Objective: The objectives of our study were to identify independent clinical, demographic, and MR imaging correlates of malignancy in patients with cirrhosis and to develop a predictive model based on identified correlates of malignancy.
Materials and methods: Sixty examinations of 58 patients with biopsy proof of lesions suggestive of hepatocellular carcinoma on MR imaging were retrospectively reviewed. The signal intensity of the lesion on T2-weighted imaging and dynamic gadolinium-enhanced imaging, the size of the lesion, and the number of suspicious lesions were recorded; in addition, patient age and sex, alpha-fetoprotein level, and hepatitis C viral genotype were noted. The association between malignancy and each predictor variable was evaluated using the chi-square test or the two-group t test. The final logistic regression model included the variables that were shown to have a significant association with malignancy and the clinically relevant predictors. We used the adjusted odds ratios to measure the strength of each association. The discriminant ability of the model for detecting hepatic malignancy was assessed using receiver operating characteristic curve analysis.
Results: The prevalence of hepatic malignancy in our study population was 64%. The area under the receiver operating characteristic curve for the logistic regression model was 0.82. Venous washout (odds ratio = 9.2), alpha-fetoprotein level (odds ratio = 3.2), and number of lesions (odds ratio = 1.5) were significant predictors for malignancy (p < 0.05). When arterial enhancement and venous washout were either both present or both absent, alpha-fetoprotein level contributed little to the prediction of malignancy.
Conclusion: The MR characteristics of hepatic lesions during the dynamic venous phase in conjunction with the serum alpha-fetoprotein level and number of lesions are predictors of hepatic malignancy. The use of these predictors can facilitate explicit estimation of malignancy in individuals with underlying cirrhosis, potentially improving clinical decision-making.