Exploration Of Prognostic Factors And Construction Of Prognostic Analysis Models For Rectal Cancer

Part ? Clinical prognostic model of pathological tumor regression grade after neoadjuvant chemoradiotherapy in locally advanced rectal cancer Objective This study intended to explore the clinical prognostic model related to pathological tumor regression grade after neoadjuvant chemoradiotherapy in locally advanced rectal cancer,so as to provide a theoretical reference for diagnosis,treatment and prognosis prediction of locally advanced rectal cancer.Methods Patients with locally advanced rectal cancer who successfully completed preoperative long-term neoadjuvant chemoradiotherapy and underwent total mesorectal resection during January 1,2010 and March 8,2018 were screened for retrospective analysis.Each patient's pathological Tumor Regression Grade(TRG)Grade was assessed in accordance with the TRG system defined by the American Joint Committee on Cancer(AJCC)and the College of American Pathologists(CAP).The studypopulation was randomly divided into training set and verification set according to 2:1.Disease-free Survival(DFS)was the main observational indicator,and outcome events were defined as disease progression,local recurrence,distant metastasis or death from any cause.Univariate Cox regression and kaplan-meier method were used to select the significant variables predicting the prognosis.The variables were screened by the Least absolute shrinkage and selection operator(LASSO)and the ten-fold cross validation method.Finally,the DFS prediction model was established by multi-factor Cox regression,and the visualization of regression equation was realized by forest map,Nomogram,and the web application.The model was evaluated from three aspects:discrimination,calibration and clinical effectiveness.Results Finally,243 subjects meeting the study requirements were obtained through inclusion and exclusion criteria,and patient data were searched and sorted out through the electronic medical record system,including gender,age,Body Mass Index(BMI),family history,smoking history,drinking history,comorbidities,clinical TNM staging,pathological grade,tumor size,CA199,CEA,pathological TNM stage,tumor vascular invasion,nerve infiltration,etc.By univariate Cox regression,BMI,drinking history,clinical TNM stage,tumor size,tumor location,CA199,CEA,tumor vascular invasion,nerve infiltration,pathological TRG grade and pathological TNM stage were all significant predicting variables of DFS.The TRG and TNM multivariate Cox regression model was constructed by variables selected through LASSO regression.The pathological TRG grade and pathological TNM stage were independent prognostic factors for DFS.The forest map,nomogram and web application of the two models were displayed.TNM model and TRG model both performed well.They both have good discrimination(C index: 0.782,0.749;1,3,5 years NRI: 7.90%,-6.90%,-3.70%;1,3,5 years IDI:-1.40%,-3.30%,-4.60%).The confidence intervals of the calibration curves all contained actual values,and the Decision Curve were higher than the extremecurves.But the discrimination and clinical effectiveness of the pathological TRG model was not as good as that of the pathological TNM model.Conclusions This study established a clinical prognostic model related to pathological tumor regression grade after neoadjuvant chemoradiotherapy in locally advanced rectal cancer.The discrimination,calibration and clinical effectiveness of the model were well-performed,but not as good as the pathological TNM prognosis model.We may need to develop a better prognosis prediction model to improve the prediction ability and clinical application value.Part ? Clinical prognostic model of magnetic resonance tumor regression grade after neoadjuvant chemoradiotherapy in locally advanced rectal cancerObjective This study intended to explore the clinical prognostic model related to Magnetic Resonance(MR)tumor regression grade after neoadjuvant chemoradiotherapy in locally advanced rectal cancer,so as to provide a theoretical reference for diagnosis,treatment and prognosis prediction of locally advanced rectal cancer.Methods Patients with locally advanced rectal cancer who successfully completed preoperative long-term neoadjuvant chemoradiotherapy,underwent preoperative pelvic enhanced MR and total mesorectal resection during January 1,2010 and March 8,2018 were screened for retrospective analysis.Magnetic resonance tumor regression grading(MR-TRG)was determined for each patient based on the tumor proportion in MR T2-weighted images.The study population was randomly divided into training set and verification set according to 2:1.Disease-free Survival(DFS)was the main observational indicator,and outcome events were defined as disease progression,local recurrence,distant metastasis or death from any cause.Univariate cox regression and kaplan-meier method were used to select the significant variables predicting the prognosis.The variables were screened by the Least absolute shrinkage and selection operator(LASSO)and the ten-fold cross validation method.Finally,the DFS prediction model was established by multi-factor Cox regression,and the visualization of regression equation was realized by forest map,Nomogram,and the web application.The model was evaluated from three aspects: discrimination,calibration and clinical effectiveness.Results Finally,243 subjects meeting the study requirements were obtained through inclusion and exclusion criteria,and patient data were searched and sorted out through the electronic medical record system,including gender,age,Body Mass Index(BMI),family history,smoking history,drinking history,comorbidities,clinical TNM staging,pathological grade,tumor size,CA199,CEA,pathological TNM stage,tumor vascular invasion,nerve infiltration,etc.By univariate cox regression,MR-TRG grade was significant predicting variables of DFS.The MR-TRG multivariate Cox regression model was constructed by variables selected through LASSO regression.The MR-TRG grade was independent prognostic factors for DFS.The forest map,nomogram and web application of the model were displayed.MR-TRG model had good discrimination(C index: 0.761;1,3,5 years NRI: compared with pathological TNM model,10.30%,19.00%,21.00%;compared with pathological TRG model,0.20%,21.10%,16.30%;1,3,5 years IDI: compared with pathological TNM model,0.20%,7.60%,8.40%;compared with pathological TRG model,1.70%,10.90%,13.00%).The confidence intervals of the calibration curves all contained actual values,and the prediction curve of clinical effectiveness of 3-year DFS was better than the other two models when the threshold probability was between 0.35 and 0.45.Conclusions This study established a clinical prognostic model related to MR-TRG after neoadjuvant chemoradiotherapy in locally advanced rectal cancer.The discrimination,calibration and clinical effectiveness of the model were well-performed.The MR-TRG model showed the best discrimination and excellent model calibration and clinical effectiveness in the prediction of 3-year DFS,which ccould provide an important reference for the prognostic risk assessment and clinical treatment decision of patients with locally advanced rectal cancer after neoadjuvant chemoradiotherapy.Part ? Identification of key tumorigenesis? related genes and comprehensive analysis of the expression and prognostic value of CXC chemokines in colorectal cancerBackground and objective Colorectal cancer(CRC)is one of the most common malignant tumors and its development involves multi? gene driven processes.This study identified tumorigenesis? associated gene signatures using microarray expression profiling data.The CXC motif chemokines were some of these genes which played an important role in inflammatory processes and angiogenesis and may also be associated with colorectal tumor development.This study was aimed to further explore the expression pattern and prognostic value of CXC chemokines in colorectal cancer,and possible genes and new prognostic prediction molecules related to the development of colorectal cancer.Methods The gene expression profiling of GSE39582,a dataset containing 566 colon cancer samples and 19 non? tumoral colorectal mucosae was downloaded from Gene Expression Omnibus(GEO)database.Differentially expressed genes(DEGs)were extracted by GEO2 R.Functional enrichment analysis of DEGs was conducted on Database for Annotation,Visualization and Integrated Discovery(DAVID)platform using Gene Ontology(GO)and Kyoto Encyclopedia of Genes and Genomes pathways(KEGG)database.The Protein Protein Interaction network(PPI network)was constructed by using Search Tool for the Retrieval of Interacting Genes(STRING)database,and the DEGs module was established using Molecular Complex Detection(MCODE)with Cytoscape software to obtain key genes.The gene enrichment in DEGs module was analyzed by KEGG pathway.We further studied the CXC chemokines in key genes,investigated data related to transcription,translation,survival and tumor immune infiltration for CXC chemokines in patients with CRC from the Oncomine,GEPIA,c Bio Portal,HPA and TIMER databases.Results This study analyzed GEO database and a total of 439 DEGs were identified by between colorectal cancer and non? tumoral colorectal mucosae.Among which up-regulated genes like CXCL3,CLDN2,FABP1,COL9A3 and SPARC,down-regulated genes like CA4,AKR1B10,CHGA,FXYD3 and TRPM6 were key tumorigenesis? related genes.We further studied the CXC chemokines in key genes,and found that the expression levels of CXCL1-3,CXCL5,and CXCL8 were higher in CRC tissues than in colorectal tissues.Expression among stages significantly varied for CXCL1-3 and CXCL9-11.The survival analysis revealed that high transcriptional levels of CXCL4 and CXCL9-11 could serve as positive prognostic factors for patients with CRC.CXCL9-11 were highly associated with CD8+ T cells and natural killer(NK)cells in the tumor immune infiltration analysis,indicating their role in the antitumor immune response.Conclusions This study identified 439 DEGs and results implied that CXCL1-3,CXCL5,and CXCL8 were important factors during CRC oncogenesis and that CXCL9-11 could be new biomarkers for the prognosis of CRC.This study may provide clues to the discovery of potential diagnostic,therapeutic and prognostic biomarkers for CRC.