Diagnostic accuracy of MRI with extracellular vs. hepatobiliary contrast material for detection of residual hepatocellular carcinoma after locoregional treatment.

Abstract

To compare the diagnostic accuracy of extracellular gadolinium-based contrast-enhanced MRI (Gd-MRI) and gadoxetic acid-enhanced MRI (EOB-MRI) for the assessment of hepatocellular carcinoma (HCC) response to locoregional therapy (LRT) using explant correlation as the reference standard. Forty-nine subjects with cirrhosis and HCC treated with LRT who underwent liver MRI using either Gd-MRI (n = 26) or EOB-MRI (n = 23) within 90days of liver transplantation were included. Four radiologists reviewed the MR images blinded to histology to determine the size and percentage of viable residual HCC using a per-lesion explant reference standard. Sensitivities, specificities, accuracies, and agreement with histology for the detection residual HCC were calculated. Gd-MRI had greater agreement with histology (ICC: 0.98 [0.95-0.99] vs. 0.80 [0.63-0.90]) and greater sensitivity for viable HCC (76% [13/17 50-93%] vs. 58% [7/12; 28-85%]) than EOB-MRI; specificities were similar (84% [16/19; 60-97%] vs. 85% [23/27; 66-96%]). Areas under ROC curves for detecting residual viable tumor were 0.80 (0.64-0.92) for Gd-MRI and 0.72 (0.55-0.85) for EOB-MRI. Gd-MRI had greater inter-rater agreement than EOB-MRI for determining the size of residual viable HCC (ICC: 0.96 [0.92-0.98] vs. 0.85 [0.72-0.92]). Gd-MRI may be more accurate and precise than EOB-MRI for the assessment of viable HCC following LRT.