The incidence of Oesophageal Adenocarcinoma (OAC) has risen 6-fold in the western world in the last forty years but survival is poor. Increased molecular understanding of this heterogeneous disease is needed to improve treatment selection and develop novel therapies. This study uses gene expression data to perform unbiased molecular subtyping and identify prognostic subgroups in OAC.
Transcriptional profiling of 274 treatment naïve OAC biopsies was performed using the Almac Diagnostics Xcel array™. All patients received platinum-based neo-adjuvant chemotherapy prior to surgical resection at four United Kingdom centres between 2004-2012. Iterative semi-supervised clustering based on gene expression level variability was performed followed by functional enrichment using DAVID. Cluster membership was assessed for independence of known prognostic factors using Cox proportional hazards regression for relapse-free (RFS) and overall survival (OS). Clustering was repeated with a published 51-gene hypoxia signature with validation in the TGCA OAC (n = 65) and oesophageal squamous cell carcinoma (n = 45) cohorts.
Patients were clustered into two groups with significantly different RFS (HR = 0.54, p = 0.05) and OS (HR = 0.52, p = 0.04). There were no significant differences in known prognostic factors such as pathological response, lymphovascular invasion and resection margin. Pathway analysis revealed the PI3K-AKT, p53, Tumour Necrosis Factor and Hypoxia Inducible Factor 1 (HIF-1) signalling pathway to be upregulated in the poor prognostic group. To further investigate the role of the HIF-1 pathway, a hypoxia 51-gene signature was applied. Patients were stratified into hypoxia low and high groups with improved RFS (HR 0.64, 95% CI 0.42-0.97; p = 0.04) and OS (HR 0.67, 95% CI 0.44-1.02; p = 0.06) in the hypoxia-low group. Increased OS for the hypoxia-low group was also observed in the TCGA cohort (HR 0.49, 95% CI 0.24-0.97; p = 0.04). There was a significant association between membership of the poor prognostic and hypoxia-low cluster groups (p < 0.001).
Molecular stratification and application of a hypoxia gene signature identifies a poor prognostic group of OAC patients characterised by upregulation of hypoxia signalling.
Clinical trial identification
Legal entity responsible for the study
Queen\'s University Belfast.
Northern Ireland Health and Social Care Research and Development Division.
R.D. Kennedy: Global VP of Biomarker Development for Almac Diagnostics and have patent declarations. All other authors have declared no conflicts of interest.