Population-specific cut-off points of fatty liver index: a study based on the National Health and Nutrition Examination Survey data

Background: Fatty liver index (FLI) is the most recognized blood biomarker for diagnosis of hepatic steatosis (HS), but lacks the reliable specific cut-off points (COPs). Therefore, we aim to investigate the population-specific COPs of FLI based on the results of liver ultrasound transient elastography (LUTE) and conventional ultrasonography in the National Health and Nutrition Examination Survey (NHANES).

Methods: 5948 participants who underwent LUTE from the NHANES 2017-2018 and 14,797 participants who underwent conventional ultrasonography from the Third NHANES (NHANES III) were recruited. FLI was calculated by using body mass index (BMI), waist circumference (WC), triglyceride, and gamma-glutamyl transferase, and its optimal COPs in a specific population (stratified by sex, BMI, and WC) were obtained from receiver operator characteristics (ROC) curve with ultrasonic-diagnosed HS as the reference standard.

Results: Based on LUTE in NHANES 2017-2018, the prevalence of HS and metabolic dysfunction-associated fatty liver disease (MAFLD) were 58.7% and 56.2%, respectively, and the optimal COP of FLI for HS diagnosis in the overall population was 45.60, with an area under ROC curve (AUROC) of 0.833 (0.822-0.844). Based on conventional ultrasonography in NHANES III, the prevalence of HS and MAFLD were 34.4% and 27. 9%, respectively, and the optimal COP of FLI for HS was 59.5, with an AUROC of 0.681 (0.671-0.691). With the increase of BMI and WC, the COPs increased gradually with significant differences between different groups. Compared with conventional ultrasonography, the COPs of FLI based on LUTE were much more precise, with higher diagnostic ability. The population-specific COPs of FLI stratified by gender, WC, and BMI were tabulated.

Conclusion: In the United States, the incidences of HS and MAFLD were high, especially when assessed by LUTE. The FLI based on LUTE is well capable of predicting HS when stratified by gender, WC, and BMI.