Evaluation of an abbreviated screening MRI protocol for patients at risk for hepatocellular carcinoma

Abstract

Purpose: In this study, we compare an abbreviated screening MRI protocol (aMRI), utilizing only dynamic contrast-enhanced images, to a conventional liver MRI (cMRI) for the characterization of observations in at-risk patients.

Materials and methods: 164 consecutive HCC screening MRIs were retrospectively analyzed. Two sets of de-identified image sets were created: one with all acquired sequences including T2- and diffusion-weighted sequences (cMRI), and one with only T1-weighted precontrast and dynamic post-contrast images utilizing an extracellular gadolinium contrast agent (aMRI). Three readers assigned a LI-RADS score based on the lesion with the highest LI-RADS category using the aMRI and cMRI datasets during separate reads.

Results: There was no change between the aMRI and cMRI LI-RADS categorization in 93%, 96%, and 96% of cases for readers 1, 2, and 3, respectively. In the majority of the discrepant cases, the score increased from LI-RADS 3 to LI-RADS 4 due to the presence of ancillary features on T2 and DWI. Kappa values for interobserver variability demonstrated fair-to-moderate LI-RADS agreement among the 3 readers.

Conclusion: There was strong agreement between the abbreviated T1-only MRI protocol and a full liver MRI, with only 5% of cases changing LI-RADS categorization due to the inclusion of T2 and DWI. The estimated time to run this abbreviated MRI is approximately 7-10 min, possibly allowing for a more cost-effective screening MRI than our cMRIs.

Keywords: Hepatocellular carcinoma; LI-RADS; MRI; Screening.

Conflict of interest statement

Conflict of interest: Jennifer Y. Lee, Eugene J. Huo, Stefanie Weinstein, Charmaine Santos, Alexander Monto, Carlos U. Corvera, Judy Yee and Thomas A. Hope declare that they have no conflict of interest.

Figures

Figure 1:

65-year-old male with hepatitis C and elevated AFP with a 2.7 cm arterially enhancing lesion in segment 5 (A, arrow) with washout (B, arrow) and capsule (C, arrow), meeting LI-RADS 5 criteria. Ancillary features including faint T2 hyperintensity (D, arrow) and moderate diffusion restriction, as evidenced by hyperintensity on the high-B-value diffusion-weighted images (E, arrow) with low value on the corresponding ADC map (F), which did not affect the LI-RADS categorization.

Figure 2:

65-year-old male with history of hepatitis C and alcohol abuse with a 1.5 cm lesion in the mid right hepatic lobe (arrow) that demonstrates arterial phase hyperenhancement (A and B, arrow) without evidence of washout (C, arrow), which would be categorized as LI-RADS 3 on an abbreviated protocol (aMRI). However, the presence of T2 hyperintensity (D, arrow) and mild diffusion restriction (E and F, arrow) upgrades the lesion to LI-RADS 4 on cMRI.

Figure 3:

Breakdown by reader of percent of studies categorized as either LI-RADS 1 and LI-RADS 2 combined (black) are compared with LI-RADS 3, 4 and 5 combined (light gray), demonstrating strong agreement between aMRI and cMRI for all three readers (kappa = 0.51 and 0.49 for aMRI and cMRI, respectively).