Evaluating Accuracy of LI-RADS Nonradiation Treatment Response Algorithm v2024 and Ancillary Features at Hepatobiliary MRI versus CT.

Abstract

Background Liver Imaging Reporting and Data System (LI-RADS) Nonradiation Treatment Response Algorithm (TRA) version 2024 (v2024) was recently introduced following version 2017 (v2017) to standardize the assessment of hepatocellular carcinoma (HCC) after nonradiation local-regional therapy (LRT). However, a comparison of their accuracy across imaging modalities has not yet been performed. Purpose To compare the diagnostic accuracy of LI-RADS Nonradiation TRA v2024 with that of LI-RADS TRA v2017 for assessing HCC viability after LRT and to evaluate the incremental value of ancillary features (AFs) and accuracy of CT versus hepatobiliary agent (HBA)-enhanced MRI. Materials and Methods This retrospective study included patients who underwent nonradiation LRT for HCC followed by HBA MRI, or HBA MRI and CT, and then hepatic surgery between January 2018 and December 2021. Three radiologists independently applied LI-RADS Nonradiation TRA v2024 (with and without AFs) and v2017 to the largest treated lesion and reached a consensus. Surgical pathologic findings (incomplete or complete necrosis) were the reference standard. Diagnostic accuracy metrics were compared with Bonferroni-adjusted McNemar tests; interobserver agreement was assessed with Fleiss κ. Results This study included 245 patients (mean age, 60 years ± 9.3 [SD]; 199 men). At HBA MRI, LI-RADS Nonradiation TRA v2024 with AFs demonstrated the highest diagnostic sensitivity (79.3%) and accuracy (81.6%), compared with v2024 without AFs (sensitivity, 65.8%; accuracy, 71.8%) and v2017 (sensitivity, 67.4%; accuracy, 73.1%) (all <i>P</i> < .001) and preserved specificity (both <i>P</i> = .47). Among 194 patients who underwent both CT and MRI, HBA MRI (v2024 with AFs) achieved higher sensitivity (78.7%) and accuracy (81.4%) than CT (64.0% and 70.6%, respectively; <i>P</i> = .001 and <i>P</i> = .002, respectively), with no evidence of a difference in specificity (90.9% vs 93.2%; <i>P</i> > .99). HBA MRI showed higher accuracy than CT after transarterial chemoembolization (82.7% vs 66.1%; <i>P</i> < .001) but not after ablation (both 79.1% [53 of 67 patients]; <i>P</i> > .99). Conclusion LI-RADS Nonradiation TRA v2024, particularly when AFs were incorporated, improved sensitivity and accuracy for detecting viable HCC after nonradiation LRT compared with v2017, and HBA MRI showed higher sensitivity and accuracy over CT. © RSNA, 2026 <i>Supplemental material is available for this article.</i> See also the editorial by Kang in this issue.