
LongevityResearchHub

A blood test for cellular aging predicts disease and mortality risk

A deep-learning based biomarker of systemic cellular senescence burden to predict mortality and health outcomes

TL;DR

Researchers created an AI-powered blood biomarker (SASP Score) that measures senescent cell burden and successfully predicted mortality, dementia, heart attacks, and stroke in large population studies. The score also changed in response to exercise intervention, suggesting it could track anti-aging treatments.

Credibility Assessment Preliminary — 40/100
Study Design
Rigor of the research methodology
11/20
Sample Size
Whether the study was sufficiently powered
13/20
Peer Review
Review status and journal reputation
3/20
Replication
Has this finding been independently reproduced?
5/20
Transparency
Funding disclosure and data availability
8/20
Overall
Sum of all five dimensions
40/100

What this means

This is promising early-stage research suggesting a blood test could measure cellular aging and predict serious disease—but it's not yet peer-reviewed and needs independent confirmation before relying on it clinically. It's worth watching as it moves through publication and validation.

Red Flags: Preprint status (not peer-reviewed). Missing critical methodological details: exact sample sizes, confidence intervals, effect sizes, and characteristics of the independent validation cohort. No mention of data availability, code availability, or preregistration. Potential reverse causality in cross-sectional associations. Exercise intervention cohort appears small with only 18-month follow-up. No disclosure of funding sources or conflicts of interest in the abstract provided. The work involves UK Biobank (legitimate resource) but funding body not specified.

Cellular senescence—the process where cells stop dividing and accumulate with age—is considered a hallmark of biological aging. These senescent cells release harmful molecules (the senescence-associated secretory phenotype, or SASP) that can spread damage to neighboring tissues and drive disease. Current methods to measure senescence burden are impractical for clinical use, creating a gap between aging research and real-world application.

The researchers tackled this by developing a 'SASP Score'—a composite biomarker derived from blood proteins associated with senescence. They trained a deep learning model (Guided autoencoder with Transformer, or GAET) on proteomics data from the UK Biobank's Pharma Proteomics Project, identifying which blood proteins best capture senescent cell burden. They then tested this score in the same UK Biobank cohort and validated it in an independent randomized controlled trial that included an exercise intervention.

The results are promising: the SASP Score independently predicted mortality and incident serious chronic diseases (dementia, COPD, myocardial infarction, stroke) even after adjusting for traditional risk factors. Notably, in the intervention cohort, multimodal exercise (combining aerobic, resistance, and flexibility training) significantly altered the SASP Score trajectory over 18 months, suggesting the biomarker is responsive to lifestyle changes.

However, this is a preprint on medRxiv, meaning it has not yet undergone peer review—a critical gatekeeping step in science. The paper lacks detail on several important points: exact sample sizes for each analysis, effect sizes with confidence intervals, and specifics about the independent validation cohort. The cross-sectional associations within UK Biobank could reflect reverse causality (sick people have more senescent cells, not the reverse). The exercise study is interesting but relatively small and short-term; we don't yet know if the SASP Score changes persist, whether they predict real health improvements, or how durable the intervention effects are.

This work is methodologically sound and uses credible data sources, but the evidence is preliminary. The SASP Score could become a valuable tool for aging research and personalized medicine, but it needs peer-reviewed publication, independent replication, and longer follow-up studies before clinical adoption. It's a promising direction, not yet a proven solution.

View Original Source

0 Comments