A New Milestone in Hepatocellular Carcinoma Surveillance: Clinical Validation of the 2023 AASLD Guidance

Commentary on Cheng M et al. Prospective multicenter evaluation of hepatocellular carcinoma surveillance under American Association for the Study of Liver Diseases version 2023 guidance. Radiology 2026;318:e252252 (https://doi.org/10.1148/radiol.252252).

SUMMARY

Surveillance for hepatocellular carcinoma (HCC) has been suffered from limited sensitivity of US for early stage HCC screening (1). The prospective multicenter study by Cheng et al. (2) represents practical and timely validation of the 2023 American Association for the Study of Liver Diseases (AASLD) guidance for HCC surveillance. The updated surveillance algorithm (AASLD v2023) incorporated US visualization score C (VIS-C, poor visualization), increasing alpha-fetoprotein (AFP) levels, and lesion growth. The study was to assess the performance of AASLD v2023 for HCC detection in comparison to Liver Imaging Reporting and Data System (LI-RADS) v2017 and AASLD v2018 in participants at high risk of HCC (3,4). In 953 participants, AASLD v2023 achieved 94% sensitivity and 99.6% negative predictive value (NPV) for HCC detection, significantly outperforming US LI-RADS v2017 (60% sensitivity, 97.6% NPV) and AASLD v2018 (76% sensitivity, 98.5% NPV) (4). Specificity was lower AASLD v2023 (83.9%) compared with US LI-RADS v2017 (90.1%) and AASLD v2018 (89.1%), but it falls within the previously reported range for HCC surveillance (70%–100%), supporting the clinical acceptability of the inherent trade-off between sensitivity and specificity (5).

COMMENT

In contrast to earlier surveillance algorithms such as US LI-RADS v2017 or AASLD v2018, AALSD v2023 particularly addresses a clinical limitation of US with its variable visualization quality (2). By integrating US visualization scores and temporal AFP changes or lesion growth characteristics, the updated algorithm aims to trigger cross-sectional imaging when surveillance conditions are suboptimal. A remarkable advance of the new algorithm is the recognition that surveillance performance depends not only on lesion detection, but also on imaging quality and temporal biomarker changes (2). Poor US visualization (VIS-C) is now considered as an independent indication for additional cross-sectional imaging. This represents a practical response to a well-known limitation of US-based HCC surveillance, particularly in patients with cirrhosis where US visualization may be compromised. It suggests a paradigm shift from pure image-based surveillance toward a more dynamic and risk-adaptive approach.