Prediction of vessels encapsulating tumor clusters pattern in hepatocellular carcinoma based on Gd-EOB-DTPA enhanced MRI / 实用放射学杂志

ABSTRACT

Objective To investigate the value of qualitative and quantitative characteristics of gadolinium ethoxybenzyl diethylenetriamine pentaacetic acid(Gd-EOB-DTPA)enhanced MRI in preoperative prediction of vessels encapsulating tumor clusters(VETC)pattern in hepatocellular carcinoma(HCC).Methods A total of 234 patients diagnosed with HCC by pathology were analyzed retrospectively.A total of 101 VETC-positive HCC patients and 133 VETC-negative HCC patients were included.All patients were divided into training group and validation group according to 7︰3.The training group data were used to construct a prediction model for VETC-positive HCC.Receiver operating characteristic(ROC)curve was drawn and the area under the curve(AUC)was calculated to verify the diagnostic efficiency of the model.Calibration curve was drawn to verify the calibration of the model.Results Multivariate logistic regression analysis predicted the independent risk factors for VETC-positive HCC:portal phase peripheral washout[odds ratio(OR)6.493],necrosis or severe ischemia(OR 4.756),targetoid transitional phase or hepatobiliary phase(OR 0.307),and lesion to liver signal intensity ratio(LLR)on arterial phase(OR 0.074).The AUC of the training group in predicting VETC-positive HCC was 0.790[95%confidence interval(CI)0.720-0.859].The AUC of the validation group in predicting VETC-positive HCC was 0.779(95%CI 0.668-0.889).The calibration curve diagram showed that the calibration curve(the slope was 0.91)almost coincides with the ideal curve,indicating that the prediction model had better calibration.Conclusion The qualitative and quantitative characteristics of Gd-EOB-DTPA enhanced MRI can be used to predict VETC-positive HCC preoperatively,the independent risk factors of VETC include portal phase peripheral washout,necrosis or severe ischemia,targetoid transitional phase or hepatobiliary phase,and LLR on arterial phase.