Development and validation of an immune-related gene signature for the prognostic and immune landscape prediction in head and neck squamous cell carcinoma by integrated analysis of machine learning and Mendelian randomization.

Background: The immune microenvironment is pivotal in cancer advancement and reappearance. Nevertheless, the study concerning the association between immune-related genes (IRGs) and outcome in head and neck squamous cell carcinoma (HNSCC) is insufficient. This investigation sought to develop an IRG prediction model for accurately assessing the prognosis and immunological patterns in HNSCC. Methods: Gene expression and clinical information of HNSCC were obtained, including 522 HNSCC and 44 normal tissue specimens from The Cancer Genome Atlas and 270 HNSCC from the Gene Expression Omnibus GSE65858 database. By employing machine learning algorithms, an innovative prognostic IRG signature was established. This model allowed for calculating a risk score for each sample, thereby enabling the stratification of individuals into low-risk and high-risk cohorts. The prognostic significance of the signature was evaluated concerning survival, tumor mutation burden, immune cell infiltration, and its capacity to predict the response to immunotherapy. Subgroup analyses were performed based on age, sex, grade, and stage. Mendelian randomization (MR) was employed to assess the causative link between model gene expression and HNSCC development. Results: Ten IRGs were identified and incorporated into the predictive signature. The area under the receiver operating characteristic curves for overall survival at 1, 3, and 5 years were 0.694, 0.731, and 0.656, respectively. Kaplan-Meier survival analysis indicated that individuals in the high-risk cohort displayed substantially inferior outcomes versus those classified as low-risk. The multivariate prognostic analysis showed that the risk score was an independent prognostic factor associated with HNSCC (hazard ratio =3.647, P<0.001). Subgroup analyses stratified by clinical parameters demonstrated that the prognostic signature was consistently effective across all subgroups, underscoring its wide applicability. Additionally, individuals with low-risk demonstrated a more favorable prognosis, which was linked to heightened immunological scores, enhanced immune-related functioning, and increased immune cell infiltration. Moreover, low-risk patients responded better to immunotherapy than high-risk individuals. MR results suggested a causal relationship between CCR7 expression and HNSCC development. Conclusions: The IRG-related signature has been developed to predict survival results and immunological features of HNSCC. The model's robustness across various clinical subgroups, coupled with its capacity to predict responses to immunotherapy, highlights its potential for clinical application. This reliable prognostic signature has the ability to guide the development of novel therapeutic strategies for HNSCC.