| Esophageal carcinoma is one of the common malignant tumors in our country.More than 90%of patients are pathologically classified as squamous cell carcinoma.The main treatment methods are surgery and radical CRT.Patients with esophageal squamous cell carcinoma(ESCC)have a huge difference in the response of CRT,and the overall prognosis is poor,which may be related to the biological heterogeneity of the tumor.It is important to find markers that reflect the biological characteristics of tumors and accurately predict the sensitivity of esophageal squamous cell carcinoma to CRT.DWI can non-invasively reflect the motion information of tumor water molecules,and evaluate the activity of cells inside the tumor with no radiation.Multiple images can be used to obtain the changes of the internal structure of the tumor in response to treatment to predict the treatment response,but the currently commonly used two-dimensional DWI cannot reflect the global information of the tumor.It cannot be accurately used for guidance.Currently,researches are mostly focused on radiomics parameter phenotyping.This study is to establish a prediction model of treatment response to CRT through the radiomics model derived from ADC map,and correlate the radiomics feature(RF)with tumor histology and pathology to guide the decision-making in patients with ESCC.Part Ⅰ:Response Prediction to Concurrent Chemoradiotherapy in Esophageal Squamous Cell Carcinoma Using Delta-Radiomics Based on Sequential Whole-Tumor ADC MapPurpose:The purpose of this study was to investigate the association between the radiomics features extracted from a whole-tumor ADC map during the early treatment course and response to concurrent CRT in patients with esophageal squamous cell carcinoma.Methods:Patients with ESCC who received concurrent CRT were enrolled in two hospitals.Whole-tumor ADC values and RFs were extracted from sequential ADC maps before treatment,after the 5th radiation,and after the 10th radiation,and the changes of ADC values and RFs were calculated as the relative difference between different time points.RFs were selected and further imported to a support vector machine classifier for building a radiomics signature.Radiomics signatures were obtained from both RFs extracted from pretreatment images and three sets of delta-RFs.Prediction models and Nomogram for different responders based on clinical characteristics and radiomics signatures were built up with logistic regression.Results:Patients(n=76)from hospital 1 were randomly assigned to training(n=53)and internal testing set(n=23)in a ratio of 7 to 3.In addition,to further test the performance of the model,data from another institute(n=17)were assigned to the external testing set.Neither ADC values nor delta-ADC values were correlated with treatment response in the three sets.It showed a predictive effect to treatment response that the AUC values of the radiomics signature built from delta-RFs over the first 2 weeks were 0.824,0.744,and 0.742 in the training,the internal testing,and the external testing set,respectively.Compared with the evaluated response,the performance of response prediction in the internal testing set was acceptable(p=0.048).A nomogram of clinical factors combined with radiomics Signature was constructed,and the concordance index was 0.872Conclusions:The ADC map-based delta-RFs during the early course of treatment were effective to predict the response to concurrent CRT in patients with ESCC.Part Ⅱ:Study on the evolution model of response markers based on DWI to chemoradiotherapy in patients with ESCCPurpose:To use RFs from DWI as markers to analyze the evolution of imaging over CRT in patients with ESCC,construct a tumor treatment evolution model based on imaging markers,find the key subtypes related to the efficacy of CRT.Methods:Patients who received radical CRT were included,and each patient underwent DWI before treatment,after the 5th radiation,after the 10th radiation and after the 20th radiation.All patients were randomly divided into training group and test group according to the ratio of 3:2,including 40 cases in the training group and 27 cases in the testing group.The mean values of 30 RFs in the training group were calculated,and the Mann-Whitney U test was used to compare whether the parameters in the training group and the validation group were consistent.The Wilcoxon rank sum test was used to compare the differences of RFs between the sensitive group and the resistant group,and to summarize the changing patterns of the parameters with statistical differences.The 30 image parameter phenotypes before treatment were used as the root node,and the image phenotypes of multiple time nodes in radiotherapy were compared with them.The Euler distance was used as the measurement parameter,and the phylogenetic tree model was constructed by the change rate of image features.The ggtree package draws CRT evolutionary trees of ESCC.Results:A total of 67 patients were enrolled.In the mean absolute error and normalized root mean square error results between the training group and the validation group,all the p values of the Mann-Whitney U test were greater than 0.05,indicating that there was no significant statistical difference between the evolution trends of the RFs between the two groups.Among the 30 selected RFs,the three parameters,original_firstorder_Uniformity,wavelet_HLL_glcm_SumEntropy and wavelet_HLL_glszm_ZoneEntropy,have a P value of less than 0.05 in the Wilcoxon rank-sum test in the sensitive group and the resistant group which show stabilize in resistance group,linear changing in sensitive group.The phylogenetic tree model based on imaging features further confirmed the regularity of imaging phenotype changes in the evolution of CRT in ESCC.Conclusion:RFs from DWI can predict the sensitivity of ESCC treated with CRT.During the treatment,the evolution of tumor leads to changes in various parameters,and shows a regular trend of changes.Part Ⅲ:Association between RFs and histological parameters in ESCCPurpose:To explore the relationship between the RFs involved in modeling and histological parameters related to sensitivity of CRT in ESCC.Methods:The newly diagnostic patients with ESCC underwent MRI examination before treatment.Specimens of tissue biopsy were obtained from patients with concurrent CRT before treatment,and surgical specimens were obtained from patients after surgery.The tissue specimens were sectioned for HE staining and immunohistochemical staining,including tumor differentiation,Ki67,P53protein and tumor-infiltrating lymphocytes(CD3+TIL,CD4+TIL and CD8+TIL).Statistical analysis was performed between RFs and histological features,and also between pathological features and treatment response.Results:A total of 70 patients were included in the study,of whom 25 received radical surgery and 45 received concurrent CRT.The RFs included in the part were features used for establishing the prediction model in the first part.No statistical association was fpound between histological features and response to chemoradiotherapy.These 30 RFs were not statistically correlated with tumor differentiation,the expression of P53 protein and the density of CD4+TIL.In the analysis of Ki67,six RFs showed statistical significance,including original_gldm_DependenceEntropy(P=0.029),wavelet_HLL_glszm_ZoneEntropy(P=0.026),wavelet_LHL_gldm_DependenceEntropy(P=0.024),wavelet_LHL_glszm_Zone_Entropy(P=0.047),wavelet_LLH_glcm_Idmn(P=0.03)and wavelet_LLL_gldm_DependenceEntropy(P=0.025).In the analysis of CD3+TIL,five RFs showed statistical significance,including original_firstorder_Uniformity(P=0.04),original_glrlm_GrayLevel_NonUniformityNormalized(P=0.041),original-glrlm-RunEntropy(P=0.042),wavelet_LLH_glszm_ZoneEntropy(P=0.024),and wavelet_LLL_firstorder_MeanAbsoluteDeviation(P=0,047).In the analysis of CD8+TIL,only the wavelet_LLH_glszm_ZoneEntropy showed statistical significance(P=0.042).Conclusions:Parts of radiomic features in the prediction model are related to ESCC tissue markers. |
PDF Full Text Request