Accuracy of the diagnostic evaluation of hepatocellular carcinoma with LI-RADS.

Background There are few studies about the Liver Imaging Reporting and Data System (LI-RADS), which was developed with the purpose of standardizing the interpretation and reporting of liver imaging examinations in patients at risk for hepatocellular carcinoma (HCC). Purpose To evaluate the diagnostic accuracy of HCC diagnosis using LI-RADS. Material and Methods The computed tomography (CT), magnetic resonance imaging (MRI), and clinical data of 297 lesions in 249 patients between June 2012 and August 2013 were retrospectively analyzed. Using LI-RADS 2014, two radiologists evaluated the lesions and a LI-RADS category was retrospectively assigned to each nodule. Results The final diagnoses of 297 nodules in 249 patients consisted of 191 malignant and 106 benign lesions. Out of 44 LI-RADS category 1 lesions, none were HCCs. However, 2/25 category 2 lesions, 3/35 category 3 lesions, 16/25 category 4 lesions, 151/156 category 5 lesions, and 3/12 category LRM/OM (probable malignancy, not specific for HCC/other malignancy) lesions were HCCs. The Kappa value was 0.44 (95% confidence interval [CI] = 0.39-0.49) between two observers during LI-RADS grading. Conclusion The negative predictive value of LI-RADS category 1 was 100%. In addition, a relevant proportion of lesions categorized as category 2 or 3, or even as other malignancies, were HCCs. LI-RADS category 5 had a high specificity for HCC. LI-RADS was not able to give a differential diagnosis for the false-positive lesions of LI-RADS category 5.