The American Association for the Study of Liver Diseases (AASLD) version 2023 (v2023) provides higher negative predictive value (NPV) and significantly higher sensitivity in surveillance of hepatocellular carcinoma (HCC), according to newly published research.
For the prospective multicenter study, recently published in Radiology, researchers compared the AASLD v2023 system, the AASLD version 2018 (v2018) criteria and the Liver Imaging Reporting and Data System (LI-RADS) version 2017 (v2017) criteria with a review of ultrasound, computed tomography (CT) and magnetic resonance imaging (MRI) data for 953 people (median age of 51) deemed to be at high risk for HCC. The AASLD v2023 system incorporated an ultrasound (US) visualization score, lesion growth and increasing α-fetoprotein (AFP) level as triggers for HCC surveillance, according to the study.
For the detection of HCC, the study authors found that the AASLD v2023 system provided a 94 percent sensitivity in contrast to 76 percent for the AASLD v2018 criteria and 60 percent for the US LI-RADS v2017 system.
The researchers also noted a slightly higher negative predictive value (NPV) for AASLD v2023 (99.6 percent) in comparison to the AASLD v2018 (98.5 percent) and US LI-RADS v2017 (98 percent) systems.
“These findings emphasize the value of integrating imaging quality and longitudinal biomarkers into surveillance strategies rather than relying solely on lesion size or static AFP thresholds,” noted lead study author Mei-Qing Cheng, MM, who is affiliated with the Department of Medical Ultrasonics in the Institute of Diagnostic and Interventional Ultrasound at the First Affiliated Hospital of Sun Yat-Sen University in Guangzhou, China, and colleagues.
While the false positive rate for HCC was higher for the AASLD v2023 system (16.1 percent) in comparison to AASLD v2018 (10.9 percent) and US LI-RADS (9.86 percent), the study authors pointed out that the AASLD v2023 criteria offered a significant reduction in the false negative rate (6 percent vs. 24 percent for AASLD v2018 and 40 percent for US LI-RADS v2017.
Three Key Takeaways
• Substantially improved sensitivity and rule-out performance. AASLD v2023 markedly outperformed prior frameworks for HCC surveillance, demonstrating 94 percent sensitivity and a 99.6 percent negative predictive value (NPV), substantially reducing missed cancers compared with AASLD v2018 and US LI-RADS v2017.
• Practice-changing role of visualization score and longitudinal markers. Incorporation of the ultrasound visualization score (VIS), lesion growth, and rising AFP meaningfully improved detection. This was particularly the case in patients with poor ultrasound visualization (VIS-C), for whom AASLD v2023 achieved 100 percent sensitivity for HCC, supporting early escalation to CT/MRI.
• Trade-off favors fewer false negatives. Although AASLD v2023 increased false positives (16.1 percent), it dramatically reduced false negatives (6 percent vs. 24–40 percent with older systems), reinforcing its value in high-risk populations where missed HCC has major clinical consequences.
Noting that VIS-C assessment — indicating severe limitations in liver imaging — was associated with 40 percent and 33 percent of false-negative cases with the US LI-RADS v2017 and AASLD v2018 criteria, respectively, the researchers said the addition of the ultrasound visualization score into the AASLD v2023 system was “novel and practice changing.” For patients with VIS-C, the study authors noted 100 percent sensitivity for HCC with the AASLD v2023 criteria.
“These findings underscore that poor visualization itself represents a critical and independent risk factor, validating the rationale of AASLD v2023 to recommend additional CT and/or MRI for participants with VIS-C, regardless of AFP or lesion findings,” added Cheng and colleagues.
(Editor’s note: For related content, see “Study Suggests Merits of PSMA PET/MRI for Detecting HCC in LI-RADS 3 Cases,” “Can AI-Powered Image Reconstruction Facilitate Lower Radiation Dosing with Dual-Energy CT Detection of Liver Metastases?” and “Meta-Analysis Examines MRI-Based AI for Predicting Microvascular Invasion in Hepatocellular Carcinoma.”)
In regard to study limitations, the authors acknowledged possible interobserver variability with visualization score assessments. They also conceded that the high prevalence of viral hepatitis-related liver disease and relatively low body mass index in the Asian cohort may curtail extrapolation of the study findings to broader populations.
