Hepatocyte-Specific Magnetic Resonance Imaging–Based Assessment of Indeterminate Hepatic Nodules in the Liver Transplant Evaluation of Patients With Cirrhosis

Abstract

We aimed to determine the identities in explants of indeterminate hepatic nodules (IDNs) that had been scanned by dynamic magnetic resonance imaging (MRI) to establish clinicoradiological parameters predicting which IDNs were hepatocellular carcinomas (HCCs). This study included 88 patients with cirrhosis who underwent gadoxetic acid–enhanced MRI in pre–liver transplantation (LT) workup followed within 90 days by primary LT. The MRI detected 168 hepatic nodules that were classified into 6 benign tumors, 49 HCCs, and 113 IDNs, in 5, 34, and 72 patients, respectively. We compared these pre-LT radiologic diagnoses and stagings with explant pathology on a per-lesion basis to enable us to identify features of IDNs related to malignancy. Of the 168 nodules seen on MRI, 119 that were classified radiologically as consisting of 1 benign nodule (33.3%), 46 HCCs (93.9%), and 72 IDNs (63.7%) all turned out to be pathological HCCs. Of 32 patients inside Milan and 54 without HCC staged by MRI, 11 progressed beyond the criteria after LT. High serum alpha-fetoprotein level (≥20 ng/mL) was the only per-patient factor significantly associated with malignant IDNs. Per-tumor analysis of the MRI signals revealed that arterial hyperintensity, hepatobiliary hypointensity, T2-weighted mild-to-moderate intensity, and restricted diffusion-weighted images were significantly correlated with malignant IDN. A model combining these 4 MRI factors with alpha-fetoprotein level had the best performance in predicting the identification of IDNs as HCCs in explanted livers. Over 60% of the IDNs seen on dynamic images of cirrhotic livers proved to be HCCs when explanted livers were examined. It may therefore be possible to identify HCCs with reasonable accuracy by means of their hepatocyte-specific MRI features when patients are being assessed for LT.