Gene Expression and Prediction of Complete Cytoreduction in Ovarian Cancer

Abstract

INTRODUCTION: The objective of this study was to investigate the role of microarray gene expression analysis in predicting the feasibility of complete cytoreduction in ovarian cancer. METHODS: Microarray expression of 507 women with ovarian cancer from The Cancer Genome Atlas was initially analyzed and a gene expression signature created. Complete cytoreduction was defined as removal of all tumors with no macroscopic residual disease. The model was applied to the Australian Ovarian Cancer Study data set with 216 ovarian cancer samples. RESULTS: One hundred fifteen women had complete cytoreduction in The Cancer Genome Atlas data set with significantly higher 5-year overall survival as compared with the 392 women with macroscopic residual disease (P<.001). A total of 12,718 genes were analyzed; 142 predicted accuracy of complete cytoreduction 69% of the time (P=.02). The model was more precise in predicting incomplete cytoreduction with macroscopic residual tumor with a sensitivity of 82% and a positive predictive value of 78%. The performance, measured by the area under the receiver operating characteristic curve was 63%. When applied to the Australian Ovarian Cancer Study set, with 58 women with complete cytoreduction and 158 with macroscopic residual, the accuracy was 72% (P<.019) and it was still more sensitive predicting incomplete cytoreduction with a sensitivity of 83% and a positive predictive value of 79%, and the area under the receiver operating characteristic curve was 68%. CONCLUSIONS: This analysis stresses again on the importance of complete cytoreductive surgery in ovarian cancer to microscopic disease and shows that gene expression analysis might have a role in predicting patients who are not amenable to complete cytoreduction and thus would have shorter survival.