Senescence-Related Diagnostic Biomarkers in Gastric Carcinoma: Identification and Comprehensive Analysis Using a Machine Learning Algorithm

Abstract

Recent research indicates that senescence plays a pivotal role in carcinogenesis. However, there is a lack of studies exploring the clinical significance and predictive capabilities of senescence-related genes (SRGs) in gastric carcinoma (GC). This study employs machine learning techniques to discern the diagnostic biomarker associated with senescence in GC. Moreover, it delves into an extensive evaluation of its immunological infiltration, biological function, and clinical relevance. Our analysis identified four SRGs (FEN1, HIF1A, PDGFRB, and PEX5) using a combination of least absolute shrinkage and selection operator regression, support vector machine recursive feature elimination, and the area under the curve metrics. Subsequently, these four SRGs were incorporated into a senescence-based prognostic signature termed “riskScore.” Notably, the riskScore demonstrated reliability and accuracy as an independent prognostic marker. We observed a robust association between the riskScore and tumor mutation burden, clinicopathological features, tumor immune microenvironment, and overall prognosis. Single-cell sequencing revealed heightened immune cell infiltration in the high-risk group. Furthermore, the riskScore emerged as a pivotal determinant guiding therapeutic decisions for GC, including immunotherapy and chemotherapy. The results strongly suggest the riskScore as the signature diagnostic biomarker for GC. These findings lay a robust foundation for GC treatments and hold promise for developing a rapid, non-invasive technique for disease monitoring and prognostic prediction.