Abbreviated magnetic resonance imaging vs ultrasound for surveillance of hepatocellular carcinoma in high-risk patients

Background & Aims: We aimed to compare the performance of gadoxetic acid-enhanced abbreviated MRI (AMRI)-based surveillance and ultrasound-only surveillance in high-risk patients for hepatocellular carcinoma (HCC). Methods: Prospectively recruited high-risk patients (>5% annual risk of HCC) who underwent one to three rounds of complete gadoxetic acid-enhanced MRI (CMRI) and ultrasound at 6-months intervals were retrospectively analysed. AMRI consisted of diffusion-weighted, T2-weighted, and hepatobiliary phase imaging. The sensitivity, specificity, and accuracy of CMRI followed by AMRI (CAA), AMRI-only (AAA), and ultrasound-only (US) were compared using generalized estimating equations. Image quality was assessed. Results: In 382 patients, HCC was diagnosed in 43 (11.3%), including 42 with early-stage HCCs. The sensitivities of CAA (90.7%, 39/43) and AAA (86.0%, 37/43) were higher than US (27.9% [12/43]; P < 0.001), whereas the sensitivities of the two MRI approaches did not significantly differ (P = 0.56). The specificity of CAA (97.1%, 983/1012) was higher than AAA (95.6% [967/1012]; P = 0.01) and not significantly different from US (96.3% [975/1012]; P = 0.59). The CAA approach had the best accuracy of 96.9% (1022/1055), higher than the AAA approach (95.2% [1004/1055]; P = 0.01) and the US approach (93.6% [987/1055]; P = 0.01). Image quality was inadequate in 33.7% (356/1055) of US examinations but in only 10.0% (105/1055) of the AAA and 11.1% (117/1055) of the CAA approach. Conclusions: In high-risk patients, AMRI-based surveillance approaches had higher sensitivities than ultrasound-only surveillance for early-stage HCC. A sequential MRI approach of CMRI followed by AMRIs showed superior accuracy than the AMRI-only or ultrasound-only approach.