Abstract 2355: Risk prediction for late-stage ovarian cancer by meta-analysis of 1,525 patient samples

Background: Ovarian cancer causes over 15,000 deaths per year in the United States. The survival of patients is quite heterogeneous, and accurate prognostic tools would help with the clinical management of these patients. However, existing ovarian cancer gene signatures did not make it to the clinic so far. Expression data of thousands of patients from a large number of different cohorts is now available in the public domain. Harvesting all this information for building signatures is a challenging “big data” problem, but has the potential to yield more robust and accurate signatures. Methods: We developed and validated two gene expression signatures of potential clinical relevance in advanced stage serous ovarian cancer, the first for predicting survival and the second for predicting the outcome of initial debulking surgery. We integrated 13 publicly available datasets totaling 1,525 subjects. We trained prediction models using a meta-analysis variation on the Compound Covariate method, tested models via a “leave-one-dataset-out” procedure, and validated models in additional independent datasets. Selected genes from the debulking signature were validated by immunohistochemistry and qRT-PCR in two further independent cohorts of 179 and 78 patients, respectively. Results: The survival signature stratified patients into high- and low-risk groups (HR=2.19; 95% CI, 1.84 to 2.61) significantly better than the best previously published overall survival signature (P = 0.039). POSTN, CXCL14, FAP, NUAK1, PTCH1, and TGFBR2) were validated by qRT-PCR (P < 0.05) and POSTN, CXCL14 and phosphorylated Smad2/3 by immunohistochemistry (P < 0.001) as independent predictors of debulking status. The sum of IHC intensities for these three proteins provided a tool that classified 92.8% of samples correctly in high- and low-risk groups for suboptimal debulking (AUC 0.89; 95% CI 0.84 to 0.93). We present the strongest evidence to date for the existence of a biologic basis of suboptimal outcome of debulking surgery. Conclusions: Our survival signature provides the most accurate and validated prognostic model for early and advanced stage high-grade serous ovarian cancer. The debulking signature accurately (92.8%) predicts the outcome of cytoreductive surgery and will have clinical utility if the accuracy of the immunohistochemistry tool observed in our initial 179-patient validation cohort is confirmed in prospective validation. This tool will potentially allow for the identificatio