Application Of CT Radiomics In Assessment Of Interstitial Microenvironment Of Pancreatic Ductal Adenocarcinoma

Part 1: The application of CT Radiomics in the evaluation of tumor stroma ratio in patients with pancreatic ductal adenocarcinoma.Objective: The tumor-interstitial ratio(TSR)is an important survival predictor for patients with pancreatic ductal adenocarcinoma(PDAC).At present,there is a lack of effective non-invasive assessment methods.This study developed and validated a machine learning classifier based on CT for preoperative prediction of patients’ TSR.Methods: In this retrospective study,227 patients with PDAC underwent an MDCT scan and surgical resection.We evaluated the TSR by using ematoxylin and eosin staining and extracted 1409 arterial and portal venous phase radiomics features for each patient,respectively.Patients were divided into TSR-low group and TSR-high group with TSR=1as the node.Moreover,we used the least absolute shrinkage and selection operator(LASSO)logistic regression algorithm to reduce the features.The extreme gradient boosting(XGBoost)was developed using a training set consisting of 167 consecutive patients,admitted between December 2016 and December 2017.The model was validated in 60 consecutive patients,admitted between January 2018 and April 2018.We determined the XGBoost classifier performance based on its discriminative ability,calibration,and clinical utility.Results: We observed low and high TSR in 91(40.09%)and 136(59.91%)patients,respectively.A log-rank test revealed significantly longer survival for patients in the TSR-low group(25.23 months,95%CI:23.00-35.63)than those in the TSR-high group(16.43 months,95%CI:14.67-20.77)(p<0.0001).The prediction model revealed good discrimination in the training(area under the curve [AUC]= 0.93)and moderate discrimination in the validation set(AUC= 0.63).While the sensitivity,specificity,accuracy,positive predictive value,and negative predictive value for the training set were94.06%,81.82%,0.89,0.89,and 0.90,respectively,those for the validation set were85.71%,48.00%,0.70,0.70,and 0.71,respectively.Conclusion: The CT radiomics-based XGBoost classifier provides a potentially valuable noninvasive tool to predict TSR in patients with PDAC and optimize risk stratification.Part 2: The application of magnetic resonance radiology in the evaluation of tumor fibroblast activating protein in patients with pancreatic ductal adenocarcinoma.Objective: Tumor fibroblast activating protein(FAP)is an important therapeutic target in tumors,which cannot be effectively evaluated by conventional imaging.In this study,we developed and validated a CT-based radiomics model for predicting FAP content in patients with pancreatic duct adenocarcinoma(PDAC).Methods: A total of 152 patients with PDAC who underwent preoperative enhanced CT scan and underwent surgical resection in our hospital from January 2017 to April 2018 were collected.Immunohistochemistry was used to stain the sections and the expression of FAP was assessed.X-tile software was used to divide the patients into low FAP group and high FAP group.In the training set,1409 radiomics features were extracted from each patient’s preoperative images.Least absolute shrinkage and selection operator(LASSO)logistic regression algorithm were used to select the optimal features,and an extreme gradient enhancement(XGBoost)classifier was developed by using radiomics features.The classifier was developed in a training set of 94 patients between January 2017 and December 2017 and validated in a validation set of 58 patients between January 2018 and April 2018.We determined the performance of the XGBoost classifier by analyzing its discrimination,calibration,and clinical utility.Results: The patients were divided into FAP-low(n=91;59.87%)and FAP-high(n=61;40.13%)groups according to the optimal cut point of FAP level(45.71%).Survival was significantly longer in the low FAP group(21.73 months,95%CI: 18.63-26.0)than in the high FAP group(12.60 months,95%CI: 10.00-19.53)(P =0.007).The XGBoost classifier comprised 13 selected radiomics features and showed good discrimination in the training set(AUC,0.97)and the validation set(AUC,0.75).In the calibration curve,the fitting degree of the model is good(p>0.05).The decision curve analysis shows that the model is practical in clinic.Conclusion: The XGBoost classifier based on CT radiomics can noninvasively predict the expression of FAP and provide more information for clinical treatment decisions of patients with PDAC.