Abstract P2-05-34: Risk stratification with EndoPredict signature for luminal subtype breast cancers: Re-analyzing microarray experiments with Han Chinese origin

Abstract

Abstract Introduction: Breast cancer is a heterogeneous disease in terms of molecular aberrations. Luminal breast cancers, most of which are estrogen receptor (ER) positive without human epidermal growth factor receptor 2 (HER2) over expression clinically, constitute the majority of human breast cancers with better prognosis compared with basal-like or HER2-enriched subtype. The aim of the study is to evaluate the prognostication of EndoPredict signature, for which high- and low-risk group is defined based on a multi-gene assay. EndoPredict signature is supposed to guide adjuvant therapy for ER+/HER2- luminal breast cancers with up to three positive lymph nodes, while the test has not been validated for Han Chinese population yet. Materials and methods: Our microarray experiments (partially published under GSE48391) and two publicly available microarray studies (GSE5460 and GSE20685) constituted the combined dataset of 565 breast cancers with Han Chinese origin, of which 280 were ER+/HER2- by immunohistochemical analysis. Transformation of Affymetrix microarray gene expression values to RT-qPCR-based expression values were conducted with the mathematical formula provided by the EndoPredict investigators, with gene-specific transformation factor and offset. Each enrolled patient was categorized into either high- or low-risk group based on the result of EndoPredict (EP) algorithm. Results: Direct adaptation of the EP algorithm for microarray gene expression data resulted in over inflation of EP scores, and most cases were categorized into the high-risk group with the predefined threshold of EP score of 5 and adjustments with rescaling and relocations of microarray-based EP scores were performed. The proportion was 88% for low-risk group and 74% for high-risk group during the 10-year follow up period, with disease-specific survival advantage reported for those with EP-predicted low-risk group patients. On the other hand, borderline overall survival discrepancy was observed (89% for low-risk and 80% for high-risk group). In addition, patients with high-risk EP scores were associated with larger tumor size, higher nuclear grade, and more involved lymph nodes. Risk stratificaiton of Taiwanese breast cancers by EP algorithm: disease-specific survivalRisk groupTotal numberFailed numberCensored numberPercentage of disease-specific survival censored patientsLow-risk6485687.5%High-risk1564011674.4%P=0.01, log-rank test Risk stratificaiton of Taiwanese breast cancers by EP algorithm: overall survivalRisk groupTotal numberFailed numberCensored numberPercentage of overall survival censored patientsLow-risk6475789%High-risk1563112580.1%P=0.05, log-rank test Discussion and conclusion: The study provides a solution to enhance the comparability between the FFPE/RT-qPCR based EP algorithms and fresh frozen microarray gene expression data. The statistical framework presented here provides an “in-silicon” validation for EP algorithm and further studies taking clinical parameters into consideration will augment the clinical applicability of EP scores and EPclin scores to ascertain the prognostic power of multi-gene assay for luminal breast cancers in Taiwan. Citation Format: Huang C-C, Tsai M-L, Tu S-H, Huang C-S. Risk stratification with EndoPredict signature for luminal subtype breast cancers: Re-analyzing microarray experiments with Han Chinese origin [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 P2-05-34.