Ultra-deep targeted sequencing of advanced oral squamous cell carcinoma identifies a mutation-based prognostic gene signature

// Shu-Jen Chen 1, 2, 3, * , Hsuan Liu 2, 3, * , Chun-Ta Liao 4, * , Po-Jung Huang 5 , Yi Huang 2 , An Hsu 3 , Petrus Tang 5 , Yu-Sun Chang 2 , Hua-Chien Chen 1, 2, 3 , Tzu-Chen Yen 6 1 Department of Biomedical Sciences, Chang Gung University, Taoyuan, 33302, Taiwan 2 Genomic Core Laboratory, Chang Gung University, Taoyuan, 33302, Taiwan 3 Graduate Institute of Biomedical Sciences, Chang Gung University, Taoyuan, 33302, Taiwan 4 Department of Otorhinolaryngology, Chang Gung Memorial Hospital, Taoyuan, 33305, Taiwan 5 Bioinformatics Core Laboratory, Chang Gung University, Taoyuan, 33305, Taiwan 6 Department of Nuclear Medicine, Chang Gung Memorial Hospital, Taoyuan, 33305, Taiwan * These authors have contributed equally to this work Correspondence to: Tzu-Chen Yen, e-mail: yen1110@adm.cgmh.org.tw Hua-Chien Chen, e-mail: hcchen@actgenomics.com Keywords: next-generation sequencing, oral squamous cell carcinoma, mutation, prognosis Received: February 08, 2015      Accepted: April 13, 2015      Published: April 25, 2015 ABSTRACT Background: Patients with advanced oral squamous cell carcinoma (OSCC) have heterogeneous outcomes that limit the implementation of tailored treatment options. Genetic markers for improved prognostic stratification are eagerly awaited. Methods: Herein, next-generation sequencing (NGS) was performed in 345 formalin-fixed paraffin-embedded (FFPE) samples obtained from advanced OSCC patients. Genetic mutations on the hotspot regions of 45 cancer-related genes were detected using an ultra-deep (>1000×) sequencing approach. Kaplan-Meier plots and Cox regression analyses were used to investigate the associations between the mutation status and disease-free survival (DFS). Results: We identified 1269 non-synonymous mutations in 276 OSCC samples. TP53 , PIK3CA , CDKN2A , HRAS and BRAF were the most frequently mutated genes. Mutations in 14 genes were found to predict DFS. A mutation-based signature affecting ten genes ( HRAS, BRAF, FGFR3, SMAD4, KIT, PTEN, NOTCH1, AKT1, CTNNB1 , and PTPN11 ) was devised to predict DFS. Two different resampling methods were used to validate the prognostic value of the identified gene signature. Multivariate analysis demonstrated that presence of a mutated gene signature was an independent predictor of poorer DFS ( P = 0.005). Conclusions: Genetic variants identified by NGS technology in FFPE samples are clinically useful to predict prognosis in advanced OSCC patients.