MR versus CEUS LI-RADS for Distinguishing Hepatocellular Carcinoma from other Hepatic Malignancies in High-Risk Patients

Abstract

Distinguishing between other hepatic malignancies (OMs) and hepatocellular carcinoma (HCC) is vital to allow clinicians to determine optimal treatment and assess prognosis in patients at high risk for HCC. This study evaluated the performance of the Liver Imaging Reporting and Data System (LI-RADS) using magnetic resonance imaging (MRI) versus contrast-enhanced ultrasonography (CEUS) for differentiating HCC from OMs in patients at high risk for HCC. This retrospective study consecutively enrolled 106 high-risk patients with HCC and 52 high-risk patients with OMs. Patients underwent both MRI and CEUS, with histologic diagnosis as a reference standard. The diagnostic performance of MR versus CEUS LI-RADS was calculated and compared. The performance of the modified CEUS LI-RADS criteria was also evaluated and compared. Our research found that MRI features significantly differed between patients with OMs and those with HCC (p < 0.05), with sensitivities of 34.6%–69.2% and specificities of 83.0%–95.3% for diagnosing OMs and an LI-RADS M (LR-M): definite or probable malignancy, not specific for hepatocellular carcinoma sensitivity of 90.4% and specificity of 83.0% for diagnosing OM. CEUS features also significantly differed between patients with OM and HCC (p < 0.05), with sensitivities of 11.5%–96.2% and specificities of 23.6%–100% for diagnosing OMs, and an LR-M sensitivity of 98.1% and specificity of 84.0% for diagnosing OMs. Accuracies of category LR-M did not significantly differ between MR and CEUS LI-RADS (85.4% vs. 88.6%, p = 0.724). After reclassification of category LR-M nodules to category LR-5 if they exhibited clear intratumoral non-enhanced area boundaries and no punched-out appearance before 5 min, accuracy increased from 88.6% to 96.8% for CEUS LR-M and from 84.8% to 91.1% for CEUS LR-5. LR-M accuracies were significantly higher for the modified version of the CEUS LI-RADS than for MR LI-RADS (96.8% vs. 85.4%, respectively, p = 0.04). CEUS LI-RADS and MR LI-RADS can effectively be used to distinguish HCC from OMs. In patients at high risk of HCC, performance may be further improved by using a modified CEUS LI-RADS classification system in which category LR-M lesions are considered LR-5 if they have clear intratumoral non-enhanced area boundaries and do not have a punched-out appearance.