Non-alcoholic fatty liver disease (NAFLD) affects a quarter of the global adult population and is emerging as an important cause of chronic liver diseases. NAFLD is defined by the presence of hepatic fat accumulation in the absence of excessive alcohol consumption, viral infection, drug-related liver injury, or autoimmune hepatitis. Currently, diagnosis of NAFLD is based on the exclusion of other causes of liver disease. There is an ongoing debate about the current negative diagnostic criteria and the nomenclature of the NAFLD [1]. The importance of metabolic risk factors may not be reflected in the term “NAFLD,” although NAFLD is strongly associated with metabolic abnormalities. In addition, there are no clear criteria for significant alcohol consumption for screening and diagnosis of NAFLD [2-4].

In 2020, the Asian Pacific Association for the Study of Liver (AP-ASL) proposed a new definition of metabolic-association fatty liver disease (MAFLD) to highlight the contribution of metabolic dysfunction and to replace the nomenclature of NAFLD [1]. The diagnosis of MAFLD is based on the positive criteria of hepatic steatosis detected either by imaging, blood biomarkers, or histology in addition to one of the following: overweight/obesity, presence of type 2 diabetes, or evidence of metabolic dysregulation [1]. It has been reported that the MAFLD criteria can better identify patients with metabolic dysfunction than NAFLD criteria [5].

Several hepatic steatosis algorithms including fatty liver index (FLI), have been widely used as a simple and non-invasive panel for detecting NAFLD [6, 7]. One recent study has validated hepatic steatosis algorithms in the MAFLD population using national survey datasets from western populations [8]. Currently, few studies regarding steatosis indices for predicting MAFLD have been performed in the Korean population. The aims of this study were to validate the hepatic steatosis algorithms in the Korean MAFLD population and to investigate the effects of the subgroups on the diagnostic performance of the indices.

The study subjects who were enrolled from the general population visited the author’s institution for a routine health checkup from January 2022 to December 2022. This study consisted of 22,372 adults who underwent abdominal ultrasonography, among 51,496 members of the Korean population visiting a comprehensive health promotion center. The following exclusion criteria were used: hepatitis virus infection, autoimmune disease, primary biliary cirrhosis, primary sclerosing cholangitis, malignancies, chronic kidney diseases, cardiovascular disease, cerebrovascular disease, history of transplantation, hypothyroidism, epileptic disease, psychiatric disorders, and/or cases with missing data. Finally, 12,962 subjects were included in this study. Hepatic steatosis was assessed using abdominal ultrasonography by experienced physicians and MAFLD was diagnosed according to APASL guidelines [1]. This study was approved by the Institutional Review Board of Samsung Changwon Hospital (2022-10-002).

Clinical and biochemical data include age, gender, past medical history, systolic blood pressure (SBP), diastolic blood pressure (DBP), body mass index (BMI), waist circumference (WC), total protein, albumin, aspartate transaminase (AST), alanine aminotransferase (ALT), gamma glutamyltransferase (GGT), fasting glucose, insulin, HbA1c, total cholesterol, high density lipoprotein–cholesterol (HDL-cholesterol), low density lipoprotein–cholesterol (LDL-cholesterol), apolipoprotein A, apolipoprotein B, triglyceride, and high-sensitivity C-reactive protein (hs-CRP).

Hypertension was diagnosed if the subjects had SBP ≥130 mmHg or DBP ≥85 mmHg or reported a history of hypertension [9]. Diabetes was diagnosed if the HbA1c ≥6.5% or fasting plasma glucose level ≥126 mg/dL or a reported history of diabetes [10, 11]. Metabolic syndrome was diagnosed as the presence of three or more of the following: WC: ≥90 cm in men and ≥80 cm in women; triglycerides ≥150 mg/dL; HDL-cholesterol <40 mg/dL in men and <50 mg/dL in women; blood pressure ≥130/85 mmHg or antihypertensive medications; fasting blood glucose ≥100 mg/dL or antidiabetic medications [9, 12].

Hepatic steatosis algorithms such as FLI, hepatic steatosis index (HSI), non-alcoholic fatty liver disease liver fat score (NLFS), triglyceride glucose (TyG), triglyceride glucose-body mass index (TyG-BMI), and TyG-waist circumference (TyG-WC) were evaluated [6, 7, 13-17]. The scores were calculated by the formulas presented in the previous studies [6, 7, 13-17]:

FLI= EXP(0.953×LN(Triglyceride)+0.139×BMI+0.718×LN(GGT)+0.053×WC-15.745)/(1+EXP(0.953×LN(Triglyceride)+0.139×BMI+0.718×LN(GGT)+0.053×WC-15.745))×100;

HSI= 8×(ALT/AST)+BMI(+2, if diabetes; +2, if female);

NLFS= -2.89+1.18×metabolic syndrome (yes=1, no=0)+0.45×diabetes (yes=2, no=0)+0.15×fasting insulin+0.04×AST-0.94× (AST/ALT);

TyG=LN((Triglyceride×(fasting glucose/2));

TyG-BMI=TyG×BMI;

TyG-WC=TyG×WC.

Clinical data between MAFLD and non-MAFLD were compared using the chi-squared test and Mann–Whitney test, as appropriate. The value was described expressed as frequencies and the median (interquartile range). A receiver operating characteristic (ROC) curve analysis was done to calculate the sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV). All analyses were performed using MedCalc version 17.5.5 (MedCalc Software, Mariakerke, Belgium).

The frequency of MAFLD was 38% (4,873/12,962): 47% (4,098/8,814) and 19% (775/4,148) for males and females, respectively. The clinical and biochemical characteristics of the population are shown in Table 1. Old age, male, hypertension, diabetes, and metabolic syndrome were more frequently observed in the subjects with MALFD than in those without MAFLD (Table 1). All biochemical data were significantly different between the MAFLD population and the non-MAFLD population (Table 1).

The TyG-BMI showed the highest area under the receiver operating characteristic curve (AUROC) for the MAFLD (0.877, 95% CI: 0.871–0.882, sensitivity of 88% and a specificity of 71% at a cutoff of 206.7), followed by the FLI (0.872, 95% CI: 0.866–0.878), TyG-WC (0.870, 95% CI: 0.864–0.876), NLFS (0.847, 95% CI: 0.841–0.854), TyG (0.769, 95% CI: 0.762–0.776), and HSI (0.595, 95% CI: 0.587–0.604) (Fig. 1, Table 2). Pairwise comparison of ROC curves show ed that there is no difference between FLI and TyG-WC (P=0.1465). However, the diagnostic performance of NLFS (P=0.0003), TyG (P<0.0001), FLI (P<0.0001), and HSI (P<0.0001) were significantly lower than those of TyG-WC (Table 2, Fig. 1). The AUROC of the indices tended to decrease in male and subgroups with old age, overweight/obesity, hypertension, diabetes, or metabolic syndrome (Table 3).

Five hepatic steatosis algorithms, which have been widely used for NAFLD, showed superior diagnostic performances for predicting MAFLD in this Korean population-based study. In particular, TyG-BMI, TyG-WC, and FLI were the most efficient surrogate markers for the identification of MAFLD. This study suggests that these indices are useful for screening MAFLD as well as NAFLD. However, compared to other indices, the diagnostic performance of HSI was inferior for MAFLD. This might be because the HSI is calculated based on AST, ALT, BMI, and sex and the presence of diabetes and the markers regarding lipid metabolism were not considered in the formula. On the other hand, evidence of metabolic dysregulation based on triglyceride and HDL-cholesterol was one of the important criteria for MAFLD. Use of HSI for MAFLD would be recommended in subgroups with consideration of these factors.

The diagnostic performance of these indices except for HSI is in line with the results from previous studies [8, 18]. One previous study showed the diagnostic performance of FLI (AUC 0.793 with a sensitivity of 57%, and a specificity of 84%) and NLFS (AUC 0.774 with a sensitivity of 59%, and a specificity of 81%) for MAFLD in Western populations [8]. Another study showed the diagnostic performance of FLI (AUC 0.791 with a sensitivity of 71%, and a specificity of 71%) for MAFLD in a total of 1,300 Korean patients who underwent CT scans [18]. Furthermore, another recent study reported that triglyceride glucose index-related parameters can be useful for early screening of NAFLD (AUC 0.804 for TyG-BMI, 0.813 for TyG-WC), MAFLD (AUC 0.822 for TyG-BMI, 0.932 for TyG-WC), and liver fibrosis (AUC 0.708 for TyG-BMI, 0.724 for TyG-WC) [19, 20]. However, the previous studies were performed either in a hospital-based setting [18] or based on the national survey datasets from western populations [8, 19]. Currently, there is no general population-based study for Korean MAFLD [8, 18]. The present study was based on a large-scale analysis of a Korean population and was the first validation of TyG-BMI and TyG-WC for predicting MAFLD in the Korean general population.

This study showed that the frequency of MAFLD in the Korean population was significantly increased in subgroups with metabolic syndrome, diabetes, hypertension, and overweight/obesity. Because the prevalence of MAFLD has been fueled by the rise of obesity, diabetes, and metabolic syndrome, it is not surprising that there is a significant positive association between diabetes, dyslipidemia, or metabolic syndrome and MAFLD. However, these factors tend to decrease the diagnostic performances of hepatic steatosis indices in the Korean population in this study. These findings are in agreement with the previous study by Liu et al., in which AUROC of the hepatic steatosis indices decreased with the combination of metabolic factors [8]. This might be attributed to the components of hepatic steatosis indices [6, 7, 13-17]. This suggested that they could not provide added value to the diagnostic performance of the indices, because some of these components are risk factors as well as the diagnostic criteria for MAFLD. In addition, further studies regarding revised algorithms for these high-risk subgroups for MAFLD are recommended.

In conclusion, hepatic steatosis indices such as TyG-BMI, TyG-WC, FLI, TyG, and NFLS can be useful for screening MAFLD as well as NAFLD in a Korean population. This study showed that risk factors such as overweight/obesity, diabetes, hypertension, and/or metabolic syndrome may affect the diagnostic performances of indices. The MAFLD subgroup should be considered to optimize the hepatic steatosis assessments by these formulas.