| PurposeThis research aims to study the structural changes of collagen in the tumor budding hotspot area and perform quantitative analysis to build a clinical prediction model based on the collagen scores;Combined with immunohistochemical staining,more analysis were performed to reveal the relationship of collagen rearrangement,epithelial mesenchymal transformation and immune infiltration.MethodsThis research included 269 patients who suffered with colorectal cancer(stageⅢ),and all the patients underwent radical resection at Nanfang Hospital between 2007 and 2011.The experiment was conducted in two stages.In the first stage,the tumor budding number of 269 patients was counted and divided into two groups according to the criterial:below 10 buds/0.758mm2 is low grade and on the contrary is high.Then multi-photon imaging at the budding hotspot area of each patient was performed.Multidimensional collagen characteristic parameters were extracted from the imaging pictures by MATLAB software,and the LASSO-COX model was used to select the collagen signatures related to the patient’s disease-free survival and establish a collagen scoring formula.Combined with the multivariate analysis results,a clinical prediction model was built to predict the early postoperative recurrence at 2 years.In the second phase of the experiment,tissue sections of 269 patients were subjected to E-cadherin,Vimentin,CD3+,and CD8+immunohistochemical staining.ImageJ was used to quantify the number of positive markers and positive cells,and explore the indicators and tumor budding,Correlation between collagen scores and collagen morphological characteristics,and visualization of the results of correlation analysis.ResultsMultiphoton imaging results suggested the collagen structure changes and appears as "spicule sign" at the tumor budding "hotspot" area.After compressing and filtering 142 collagen signatures,the LASSO-COX regression model obtained 8 features according to the patient’s disease-free survival and constructed a collagen scoring formula.After univariate and multivariate analysis,collagen score was confirmed as an independent risk factor for survival prognosis in stage III colorectal patients,and patients with high scores had worse outcome(OS,HR:11.548,95%CI:6.522-20.449,p<0.001;DFS,HR:14.955,95%CI:8.684-25.755,p<0.001).Based on the collagen score,a clinical prediction model for postoperative recurrence of patients is constructed.The evaluation results demonstrated the model has excellent predictive performance and differentiation,which can provide an accurate reference for clinical decisionmaking.In addition,the results of correlation analysis suggested that the collagen score was positively correlated with tumor budding and negatively correlated with the expression of E-cadherin.Besides,the collagen area,collagen number,collagen length,collagen width,cross-linking space of collagen were all decreased at the tumor budding hotspot area,while the density and rigidity of collagen increased,the E-cadherin expression was also down-regulated.ConclusionsThe collagen score is an effective predictor of postoperative recurrence in patients with stage Ⅲ colorectal cancer.The EMT process at the tumor budding area is accompanied by remodeling of collagen structure.Patients with a large number of budding have high collagen density and rigidity,high tumor interstitial degree,and poor prognosis. |