Preoperative prediction of pathologic grade of HCC on gadobenate dimeglumine-enhanced dynamic MRI

Purpose To evaluate the value of gadobenate dimeglumine-enhanced MRI in predicting the pathologic grade of hepatocellular carcinoma (HCC). Materials and methods Patients with pathologically proven HCC who underwent preoperative gadobenate dimeglumine-enhanced dynamic MRI were included. Two radiologists blinded to pathology results evaluated images in consensus. Lesions were evaluated quantitatively in terms of ratio of enhancement (RE), and qualitatively based on image features related to tumor aggressiveness. Logistic regression and ROC analyses were used to determine the value of these parameters to predict pathologic grade. Results In total, 221 patients (194 males, 27 females, aged 52.9 +/- 11.7 years) with 49 poorly differentiated HCCs and 172 well/moderately differentiated HCCs were evaluated. Features significantly related to poorer pathologic grade at univariate analysis included lower RE in the early arterial phase (EAP) (p = 0.001), nonsmooth margins (p = 0.001), absence of capsule (p < 0.001), arterial peritumoral hyperenhancement (p < 0.001), higher AFP (p = 0.004), multiple tumors (p = 0.026), and larger tumor size (p = 0.028). At multivariate analysis, lower RE (EAP) (OR = 0.144, p = 0.002), absence of capsule (OR = 0.281, p = 0.004), and arterial peritumoral hyperenhancement (OR = 4.117, p < 0.001) were independent predictive factors for poorer pathologic grade. ROC analysis showed lower RE (EAP) was predictive of poorer pathologic grade (AUC = 0.667). AUC increased to 0.797 when combined with absence of capsule and presence of peritumoral hyperenhancement. Conclusions Lower RE (EAP), absence of capsule, and arterial peritumoral hyperenhancement were predictive biomarkers for poorer pathologic grade of HCC on gadobenate dimeglumine-enhanced dynamic MRI.