Computational study reveals a machine learning-based cellular senescence metric in lung adenocarcinoma, highlighting pro-senescence therapeutic strategies.
Key Points
To develop an objective machine learning framework for quantifying cellular senescence burden across cancer transcriptomes and evaluate its therapeutic relevance in lung adenocarcinoma.
Curated 888 transcriptomic profiles across various cell types and senescence conditions, applying the Boruta algorithm to identify a consensus cellular senescence-related gene signature.
Benchmarked ten machine learning algorithms with cross-validation to construct the Predictive Cellular Senescence Model (PreCSenM) for continuous senescence scoring.
Assessed model performance across lung adenocarcinoma datasets and experimentally validated drug-induced senescence mechanisms using epigenetic and transcriptional profiling.
PreCSenM demonstrated superior accuracy and consistency in predicting cellular senescence compared to existing methods across normal and neoplastic transcriptomic datasets.
High senescence scores in lung adenocarcinoma significantly associated with enhanced genomic stability, elevated immune infiltration features, and favorable patient outcomes.
Histone deacetylase inhibitors acted as potent senescence inducers in lung adenocarcinoma by upregulating the core transcription factor FOSB, with FOSB knockdown mitigating this pro-senescence effect.