Whole-exome sequencing identifies biosignatures that predict adverse survival outcomes in surgically treated patients with oral cavity squamous cell carcinoma.

<bold>Objectives: </bold>The postoperative outcomes of patients with oral cavity squamous cell carcinoma (OCSCC) vary greatly. To improve risk stratification, we sought to identify genetic biosignatures by whole-exome sequencing (WES).<bold>Materials and Methods: </bold>We retrieved patients with OCSCC patients with paired freshly frozen malignant and non-malignant tissue specimens and performed WES by Illumina HiSeq4000 platform. We further applied a tree-based method to analyze copy number variations and obtain signature classification and driver-gene identification. We further confirmed the prognostic impact of the WES biosignature in an external independent validation set.<bold>Results: </bold>We examined 168 paired samples from patients with surgically treated OCSCC. Similar to the literature, the most commonly mutated genes were TP53 (66%), FAT1 (32%), and NOTCH1 (24%). The signatures 13 (APOBEC Cytidine deaminase [C > G]), 1 (spontaneous deamination of 5-methylcytosine), and 7 (UV exposure) showed the highest concordance rates. Using the MutSigCV, MuSiC, 20/20+, OncodriveFML, e-Driver, OncodriveCLUST, and tree-based methods, we identified a nine-gene OCSCC panel (RYR1, HLA-B, TSHZ2, PCDH17, DNAH17, GRID1, SBNO2, KSR2, and GCN1L1) predicting survival outcomes in our sample. We used the TCGA database to validate the prognostic value of the panel independently. Furthermore, gene-gene covariance analysis confirmed the coexistence of several gene alterations.<bold>Conclusion: </bold>We identified and independently validated a WES biosignature that predicts outcomes in surgically treated OCSCC in Taiwan, a betel-quid-chewing-prevent area. We proposed that the panel might help clinical trial designation for adjuvant therapy based on the risk stratification from the novel gene panel and identify targets for liquid biopsy monitoring during surveillance. [ABSTRACT FROM AUTHOR]