New MR imaging criteria with a diffusion-weighted sequence for the diagnosis of hepatocellular carcinoma in chronic liver diseases

Background & Aims: To propose MRI criteria with a diffusion-weighted imaging (DWI) sequence for the diagnosis of hepatocellular carcinoma (HCC). Methods: Patients, who underwent liver MRI with contrast-enhanced sequences and DWI between 2004 and 2008 and who had at least one confirmed HCC of at least 10 mm, were included. Index diagnostic criteria were: (1) enhancement in the arterial-dominant phase and washout in the portal venous and/or equilibrium phases; (2) enhancement in the arterial-dominant phase and hyperintensity on DWI; (3) enhancement in the arterial-dominant phase and washout in the portal venous and/or equilibrium phases or hyperintensity on DWI. Two radiologists independently reviewed the corresponding sets of sequences (DWI alone; T1-weighted sequence before and after dynamic injection of gadolinium chelates; combined DWI-T1-weighted sequence). Interobserver agreement and sensitivity were determined per nodule. Results: Ninety-one patients were included (109 HCCs). The sensitivity of conventional MRI criteria for the diagnosis of HCC was 59.6% for both radiologists. The sensitivity of enhancement in the arterial-dominant phase and hyperintensity on DWI was 77.1% or 76.1%, depending on the radiologist. The sensitivity of enhancement in the arterial-dominant phase and washout in the portal venous and/or equilibrium phases or hyperintensity on DWI was 84.4% or 85.3%, depending on the radiologist. The inter-observer agreement for the latter was very good (kappa coefficient 0.82). These results were consistent in HCCs smaller than 20 mm. Conclusions: The proposed criteria, based on the characteristics of lesions after gadolinium chelate administration and hyperintensity on DWI, significantly increased the sensitivity for the diagnosis of HCC compared to conventional criteria, regardless of tumor size. (C) 2010 European Association for the Study of the Liver. Published by Elsevier B.V. All rights reserved.