Construction And Validation Of A Prediction Model For Lymph Node Metastasis In Locally Advanced Rectal Cancer

Objective:This study aims to construct a prediction model for lymph node metastasis of locally advanced rectal cancer,analyze the influence of lymph node metastasis of locally advanced rectal cancer on the therapeutic effect of neoadjuvant chemoradiotherapy,and analyze the biological functional pathways related to lymph node metastasis of locally advanced rectal cancer,so as to provide scientific basis for guiding clinical accurate diagnosis and treatment of lymph node metastasis.Methods:In this study,gene expression profiles of locally advanced rectal cancer patients with lymph node metastasis and non-lymph node metastasis were collected as training data Based on gene expression order relationship algorithm,randomforest algorithm and ten fold cross validation method were used to identify relevant characteristic variables and construct a prediction model for lymph node metastasis of locally advanced rectal cancer.The area under the subject operating characteristic curve was used to evaluate the accuracy and prognostic consistency of the prediction model.The independent cohort data set and the patient gene expression profile collected and tested by our laboratory were used as the validation set Chisquare test was used to analyze the association between lymph node metastasis of locally advanced rectal cancer and the effect of neoadjuvant chemoradiotherapy.Finally,enrichment analysis and protein interaction network were used to analyze the related biological functions of differentially expressed gene sets,and biological functional pathways related to lymph node metastasis were identified.Results:In this study,we constructed a prediction model of lymph node metastasis of locally advanced rectal cancer composed of 14 pairs of genes.The AUC of the training set was 0.972(95%CI,0.940-1),the accuracy was 0.986,and the verification set was 0.930(95%CI,0.887-0.974)and 0.747(95%CI,0.5730.920),respectively.The prediction model had a good prediction effect.In both the training set and the validation set,there was a statistically significant difference in disease-free survival between patients predicted to have lymph node metastasis and those in the non-metastatic group.There was no statistically significant association between lymph node metastasis in locally advanced rectal cancer and complete response after neoadjuvant chemoradiotherapy.Through enrichment analysis and protein interaction network analysis of differentially expressed gene sets,genes were found to be enriched in immune response,inflammatory response,monocyte chemotaxis,lymphocyte chemotaxis,chemokine signaling pathways,and complement and coagulation cascades.Condusion:In this study,we successfully constructed a prediction model of lymph node metastasis of locally advanced rectal cancer composed of 14 pairs of genes,which has good accuracy and clinical value.There was no significant difference in the rate of complete response between patients with locally advanced rectal cancer with lymph node metastasis and those without lymph node metastasis who received neoadjuvant chemoradiotherapy.During the occurrence of lymph node metastasis in locally advanced rectal cancer,the relevant functional pathways may involve immune response,inflammatory response,monocyte chemotaxis,lymphocyte chemotaxis,chemokine signaling pathways,and complement and coagulation cascades.