Abstract A36: Network-based approach elucidates critical genes in BRCA subtypes

Intratumor heterogeneity is a common characteristic across cancer types and poses a major therapeutic challenge. Breast cancer is no exception. In 2020 alone, more than 2.3 million women were diagnosed with breast cancer, resulting in 685,000 deaths globally. Breast cancer is broadly classified into four subtypes, viz. Basal, Her2, Luminal-A & Luminal-B and substantial transcriptional heterogeneity has been observed, both across subtypes, as well as within each subtype. However, the observed heterogeneity could be partly due to stochastic expression variability and the extent to which the transcriptional heterogeneity reflects phenotypic variability is unclear. A key insight exploited by our recent network-based approach, PathExt (and others), is that the global transcriptional changes in a disease context is mediated by a small number of key genes and identifying such functional mediators of the global transcription is more likely to provide functional insights into the disease process and better reflect functional heterogeneity. Here we apply PathExt to 1059 BRCA tumors and 112 healthy control samples across 4 subtypes to identify frequent key mediator genes in each subtype. Relative to conventional differential expression approach, PathExt-identified genes exhibit greater commonality across tumors, revealing key biological processes such as cell cycle, response to hormone, regulation of apoptotic processes, etc. Subtype specific biological processes were also observed, for instance, regulation of DNA binding in Basal, positive regulation of inflammatory response in Her2, positive regulation of protein phosphorylation in Luminal-A and second messenger-mediated signaling in Luminal B. Importantly, PathExt-identified genes exhibit much greater dependency scores in subtype-specific cancer cell lines. Compared to the differential expression approach as well as a previous systems approach, PathExt better recapitulates potential driver genes in multiple benchmarks. For instance, PathExt recapitulates many of the significantly mutated protein interaction modules from the NEST database, such as Cell Cycle, Immune systems, Ribonucleo-protein complexes, and Regulation of Transcription. PathExt-identified genes also show a significant overlap with BRCA driver genes in DriverDBv3 and IndoGen databases. Furthermore, PathExt recapitulates the gene sets previously associated with BRCA treatment response, such as immune response, proliferation, and DNA repair deficiency. Lastly, analysis of subtype-specific BRCA scRNA-seq data reveals that while a majority of PathExt-identified genes are highly expressed in malignant cells, many other genes are predominantly expressed in immune and stromal cells, and the distribution of key genes across cell types varies substantially across the four BRCA subtypes. Overall, application of PathExt to BRCA cohort tempers and refines previous view of inter-tumor heterogeneity, and identifies potential functional mediators of BRCA subtypes, paving way for further research.

Citation Format: Piyush Agrawal, Navami Jain, Vishaka Gopalan, Sridhar Hannenhalli. Network-based approach elucidates critical genes in BRCA subtypes [abstract]. In: Proceedings of the AACR Special Conference: Tumor Immunology and Immunotherapy; 2022 Oct 21-24; Boston, MA. Philadelphia (PA): AACR; Cancer Immunol Res 2022;10(12 Suppl):Abstract nr A36.