Methodological Study On The Sensitivity Prediction Of Neoadjuvant Chemotherapy For Breast Cancer And Preliminary Exploration Of Related Mechanisms

Breast cancer is now the leading cause of cancer morbidity and mortality among women worldwide.Breast cancer is a clinically and biologically heterogeneous malignancy,with different molecular subtypes having distinct clinical features,therapy responses,and prognosis.But the underlying molecular mechanisms responsible for these differences are still largely unknown and its sensitivity remains difficult to predict for clinical use.In our study,we adopted the powerful LASSO method to develop a new 25-gene signature classifier by using five microarray expression profile datasets in the Gene Expression Omnibus(GEO)database with patients who received neoadjuvant paclitaxel and anthracycline-based chemotherapy.The 25-gene signature classifier effectively predicts pathologic complete response(pCR)to neoadjuvant paclitaxel and anthracycline-based chemotherapy in breast cancer.The predictive accuracy of the signature classifier was further evaluated using four other independent test sets across different platforms and molecular subtypes.The univariate and multivariate logistic regression analyses showed that 25-gene signature was an independent predictor for the rate of pCR to neoadjuvant paclitaxel and anthracycline-based chemotherapy in breast cancer.Functional enrichment analysis indicated that 25 genes in the signature are mainly enriched in immune-related biological processes.These results suggest that the immune microenvironment may contribute to the sensitivity of breast cancer to chemotherapy.Further analysis indicated the prediction score of the classifier was significantly positively correlated with the immune score and 24 immune checkpoints.ESTIMATE analysis showed that the immune scores in pCR samples were significantly higher than that in residual disease(RD)samples,indicating that the immune cells in the pCR samples were relatively enriched.There were also significant differences in seven immune cell types between pCR and RD samples by CIBERSORT analysis.The difference in the composition of pretreatment TILs may be the important determinants of the chemosensitivity in breast cancer.To further optimize the model,remove the batch effect,and facilitate the clinical application,our study used the gene pair algorithm to construct a 12 gene pair model.The area under the receiver operating characteristic(ROC)curve(AUC)showed the gene pair model has good performance,but lower than the 25-gene signature.Then we used the random forest algorithm(RF)to screen 4 characteristic genes and construct a neural network model.ROC curves of the training set and test sets showed the neural network model achieved better performance,fewer genes,no batch effect,no less accuracy than the 25-gene signature.Enrichment analysis suggests that immune regulation may be responsible for the differential sensitivity to neoadjuvant chemotherapy in breast cancer both in gene pair signature and RF important genes,consistent with the 25-gene signature.In order to validate the pathological link between the immune regulation revealed by the model and the sensitivity of paclitaxel and anthracycline-based neoadjuvant chemotherapy,we further showed that pCR samples had significantly more TILs than RD samples in pretreatment breast biopsy tissue by Haematoxylin and Eosin(HE)staining according to the TILs evaluation criteria established by the International TILs Working Group.The pCR samples were predominantly immune infiltrated phenotypes,while RD samples were predominantly immune desert and immune excluded phenotypes.TILs were highly correlated with immune phenotype,MP grade,ER,PR and HER2 status.These analyses further confirmed that the pretreatment TILs in the tumor immune microenvironment may be involved in modulating the response to neoadjuvant chemotherapy in breast cancer.The above results indicates that the immune regulation invoked by the paclitaxel and anthracycline-based neoadjuvant chemotherapy is also actively involved in chemotherapeutic sensitivity.We further investigated the underlying mechanism using cytological experiments.The results showed that paclitaxel and doxorubicin could induce the activation of caspase 3,the cleavage of GSDME,the release of LDH,and the cells presented typical features of pyroptosis,suggesting that paclitaxel and doxorubicin could induce dose-dependent pyroptosis in breast cancer cells.At the same time,paclitaxel and doxorubicin can induce breast cancer cells to release immunostimulatory damage-associated molecules(DAMPs)such as HMGB1 and ATP.The DAMPs can further induce immunogenic death(ICD).These results showed that paclitaxel and doxorubicin may promote release of DAMPs and induce ICD by pyrolysis.The pretreatment immune infiltrating status can act synergistically with neoadjuvant chemotherapeutic agents to activate immune-mediated ICD.This may be the cellular and molecular biological basis of neoadjuvant chemosensitivity.Conclusively,we developed good gene signature classifiers that can effectively predict pCR to paclitaxel and anthracycline-based neoadjuvant chemotherapy in breast cancer so far.Gene enrichment and tumor immune infiltration analysis also suggest that the immune ecosystem is actively involved in modulating clinical response to neoadjuvant chemotherapy.Pathology,cytological and molecular biology further verified that paclitaxel and doxorubicin can promote release of DAMPs and induce ICD by pyrolysis.These results indicate that chemotherapy can act with pretreatment TILs to activate the immune-mediated ICD,promoting pCR outcomes to paclitaxel and anthracycline-based neoadjuvant chemotherapy in breast cancer.