Contrast-Enhanced Ultrasound With Perfluorobutane for Hepatocellular Carcinoma Diagnosis: Comparison of Imaging Phases and Diagnostic Criteria.

Background: Contrast-enhanced ultrasound (CEUS) with perfluorobutane has used varying protocols and diagnostic criteria for hepatocellular carcinoma (HCC). Objective: To assess diagnostic performance for HCC of CEUS with perfluorobutane in high-risk patients using various criteria. Methods: This retrospective post-hoc study, evaluating individual patient data from three earlier prospective studies from one hospital, included 204 patients (136 men, 68 women; mean age, 63±11 years) at high risk of HCC with 213 liver observations. Patients underwent CEUS using perfluorobutane from March 2019 to June 2022. Three radiologists (examination's operator; two subsequent reviewers) independently interpreted examinations, assessing arterial, portal-venous (arterial phase completion through 2 minutes), transitional (2-5 minutes), and Kupffer (≥10 minutes) phase findings. Six criteria for HCC were tested: 1-any arterial phase hyperenhancement (APHE), Kupffer-phase hypoenhancement; 2-non-rim APHE, Kupffer-phase hypoenhancement; 3-non-rim APHE, portal-venous washout; 4-non-rim APHE, portal-venous washout and/or Kupffer-phase hypoenhancement; 5-non-rim APHE, portal-venous and/or transitional washout; 6-non-rim APHE, any of portal-venous washout, transitional washout, or Kupffer-phase hypoenhancement. Depending on the criteria, observations were instead deemed non-HCC malignancy if showing rim APHE, early washout (<1 minute), or marked washout (at 2 minutes). Reference was pathology for malignant, and pathology or imaging follow-up for benign, observations. Diagnostic performance was assessed, pooling readers' data. Results: Criteria 1 (not recognizing features of non-HCC malignancy) had highest sensitivity (86.9%) but lowest specificity (43.2%) for HCC. Compared with nonrim APHE and portal-venous washout (criteria 3), addition of Kupffer-phase hypoenhancement (criteria 4), transitional washout (criteria 5), or either feature (criteria 6) significantly increased sensitivity (34.4% vs 62.6-64.2%) and accuracy (61.8% vs 75.1-76.4%), although significantly decreased specificity (98.5% vs 91.9-94.1%). Criteria 2, 4, 5, and 6 (all incorporating transitional washout and/or Kupffer-phase hypoenhancement) showed no significant differences in sensitivity (62.6-64.2%), specificity (91.9-94.1%), or accuracy (75.1-76.4%). Conclusion: Recognition of features of non-HCC malignancy improved specificity for HCC. Incorporation of transitional washout and/or Kupffer-phase hypoenhancement improved sensitivity and accuracy, albeit lowered specificity, versus arterial and portal-venous findings alone, without further performance variation among criteria incorporating those two findings. Clinical Impact: Kupffer-phase acquisition may be optional for observations classified as HCC or non-HCC malignancy by arterial, portal-venous, and transitional phases.