Abstract Background: The expression of ER, PR and HER2 in breast cancer is determined by routine pathology assessment to indicate systemic treatment. Here, we investigated the ability of a standardized mRNA-based assay to predict ER, PR and HER2 status and treatment response. Methods: ESR1, PGR and ERBB2 expression was analyzed from the standardized nCounter-based PAM50 assay in 1,544 FFPE breast tumors obtained from 13 independent studies (NeoEribulin, GEICAM 2012-09, TBCRC023/006, LPT109096, EGF117165, PAMELA, PerELISA, GEICAM 2003-11, VENTANA, IBIMA and HCB). All immunohistochemical and in situ hybridization analyses followed ASCO/CAP criteria and were performed in central labs except for Neoeribulin and VENTANA studies. To explore the best cutoff for each gene, we used Monte Carlo cross validation (repeated 1000 times, 2/3 training, 1/3 testing) to achieve the highest kappa values when gene expression was compared to pathology assessment. Receiver operating characteristic analysis and area under the ROC curve (AUC) was used to evaluate the performance of each gene to predict ER, PR and HER2 status. Finally, the association of each gene (using the pre-established cutoffs) with pathological complete response (pCR) was evaluated using univariate logistic regression analyses in 2 neoadjuvant cohorts: 1) A combined HER2+ cohort using 191 tumor samples from PAMELA/PerELISA phase II trials, where patients received neoadjuvant dual HER2 blockade without chemotherapy for 15-18 weeks and 2) a combined cohort of hormone receptor-positive/HER2-negative using 205 tumor samples from IBIMA/HCB consecutive series, where patients received anthracycline/taxane-based chemotherapy. Results: Concordance between ESR1 and ER was 95.3% (95% confidence interval [95CI] 94.0-96.4%; AUC=0.98). ER+ and ER- cases were classified as ESR1- and ESR1+ in 5.5% and 4.8% of the cases, respectively. Concordance between PR and PGR was 85.2% (95CI 83.1-87.1%; AUC=0.95), and between ERBB2 and HER2 was 92.8% (95CI 91.4-94.1%; AUC=0.95). HER2+ and HER2- cases were classified ERBB2- and ERBB2+ in 16.7% and 2.9% of the cases, respectively. In the neoadjuvant HER2+ cohort, the pCR rates were 44.6% in ESR1- and 18.8% in ESR1+ (odds ratio [OR] 3.9; 95CI 1.9-8.4; p<0.001), 38.5 in PGR- and 15.8% in PGR+ (OR 3.2; 95CI 1.6-6.7; p=0.001), 2.6% in ERBB2- and 35.3% in ERBB2+ (OR 19.7; 95CI 2.6-149.1; p<0.001). In the neoadjuvant HR+/HER2- cohort, the pCR rates were 40% in ESR1- and 5.12% in ESR1+ (OR 11.5; 95IC 2.7-48.5; p<0.001), 15.5% in PGR- and 4.3% in PGR+ (OR 3.8; 95IC 1.3-11.7; p=0.019), 7.36% in ERBB2- and 0% in ERBB2+ (OR NA). Finally, ESR1, PGR and ERBB2 expression as continuous variables were found significantly associated with pCR in both cohorts. Conclusion: ESR1, PGR and ERBB2 standardized mRNA levels show high concordance with ER, PR and HER2, and might provide a more objective and quantitative prediction of response to chemotherapy and anti-HER2 therapy. The level of concordance between GE and pathology-based assessments is similar to 2 FDA 510(k) cleared pathology-based ER assays (e.g. SP1 versus 6F11 antibody clones), local versus central HER2 testing, and ER or PR status when determined by the ligand-binding assay versus immunohistochemistry. Citation Format: Pascual T, Tsai YS, Martín M, Lluch A, Cortes J, Llombart A, Conte P, Guarneri V, Rimawi MF, Alba E, Ruiz-Borrego M, Rojo F, de la Haba J, Schiff R, Adamo B, Vidal M, Paré L, Chic N, Muñoz M, Galvan P, Gonzalez-Farre B, Brauer HA, Sullivan A, Nuciforo P, Parker JS, Prat A. Standardized nCounter-based determination of estrogen receptor (ER), progesterone receptor (PR) and HER2 receptor status in breast cancer [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P2-08-06.
Read full abstract