Bioinformatics and machine learning reveal novel prognostic biomarkers in head and neck squamous cell carcinoma.

Abstract

Head and neck squamous cell carcinoma (HNSCC), the seventh most common cancer worldwide, has become more closely linked to poor lifestyle habits. Despite improvements in cancer treatment approaches, patients with stage I-II HNSCC have a 70-90% 5-year survival rate, and for patients with advanced stages III-IV, this rate falls to about 40%. This controversy is all about the heterogeneity of HNSCC. Finding diagnosis and prognosis biomarkers has the potential to make significant improvements in the life expectancy and overall health of these patients. The combination of bioinformatics and machine learning has facilitated the finding of the best markers for HNSCC. In this regard, RNA expression data were obtained to identify genes that were expressed differently (DEGs) and utilize a deep learning algorithm to identify genes that exhibited significant variability. In addition, correlations between clinical data and DEGs, the building of a Receiver Operating Characteristic (ROC) curve, and the prediction of tumor-infiltrating immune cells were analyzed. Deep learning analysis identified diagnostic and prognostic biomarkers strongly associated with carcinogenesis, such as KRT33B, KRTAP3-3, C14orf34, and ACADM. In addition, after analyzing the ROC curve, it was found that the combination of ACADM, KRT33B, and C14orf34 is the most practical combination of diagnostic markers. This combination achieved sensitivity, specificity, and Area Under the Curve (AUC) values of 0.92, 0.86, and 0.93, respectively.