Abstract 2249: Integrated multi-level omics-based biomarker models for ovarian cancer survival

Abstract

Abstract Background: Although a number of prognostic biomarkers for ovarian cancer have been proposed, the pool of clinically useful biomarkers remains small. While the predictive power of individual biomarkers may be weak, the advent of more affordable high-throughput technologies and readily available “omics” level data point toward the possibility of high performance biomarker panels. In this study, we used tumor-based somatic mutation, mRNA and miRNA expression, and DNA methylation data from The Cancer Genome Atlas (TCGA) to construct integrated biomarker models for overall and progression-free survival. Methods: The somatic mutations, mRNA expression, and DNA methylation of 451 candidate genes, as well as the expression of the miRNAs targeting them, were individually analyzed for association with overall and progression-free survival in 488 TCGA serous cystadenocarcinoma patients of predominantly Caucasian descent. Following correction for multiple comparisons, molecular variations associated with survival were combined to construct integrated cross-validated prediction models for overall and progression-free survival. The discriminative power of these models was then evaluated for 1-, 3-, and 5-year overall and progression-free survival using ROC analysis. Results: Our cross-validated integrated molecular models were demonstrated to predict 5-year overall survival in the total patient pool with a ROC AUC of 0.801 and 5-year overall and progression-free survival in Caucasian patients with ROC AUCs of 0.866 and 0.809, respectively. Conclusions: Our findings suggest the potential utility of our multi-omics-based biomarker models in helping to inform clinical decisions following ovarian cancer diagnosis. Although extensive efforts were made to maximize the predictive potential of our models, further retrospective and prospective validation efforts are necessary before their clinical utility can be accurately assessed. Citation Format: Alan Fu, Shen-Chih Chang, Aileen Baecker, Helena R. Chang, Zuo-Feng Zhang. Integrated multi-level omics-based biomarker models for ovarian cancer survival [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 2249. doi:10.1158/1538-7445.AM2017-2249