Background

Metabolic dysfunction-associated steatotic liver disease (MASLD) is a newly adopted term to replace NAFLD as a more inclusive and positive term that also better reflects the underlying mechanism of the disease. We aimed to develop a novel noninvasive assessment index to characterize MASLD with its new diagnostic criteria.

Methods

Using data from National Health and Nutrition Examination Surveys 2017-2020 (NHANES), we included participants aged 20-74 years old who met the diagnostic criteria of MASLD. Participants were randomly grouped into a training cohort and an internal validation cohort, with an over-sampling technique to balance the training cohort. We performed a forward stepwise logistic regression to create a MASLD index in the training cohort. Area under the curves (AUC), sensitivity analysis, Delong test and Brier Score were performed and compared with previous diagnostic indices including the Hepatic Steatosis Index (HSI), Fatty Liver Index (FLI), US FLI, and the Metabolic Dysfunction Associated Fibrosis 5 score (MAF-5) in the internal validation cohort as well as two external validation cohorts from 5 centers from Japan and UK Biobank.

Results

The study included 1,853 eligible participants from NHANES (MASLD prevalence 40.9%) and external validation cohorts of 33,627 patients from 5 Japan centers (MASLD prevalence 28.1%), 774 participants from UK Biobank (MASLD prevalence 19.6%) (Figure 1). Using data derived from the NHANES cohort, we developed a forecast formula named MASLD index with four variables (waist circumference [WC], triglyceride [TG]s, fasting glucose [FG], and race and ethnicity), which performed well in both the training (n=1,062) and internal validation cohort (n=556), with AUCs of 0.87 (95%CI 0.84-0.89) and 0.84, respectively. The AUCs in the training and internal validation cohorts for the US FLI for MASLD were (0.84, 0.85), followed by the FLI (0.84, 0.83), HSI (0.82, 0.79) and MAF-5 (0.81, 0.79), respectively (Figure 2). In both the training and internal validation cohorts, the curve of MASLD index was closest to the reference line, with the lowest Brier Score, indicating best agreement between the predicted and observed probabilities and with superior performance over other indices across subgroups by sex, age, race and ethnicity, and abdominal obesity subgroups (Table 1). The MASLD index also achieved AUCs of 0.88 (95%CI 0.87-0.88) and 0.84 (95%CI 0.80 - 0.87) for the two external validation cohorts in the Japan and UK Biobank, respectively.

Conclusions

Using data derived from a large population-based U.S. cohort and two large multicenter external validation cohorts from 5 centers in Japan and UK Biobank, we proposed a simple yet accurate novel diagnostic index for MASLD outperforming traditional indices.