A 6-gene signature for loco-regional control prognosis in HNSCC patients treated by PORT-C

Purpose: The aim of this study was to identify and validate a gene signature combining machine learning approaches and biological information in order to predict loco-regional control (LRC) in patients with HPV-negative, locally advanced HNSCC who received postoperative radio(chemo)therapy (PORT(-C)). Materials and methods: Gene expression analysis was performed using the GeneChip Human Transcriptome Array 2.0 on a multicentre retrospective training cohort of 128 patients and an independent validation cohort of 114 patients of the German Cancer Consortium Radiation Oncology Group (DKTK-ROG) treated with PORT(-C) (figure A). Genes were filtered based on differential gene expression analysis and Cox regression. The identified gene signature was combined with clinical features and with previously identified genes related to cancer stem cells [1-2] and hypoxia [3]. Model performance was evaluated by the concordance index (ci) and Kaplan-Meier analyses. Results: We identified a 6-gene signature consisting of four individual genes CAV1, GPX8, IGLV3-25, TGFBI and one metagene combining the highly correlated genes INHBA and SERPINE1. A multivariable Cox model combining the 6-gene classifier and clinical parameters was fit to the training data (ci=0.81) and was successfully validated (ci=0.66). It stratified patients into two risk groups that significantly differed in the primary endpoint LRC in training (p