Radiomics For Predicting Lymph Nodemetastasis And Response Toneoadjuvant Chemoradiotherapy In Colorectal Cancer

Objective: The lymph node metastasis(LNM)status of colorectal cancer(CRC)and the evaluation of neoadjuvant chemoradiotherapy(n CRT)efficacy before surgery are crucial in making treatment decisions for CRC patients.The objective of this study is to investigate the value of multiphase CT-based radiomics and multiparameteric MRIbased radiomics for the prediction of LNM and therapeutic responses to nCRT in CRC patients.Methods: We divided this retrospective study into two parts.The first part included 191 patients who underwent non-contrast,arterial and portal venous phase CT scans(103LNM+ and 88 LNM-).The second part enrolled 193 patients with locally advanced rectal cancer who underwent baseline T1 weighted(T1W),T2 weighted(T2W)and T2 weighted fat saturation(T2WFS)MRI scans,of whom 106 patients responded to nCRT(including 31 pathologic complete response patients and 75 patients in partial pathological remission)and 87 patients were non-responders.In both parts of the study,we extracted radiomic features from the region of interest(ROI)of the CT or MRI images.Then,redundant features were ruled out using the intraclass correlation coefficient(ICC),Mann-Whitney U test,least absolute shrinkage and selection operator(LASSO)regression and akaike information criterion(AIC).Finally,logistic regression was used to build models.The predictive performance of each model was analyzed using the receiver operating characteristic curve(ROC),and the nomogram was evaluated using calibration and decision curve analysis(DCA).Result: The results of the first study showed that the area under the ROC curve(AUC)of the multi-phase CT combined radiomics signature(Com-RS)was better than the single-phase CT model,with AUCs of 0.825 and 0.709 on the training and validation cohorts,respectively.A nomogram was construct by integrating the Com-RS with tumor length,and generated good performancewith AUCs of 0.830 and 0.712 in thetraining andvalidation cohorts,respectively.The results of the second study,for the prediction of good response(GR)and pathologic complete response(p CR),the nomogram achieved favorable prediction efficacy compared to the radiomics modelandclinical model,with AUCs of 0.918 and 0.944 in thetraining andvalidation cohort for predicting GR,respectively;with AUCs of 0.959 and 0.912 in the training and validation cohort for predicting p CR,respectively.Calibration curves and DCA comfirmed potential clinical relevance andapplicability of ournomogram.Conclusion: The developed multiphase CT-based radiomics nomogram and multiparameteric MRI-based radiomics nomogram can be considered as a promising tool for predicting the preoperative lymph node status and therapeutic responses to nCRT in CRC,and assist clinicians in developing appropriate individualised treatment plansfor CRC patients.