LI-RADS v2018 versus KLCA-NCC v2022: comparison of probability-based HCC categories.

Abstract

Objectives: To compare the diagnostic performance of corresponding probability-based categories between LI-RADS v2018 and Korean Liver Cancer Association-National Cancer Center (KLCA-NCC) v2022.

Materials and methods: This retrospective multicenter study included treatment-naïve patients at risk of hepatocellular carcinoma (HCC) who underwent gadoxetic acid-enhanced MRI between January 2015 and June 2018. Four radiologists independently evaluated MRI and CT when available; observations were categorized according to both criteria. Reference standards were pathology or clinical diagnosis. Generalized estimating equations were used for lesion clustering.

Results: The study included 2237 patients (1666 men; mean age, 59 ± 11 years) with 2445 lesions. On MRI, LR-5 showed a higher positive predictive value (PPV) than KLCA-NCC "definite HCC" (93.9% vs. 91.7%, p < 0.001). PPV was not different between LR-4 and KLCA-NCC "probable HCC" (68.2% vs. 64.1%, p = 0.141). HCC proportion was 13.2% in LR-3 and 40.8% in KLCA-NCC "indeterminate nodule" (p < 0.001). In 1455 patients with 1590 lesions on CT, LR-5 and KLCA-NCC had comparable PPV (92.9% in each). PPV of KLCA-NCC "probable HCC" was higher than LR-4 (83.8% vs. 73.9%, p = 0.012). HCC proportion was 52.0% in LR-3 and 46.2% in KLCA-NCC "indeterminate nodule" (p = 0.001).

Conclusion: Both LI-RADS v2018 and KLCA-NCC v2022 demonstrated effective probability-based risk stratification, but corresponding probability categories were not interchangeable. On CT, category performance differed from MRI, highlighting the need for modality-specific refinement to improve diagnostic precision.

Key points: Question Do conceptually corresponding diagnostic categories in LI-RADS v2018 and KLCA-NCC v2022 demonstrate equivalent diagnostic performance for HCC? Findings Both systems demonstrated effective probability-based stratification, but corresponding categories were not equivalent, and the performance was modality-specific. Clinical relevance Corresponding categories are not interchangeable between the two systems. The heterogeneity within "probable HCC" and modality-specific performance differences suggest potential benefits from subcategorization and refinement of CT-specific criteria.