Preliminary Study On Genes Related To Predicting The Efficacy Of Neoadjuvant Radiochemotherapy In Locally Advanced Rectal Cancer

Objective:Neoadjuvant radiochemotherapy(NRCT)followed by total mesorectal excision(TME)is the standard treatment for locally advanced rectal cancer(LARC).However,after NRCT,only about 20% of patients can achieve complete pathological remission(pRC).Therefore,how to select suitable patients for neoadjuvant therapy has become a research hotspot.In this study,gene expression profiling was used to predict the target genes of pathological remission in locally advanced rectal cancer after neoadjuvant radiotherapy and chemotherapy,and to provide a target for future clinical verification.Methods : Four dataset GSE35452,GSE46862,GSE68204 and GSE53781 was obtained from Gene Expression Omnibus which contain mRNA expression matrix and tumor regression grading.The first three datasets was combined into one dataset(n=174)and divided into training set(n=121)and internal validation set(n=53)after batch effect removal and the last dataset was used as external validation set(n=26).Univariate logistic regression and t test was separately used to identify genes associated with response to neoadjuvant radiochemotherapy(crude P<0.05).Genes was ranked with P values.All genes with P<0.05 in univariate logistic regression was used as covariates with least absolute shrinkage and selection operator(LASSO)algorithm and first 50 genes was used with support vector machine algorithm(SVM)to develop models which was verified in internal validation set.This procedure was repeated 500 times to determine the reproducibility of selected gene signatures and models.The 21 most important genes revealed by LASSO was utilized as candidate genes to generate sensitive index(SI)for neoadjuvant radiochemotherapy that was calculated as total sum of production of coefficients in logistic regression and expression values in combined dataset and external validation set separately.The differentially expressed genes were identified between responsive and non-responsive groups in combined dataset(n=174)and subjected into regulatory network analysis.Results:1.A total of 12803 genes in combined dataset from GSE35452,GSE46862 and GSE68204 was subjected for screening.The accuracy,specificity and sensitivity of LASSO to predict pathological response in internal validation set were 0.523(95% CI: 0.396-0.642,0.578(95% CI: 0.373-0.762)and 0.464(95% CI: 0.258-0.700)respectively.The accuracy,specificity and sensitivity of SVM were 0.504(95% CI: 0.377-0.623),0.596(95% CI: 0.393-0.830),0.405(95% CI: 0.182-0.650)respectively.The 21 genes that were included in the model most frequently by LASSO algorithm had AUC of 0.863(95%ci: 0.811 ~ 0.912)and 0.925(95%ci: 0.817 ~ 1.000)for the combined data set of 174 cases and the external independent validation set of the radiochemotherapy sensitive indicators.2.In 174 combined datasets,505 differentially expressed genes were mainly regulated by NF-kB,CTNNB1,IL6/STAT3,E2F1,GLI2 and MYC transcription factors.Regulatory network analysis suggested that the mechanisms responsible for invasion,metastasis and stem cell transformation was potential mechanism participating in resistance to radiochemotherapy in rectal cancer.Conclusion:1.This study screened 21 predictive genes,which may be closely related to radiotherapy and chemotherapy sensitivity of rectal cancer.Their mechanisms include glycometabolism,serine/glycine synthesis and metabolism,procoagulant mechanism,transcriptional regulation and cell cycle regulation.2.The predictive model of neoadjuvant chemoradiotherapy for locally advanced rectal cancer using selected related genes has certain predictive ability,which still needs to be further improved..