Identification of diagnostic senescence-associated marker genes in asthma based on machine learning and experimental...

Asthma imposes a significant global health burden; however, the mechanistic role of cellular senescence in its pathogenesis remains poorly defined.

Transcriptomic profiles of lung tissues from asthmatic patients and healthy controls were retrieved from the Gene Expression Omnibus (GEO) database. Raw RNA-seq data were processed using R packages including limma. Ensemble machine learning approaches-least absolute shrinkage and selection operator (LASSO) regression, Random Forest (RF), and support vector machine-recursive feature elimination (SVM-RFE) - were applied to identify hub senescence-associated genes (SAGs). Immune cell infiltration was quantified using deconvolution algorithms to compare asthmatic and control cohorts. Additionally, both animal and cellular models of asthma were established to validate the expression of these candidate genes.

Transcriptomic data from 88 asthmatic and 20 control lung samples (from the GEO database) were integrated with the machine learning analyses (LASSO, RF, and SVM-RFE) to identify six hub SAGs (EGR2, FLRT2, KIT, NANOG, NOX1, and RUNX1T1). The combined gene signature demonstrated high diagnostic accuracy (area under the curve (AUC) = 0.935) and robust external validation performance (AUC = 0.902). Nomogram, decision curve, and calibration analyses confirmed its clinical utility. Functional enrichment analyses computationally predicted associations between these genes and several signaling pathways, including the MAPK cascades, cytokine signalling, and AGE-RAGE pathways. Finally, the expression of the hub SAGs was validated in mouse and cellular asthma models using quantitative reverse transcription-polymerase chain reaction.

Collectively, these findings identify a novel SAG signature with diagnostic potential for asthma, and provide mechanistic insights into the role of cellular senescence in asthma development.