Researchers develop deep-learning biomarker of systemic cellular senescence burden

Researchers have developed and validated a deep-learning-based composite SASP Score that predicts mortality and chronic disease risk, offering a potential tool to monitor systemic cellular senescence burden and the efficacy of longevity interventions.

Researchers from the UConn Center on Aging and partner institutions developed a composite senescence-associated secretory phenotype (SASP) Score to quantify systemic cellular senescence burden.

The model utilizes a Guided autoencoder with Transformer (GAET) framework trained on large-scale proteomics data from the UK Biobank Pharma Proteomics Project (UKB-PPP).

The deep-learning-derived SASP Score was validated as an independent predictor of mortality risk and incident chronic medical conditions, including dementia, COPD, myocardial infarction, and stroke.

In an independent clinical trial cohort, the SASP Score trajectory demonstrated sensitivity to change over an 18-month multimodal exercise intervention.

Measuring systemic cellular senescence in vivo has historically been difficult due to the lack of single-factor consensus biomarkers; a composite, multi-protein score solves this by serving as a proxy for whole-body senescence burden.

The biomarker's predictive capacity for major age-related chronic illnesses highlights the potential of geroscience-guided metrics in clinical risk assessment.

Validation of score trajectory changes in response to physical exercise demonstrates that the biomarker is dynamic and capable of tracking physiological responses to interventions.

The introduction of a validated, non-invasive biomarker of systemic cellular senescence could streamline cohort selection and provide a surrogate endpoint for clinical trials investigating senotherapeutic candidates.

Developers of senolytic therapies may leverage similar composite proteomic tools to demonstrate target engagement and therapeutic efficacy to regulatory bodies.

The study highlights the utility of large-scale biobank data paired with semi-supervised deep learning architectures to generate clinically actionable biological age metrics.

Further clinical evaluations applying the SASP Score to assess drug candidates, particularly senolytics, to confirm if pharmacological clearance of senescent cells registers as a reduction in the composite score.