The discovery and evaluation of treatment stratification biomarkers in epithelial ovarian cancer

Abstract

Background: Epithelial ovarian cancer (EOC) has the poorest survival outcomes for all gynaecological malignancies. Despite emerging knowledge on the heterogeneity of the disease the majority of patients are initially treated the same. Biomarkers which could predict treatment response would greatly enhance patient care. Results: Differential DNA methylation analysis of serous EOC tumours identified 180 loci significantly associated with residual disease status (n=297, p<0.05, FDR <5%), with 23 corresponding genes also significantly associated in gene expression array analysis (p<0.05, FDR <10%). Differential DNA methylation of 27 loci were significantly associated with overall survival in patients receiving optimal debulking (n=78, p<0.05). Patients optimally debulked but with poor prognosis markers were found to have the same survival as those suboptimally debulked suggesting that supraradical surgery is not beneficial in these patients. Hypomethylation at intragenic regions of the homeobox gene MSX1 was associated with primary platinum resistance significantly (n=61, p<0.05, FDR<5%) and these findings were validated in an independent dataset (n=252, p<0.05). DNA methylation was also significantly correlated to gene expression. Platinum resistant ovarian cancer cell lines demonstrated significantly lower gene expression of MSX1 compared to sensitive pairs. Proliferation and apoptosis cell-line assays demonstrated sensitisation to cisplatin when cisplatin resistant A2780/CP70 cells re-expressed MSX1 following gene transfection. An increase in p53 downstream transcripts, CDKN1A (p21) and BAX was also demonstrated in these MSX1 transfectants. The detection of disease at 5 distinct anatomical sites determined by preoperative computed tomography was significantly associated with surgical debulking outcomes in a test (n=111) and validation (n=70) cohort of patients with EOC (sensitivity 64.7-69.2%, specificity 67.9-71.4%, AUC 0.721-0.749). Conclusions: DNA methylation is a potential rich source of biomarkers predicting cytoreductive outcome and survival. The discovery and validation of a novel DNA methylation biomarker of chemotherapy resistance is demonstrated with exciting findings related to its biological function in cisplatin sensitive assays.