Multi omics analysis of mitophagy subtypes and integration of machine learning for predicting immunotherapy responses in head and neck squamous cell carcinoma.

Mitophagy serves as a critical mechanism for tumor cell death, significantly impacting the progression of tumors and their treatment approaches. There are significant challenges in treating patients with head and neck squamous cell carcinoma, underscoring the importance of identifying new targets for therapy. The function of mitophagy in head and neck squamous carcinoma remains uncertain, thus investigating its impact on patient outcomes and immunotherapeutic responses is especially crucial. We initially analyzed the differential expression, prognostic value, intergene correlations, copy number variations, and mutation frequencies of mitophagy-related genes at the pan-cancer level. Through unsupervised clustering, we divided head and neck squamous carcinoma into three subtypes with distinct prognoses, identified the signaling pathway features of each subtype using ssGSEA, and characterized subtype B as having features of an immune desert using various immune infiltration calculation methods. Using multi-omics data, we identified the genomic variation characteristics, mutated gene pathway features, and drug sensitivity features of the mitophagy subtypes. Utilizing a combination of 10 machine learning algorithms, we have developed a prognostic scoring model called Mitophagy Subgroup Risk Score (MSRS), which is used to predict patient survival and the response to immune checkpoint blockade therapy. Simultaneously, we applied MSRS to single-cell analysis to explore intercellular communication. Through laboratory experiments, we validated the biological function of SLC26A9, one of the genes in the risk model. In summary, we have explored the significant role of mitophagy in head and neck tumors through multi-omics data, providing new directions for clinical treatment.