Critical analysis of major and ancillary features of LI-RADS v2018 in the differentiation of small (< 2 cm) hepatocellular carcinoma from dysplastic nodules with gadobenate dimeglumine-enhanced magnetic resonance imaging

OBJECTIVE: To evaluate the performance of major features, ancillary features, and categories of Liver Imaging Reporting and Data System (LI-RADS) version 2018 at magnetic resonance (MR) imaging in the differentiation of small hepatocellular carcinoma (HCC) from dysplastic nodules (DNs). PATIENTS AND METHODS: This retrospective study included cirrhotic patients with pathologically proven untreated HCCs and DNs (&lt; 2 cm) and liver MR imaging performed with gadobenate dimeglumine contrast agent within 3 months before pathological analysis, between 2015 and 2018. 37 patients with 43 observations (17 HCCs and 26 DNs) met the inclusion criteria. Two radiologists assessed major and ancillary imaging features for each liver observation and assigned a LI-RADS v2018 category in consensus. Estimates of diagnostic performance of major features, ancillary features, and LI-RADS categories were assessed based on their sensitivity, specificity, positive (PPV), and negative predictive values (NPV). RESULTS: Major features (nonrim arterial phase hyperenhancement, nonperipheral "washout", and enhancing "capsule") had a sensitivity of 94.1%, 88.2%, and 41.2%, and a specificity of 57.7%, 42.3%, and 88.5% for HCC, respectively. Ancillary features (hepatobiliary phase hypointensity, mild-moderate T2 hyperintensity, restricted diffusion, and fat in the lesion more than adjacent liver) had a sensitivity of 94.1%, 64.7%, 58.8%, and 11.8%, and a specificity of 26.9%, 61.5%, 65.4%, and 76.9% for HCC, respectively. The LR-5 category (determined by using major features only vs. the combination of major and ancillary features) had a sensitivity of 88.2% at both evaluations and a specificity of 76.9% and 80.8% for HCC, respectively. The combination of LR-4, LR-5 categories (determined by using major features only vs. the combination of major and ancillary features) had a sensitivity of 94.1% at both interpretations and a specificity of 65.4% and 26.9% for HCC, respectively. The use of ancillary features modified LI-RADS category in 25.6% of observations (11/43), predominantly upgraded from LR-3 to LR4 (10/11), increasing the proportion of low-grade DNs and high-grade DNs categorized as LR-4 (from 15.4% to 61.5% and from 7.7% to 46.1%, respectively). CONCLUSIONS: The added value of ancillary features in combination with major features is limited for the non-invasive diagnosis of small HCC; however, their use modifies the final category in a substantial proportion of observations from LR-3 to LR-4, thus allowing possible changes in the management of patients at risk for HCC.