OBJECTIVES:

Non-invasive tests (NITs) are used for staging risk in nonalcoholic steatohepatitis (NASH). These include biomarker-based measures such as the Fibrosis-4 index (FIB-4) and AST-to-Platelet Ratio Index (APRI), and more recently, the FibroScan+AST (FAST) score (which uses liver stiffness and steatosis measurements obtained via vibration-controlled transient elastography [VCTE] imaging). While NITs are primarily used for staging risk, they have also been considered for identification of presumed NASH in observational research. Existing research does not address how different NITs may impact the estimated prevalence of presumed NASH. This analysis estimated the prevalence of presumed NASH among US adults, and assessed variation arising from use of different NITs.

METHODS:

A cross-sectional analysis was conducted using the 2017-March 2020 National Health and Nutrition Examination Survey (NHANES) cycle. The analysis was weighted to provide nationally-representative estimates for US adults. NASH was identified using a multi-stepped approach by which participants were restricted to those with steatosis then to those without common alternative causes of liver disease. Presumed NASH was distinguished based on FIB-4, APRI, and FAST score cut-offs across 16 scenarios to assess the impact on prevalence estimates.

RESULTS:

Among NHANES participants with complete data for the analysis (N=6,789), prevalence of presumed NASH identified using FAST score was estimated to range from 1.2%-4.7% (FAST score ≥0.67 and ≥0.35, respectively). Use of non-imaging NITs resulted in a wide range of prevalence estimates from 1.1%-1.6% for APRI ≥0.70, 7.8%-11.2% for FIB-4 ≥1.59, and 28.5%-39.2% for FIB-4 ≥0.90. Fibrosis stage distributions differed when imaging versus non-imaging NITs were used (26%-36% vs 47%-95%, respectively, for F2-F3 fibrosis).

CONCLUSIONS:

Prevalence of presumed NASH estimated using FAST score aligned with historical biopsy-based estimates. Analyses using non-imaging NITs (FIB-4, APRI) yielded wide ranges of prevalence estimates, suggesting these measures should be supplemented with additional information for identification of presumed NASH.