Analysis of comparative performance of CEUS and CECT/MR LI-RADS classification: Can CEUS dichotomize LI-RADS indeterminate lesions on CT or MRI?

PURPOSE

To study the comparative performance of contrast-enhanced ultrasound (CEUS) and contrast-enhanced CT or MRI (CECT/MR) in evaluating liver lesions using the LI-RADS guidelines.

METHODS

Retrospective analysis of radiology database from July 2010 to April 2017 revealed 228 patients who had CECT/MR and CEUS. Patients at risk of hepatocellular carcinoma (HCC), had contemporaneous CEUS and CECT/CEMR studies within 3 months and adequate follow up were included; reviewed (2 reviewers) and graded according to the 2017 CEUS and 2018 CECT/MR LI-RADS guidelines. Reference standard was multidisciplinary clinical decisions, histology or follow-up imaging.

RESULTS

The study cohort consisted of 45 patients with 46 lesions. HCC were significantly larger than non-malignant (mean sizes of 2.5 and 1.4 cm, respectively, p<0.001). Intraclass correlation coefficient for CEUS review (0.941) was higher than of CECT/MR review (0.643). Mean area-under-ROC curve (AUC) for CEUS (0.994) was significantly higher than of CECT/MR (0.760) for all lesions (p=0.01). For lesions scored LR-3 by CECT/MR, the AUC was significantly higher for CEUS (0.978) than CECT/MR (0.500) (p<0.001). Twenty-one (of 27) lesions, classified LR-3 or LR-4 by CECT/MR were upgraded by CEUS and 20 were found to be HCC. Six lesions that were LR-3 on both CECT/MR and CEUS were found to be non-malignant. There was good concordance for LR-5 lesions between both techniques.

CONCLUSION

CEUS is useful for reassessment of lesions with intermediate probability (LR-3) or probable for HCC (LR-4) on CECT/MR. Lesions upgraded by CEUS tend to be HCC. Lesions that remain LR-3 on CEUS tend to be non-malignant.