Abstract AS02: Network-based stratification: combining genome-scale somatic mutation profiles with genetic interaction networks to identify clinically relevant subtypes in high grade serous ovarian carcinoma

Abstract

Abstract Background: Like many solid tumors, high grade serous ovarian carcinoma (HGS-OvCa) is a heterogeneous entity with a widely variable clinical course even among patients of the same stage and histological subtype. We have recently developed a method called Network-based Stratification (NBS), which combines genome-scale somatic mutation profiles with genetic interaction networks and performs unsupervised clustering of patients into subtypes. Remarkably, patients in these subtypes display distinct biological and clinical characteristics. Methods: Briefly, somatic mutations for each patient are represented as a profile of binary (1,0) states on genes, in which a ‘1’ indicates a mutated gene. For each patient we project mutation profiles onto a human gene interaction network, then apply network propagation to spread the influence of each subsampled mutation profile over its network neighborhood. These ‘network-smoothed’ profiles are then clustered into a predefined number of subtypes using the unsupervised technique of non-negative matrix factorization (NMF). Results: Using the NBS technique four distinct subtypes were identified from the TCGA HGS-OvCa cohort of 330 exome sequenced and clinically-annotated tumors. The median overall survival (OS) in months for these four subtypes was 36, 48, 56, and not-yet-reached respectively, which was highly significant (Logrank p = 1.59 x 10-5). No significant differences between subtypes were found in age, tumor stage at presentation, or percentage of patients with an optimal surgical resection. Furthermore, the survival difference is significant when comparing a Cox proportional hazards model of NBS subtypes together with the above factors and mutation rate is compared to a model of these factors alone (Likelihood ratio test p = 3.3x10-4). Each subtype exhibits a distinct set frequently mutated pathways, with subtype 1 (which had the worst OS) enriched for mutation in the fibroblast growth factor receptor, caspase and nucleoskeleton pathways, subtype 2 enriched for ‘classic’ HGS-OvCa mutations including mutation in DNA damage response genes. Recognizing that full exome sequencing is not yet standard of care, we have built a limited classifier that only requires as input the mutational status of 50 genes. This 50 gene classifier performed with 94% accuracy and 80% precision relative to full exome. Using this 50 gene classifier we have stratified an independent cohort of 93 Australian HGS-OvCa patients from the ICGC and found survival relationships similar to what was seen in the TCGA cohort (median OS of 22 months for subtype 1 vs. not-yet-reached for other subtypes, Logrank p = 3.6x10-4). The 50 gene NBS classifier was also applied to a panel of cell lines from The Genomics of Drug Sensitivity in Cancer Project. Cell lines classified as subtype 1 had significantly higher cisplatin IC50 relative to other subtypes, whereas there was no difference in docetaxel or paclitaxel IC50. Citation Format: John Paul Shen, Matan Hofree, Ana Bojorquez-Gomez, Cheryl Saenz, Trey Ideker. Network-based stratification: combining genome-scale somatic mutation profiles with genetic interaction networks to identify clinically relevant subtypes in high grade serous ovarian carcinoma [abstract]. In: Proceedings of the 10th Biennial Ovarian Cancer Research Symposium; Sep 8-9, 2014; Seattle, WA. Philadelphia (PA): AACR; Clin Cancer Res 2015;21(16 Suppl):Abstract nr AS02.