The Application Of CT-based Radiomic In Predicting The Efficacy Of Chemoradiotherapy For Loco-regional Lymph Node Recurrence After Radical Resection Of Oesophageal Squamous Cell Carcinoma

Part Ⅰ Prognostic factors of chemoradiotherapy for loco-regional lymph node recurrence after radical resection of oesophageal squamous cell carcinomaObjective:There are no standard therapeutic strategies for local lymph node(LN)recurrence after radical resection of oesophageal squamous cell carcinoma(ESCC),and prognostic risk factors remain controversial.We assessed clinical outcomes and prognostic factors of chemoradiotherapy(CRT)or radiotherapy(RT)for LN recurrence of ESCC after curative resection.Methods:A total of 117 ESCC patients with LN recurrence after radical resection receiving salvage treatment at our hospital were retrospectively reviewed from 2014 to 2017.Overall survival(OS)was estimated using the Kaplan-Meier method and the clinical characteristics were assessed using the log-rank test in the univariate analysis.Multivariate prognostic analysis was performed using the Cox proportional hazard model.Results:With a median follow-up of 19.0 months,the 1-,2-and 3-year OS rates were 75.2%,40.2%and 27.4%,respectively.The median survival time(MST)was 19.0 months.On univariate analysis for OS,pathological TNM stage,number of LN metastasis,LN maximum(Max)diameter,salvage treatment mode and tumor response were significantly associated with OS(P=0.0074,0.015,0.0011,0.028,<0.0001,respectively).On multivariate analysis,tumor response[Response VS.No-response hazard ratio(HR):2.43;95%confidence interval(CI):1.53-3.90,P<0.0001]and LN Max diameter(≤28 mm VS.>28 mm HR:2.07;95%CI:1.33-3.32,P=0.012)were independent prognostic factors.Conclusion:Salvage CRT or RT was safe and effective for treating LN recurrence after radical resection in ESCC.Patients with the small LN Max diameter(≤28 mm)and obtained response after salvage therapy appeared to achieve long-term OS.Part Ⅱ CT-based radiomic analysis for prediction of local efficacy of salvage chemoradiotherapy for loco-regional lymph node recurrence after curative esophagectomyObjective:To investigate the capability of computed tomography(CT)radiomic features to predict the therapeutic response and local control of the loco-regional recurrence lymph node(LN)after curative esophagectomy by chemoradiotherapy(CRT).Methods:This retrospective study included 124 LN from 75 patients with postoperative esophageal squamous cell carcinoma(ESCC).There were 97 LN from 57 patients in training cohort and 27 LN from 18 patients in validation cohort.The region of the tumor was contoured in pretreatment contrast-enhanced CT images.The least absolute shrinkage and selection operator(LASSO)was used to identify radiomic predictors in the training cohort.Model performance was evaluated using the area under the receiver operating characteristic curves(AUC)in both cohorts.The Kaplan-Meier method was used to determine the local recurrence time of cancer.Log-rank test and Cox proportional hazards were used for univariate and multivariate analysis.Results:The radiomics model composed by nine features could use to predict treatment response.The AUCs in the training and validated cohorts were 0.765(95%CI:0.556-0.975)and 0.806(95%CI:0.631-0.980)respectively.A significant difference of radiomic score(Rad-score)between the response and non-response was observed in the two cohorts(P<0.001 and 0.008,respectively).A model composed of eight radiomics features was used to predict the local control time.In the training and validate group,the AUCs of predicting the 1-year and 2-year local control rate were 0.744(95%CI:0.628-0.860),0.872(95%CI:0.792-0.951)and 0.708(95%CI:0.486-0.930),0.656(95%CI:0.303-1.0)respectively.The local control time of LN with higher Rad-score(≥-3.894)was significantly shorter than that of lymph nodes with lower Rad-score(<-3.894,P<0.0001).After the Cox regression analysis,the Rad-score was one of the high-risk factors for local recurrence within two years.Conclusions:The model which was established by the CT-based radiomic analysis of lymph nodes could effectively predict and evaluate the treatment response and local control of salvage radiotherapy or chemoradiotherapy for regional lymph node recurrence after curative esophagectomy in patients with esophageal squamous cell carcinoma.Part Ⅲ Building CT radiomics based nomogram to predict overall survival of salvage chemoradiotherapy for loco-regional lymph node recurrence after curative esophagectomyObjective:To explore the application of CT radiomics to predict the survival after salvage chemoradiotherapy(CRT)in patients with esophageal squamous cell cancer(ESCC)for loco-regional lymph node(LN)recurrence after esophagectomy.To develop and validate of the nomogram model including CT radiomics and clinical features to predict overall survival(OS)of ESCC patients with LN recurrence after esophagectomy.Methods:A total of 114 ESCC patients with LN recurrence after radical resection receiving salvage CRT at our hospital were analyzed retrospectively and divided into training cohort(72 patients)and validation cohort(42 patients).The recurrence LN in enhanced CT images before treatment were analyzed by radiomic.The least absolute shrinkage and selection operator(LASSO)with COX regression was performed to select optimal radiomic features to built the radiomic model for predicting OS in the training cohort,and the univariate and multivariate analyses were performed to identify the predictive clinical factors for OS.Finally,a nomogram model including radomics model and clinical features was established.In the training cohort and validation cohort,the performance of nomogram model was assessed by the time-dependent area under the receiver operating characteristic curves(AUC),C-index,Akachi information criterion(AIC),likelihood ratio test and decision curve.Results:Four features from 427 radiomic features were selected by LASSO-COX regression analysis to built the radiomic model for OS.The AUC values of radiomic model for predicting the 1-,2-and 3-year survival rates in the training cohort and validation cohort were 0.7006,0.7738,0.7627 and 0.6538,0.7012,0.7824 respectively.In the training cohort,the age,postoperative TNM stage,number of LN metastasis,LN maximum diameter,treatment response and radiomic score(Rad-score)were significantly associated with OS on multivariate analysis.The combined nomogram model with Rad-score and the clinical features and the clinical nomogram model without Rad-score were established respectively.In the training cohort,the AUCs for predicting the 1-,2-and 3-year survival rates,C-index and AlC of the combined model and clinical model were 0.7767,0.8619,0.8600,0.760,348.6 and 0.7788,0.8389,0.8409,0.734,352.63 respectively.In the validation cohort,the AUCs,C-index and AIC were 0.8010,0.8395,0.7546,0.701,212.0 and 0.8037,0.8320,0.7268,0.675,213.2 respectively.Likelihood ratio and decision curve analysis also suggested that the combined nomogram model is better than the clinical nomogram model.Conclusion:The combined nomogram model incorporated the CT radiomic model could more effectively predict the OS of patients with LN recurrence after radical resection receiving salvage CRT than the clinical feature nomogram model and radiomic model,and provided a better prediction tool for clinical work.