A Study Of Contrast-Enhanced MDCT And Radiomics In Prediction Of Lymph Node Metastasis And Overall Survival Of Patients With Gastric Cancer

Part One Lymph node count on contrast-enhanced MDCT based on different criterion predicts lymph node metastasis of gastric cancer Objective:MDCT has been widely used for the diagnosis of lymph node metastasis in gastric cancer.However,diagnostic accuracy in true nodal status is unsatisfied due to the size-dependent diagnostic criterion for MDCT.The diagnostic value of lymph node count on preoperative MDCT in predicting lymph node metastasis in patients with gastric cancer is not clear.This study aims to evaluate the usefulness of the lymph node count on MDCT based on different size criterion in predicting the presence of lymph node metastasis of gastric cancer.Methods:Preoperative noncontrast and contrast-enhanced MDCT imaging data and medical records from 195 gastric cancer patients treated with gastrectomy were retrospectively analyzed.MDCT scan with slice thickness of 5mm.A slice thickness of 1.5mm axial images from isotropic voxels at MDCT were reconstructed.The regional lymph nodes were viewed with the axial venous phase CT images.Lymph node count on MDCT were recorded,taking 5mm and 8mm in nodal diameter as size criterion respectively.The optimal cut-off values of lymph node count on MDCT based on different criterion for diagnosing lymph node metastasis were identified using Youden’s index,calculating by the equation J=sensitivity+specificity-1.Diagnostic performance of lymph node count on MDCT based on different criterion was assessed by the area under the ROC curve,compared with Delong test.Results:Lymph node count on MDCT was significantly correlated with pathologic metastatic nodal count.Lymph node count on MDCT increased with pathologic metastatic nodal count increasing.When 5mm as the threshold value of nodal diameter,the optimal cut-off value of lymph node count on MDCT identified by Youden’s index for predicting the presence of lymph node metastasis was > 4,and the area under the curve(AUC)was 0.742.When 8mm as the threshold value of nodal diameter,the optimal cut-off value of lymph node count on MDCT was ≥1(presence of lymph node greater than 8mm),and AUC was 0.714.There was no significant difference in the diagnostic accuracy,sensitivity,and specificity between two methods(68.2% vs.68.2%,63.7% vs.68.9%,78.3% vs.66.7%,respectively).Compared with the marker(presence of lymph node greater than 8mm),the accuracy and sensitivity were improved(73.9% vs.68.2%,78.5% vs.68.9%,respectively)when combining the two methods(lymph node count on MDCT >4 when 5mm as the threshold value and presence of lymph node greater than 8mm).Conclusion:Lymph node count on MDCT >4 when 5mm as the threshold value,lymph node count on MDCT ≥ 1 when 8mm as the threshold value can be used as adjuvant diagnostic markers for the presence of lymph node metastasis in patients with gastric cancer.The diagnostic performance of two methods was comparable.The diagnostic accuracy and sensitivity can be improved,when combining the two methods as diagnostic criterion.Part Two A study on the correlation between lymph node count based on different criterion on contrast-enhanced MDCT and overall survival of patients with gastric cancer Objective:The correlation between lymph node count on preoperative MDCT and the overall survival of patients with gastric cancer remains unclear.This study aimed at evaluating the prognostic implications of lymph node count on MDCT in gastric cancer patients according to Tumor,Node,and Metastasis(TNM)stage,using different size criterion for counting.METHODS:126 patients who underwent gastrectomy for gastric adenocarcinoma between2013 and 2016 were evaluated retrospectively.MDCT scan with slice thickness of5 mm.A slice thickness of 1.5mm axial images from isotropic voxels at MDCT were reconstructed.The regional lymph nodes were viewed with the axial venous phase CT images.Lymph node count on MDCT were recorded,taking 5mm and 8mm in nodal diameter as size criterion respectively.The optimal cut-off values of lymph node count on MDCT based on different criterion for the impact on the overall survival(OS)were identified using the X-tile program.The prognostic implications of lymph node count on MDCT for the survival were analyzed according to different size criterion and tumor T,N,TNM stage.RESULTS:When 8mm as the threshold value of nodal diameter,the optimal cut-off value of lymph node count on MDCT identified by the X-tile program was 1(presence or absence of lymph node greater than 8mm).There was no impact of lymph node count on MDCT on the OS of the entire cohort.According to T staging,the OS of T1 to T2 gastric cancer patients with lymph node count on MDCT ≥ 1(presence of lymph node greater than 8mm)was significantly worse than the OS of those with lymph node count on MDCT < 1(absence of lymph node greater than 8mm)(P=0.002).There was no impact of lymph node count on MDCT on the OS of T3 to T4 patients.According to N staging,there was no impact of lymph node count on MDCT on the OS of N0 or N+ patients.According to TNM staging,there was no impact of lymph node count on MDCT on the OS of I to II or III patients.When 5mm as the threshold value of nodal diameter,the optimal cut-off value of lymph node count on MDCT was7.Patients of the entire cohort with lymph node count on MDCT > 7 had a significantly worse overall survival than those with lymph node count on MDCT ≤7(P = 0.005).According to T staging,patients with lymph node count on MDCT >7 had a significantly worse overall survival than those with lymph node count on MDCT ≤ 7 for T3 to T4 gastric cancer patients(P = 0.019).There was no impact of lymph node count on MDCT on the OS of T1 to T2 patients.According to N staging,patients with lymph node count on MDCT > 7 had a significantly worse overall survival than those with lymph node count on MDCT ≤ 7 for N+ patients(P =0.042).There was no impact of lymph node count on MDCT on the OS of N0 patients.According to TNM staging,patients with lymph node count on MDCT > 7had a significantly worse overall survival than those with lymph node count on MDCT ≤ 7 for stage III patients(p = 0.048).There was no impact of lymph node count on MDCT on the OS ofⅠ-Ⅱpatients.Lymph node count on MDCT > 7when 5mm as the threshold value and stage III were independent poor risk factors.CONCLUSION:The prognostic implications of lymph node count on MDCT in patients with gastric cancer were correlated with the size criterion of lymph node and tumor TNM stage.Evaluating the prognosis with lymph node count on MDCT might be tailored to individual gastric cancer patients.We should alert the count of lymph node greater than 8mm on MDCT to evaluate the prognosis for relatively early stage of T1 to T2 gastric cancer patients.We recommend taking the smaller size(5mm)as counting criterion of lymph node count on MDCT to evaluate the prognosis for relatively advanced stage of T3 to T4,N+,and III gastric cancer patients.Part Three The value of multiphase contrast-enhanced MDCT-based radiomics model for assessing lymph node metastasis and overall survival of patients with gastric cancer Objective:The aim of this study was to develop and validate a multiphase contrast-enhanced MDCT-based radiomics model for preoperative prediction of lymph node metastasis in gastric cancer,and to assess the value of the model for prediction of lymph node metastasis and overall survival of patients with gastric cancer.METHODS:Preoperative noncontrast and contrast-enhanced MDCT imaging data and medical records from 151 gastric cancer patients treated with gastrectomy were retrospectively analyzed(train cohort: 106,validation cohort: 45).MDCT scan with slice thickness of 5mm.A slice thickness of 1.5mm axial images from isotropic voxels at MDCT were reconstructed.Tumor regions of interest(ROIs)were manually delineated with the axial images of largest tumor area.Radiomics features were extracted from non-contrast,arterial-phase,venous-phase CT images.The least absolute shrinkage and selection operator(LASSO)method was used to select features and build a radiomics signature generating radiomic score.Multivariate logistic regression analysis was performed to determine the independent risk factors of lymph node metastasis.Performance and clinical utility of the multiphase contrast-enhanced CT-based radiomics model was evaluated by receiver operating characteristic curve(ROC),calibration curve,Hosmer-Lemeshow test and decision curve analysis(DCA)in training and validation cohort.Moreover,we investigated the impact of the model on the overall survival(OS)of gastric cancer patients.RESULTS:The radiomics signature comprised a total of 7 robust features.They were two features of noncontrast images(wavelet.HLH_glcm_Inverse Variance、 wav elet.LLH_ngtdm_Coarseness),three features of arterial phase images(log.sigma.1.0.mm.3D_ngtdm_Busyness、 wavelet.LHH_glrlm_Short Run Low Gray Level Empha sis、 wavelet.HHL_gldm_Dependence Non Uniformity Normalized)and two feature s of venous phase images(original_ngtdm_Coarseness、wavelet.HHH_glcm_Id).The radiomics signature showed favorable prediction efficacy in train cohort(AUC,0.841;95% confidence interval(CI),0.757-0.905)and validation cohort (AUC,0.804;95% CI,0.673-0.899).Multivariate logistic regression analysis showed the radiomics signature,tumor differentiation and lymph node count o n MDCT were the independent risk factors of lymph node metastasis in gastric cancer.The MDCT-based radiomics model based on these three factors was b uilt.The area under curve(AUC)of the model were 0.884(95% CI:0.807-0.938)in train cohort and 0.857(95% CI:0.734-0.937)in validation cohort,which were higher than the models only used tumor differentiation,lymph nod e count on MDCT or the radiomics signature.The decision curve analysis indi cated its clinical usefulness.Besides,the model was significantly associated wit h the overall survival of gastric cancer patients.CONCLUSION:The multiphase contrast-enhanced MDCT-based radiomics model showed favorable accuracy for prediction of lymph node metastasis in gastric cancer.In addition,the model presented potential of prognostic risk stratification in gastric cancer patients.