Small (≤1-cm) hepatocellular carcinoma: diagnostic performance and imaging features at gadoxetic acid-enhanced MR imaging.

Abstract

To assess diagnostic performance and imaging features of gadoxetic acid-enhanced magnetic resonance (MR) imaging in small (≤1-cm) hepatocellular carcinoma (HCC) detection in patients with chronic liver disease. The institutional review board approved this retrospective study and waived informed consent. Sixty patients (56 men, four women; mean age, 60.1 years) with HCC (146 lesions; 70 > 1 cm, 76 ≤ 1 cm) underwent gadoxetic acid-enhanced MR imaging. HCC was confirmed at surgical resection (72 lesions; 30 > 1 cm, 42 ≤ 1 cm) or by showing interval growth with typical enhancement patterns at follow-up dynamic computed tomography or MR imaging (74 lesions; 40 > 1 cm, 34 ≤ 1 cm). Two radiologists assessed MR imaging features and graded likelihood of HCC with a five-point confidence scale. Jackknife alternative free-response receiver operating characteristic (JAFROC) method was used. Mean JAFROC figure of merit for small HCC was 0.717; that for large (>1-cm) HCC was 0.973 with substantial agreement (κ = 0.676). Mean sensitivity and positive predictive value (PPV) were 46.0% (70 of 152) and 48.3% (70 of 145) for small HCC versus 95.0% (133 of 140) and 78.2% (133 of 170) for large HCC, respectively. Eleven of 76 small HCCs (14%) were not seen on MR images, even after careful investigation. MR imaging features of small HCC included arterial enhancement (79%, 60 of 76), hypointensity on hepatobiliary phase (HBP) images (68%, 52 of 76), washout on 3-minute delayed phase images (50%, 38 of 76), hyperintensity on T2-weighted images (43%, 33 of 76), hypointensity on T1-weighted images (32%, 24 of 76), and restriction on diffusion-weighted images (28%, 20 of 72). Arterial enhancement and washout on 3-minute delayed phase images or hypointensity on HBP images occurred in 66% of small HCCs (50 of 76). Diagnostic performance of gadoxetic acid-enhanced MR imaging for small HCC detection is still low, with mean sensitivity of 46.0% (70 of 152) and mean PPV of 48.3% (70 of 145). By adding hypointensity on HBP images as washout, diagnostic performance for small HCC detection can be improved.