Related Experiment Video
Updated: Jul 16, 2025
Novel In Vivo Micro-Computed Tomography Imaging Techniques for Assessing the Progression of Non-Alcoholic Fatty Liver Disease
Published on: March 24, 2023
Poor Performance of Non-invasive Tests for Advanced Fibrosis in Nonalcoholic Fatty Liver Disease: A Multicentric
Umang Arora1, Ritik M Goyal1, Kevin K J Teh2
1Department of Gastroenterology and Human Nutrition, All India Institute of Medical Sciences, Delhi, 110029, India.
Background:
Non-invasive tests (NITs) are useful to assess advanced fibrosis (AF) in nonalcoholic fatty liver disease (NAFLD). Data from Asian countries suggest that these tests have poor performance. We aimed to assess diagnostic accuracy of established thresholds of biomarker-based NITs and Transient Elastography (TE) in identifying AF and evaluated the utility of a two-step test approach.
Methods:
Biopsy-proven 641 NAFLD patients (55.2% males, median age 42 years) were included from three different centers of Asia. AF (≥ F3) was identified as per histological staging (24.8%).
Results:
TE had the highest area under the receiver operating characteristic curve (AUROC) 0.82 (0.79-0.86), and all other biomarker-based NITs had low AUROC (< 0.7). NITs performed poorly at established thresholds. The combination of NITs utilizing liver stiffness measurement (LSM) and biomarkers, Agile 3+ and FAST, demonstrated acceptable diagnostic accuracy (AUROC 0.82 and 0.78, respectively), but none were superior to LSM alone. LSM measured using appropriate M and XL probes remained accurate regardless of body mass index (BMI); NFS and APRI scores were less accurate at higher BMI ranges. A two-step approach using NFS rule-out criteria (< - 2.97 to rule out) followed by LSM (< 7.3 kPa to rule out and ≥ 12.7 kPa to rule in) correctly classified 62.4% of patients, with only 10.2% of patients incorrectly classified.
Conclusion:
NITs have not been validated to identify AF in the Asian NAFLD population, and internationally accepted thresholds yield high false-negative rates. LSM and LSM-based combination tests remain the most accurate.
