Abstract P1-07-01: A pan-cancer perspective of functional proteomics provides novel information content for uncommon breast cancer subtypes

Abstract

Abstract Cancer cell lines (CCLs) serve as models to study the functional consequences of the genomic lesions in patients and as screening platforms for prediction of drug response. While genomic and transcriptomic data have proven to be useful predictors, the ability of these omics platforms to predict protein level and function is limited. Furthermore, since proteins are the targets of the majority of the targeted therapies, protein levels and importantly protein function would be expected to provide more powerful predictions than DNA or RNA data. While large scale genomic and transcriptomic data linked to drug sensitivity are available for over a thousand CCLs, proteomic data is available for only a small subset of lines. Here we performed proteomic profiling of 736 cell lines using reverse-phase protein arrays (RPPAs) with approximately 300 antibodies providing an unbiased sparse representation of the majority of signaling pathways. The functional proteomic analysis revealed 10 protein-based clusters across all cell lines. Similar to human tumors, the breast cell lines fell into three major clusters representing basal-like, luminal/Her2-amplified and claudin-low breast cancer subtypes. The basal-like and claudin-low clusters contained all of the representative breast cancer cell lines as well as a much larger number of other CCLs. For example, the 6 claudin-low breast cancers analyzed reside in an EMT cluster, in which only 8/126 are breast cell lines. However, the complete cluster including multiple non-breast cancer cell lines recapitulated mRNA and protein features of claudin-low breast tumors, including a high EMT signature and low level of hormone receptor pathway activity. We thus explored whether we could gain power for linking the limited number of basal and claudin-low breast cancer cell lines to therapeutic sensitivity by assessing patterns of drug sensitivity in each cluster for both the breast and non-breast cancer cell lines in the cluster. We explored drug sensitivity of 481 therapeutic compounds from the Cancer Therapeutic Response Portal (CTRP v2) and demonstrated that the non- breast cancer and breast cancer cell lines in each cluster provided similar patterns of drug sensitivity. For example, Claudin-low/EMT cell lines of both breast cancer and non-breast cancer origin showed decreased sensitivity to PI3K/mTOR inhibitors compared to luminal breast cancers (p<0.05 for 4 mTOR inhibitors) and drugs targeting EGFR family compared to basal cell lines (p<0.05 for 7 EGFR/ERBB2 inhibitors). Thus it is possible to gain information by characterizing cell lines with similar patterns of protein expression and provide important information related to drug sensitivity of uncommon breast cancer lineages. The functional proteomic analysis provides a wealth of information that complements the genomic and transciptomic studies of cancer cell lines, and demonstrates the opportunity to leverage cell line 'pan-cancer' proteomic patterns to improve characterization of specific breast cancer subtypes. To facilitate broad access to these data, we developed a user-friendly data portal, the MD Anderson Cell Lines Project (MCLP), that provides both data analysis and download (http://ibl.mdanderson.org/mclp/). Citation Format: Zhao W, Li J, Lu Y, Akbani R, Liang H, Mills GB. A pan-cancer perspective of functional proteomics provides novel information content for uncommon breast cancer subtypes [abstract]. In: Proceedings of the 2016 San Antonio Breast Cancer Symposium; 2016 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2017;77(4 Suppl):Abstract nr P1-07-01.