Combination of CT/MRI LI-RADS with CEUS can improve the diagnostic performance for HCCs

Abstract

Background & AimsTo compare the classifications for focal liver lesions smaller than 3 cm by the CEUS and CT/MRI LI-RADS and explore the discrepancy between the two classifications strategies. MethodsFocal liver lesions with a size smaller than 3 cm undergoing CEUS and CT/MRI (MR required on a 3.0 T magnet) within 1 month were enrolled. Each nodule was categorized according to the CEUS LI-RADS v2017 (usLI-RADS) and CT/MRI LI-RADS v2018. Intermodality agreement between the usLI-RADS and LI-RADS for CT/MRI was assessed. The reasons for inconsistent classifications by the CEUS and CT/MRI LI-RADS were clarified. ResultsA total of 213 lesions were included. The positive predictive value (PPV) for HCCs of LR-3, 4 and 5 were 57.1% (95% CI: 32.6–78.6), 75.0% (95% CI: 62.3–84.5) and 94.2% (95% CI: 88.9–97.0) for CT/MRI and 37.5% (95% CI: 18.5–61.4), 56.0% (95% CI: 37.1–73.3) and 97.9% (95% CI: 94.1–99.3) for CEUS, respectively. CECT/MRI and CEUS LI-RADS had a poor agreement in classification with a kappa value of 0.254 (P < 0.001). 84 lesions (39.4%) were classified inconsistently in CT/MRI and CEUS LI-RADS. 40 lesions classified as LR-3 or LR-4 by CT/MRI were upgraded by CEUS. 21 lesions classified as LR-4 or LR-5 by CT/MRI were downgraded by CEUS. We reclassified the lesion in CT/MRI LR-3 and LR-4 by CEUS. The PPVs for HCCs in reclassified LR-3, LR-4 and LR-5 were 27.3% (95% CI: 9.7–56.6), 41.7% (95% CI: 26.2–69.0) and 94.2% (95% CI: 89.3–97.0), respectively, which decreased the incidences of HCCs in CT/MRI LR-3 and LR-4. ConclusionLesions in CT/MRI LR-3 and 4 have a higher probability of being HCCs than those in usLR-3 and 4, respectively. Reclassification of lesions in CT/MRI LR-3 and 4 using CEUS can help the corresponding HCCs get a definite diagnosis.