Abstract 5287: RNA DNA divergences: An unsuspected marker of cancer genomic instability accurately predicts triple-negative breast cancer severity

Abstract

Abstract Next generation sequencing provides 3 measures of cancer genomic instability i.e. somatic DNA variations, differential gene expression and RNA DNA divergences (RDD). The latter reflects changes in RNA sequences that are not present at the DNA level. Triple negative breast cancers (TNBC) represent the most severe form of the disease and are currently not amenable to targeted therapies nor to prognosis testing. We used the 3 measures of genomic instability to construct specific optimal algorithms that effectively separated 20 TNBC patients with poor or good clinical outcomes: 11 patients died from the disease within 1000 days following diagnosis of non metastatic TNBC while 9 were alive after 2500 days of follow up. All 3 models efficiently separated these 2 clinically polarized groups. However, only RDD based algorithms and not those relying on somatic mutations and expression profiles retains performances in excess of 90% accuracy after statistical cross validation. The 3 models were then applied in blind to 45 unknown patients with the same inclusion criteria i.e. non metastatic TNBC diagnosed before the age of 65 irrespective of menopausal, lymph node, tumor size and ethnic origin or recruitment centers. Kaplan-Meier analysis showed that the RDD based algorithm was highly predictive of clinical outcome i.e. 100% of patients predicted with good outcome were alive while 80% of patients predicted with poor outcome died in the same time interval (p<10-5). Algorithm based on somatic mutations and expression failed to predict clinical outcome in Kaplan-Meier analysis of validation set (p = NS). We then applied RDD predictive algorithm to TNBC cell lines and found that cells with RDD rate leading to poor clinical outcome prediction were statistically significantly associated with higher invasion and migration capacity (p<0.001). Inactivation of the expression of the most important transcript contributing to the RDD model reversed the in vitro phenotype. Comparison of RDD events occurring in good an poor outcome patients showed statistically significant differences in affected and replacement bases between the 2 groups. Therefore, RDD originate in non-random defects in the transfer of information between DNA and RNA that strongly contribute to tumor severity, further identification of specific transcripts differentially affected by RDD rate between poor and good clinical outcome allows development of novel strategies for selective therapeutic intervention. Citation Format: Bernard E. Bihain, Stéphane Verdun, Julie Tomasina, Benoit Hilselberger, Marie Brulliard, Lionel Bonnard, Marina Trarbach, Olivier Roitel, Sandrine Jacquenet, Virginie Ogier, Jean-Pierre Armand, Benoit Thouvenot. RNA DNA divergences: An unsuspected marker of cancer genomic instability accurately predicts triple-negative breast cancer severity. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 5287. doi:10.1158/1538-7445.AM2015-5287