Integrating Multiomics and Machine Learning: Senescence-Regulated ALMS1-IT1/miR-7c-5p/HMGA2 Axis as a Novel Therapeutic Target for Head and Neck Squamous Cell Carcinoma.

Head and neck squamous cell carcinoma (HNSCC) is distinguished by the absence of definitive diagnostic markers and effective treatment modalities, which results in its poor clinical outcomes. Although senescence-targeted therapies have emerged as promising interventions, the functional complexity of cellular senescence in HNSCC pathogenesis remains poorly characterized. This study investigated the role of senescence levels in HNSCC through an extensive multiomics analysis coupled with a diverse array of machine learning algorithms. Results indicate that elevated senescence levels in HNSCC, alongside a novel subtyping based on senescence-associated long noncoding RNAs, highlight significant variations in the tumor microenvironment and treatment responses across different subtypes. We used a set of eight machine learning algorithmsgradient boosting, gradient boosting with cross-validation, gradient boosting machines, linear discriminant analysis, linear discriminant analysis with cross-validation, random forest, support vector machine, and support vector machine with cross-validationto identify the significant ALMS1-IT1/miR-7c-5p/HMGA2 regulatory axis. Single-cell analysis of dysregulated expression within the ALMS1-IT1/miR-7c-5p/HMGA2 axis predominantly highlighted aberrant interactions between immune cells and epithelial cells. Drug sensitivity evaluations suggested that inhibitors targeting ALMS1-IT1, such as sulforaphane and cloxacillin, alongside HMGA2 inhibitors such as ixabepilone, provide innovative, personalized, and combined targeted therapeutic strategies for patients with HNSCC.