Evolution of gastro-oesophageal cancers and selection of therapy resistant subclones through metastatic dissemination


Session type:


Matthew Davies1
1Institute of Cancer Research



Gastro-oesophageal adenocarcinomas (GOAs) are the third most common cause of cancer mortality worldwide. Many recent efforts to improve GOA survival through biomarkers or targeted drugs have been unsuccessful. We analysed tumour regions and metastases from 9 treatment naïve GOAs to assess intratumour heterogeneity (ITH) and its contribution to poor outcomes.


Samples were systematically collected from 4+ regions (centre, invasive margins, nodal metastasis) of surgically resected tumours and analysed by targeted deep sequencing and SNP arrays.


ITH was identified in all 9 tumours. A single microsatellite instable (MSI) tumour was identified in which 19/33 mutations and 11/23 likely driver mutations were heterogeneous while copy number aberration (CNA) heterogeneity was entirely absent. Four distinct mutations in the tumour suppressor gene ARID1A were found in the MSI case. Such parallel evolution indicates that subclonal expansion was driven by Darwinian selection. In contrast, CNA heterogeneity predominated in the remaining 8 tumours, which were chromosomally instable (CIN). Subclonal CNAs affected 17%-76% (median:51%) of the genomes of individual cancers. Mutational heterogeneity was lower with 22% of all somatic mutations and 29% of the likely driver mutations heterogeneous across the CIN cases. Importantly, heterogeneous amplifications of KRAS, NRAS, ERK2 and ERBB3 were identified in 4/8 CIN tumours, localized exclusively in nodal metastases.


Genomic instability processes influence evolution as shown by the predominance of mutational heterogeneity in MSI, and of CNA heterogeneity in CIN tumours. Evolutionary convergence through acquisition of MAP-Kinase pathway activating aberrations in metastases indicates that this pathway is important for metastatic dissemination and niche survival. These candidate metastasis drivers are predicted to confer resistance to upstream tyrosine kinase-targeting drugs. The evolution of metastases may hence foster the expansion of subclones pre-adapted to future targeted therapy. Pervasive ITH in GOA, characterized by parallel and convergent evolution, could be major contributors to biomarker and targeted therapy failures.