Interobserver agreement of LR-TRA 2017 and 2024 and mRECIST criteria on MRI after locoregional therapy: Impact of ancillary features and reader experience
Year: 2026
Type: preprint
Abstract: Purpose: To compare interobserver agreement of the Liver Imaging Reporting and Data System Treatment Response Algorithm (LR-TRA) v2017 and v2024, with and without ancillary features, and of modified Response Evaluation Criteria in Solid Tumors (mRECIST) for hepatocellular carcinoma (HCC) after nonradiation locoregional therapy, and to assess the effects of reader experience, ancillary features, and treatment type. Methods: In this retrospective, single-center study, six radiologists, four of them subspecialized in abdominal imaging, independently classified 100 treated HCC lesions on MRI using LR-TRA v2017, v2024 without and with ancillary features, and mRECIST. Agreement was assessed with Fleiss' kappa globally and by category, reader group, and treatment type. Results: Global agreement was moderate for LR-TRA v2017 (κ = 0.508), v2024 without (κ = 0.538) and with ancillary features (κ = 0.554), and substantial for mRECIST (κ = 0.727). Among subspecialized readers, agreement increased for all LR-TRA variants (κ = 0.639–0.646; p ≤ 0.016) but not for mRECIST (p = 0.966). The equivocal category was the least reproducible and the most dependent on reader experience (all p ≤ 0.025). Ancillary features reduced equivocal classifications by 44.4%, with restricted diffusion present in 89.3% of reclassified observations, although agreement on reclassification was low (κ = 0.281). A treatment-related difference in agreement observed with v2017 was absent with v2024. Conclusion: LR-TRA v2024 showed numerically higher interobserver agreement than v2017, but reproducibility remained moderate. Reader experience had a greater influence on agreement than the LR-TRA version, particularly for the equivocal category. Ancillary features reduced equivocal classifications without improving agreement within this category.
Source: Research Square
Language: English
Subfield: Hepatology
Field: Medicine
Domain: Health Sciences
Open Access status: green
OpenAlex ID: W7236618988
