Abstract 2353: A comparative meta-analysis of prognostic gene signatures for late-stage ovarian cancer

Abstract

Abstract Numerous gene signatures of patient prognosis for late-stage, high-grade serous ovarian cancer (HSOC) have been proposed, but diverse data and methods make these difficult to compare or prioritize for clinical application. However, the corresponding volume of publicly available expression data creates an opportunity to validate previous findings and to develop more robust signatures. To establish the current state of HSOC prognostic gene signatures, we undertook a systematic review and meta-analysis of the publicly available microarray data and published prognostic gene signatures. This effort generated a curated database of 1,319 clinically annotated HSOC microarray assays from 10 studies (Ganzfried et al., DATABASE 2013) and implementations of 14 published prognostic signatures for evaluation in independently available data. For each published prognostic signature we evaluated 1) its reproducibility, 2) its prognostic accuracy for overall survival in independent datasets, and 3) the prognostic value of the genes used in each signature relative to random gene sets. Twelve published models performed better than 97.5% of randomized risk scores, and six out-performed 97.5% of random signatures of the same size trained on the same data. The four top-ranked models achieved overall validation C-Indices of 0.56 to 0.60, and shared anti-correlation with expression of immune response pathways. Most models demonstrated lower accuracy in new datasets than in validation sets presented in their publication. This analysis provides definitive support for several prognostic models, but confirms that these require improvement to be of clinical value. This work addresses outstanding controversies in the ovarian cancer literature, including whether the prognostic signature proposed by The Cancer Genome Atlas Consortium (Nature 2011) improved on existing signatures, the impact of batch effects on a frequently re-used, high-profile dataset (Dressman et al., J Clin Oncol 2007), and the prognostic value of a recently proposed DNA damage repair signature (Kang et al., J Natl Cancer Inst 2012). Finally, this work provides a reproducible framework for meta-analytic evaluation of gene signatures. Citation Format: Levi Waldron, Markus Riester, Michael Birrer, Giovanni Parmigiani. A comparative meta-analysis of prognostic gene signatures for late-stage ovarian cancer. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 2353. doi:10.1158/1538-7445.AM2014-2353