RESUMO
OBJECTIVES: Resistance to cancer therapy is an enduring challenge and accurate and reliable preclinical models are lacking. We interrogated this unmet need using high grade serous ovarian cancer (HGSC) as a disease model. METHODS: We created five in vitro and two in vivo platinum-resistant HGSC models and characterised the entire cell panel via whole genome sequencing, RNASeq and creation of intraperitoneal models. RESULTS: Mutational signature analysis indicated that platinum-resistant cell lines evolved from a pre-existing ancestral clone but a unifying mutational cause for drug resistance was not identified. However, cisplatin-resistant and carboplatin-resistant cells evolved recurrent changes in gene expression that significantly overlapped with independent samples obtained from multiple patients with relapsed HGSC. Gene Ontology Biological Pathways (GOBP) related to the tumour microenvironment, particularly the extracellular matrix, were repeatedly enriched in cisplatin-resistant cells, carboplatin-resistant cells and also in human resistant/refractory samples. The majority of significantly over-represented GOBP however, evolved uniquely in either cisplatin- or carboplatin-resistant cell lines resulting in diverse intraperitoneal behaviours that reflect different clinical manifestations of relapsed human HGSC. CONCLUSIONS: Our clinically relevant and usable models reveal a key role for non-genetic factors in the evolution of chemotherapy resistance. Biological pathways relevant to the extracellular matrix were repeatedly expressed by resistant cancer cells in multiple settings. This suggests that recurrent gene expression changes provide a fitness advantage during platinum therapy and also that cancer cell-intrinsic mechanisms influence the tumour microenvironment during the evolution of drug resistance. Candidate genes and pathways identified here could reveal therapeutic opportunities in platinum-resistant HGSC.