The tumor immune microenvironment plays an important role in head and neck squamous cell carcinoma (HNSCC). Reliable prognostic signatures able to accurately predict the immune landscape and survival rate of HNSCC patients are crucial to ensure an individualized/effective treatment. Here, we used HNSCC transcriptomic and clinical data retrieved from The Cancer Genome Atlas and identified differentially expressed immune-related long non-coding RNAs (DEirlncRNAs). DEirlncRNA pairs were recognized using univariate analysis. Cox and Lasso regression analyses were used to determine the association between DEirlncRNA pairs and the patients' overall survival and build the prediction model. Receiver operating characteristic curves and Kaplan-Meier survival curves were used to validate the prediction model. We then reevaluated the model based on the clinical factors, tumor-infiltrating immune cells, chemotherapeutic efficacy, and immunosuppression biomarkers. We built a risk score model based on 18 DEirlncRNA pairs, closely related to the overall survival of patients (hazard ratio: 1.376; 95% confidence interval: 1.302-1.453; P < 0.0001). Compared with two recently published lncRNA signatures, our DEirlncRNA pair signature had a higher area under the curve, indicating better prognostic performance. Additionally, the signature score positively correlated with aggressive HNSCC outcomes (low immunity score, significantly reduced CD8 + T cell infiltration, and low expression of immunosuppression biomarkers). However, high-risk patients might have high chemosensitivity. Overall, the lncRNAs signature established here shows promising clinical prediction and the effective disclosure of the tumor immune microenvironment in HNSCC patients; therefore, such signature might help distinguish patients that could benefit from immunotherapy.
Keywords: Head and neck squamous cell carcinoma; TCGA; long non-coding RNAs; prognosis; tumor-infiltrating immune cells.