Rate of observation and inter-observer agreement for LI-RADS major features at CT and MRI in 184 pathology proven hepatocellular carcinomas

Purpose: To compare frequency and inter-reader agreement for LI-RADS v2014 major features at CT vs. MRI in pathology-proven cases of hepatocellular carcinoma. Methods: Pathology reports and imaging studies from patients having undergone liver transplant or hepatectomy for hepatocellular carcinoma were reviewed. Size, location, washout, and capsule appearance for each lesion were recorded by two radiologists. Cohen's kappa and intraclass correlation coefficients (ICC) were calculated. Results: One hundred and thirty-four patients with 184 tumors were reviewed. Seventy-seven percentage of lesions were imaged by CT and 23% by MRI. No lesions were evaluated with both modalities. Mean lesion diameter was 2.6 +/- 1.3 cm (ICC = 0.92). Arterial phase hyperenhancement was seen in 86% of lesions (kappa = 0.75). Washoutwas seen in 82% of studies (kappa = 0.61). Arterial phase hyperenhancement and washout were seen equally at CT and MRI (p = 1.00 and 0.46, respectively). Capsule was infrequently observed (27%) but was seen more commonly at MRI (44%) than at CT (17%) with p = 0.002 and (kappa = 0.56). Forty-seven percent of lesions with at least one prior study met LI-RADS criteria for threshold growth. The rates of LI-RADS categories 3, 4, and 5 were 9%, 37%, and 54%, respectively. More 1-2 cm LI-RADS 5 lesions were seen at MRI (43%) than atCT (8%), p = 0.01. Conclusion: A combined LI-RADS 4/5 group was 91% sensitive for hepatocellular carcinoma. Arterial enhancement and washout were seen more frequently than capsule, the sole finding seen more frequently at MRI than at CT. Inter-reader reliability was substantial for arterial hyperenhancement and washout but moderate for capsule. Capsule remains an important finding in small arterially enhancing lesions (1-2 cm) which require a second major criterion to upgrade to a LI-RADS 5 lesion.