Abstract A13: Implication of immune cell composition in biopsy specimens of triple-negative breast cancer for responsiveness to neoadjuvant chemotherapy

Abstract

Abstract Triple-negative breast cancer (TNBC) is a subtype of breast carcinoma that has little or no expression of hormone receptors and human epidermal growth factor receptor 2 (HER2). Most patients with TNBC receive chemotherapy, since TNBC is an aggressive type of cancer and lacks targeted therapy. Recently, possible effectiveness of immunotherapy for TNBC has been identified. Here, we aim to elucidate predictive power of immune cell composition and immune-related gene expression in neoadjuvant chemotherapy (NAC) setting of TNBC. Expression profiles of 579 immune-related genes were examined in biopsy tissues obtained from 55 patients with primary TNBC treated with NAC (anthracycline, cyclophosphamide, and docetaxel) using NanoString nCounter GX Human Immunology Panel. Then, the composition of 22 subtypes of immune cells was estimated from the gene expression profiles using CIBERSORT. The optimal combination of immune cell subsets as well as immune-related genes to predict pathologic complete response (pCR) and clinical outcome (death) was inferred by univariate receiver operating characteristic (ROC) curve and subsequent multivariate ROC curve analyses. Among the 55 patients, 17 cases showed pCR whereas the others showed non-pCR. P values were provided by CIBERSORT for all the specimens but two were less than 0.05. Immune cell composition varied by patients, and only the proportion of resting dendritic cells was significantly different between the two groups, pCR vs. non-pCR (p=0.038). In our univariate ROC curve analysis, follicular helper T cell (area under the curve, AUC 0.646) and monocyte (AUC 0.625) were the most relevant immune cell subtypes to patients’ clinical outcome and pCR, respectively. Multivariate ROC curve-based exploratory analysis identified the highest AUC (0.647) for the prediction of clinical outcome from the combination of follicular helper T cell and naive B cell proportions. The combination of 18 out of 22 immune cell subsets showed highest AUC (0.745) for pCR. In terms of immune-related genes, univariate ROC curve analysis identified TNFSF13B (AUC 0.756) and MAPK1 (AUC 0.718) as the most relevant genes to patients’ survival and pCR, respectively. From our multivariate ROC curve analysis, the combination of 113 genes including RUNX1 had highest AUC (0.915) for survival outcome prediction. For the prediction of pCR, the combination of 107 genes including MAPK1, NFKB1, and TRAF1 showed highest AUC (0.946). Forty-one genes were overlapped between the two sets of genes. Our study showed not only immune-related gene expression but also composition of immune cell subtypes could predict responsiveness to neoadjuvant chemotherapy in biopsy specimens of TNBC, though its predictive power seemed to be lower than that of combination of gene expression. Further analysis of immune-related gene expression in each immune cell type might improve our understanding of immune microenvironment and provide higher predictive values in clinical response to chemotherapy in TNBC patients. Citation Format: In Ah Park, Jinho Jang, Hyoung-oh Jeong, Gyungyub Gong, Semin Lee, Hee Jin Lee. Implication of immune cell composition in biopsy specimens of triple-negative breast cancer for responsiveness to neoadjuvant chemotherapy [abstract]. In: Proceedings of the AACR Special Conference on Tumor Immunology and Immunotherapy; 2018 Nov 27-30; Miami Beach, FL. Philadelphia (PA): AACR; Cancer Immunol Res 2020;8(4 Suppl):Abstract nr A13.