To assess the value of Hepatobiliary phase MRI (HPB-MRI) to differentiate FNH and HCA, and evaluate its impact on diagnostic accuracy, diagnostic confidence, inter-observer variability, and patient clinical management.
Forty-nine patients referred for Gd-BOPTA-enhanced MRI were retrospectively included in this IRB-approved study, with a total of 119 lesions-90 FNH and 29 HCA. Two observers separately assessed in 2 distinct randomized reading sessions the performance of MRI with (HBP-MRI) or without (conventional MRI) the use of HBP images. Each lesion was ranked with a 5-point scale (from 1 Typical FNH to 5 Certainly not a FNH). Sensitivity, specificity, overall accuracy, and inter-observer agreement for the differentiation of FNH from HCA were calculated and compared between conventional and HBP-MRI.
Both sensitivity (respective values of 38.9% and 97.8%), overall accuracy (respective values of 53.8% and 98.3%), and inter-observer agreement (respective values of Kappa 0.56 and 0.88) were significantly higher using HBP-MRI than with conventional MRI, with unchanged specificity (100%). The sensitivity of conventional MRI for the diagnosis of FNH was significantly lower in lesions ≤ 3 cm (20% vs. 88%). Overall, HBP could have changed lesion management in 59/119 cases (49.5%), including 53 FNH and 6 HCA with no impact in 60/119 lesions (50.5%) including all 35 lesions classified as scores 1 and 2 for the diagnosis of FNH.
The clinical impact of HBP-MRI is mostly important for smaller than 3-cm FNH, and more limited in larger FNH lesions as well as for HCA diagnosis for which conventional MRI is already accurate. The use of extracellular contrast agents upfront could limit the required use of linear HBP contrast agents for benign hepatocellular lesion characterization. On HBP, all FNH appeared hypointense compared to adjacent liver while close to 97% of HCA appeared hypointense.