Diagnostic Accuracy of Extracellular versus Hepatobiliary Contrast-enhanced MRI in LI-RADS Nonradiation Treatment...

Background Whether the diagnostic accuracy of the Liver Imaging Reporting and Data System (LI-RADS) Treatment Response Algorithm (TRA) differs according to the type of MRI contrast agent has been unclear. Purpose To compare the diagnostic accuracy of LI-RADS Nonradiation TRA version 2024 for detecting viable hepatocellular carcinoma (HCC) after locoregional therapy (LRT) using either extracellular contrast agent (ECA) MRI or hepatobiliary contrast agent (HBA) MRI. Materials and Methods This retrospective study included consecutive patients who underwent nonradiation LRT for HCC and liver transplant between January 2020 and March 2024. Patients underwent post-LRT multiphasic MRI using ECA or HBA within 90 days before transplant. Two radiologists independently assigned LI-RADS TRA categories in accordance with the LI-RADS Nonradiation TRA version 2024. Explant pathology served as the reference standard. Propensity score matching was conducted to balance the baseline characteristics between the two groups. After matching, the sensitivity, specificity, and accuracy of LI-RADS TRA Viable category-with and without ancillary features (AFs)-were compared between ECA MRI and HBA MRI using cluster bootstrap resampling. Results A total of 216 patients (mean age, 60 years ± 6.7 [SD]; 171 men; 77 underwent ECA MRI and 139 underwent HBA MRI) with 414 treated observations were included, and 143 matched pairs of treated observations were analyzed. Without AFs, ECA MRI showed higher sensitivity for helping detect viable HCC than HBA MRI (74.0% [54 of 73 observations; 95% CI: 63.6, 84.2] vs 57.5% [42 of 73 observations; 95% CI: 45.8, 68.5]; P = .02). With AFs, sensitivity improved for both contrast agents, with no evidence of a difference between groups (76.7% [56 of 73 observations; 95% CI: 66.7, 86.3] vs 67.1% [49 of 73 observations; 95% CI: 56.0, 77.5]; P = .16). The impact of AFs on sensitivity was more pronounced in HBA MRI (57.5% [42 of 73 observations; 95% CI: 45.8, 68.5] vs 67.1% [49 of 73 observations; 95% CI: 56.0, 77.5] with AFs; P = .002) than ECA MRI (74.0% [54 of 73 observations; 95% CI: 63.6, 84.2] vs 76.7% [56 of 73 observations; 95% CI: 66.7, 86.3] with AFs; P = .25). Conclusion ECA MRI demonstrated higher sensitivity than HBA MRI when AFs were not applied, whereas using AFs resulted in comparable sensitivities between groups. © RSNA, 2026 Supplemental material is available for this article.