A New Reporting System for Diagnosis of Hepatocellular Carcinoma in Chronic Hepatitis B With Clinical and Gadoxetic Acid-Enhanced MRI Features

Abstract

Background Current major guidelines for diagnosis of hepatocellular carcinoma (HCC) based on imaging findings are different from each other and do not include clinical risk factors as a diagnostic criteria. Purpose To developed and validated a new diagnostic score system using MRI and clinical features as applied in chronic hepatitis B patients. Study Type Retrospective observational study. Subject A total of 418 treatment-naive patients (out of 902 patients) with chronic hepatitis B having 556 lesions suspected for HCC which were eligible for curative treatment. Field Strength/sequence T1W GRE in- and opposed-phase, T2W FSE, DWI, and T1W 3D-GRE dynamic contrast-enhanced sequences at 1.5 T and 3 T. Assessment Six radiologists with 7-22 years of experience independently evaluated MR images based on Liver Imaging Reporting and Data System (LI-RADS) version 2018. Statistical Tests Based on logistic regression analysis of MRI features and clinical factors, a risk score system was devised in derivation cohorts (268 patients, 352 lesions) and externally validated (150 patients, 204 lesions). The performance of the new score system was assessed by Harell's c-index. Using cutoff value of 12, maintaining positive predictive value >= 95%, the diagnostic performances of the score system were compared with those of LR-5. Results The 15-point diagnostic scoring system used MRI features (lesion size, nonrim arterial phase hyperenhancement, portal venous phase hypointensity, hepatobiliary phase hypointensity, and diffusion restriction) and clinical factors (alpha-fetoprotein and platelet). It showed good discrimination in the derivation (c-index, 0.946) and validation cohorts (c-index, 0.907). Using a risk score of 12 as a cut-off, this system yielded higher sensitivity than LR-5 (derivation cohort, 76.8% vs. 52.1%; validation cohort, 73.4% vs. 49.5%) without significant decrease in specificity (derivation cohort, 93.1% vs. 97.2%, P = 0.074; validation cohort, 91.7% vs. 96.1%, P = 0.299). Data Conclusion A new score system showed improved sensitivity in chronic hepatitis B patients compared to LI-RADS without significant compromise in specificity. Evidence Level 3 Technical Efficacy Stage 2.