Noncontrast Abbreviated MRI for Post-TACE Treatment Response Monitoring of Hepatocellular Carcinoma Based on Ancillary Features from LI-RADS.

Abstract

Background Posttreatment imaging evaluation after local-regional treatment for hepatocellular carcinoma (HCC) is essential for detecting recurrence and guiding retreatment. Purpose To assess the utility of noncontrast abbreviated MRI (NC-AMRI) using ancillary features from the Liver Imaging Reporting and Data System (LI-RADS) algorithms for detecting intrahepatic recurrence of HCC during treatment response monitoring after transarterial chemoembolization (TACE). Materials and Methods This single-center retrospective study included patients with HCC who underwent TACE followed by gadoxetic acid-enhanced liver MRI between January 2020 and May 2023. NC-AMRI simulated from full-sequence MRI was assessed using two ancillary features (mild-moderate T2 hyperintensity or restricted diffusion) permitted in the LI-RADS CT/MRI nonradiation Treatment Response Assessment algorithm version 2024 and LI-RADS CT/MRI diagnostic algorithm version 2018, whereas full-sequence liver MRI was interpreted with complete criteria (LI-RADS treatment response [LR-TR] Viable category for treated lesions; LR-5 or hypervascular LR-4 or LR-M for new lesions). Diagnostic accuracy for detecting viable tumors (per lesion) and overall intrahepatic recurrence (per patient) was evaluated for both NC-AMRI and full-sequence MRI. Results Among 195 patients with 267 TACE-treated lesions (mean age, 64 years ± 9.8 [SD]; 143 men), 92 patients (47.2%) had intrahepatic recurrence, including 76 (39.0%) with viable tumors in 81 treated lesions and 31 patients (15.9%) with newly developed tumors, whereas 15 patients (7.7%) had both. For viable tumors in treated lesions, compared with full-sequence MRI, NC-AMRI demonstrated both lower per-lesion sensitivity (76.5% [62 of 81] vs 87.7% [71 of 81]; <i>P</i> = .04) and specificity (88.7% [165 of 186] vs 96.2% [179 of 186]; <i>P</i> = .004). For overall intrahepatic recurrence per patient, compared with full-sequence liver MRI, NC-AMRI demonstrated both lower sensitivity (78.3% [72 of 92] vs 92.4% [85 of 92]; <i>P</i> = .004) and specificity (79.6% [82 of 103] vs 95.1% [98 of 103]; <i>P</i> < .001). Conclusion NC-AMRI demonstrated moderate diagnostic accuracy for post-TACE treatment response monitoring in patients with HCC, including viable tumors in treated lesions and newly developed tumors, but lower diagnostic accuracy than full-sequence liver MRI. © RSNA, 2026 <i>Supplemental material is available for this article.</i> See also the editorial by Kozaka in this issue.