Hepatobiliary phase hypointensity on gadobenate dimeglumine- enhanced magnetic resonance imaging may improve the diagnosis of hepatocellular carcinoma

BackgroundTo determine the clinical value of hepatobiliary phase (HBP) hypointensity for noninvasive diagnosis of hepatocellular carcinoma (HCC).MethodsA total of 246 high-risk patients with 263 selected nodules (126 HCCs, 137 non-HCCs) undergoing gadobenate dimeglumine (Gd-BOPTA)-enhanced magnetic resonance imaging (MRI) were included in the study. Imaging-based diagnoses of small (≤3 cm) and large (>3 cm) HCCs were made using the following 4 criteria: (I) non-rim arterial phase hyper-enhancement (APHE) plus hypointensity on the portal venous phase (PVP); (II) non-rim APHE plus hypointensity on the PVP and/or transitional phase (TP); (III) non-rim APHE plus hypointensity on the PVP and/or TP and/or HBP; (IV) criterion 3 plus non-LR-1/2/M. Based on typical imaging features, LR-1, LR-2, or LR-M (if definitely benign, probably benign, malignant but not HCC specific, respectively) were defined according to the Liver Imaging Reporting and Data System (LI-RADS). Sensitivities and specificities of imaging criteria were calculated and compared using McNemar's test.ResultsAmong the diagnostic criteria for small HCCs, criterion 3 and 4, which included HBP hypointensity, showed significantly higher sensitivities (96.4% and 94.6%, respectively) than criterion 1 (58.9%, P<0.001 for both). Moreover, criterion 4, which included HBP hypointensity and ancillary features, showed significantly higher specificity (94.7%) than criterion 3 (66.7%, P<0.001) and comparable specificity to criterion 1 (97.4%, P=0.375), achieving the highest accuracies (94.7%). The diagnostic performance of criterion 4 for large HCCs was similar to that for small HCCs.ConclusionsHBP hypointensity acquired from Gd-BOPTA-MRI can improve sensitivity and maintain high specificity in the diagnosis of both small and large HCCs after excluding benignities or non-HCC malignancies according to characteristic imaging features.