At the crossroads of senescence and immunosuppression: Machine learning and multi-omics unveil the central role of FAP in HNSCC.

The global incidence of head and neck squamous cell carcinoma (HNSCC) is increasing, and the failure rate of current treatments remains substantially high. Therefore, there is an urgent need to find effective targets and formulate precise personalized treatment strategies. HNSCC is characterized by an immunosuppressive environment, and the accumulation of senescent cells can promote aging by reshaping immune system function. However, in the context of HNSCC, the understanding of aging-related immunosuppressive genes (ARIS) remains limited. In this study, we identified 13 prognosis-associated ARIS and, using machine learning, pinpointed Fibroblast Activation Protein (FAP) as a key ARIS. In vitro experiments confirmed functional inhibition of FAP via siRNA-mediated silencing and pharmacological blockade with UMAC-1110 significantly suppressed the proliferation, migration, and invasion of CAL-27 cells. Furthermore, FAP was predominantly enriched in the FAP-positive myofibroblastic cancer-associated fibroblast (FAP + myCAF) subpopulation and was associated with altered inferred FAP + myCAF-neutrophil communication, consistent with a potential contribution to an immunosuppressive tumor microenvironment (TME). We subsequently performed in silico knockout analyses in FAP + myCAFs and constructed a FAP-centered regulatory network, thereby identifying candidate FAP-associated signaling modules for further functional validation. Notably, FAP expression increased following anti-programmed death-ligand 1 (PD-L1) treatment. We also identified three FAP targeting inhibitors, Pazopanib, Talazoparib and Ursodiol. And Pazopanib that exhibited a more stable binding affinity with FAP was selected for further in vitro validation. Collectively, these findings may offer new avenues for clinical intervention and precision therapy in HNSCC.