| Part Ⅰ:Preliminary study on the assessment of molecular type and related protein expression of gastric cancer by whole tumor histogram analysis derived from DCE-MRIObjectiveTo explore the feasibility of whole-tumor histogram analysis parameters derived from dynamic contrast-enhanced MRI(DCE-MRI)to predict the molecular typing of gastric cancer,and to analyze the correlation between quantitative DCE parameters and related protein expressions.Material and MethodsForty-three patients confirmed as gastric adenocarcinoma by histopathology were enrolled in this prospective study.DCE-MRI were performed before surgery,and mean values,10th,25th,50th,75th,90th percentile values of quantitative DCE parameters(Ktrans,Kep,Ve)and histogram parameters(Skewness,Kurtosis and Entropy)values were manually extracted by Omni-Kinetics software.The specimens were stained with biomarkers EBER-ISH,MLH1,PMS2,E-Cadherin,P53 and HER2.According to the different dyeing results,they were divided into five molecular types and four HER2 expression grades.The abnormal E-Cadherin group and abnormal P53 group were combined into a high-grade malignant group,and the EBV group,MSI group and normal P53 group were combined into a low-grade malignant group.The quantitative DCE parameters or histogram parameters values between two groups were compared using two independent samples t-tests or Mann-Whitney U tests.ROC were performed to determine the best parameters and performance for predicting the two malignant groups.Multivariate logistic regression analysis was used to determine the independent predictors of two malignant groups,and a diagnostic model was established.ROC curve was drawn to confirm the diagnostic efficiency of the model.Goodness of fit of diagnostic models was assessed using the Hosmer-Lemeshow test.The quantitative DCE parameters between positive and negative expression groups of related proteins were compared using the Mann-Whitney U test.The correlations between quantitative DCE parameters and HER2 expression grades were evaluated using Spearman rank correlation test.Results1.Among 43 cases,3 cases were EBV-positive groups,9 cases were MSI groups,2 cases were aberrant E-Cadherin groups,23 cases were aberrant P5 3 groups,and 6 cases were normal P53 groups.There were significant differences in Ktrans mean、Ktrans 25%、Ktrans 50%、Ktrans 75%、Ktrans 90%、Kep mean、Kep 10%、Kep 25%、Kep 50%、Kep 75%、Kep 90%、Ve 10%、Ve 25%between two malignant groups(P=0.006,0.044,0.007,0.007,0.009,0.004,0.032,0.024,0.004,0.005,0.016,0.021,0.028,respectively),and each quantitative parameter was negatively correlated with two different malignant groups.The remaining quantitative parameters and histogram parameters were not statistically significant in differentiating the two different malignant groups(all P>0.05).The ROC curve showed that Kep 50%was the best parameter to distinguish two different malignant groups,with AUC of 0.761,sensitivity,specificity,accuracy,positive predictive value,negative predictive value were 72.2%,84.0%,78.1%,81.9%,75.1%,respectively.In logistic regression analysis,Ktrans 50%was an independent risk factor for predicting gastric cancer in the low-grade group and high-grade group.The established prediction model was not statistically different from the ideal model by Hosmer-Lemeshow test(P=0.474).The AUC value of the prediction model was 0.746,and the sensitivity,specificity,accuracy,positive predictive value,and negative predictive value were 66.7%,80.0%,73.4%,76.9%,and 70.6%,respectively.2.There were no significant differences in quantitative DCE parameters between positive and negative expressions of all related proteins(all P>0.05).3.There were no significant correlation between HER2 expression grades and Ktrans,Kep,Ve values(r=-0.08,-0.03,-0.16,P=0.63,0.84,0.31,respectively).ConclusionWhole-tumor histogram analysis parameters derived from DCE-MRI can assess different malignant groups based on molecular types of gastric cancer in a certain extent,which may provide a new direction for novel evaluation and treatment of gastric cancer.Part Ⅱ:Feasibility study of whole tumor histogram analysis derived from DCE-MRI in predicting pathological grade,Lauren classification,nerve invasion,vascular tumor thrombus of resectable gastric cancerObjectiveTo explore the feasibility of the whole tumor histogram analysis parameters derived from DCE-MRI to predict pathological grade,Lauren classification,nerve invasion,vascular tumor thrombus of resectable gastric cancer.Material and MethodsNinty-seven patients confirmed as GC by histopathology were enrolled in this prospective study.DCE-MRI were performed before surgery,and mean values,10th,25th,50th,75th,90th percentile values of quantitative DCE parameters(Ktrans,Kep,Ve)and histogram parameters(Skewness,Kurtosis and Entropy)values were measured by Omni-Kinetics software.The pathological grades(high,medium and low differentiation),Lauren classification(intestinal type,mixed type,diffuse type),nerve invasion,and vascular tumor thrombosis were evaluated by surgical specimens.Intraclass correlation coefficient(ICC)testing was used to determine the consistency of quantitative DCE parameters and histogram parameters values between two radiologists using Bland-Altman analysis.The quantitative DCE parameters or histogram parameters values between pathological grade,Lauren classifications were compared using ANOVA or Kruskal-Wallis testing.The quantitative DCE parameters or histogram parameters values between the vascular tumor thrombus positive negative group and the nerve invasion positive negative group were compared using Mann-Whitney U testing.Receiver operating characteristic(ROC)analyses was performed to find out the best parameters for identifying pathological grades,Lauren classification,nerve invasion,and vascular tumor thrombosis.Multivariate logistic regression analysis was used to determine the independent risk factors of poorly differentiated gastric cancer and moderately and well-differentiated gastric cancer for the above parameters with statistical significance,and a diagnostic model was established.ROC curve was drawn to confirm the diagnostic efficiency of the model.The Hosmer-Lemeshow test was used to evaluate the goodness of fit of the diagnostic model.Results1.There were statistically difference in Ktrans mean Ktrans 10%,Ktrans 75%,Ktrans 90%,Kep mean,Kep 50%,Kep 75%,Kep 90%and Entropy to identify pathological grades(P=0.017,0.021,0.018,0.004,0.008,0.035,0.011,0.001 and 0.003,respectively),and with the increase of differentiation degree,most of the quantitative DCE parameters or histogram parameter values tended to increase gradually.The remaining quantitative DCE parameters and histogram parameters were not significantly different in identifying the pathological classification(all P>0.05).In ROC analysis,Entropy was the best parameter to identify the pathological grade(The AUC was 0.686,and the sensitivity,specificity,accuracy,positive predictive value,and negative predictive value were 65.3%,72.9%,69.1%,70.7%,and 67.8%,respectively).In logistic regression analysis,Ktrans 10%and Entropy were independent risk factors for predicting poorly differentiated gastric cancer and moderately and well-differentiated gastric cancer.The established diagnostic model was not statistically different from the ideal model by Hosmer-Lemeshow test(P=0.854).The AUC value of the diagnostic model was 0.766,and the sensitivity,specificity,accuracy,positive predictive value,and negative predictive value were 69.4%77.1%,73.3%,75.2%,and 71.6%,respectively.2.There was statistically difference in Entropy to identify the Lauren types(P=0.005).The remaining quantitative DCE parameters and histogram parameters were not significantly different in identifying the Lauren classification(all P>0.05).In ROC analysis,the AUC value of Entropy in identifying the Lauren classification of gastric adenocarcinoma was 0.701,and the sensitivity,specificity,accuracy,positive predictive value,and negative predictive value were 43.3%,91.0%,67.2%,82.8%,and 61.6%,respectively.3.There was statistically difference in Entropy to identify nerve invasion and vascular tumor thrombus(P=0.001,and<0.001,respectively),In addition,Ktrans 25%was also statistically different in identifying vascular tumor thrombi(P=0.038).The remaining quantitative DCE parameters and histogram parameters were not significantly different in determining nerve invasion and vascular tumor thrombus(all P>0.05).In ROC analysis,Entropy was the best parameter to identify neural invasion(AUC:0.704,sensitivity,specificity,accuracy,positive predictive value,negative predictive value were 53.4%,82.1%,67.8%,74.9%,63.8%,respectively)and vascular tumor thrombus(AUC:0.741,sensitivity,specificity,accuracy,positive predictive value,and negative predictive value were 66.2%,79.1%,72.7%,76.0%,and 70.0%,respectively)of gastric adenocarcinoma.ConclusionWhole-tumor histogram analysis parameters derived from DCE-MRI may be able to quantitatively distinguish the pathologic grade,Lauren classification,nerve invasion and vascular tumor thrombus of gastric cancer and thus help clinicians predict the patient’s condition and optimize treatment decision.Part Ⅲ:Feasibility study of whole tumor histogram analysis derived from DCE-MRI in predicting T and N staging of resectable gastric cancerObjectiveTo explore the feasibility of the whole tumor histogram analysis parameters derived from DCE-MRI to predict T and N staging of resectable gastric cancer(GC).Material and MethodsSeventy-eight patients confirmed as GC by histopathology were enrolled in this prospective study.DCE-MRI were performed before surgery,and quantitative DCE parameters(Ktrans,Kep,Ve)and histogram parameters(Skewness,Kurtosis and Entropy)were measured by Omni-Kinetics software.Intraclass correlation coefficient(ICC)testing was used to determine the consistency of Ktrans,Kep and Ve values and histogram metrics values between two radiologists using Bland-Altman analysis.The quantitative DCE parameters or histogram parameters values between T stage or N stage were compared using ANOVA or Kruskal-Wallis testing.Receiver operating characteristic(ROC)analyses was performed to find out the best parameters for identifying T and N staging.Multivariate Logistic regression analysis was used to determine the independent risk factors of T1+2 and T3+4 gastric cancer,and a diagnostic model was established.ROC curve was drawn to confirm the diagnostic efficiency of the model.The Hosmer-Lemeshow test was used to evaluate the goodness of fit of the diagnostic model.Results1.There was statistically difference in Ktrans,Kep,Ve and entropy to identify T staging(P=0.015,0.033,<0.001,and 0.007,respectively),and in pairwise comparisons of Ve values showed statistically difference between T1+2 and T3 group(P<0.001),T1+2 and T4 group(P<0.001).There were statistically differences in Ve to identify N staging(P=0.041).In ROC analysis,Ve was the best parameter for identifying T staging,the AUC value was 0.788,and the sensitivity,specificity,accuracy,positive predictive value,and negative predictive value were 92.9%,57.8%,72.4%,67.7%,and 80.5%,respectively.In logistic regression analysis,Ve and Entropy were independent risk factors for predicting T1+2 and T3+4 gastric cancer.The established diagnostic model was not statistically different from the ideal model by Hosmer-Lemeshow test(P=0.384).The AUC value of the diagnostic model was 0.866,and the sensitivity,specificity,accuracy,positive predictive value,and negative predictive value were 85.7%,75.0%,80.4%,77.4%,and 84.0%,respectively.2.There was statistically difference in Ve to identify N staging(P=0.041).In the ROC analysis,the AUC value of Ve to identify N stage was 0.590,and the sensitivity,specificity,accuracy,positive predictive value,and negative predictive value were 71.4%,89.9%,80.7%,87.6%,and 75.9%,respectively.ConclusionThe whole tumor histogram analysis parameters derived from DCE-MRI may be able to quantitatively evaluate T and N staging of GC,so as to help clinicians predict the patient’s condition and optimize treatment decision. |
PDF Full Text Request