Abstract 5467: Outlier analysis to identify determinants of therapeutic resistance in breast cancer

Abstract

Listen

Translate article

Abstract Tumors are typically characterized by variable responses to targeted therapy driven by the heterogeneity of drivers among tumors. Inter-tumor heterogeneity also results in different resistance determinants to the same therapeutic agent. Outlier analysis has been applied extensively to identify drivers of cancer when comparing a heterogenous population of tumor samples to a relatively homogenous population of normal samples. However, the application of outlier analysis has rarely been applied to identify intrinsic determinants of resistance to targeted cancer therapies. We recently applied outlier analysis to baseline proteomics data collected from HER2/ERBB2 positive breast cancer patients treated with anti-HER2 therapy, comparing protein levels in each patient that did not show pathological complete response (pCR) to the population of patients that did. Three of the five non-pCR patients were outliers for low expression of the ERBB2 protein relative to the pCR patients, suggesting that lack of protein expression of the direct therapeutic target drives resistance in some cases whereas other mechanisms contribute to resistance in other cases. Here, we systematically evaluate the ability of multiple outlier analysis methods, including the method we implemented for the HER2 study (outlieR), to identify molecular genes associated with poor prognosis in estrogen receptor (ESR1) positive patients receiving hormone therapy (targeting ESR1) using a large-scale publicly available dataset which contains baseline gene expression and survival data for ~2000 breast cancer patients (METABRIC). Specifically, we compared expression data from tumors of 72 patients that showed poor prognosis (died of disease within 5 years) to data from 226 tumors from patients with good prognosis (did not die of disease, with at least five years of follow-up time). For this evaluation, we focused on the direct target of hormone therapy, ESR1, and growth factor receptors associated with endocrine therapy resistance. Scores from outlier analysis methods consistently ranked ESR1, EGFR, and genes associated with the ERBB2 locus (amplification of the locus drives ERBB2 gene expression in HER2 positive breast cancer) more highly than the the T-test metric did, with the outlieR method outperforming another established method, Outlier Sums (OS), in most cases. Furthermore, four targets of FDA-approved drugs, including two genes in the MAPK pathway, were amongst the top 200 genes identified by the outlieR method, whereas TOP2A was the only approved target in the top 200 genes identified by the T-test, a target of cytotoxic chemotherapy treatments. Finally, the outlieR method generates outlier scores for each gene in each non-responder relative to the set of non-responders, allowing for the evaluation of genes associated with resistance on a patient-by-patient basis. These observations suggest that outlier analysis can be used to prioritize molecular features as potential mechanisms of resistance and alternative drug targets by accounting for heterogeneity between resistant tumors. Citation Format: Eric James Jaehnig, Meenakshi Anurag, Jonathan T. Lei, Bing Zhang. Outlier analysis to identify determinants of therapeutic resistance in breast cancer [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 5467.