Abstract
T-helper 17 (Th17) cells significantly influence the onset and advancement of malignancies. This study endeavor focused on delineating molecular classifications and developing a prognostic signature grounded in Th17 cell differentiation-related genes (TCDRGs) using machine learning algorithms in head and neck squamous cell carcinoma (HNSCC). A consensus clustering approach was applied to The Cancer Genome Atlas-HNSCC cohort based on TCDRGs, followed by an examination of differential gene expression using the limma package. Machine learning techniques were utilized for feature selection and model construction, with validation performed using the GSE41613 cohort. The interplay between the predictive marker, immune landscape, immunotherapy response, drug sensitivity, and clinical outcomes was assessed, and a nomogram was constructed. Functional evaluations of TCDRGs were conducted through colony formation, transwell invasion, and wound healing assays. Two distinct HNSCC subtypes with significant differences in prognosis were identified based on 87 TCDRGs, indicating different levels of Th17 cell differentiation. Thirteen differentially expressed TCDRGs were selected and used to create a risk signature, T17I, using the random survival forest algorithm. This signature was associated with grade, chemotherapy, radiotherapy, T stage, and somatic mutations. It was revealed that there were differences in the immune response-related pathways between the high- and low-risk groups. Inflammatory pathways were significantly activated in the low-risk group. The T17I signature was associated with immune infiltration. Specifically, there was a higher infiltration of immune activation cells in the low-risk group, whereas the high-risk group had a higher infiltration of M2 macrophages. In addition, the T17I signature was significantly associated with drug sensitivity. A nomogram combining age, radiotherapy, and the T17I signature accurately predicted the prognosis of patients with HNSCC. Finally, in vitro experiments confirmed that knockdown of LAT gene expression promotes proliferation, metastasis, and invasion of HNSCC cells. In conclusion, this study successfully identified molecular subtypes and constructed a prognostic signature and nomogram based on TCDRGs in HNSCC, which may aid in personalized treatment strategies.