AbstractNeoadjuvant chemotherapy is the foundation treatment for triple-negative breast cancer (TNBC) and frequently results in pathological complete response (pCR). However, there are large differences in clinical response and survival after neoadjuvant chemotherapy of TNBC patients. The aim was to identify genes whose expression significantly associates with the efficacy of neoadjuvant chemotherapy in patients with TNBC. Transcriptomes of 46 formalin-fixed paraffin-embedded (FFPE) tumor samples from TNBC patients were analyzed by RNA-seq by comparing 26 TNBCs with pCR versus 20 TNBCs with pathological partial remission (pPR). Subsequently, we narrowed down the list of genes to those that strongly correlated with drug sensitivity of 63 breast cancer cell lines based on Dependency Map Consortium data re-analysis. Furthermore, the list of genes was limited to those presenting specific expression in breast tumor cells as revealed in three large published single-cell RNA-seq breast cancer datasets. Finally, we analyzed which of the selected genes were significantly associated with overall survival (OS) in TNBC TCGA dataset. A total of 105 genes were significantly differentially expressed in comparison between pPR versus pCR. As revealed by PLSR analysis in breast cancer cell lines, out of 105 deregulated genes, 42 were associated with sensitivity to docetaxel, doxorubicin, paclitaxel, and/or cyclophosphamide. We found that 24 out of 42 sensitivity-associated genes displayed intermediate or strong expression in breast malignant cells using single-cell RNAseq re-analysis. Finally, 10 out of 24 genes were significantly associated with overall survival in TNBC TCGA dataset. Our RNA-seq-based findings suggest that there might be transcriptomic signature consisted of 24 genes specifically expressed in tumor malignant cells for predicting neoadjuvant response in FFPE samples from TNBC patients prior to treatment initiation. Additionally, nine out of 24 genes were potential survival predictors in TNBC. This group of 24 genes should be further investigated for its potential to be translated into a predictive test(s).
Read full abstract