Enhancing Capsule on Dynamic Liver CT: Association With Hepatocellular Carcinoma Diagnosis Using LI-RADS Version 2018

BACKGROUND. The LI-RADS version 2018 (v2018) diagnostic algorithm prioritizes an enhancing capsule differently from other major features. The role of an enhancing capsule in hepatocellular carcinoma (HCC) diagnosis remains uncertain, particularly for CT. OBJECTIVE. The purpose of this study was to compare the diagnostic performance of CT for diagnosis of HCC using original LI-RADS v2018 criteria for LR-5 versus modified LR-5 criteria that treat an enhancing capsule the same as other major features. METHODS. This retrospective study included 546 patients (432 men and 114 women; mean age, 59.8 ± 9.6 [SD] years) at risk for HCC who underwent dynamic liver CT within 1 month before partial hepatectomy or liver transplant between January 2022 and December 2023. Two radiologists independently assessed hepatic observations (n = 645) for LI-RADS major features; discrepancies were resolved through a consensus discussion with a third radiologist. LI-RADS categories were assigned using original LR-5 criteria and modified criteria giving an enhancing capsule the same priority as other major features (nonperipheral washout and threshold growth). The reference standard was pathologic assessment or 2 or more years of imaging follow-up. Criteria were compared using generalized estimating equations. RESULTS. For reader 1, sensitivity, specificity, PPV, NPV, and accuracy were, respectively, 72.7%, 89.2%, 96.8%, 43.0%, and 75.2% with the use of original LR-5 criteria and 75.7%, 88.4%, 96.7%, 45.5%, and 77.5% with the use of modified criteria, respectively. For reader 2, these values were, respectively, 72.8%, 91.1%, 97.3%, 43.6%, and 75.6% with the use of original criteria and 75.0%, 90.3%, 97.1%, 45.4%, and 77.3% with the use of modified criteria. For both readers, sensitivity and accuracy were significantly higher for modified criteria than for original criteria (p < .05); specificity, PPV, and NPV were not significantly different between the criteria (p > .05). Among observations measuring 10-19 mm with nonrim arterial phase hyperenhancement, when consensus assessments were used, sensitivity, specificity, PPV, NPV, and accuracy were 52.5%, 84.0%, 95.5%, 19.4%, and 56.3%, respectively, when original LR-5 criteria were used and 64.9%, 78.3%, 95.1%, 23.2%, and 66.4%, respectively, when modified criteria were used. CONCLUSION. When LI-RADS v2018 was applied to CT, treating an enhancing capsule the same as other major features provided significantly increased sensitivity for HCC detection without a significant change in specificity or PPV. CLINICAL IMPACT. Treating enhancing capsule the same as other major features for CT-based LI-RADS could improve noninvasive HCC diagnosis while simplifying interpretations.