Molecular predictors of residual disease in patients with ovarian cancer

Abstract

Objectives: Gross residual disease (RD) following primary cytoreduction is the best predictor of overall survival in patients with high-grade serous ovarian cancer (HGSOC). Accurate identification of patients who will have RD has been elusive, prompting many to undergo unnecessary surgical exploration. Our goal was to identify and validate molecular markers associated with high rates of RD. Methods: We interrogated two publicly available genomic datasets of primary HGSOC for genes consistently differentially expressed in RD and no residual disease (R0) cohorts and significant at a false-discovery rate (FDR) of 10%. Genomic expression was further validated in an independent cohort (chemonaive ovarian tumor tissues) using quantitative reverse-transcriptase polymerase chain reaction followed by a blinded prediction of RD. A one-sided Fisher’s exact test was used to compare RD rates in predicted highand low-risk groups. Results: In the TCGA and Tothill datasets, 491 and 189 patients, respectively, met the criteria for inclusion. As expected, survival was significantly better for patients with R0 resection compared to those with any RD. We identified 47 probesets, representing 38 different genes, significant in both datasets at 10% FDR. These included probesets for FABP4 and ADH1B, which tracked tightly and showed dynamic ranges N16-fold. For these genes, there is a level of expression above which nearly all patients have RD; the distribution of expression values is roughly bimodal. In the validation cohort, using the top quartile of FABP4 polymerase chain reaction values as a prespecified cutoff, we found 30/35 RD cases in the high-expression group and 54/104 in the low-expression group (P= 0.0002). Our predictive method based on FABP4, therefore, correctly identified a cohort of patients with significantly increased rates of RD in an independent test set. Further examination of the ADH1B results showed that predictions using either ADH1B alone or in combination with FABP4 would have produced similar results. Examining the reasons for RD in the high-risk cohort suggests a preponderance of cases with either innumerable sites of disease or unresectable disease involving vital regions. Conclusions: High FABP4 and ADH1B expression are associated with significantly higher risk of RD in HGSOC patients. Patients with high tumoral FABP4 levels may be candidates for neoadjuvant chemotherapy.