Novel Four-gene Panel for Detecting Senescence-associated Cell States in Skeletal Muscle Tissue

Sarcopenia, characterized by age-associated loss of skeletal muscle mass, function, and physical performance, is a major challenge in aging societies because of its association with a decreased lifespan. Existing gene expression-based diagnostic methods often rely on large gene sets, requiring high costs and analytical complexity. In this study, we developed a machine learning-driven framework to identify senescence-associated cell states in skeletal muscle tissue.

Publicly available single-cell RNA sequencing data from 2- and 24-month-old C57BL/6J male mice from single-cell and single-nucleus RNA sequencing datasets comprising over 365,000 cells from skeletal muscle were obtained from the DRYAD Repository and consolidated into 15 cell populations. Candidate genes for machine learning were selected from differential expression analysis. Multiple machine learning algorithms, including logistic regression, support vector machines, and random forest, were trained with recursive feature elimination. Model performance was evaluated using the area under the receiver operating characteristic curve.

Differential expression analysis across 15 distinct cell populations yielded 30 candidate genes, to which five machine learning models were applied to select biomarkers. Using our approach, we identified a four-gene panel (Malat1, Wdr89, Zfp36, and Jund) exhibiting high predictive accuracy. This panel was validated using additional aging datasets and compared with existing models, which highlighted its potential as a reliable tool for detecting senescence-associated cells.

In this study, we established a four-gene biomarker panel for detecting senescence-associated cells in skeletal muscle, providing a practical tool for investigating sarcopenia pathophysiology and identifying therapeutic targets.