Validation of a hypoxia related gene signature in multiple soft tissue sarcoma cohorts.

// Lingjian Yang 1 , Laura Forker 1 , Joely J. Irlam 1 , Nischalan Pillay 2, 3 , Ananya Choudhury 1 and Catharine M. L. West 1, 4 1 Translational Radiobiology Group, Division of Cancer Sciences, University of Manchester, Manchester Academic Health Science Centre, Christie Hospital, Manchester, UK 2 Cancer Institute, University College London, London, UK 3 Histopathology, Royal National Orthopaedic Hospital, Stanmore, UK 4 NIHR Manchester Biomedical Research Centre, Central Manchester University Hospitals NHS Foundation Trust, Manchester Academic Health Science Centre, Manchester, UK Correspondence to: Catharine M. L. West, email: catharine.west@manchester.ac.uk Keywords: soft tissue sarcoma; tumor hypoxia; gene expression signature; prognostic biomarker Received: October 06, 2017      Accepted: November 13, 2017      Published: December 12, 2017 ABSTRACT Purpose: There is a need for adjuvant/neo-adjuvant treatment strategies to prevent metastatic relapse in soft tissue sarcoma (STS). Tumor hypoxia is associated with a high-risk of metastasis and is potentially targetable. This study aimed to derive and validate a hypoxia mRNA signature for STS for future biomarker-driven trials of hypoxia targeted therapy. Materials and Methods: RNA sequencing was used to identify seed genes induced by hypoxia in seven STS cell lines. Primary tumors in a training cohort (French training) were clustered into two phenotypes by seed gene expression and a de novo hypoxia signature derived. Prognostic significance of the de novo signature was evaluated in the training and two independent validation (French validation and The Cancer Genome Atlas) cohorts. Results: 37 genes were up-regulated by hypoxia in all seven cell lines, and a 24-gene signature was derived. The high-hypoxia phenotype defined by the signature was enriched for well-established hypoxia genes reported in the literature. The signature was prognostic in univariable analysis, and in multivariable analysis in the training ( n = 183, HR 2.16, P = 0.0054) and two independent validation ( n = 127, HR 3.06, P = 0.0019; n = 258, HR 2.05, P = 0.0098) cohorts. Combining information from the de novo hypoxia signature and a genome instability signature significantly improved prognostication. Transcriptomic analyses showed high-hypoxia tumors had more genome instability and lower immune scores. Conclusions: A 24-gene STS-specific hypoxia signature may be useful for prognostication and identifying patients for hypoxia-targeted therapy in clinical trials.