This study aimed to evaluate a non–gadolinium-enhanced magnetic resonance imaging (MRI) protocol including T2-weighted, T2*-weighted, and diffusion-weighted MRI sequences for identifying hepatocellular carcinoma (HCC) with liver explantation as the reference standard. Also, a stand-alone pre– and dynamic post–gadolinium-enhanced liver MRI data set was interpreted from the available patient data for relative comparison purposes.
A retrospective review identified 37 appropriately selected liver transplant patients who had had preoperative MRI. Two data sets were created from the MRI studies: (1) non–gadolinium-enhanced (including T2-weighted, T2*-weighted, and diffusion-weighted sequences) and (2) pre– and dynamic post–gadolinium-enhanced (3-dimensional T1-weighted gradient recalled echo) and were presented to 2 independent, blinded observers. A separate blinded observer assessed the pathologic results from liver explantation to establish the reference standard.
On explant pathology, 21 of 37 patients had 31 HCC (mean [SD] largest diameter, 19  cm; range, 7–40 mm). Per-lesion sensitivity of non–gadolinium-enhanced MRI for identifying HCC was 52% (reader 1) and 55% (reader 2), and specificity was 90% (reader 1) and 88% (reader 2). Per-lesion sensitivity of the stand-alone pre– and dynamic post–gadolinium-enhanced MRI was 84% (reader 1) and 81% (reader 2), and specificity was 62% (reader 1) and 65% (reader 2).
Non–gadolinium-enhanced MRI had a moderate sensitivity for HCC but had a high specificity. Although non–gadolinium-enhanced MRI cannot be recommended as a primary imaging approach for HCC, the results demonstrate the contribution of non–gadolinium-enhanced sequences to imaging of HCC. A non–gadolinium-enhanced MRI protocol may have a diagnostic value when gadolinium cannot be administered.
From the Department of Radiology and Radiological Science, Medical University of South Carolina, Charleston, SC.
Received for publication June 3, 2011; accepted August 23, 2011.
Reprints: Andrew D. Hardie, MD, Department of Radiology and Radiological Science, Medical University of South Carolina, 25 Courtenay Dr, MSC 226, Charleston SC 29401 (e-mail: firstname.lastname@example.org).
The authors report no conflicts of interest.