Objective To investigate the value of pre-treatment MRI features in predicting cachexia in hepatocellular carcinoma(HCC).Methods A retrospective analysis was conducted on 399 patients with hepatocellular carcinoma,recording their pre-treatment clinical and MRI data.All patients underwent MRI plain and enhanced scan,and their weight was followed up 6 months after the MRI examination.According to the diagnostic criteria for cachexia,patients were divided into cachexia group and non-cachexia group.They were randomly divided into the training set(n=279)and the validation set(n=120).Univariable and multivariable logistic regression analyses were used to screen variables associated with cachexia in hepatocellular carcinoma and to establish a predictive model.The receiver operating characteristic(ROC)curve was used to evaluate the predictive performance of different models.The DeLong test was used to compare the AUC values of different models,and the best-performing model was used to establish a predictive nomogram for cachexia in hepatocellular carcinoma.Results Multivariable logistic regression analysis showed that serum albumin<40 g/dL,serum alpha-fetoprotein>100 ng/mL,tumor diameter>5 cm,portal vein tumor thrombus,intratumoral arterial enhancement,and arterial phase peritumoral enhancement were independent predictors of cachexia in hepatocellular carcinoma.The clinical-imaging model showed the best predictive performance,with an AUC of 0.843 in the training set and 0.854 in the validation set.Conclusion The nomogram based on MRI features can predict cachexia in hepatocellular carcinoma 6 months earlier than clinical diagnosis,which has important clinical guidance significance.
维普
万方数据
