Gd-EOB-DTPA-enhanced MR imaging features of hepatocellular carcinoma in non-cirrhotic liver

To evaluate clinical, pathological and gadolinium-ethoxybenzyl-diethylenetriamine pentaacetic acid-enhanced magnetic resonance imaging (Gd-EOB-DTPA-enhanced MRI) findings of hepatocellular carcinoma (HCC) in non-cirrhotic livers and compare with HCC in cirrhotic livers.

This retrospective study included patients with pathologically confirmed HCC who underwent preoperative Gd-EOB-DTPA-enhanced MRI between January 2015 and October 2021. Propensity scores were utilized to match non-cirrhotic HCCs (NCHCCs) patients with cirrhotic HCCs (CHCCs) patients. The clinical, pathological and MR imaging features of NCHCCs were compared with CHCCs. Correlation between these features and the presence of NCHCCs were analyzed by logistic regression analysis. The predictive efficacy was evaluated using receiver operating characteristic (ROC) analysis. The area under the receiver operating characteristic curve (AUC) was used to compare performance, and the Delong test was used to compare AUCs.

After propensity score matching (1:3), a total of 144 patients with HCCs (36 NCHCCs and 108 CHCCs) were included. NCHCCs were larger in tumor size than CHCCs (P < 0.001, Cohen's d = 0.737). NCHCCs were more common in patients who have hepatitis C (5.6 % vs 1.9 %, P > 0.05) or have no known liver disease (11.1 % vs 0.9 %, P = 0.004), while hepatitis B was more common in CHCC patients (83.3 % vs 97.2 %, P = 0.003). Compared with CHCCs, NCHCCs more frequently demonstrated non-smooth tumor margin (P = 0.001, Cramer's V = 0.273), peri-tumoral hyperintensity (P < 0.05, Cramer's V = 0.185), hyperintense and heterogeneous signals in hepatobiliary phase (HBP) (P < 0.05). CHCCs were more likely to have satellite nodules compared to NCHCCs (33.3 % vs 57.4 %, P < 0.05, Cramer's V = 0.209). Based on the univariate and multivariate logistic regression analysis, the tumor size, non-smooth tumor margin, heterogeneous intensity in HBP and satellite nodule were significantly correlated to NCHCCs (P all <0.05). ROC curve analysis demonstrated that tumor size and non-smooth tumor margin were potential imaging predictors for the diagnosis of NCHCC, with AUC values of 0.715 and 0.639, respectively. The combination of the two imaging features for identifying NCHCC achieved an AUC value of 0.761, with a sensitivity of 0.889 and a specificity of 0.630.

NCHCCs were more likely to show larger tumor size, non-smooth tumor margin, peri-tumoral hyperintensity, as well as hyperintense and heterogeneous signals in HBP at Gd-EOB-DTPA-enhanced MR imaging compared with NCHCCs. Tumor size and non-smooth tumor margin in HBP may help to discriminate NCHCCs.