A total of 389 individuals attending the Vascular Medicine and Metabolism Unit from the Saint Joan University Hospital (Reus, Spain) due to lipid metabolism disturbances and diabetes, overweight/obesity, or metabolic syndrome and willing to participate in the study were sequentially recruited. Anamnesis, anthropometric, and physical examination was performed using standardized procedures at the time of clinical visits. Individuals taking lipid-lowering medications underwent a wash-out period of at least 6 weeks (8 weeks if they were on fibrates) to avoid bias. Subjects with known chronic lung, renal, or chronic liver disease, cancer, or any other serious disease were excluded (Girona et al., 2019). Individuals with an alcohol consumption above 20 g/day (females) or 40 g/day (males) were not included. Viral or autoimmune hepatic diseases were excluded when clinically indicated. None of the included patients had clinical evidence of viral or autoimmune liver disease. Diabetes (N = 264), obesity (N = 202), and metabolic syndrome (N = 291) were diagnosed according to standard clinical criteria. Specifically, the presence of diabetes was defined as fasting concentration of blood sugar ≥126 mg/dl, glycated hemoglobin A1c > 6.5, glucose after a 2 h Oral Tolerance Tests ≥200 mg/dl or being treated with antidiabetic drugs, according to international criteria (American Diabetes Association, 2019), obesity as body mass index (BMI) ≥ 30 Kg/m², and the presence of metabolic syndrome was defined according to the Adult Treatment Panel III criteria (National Cholesterol Education Program [NCEP] Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults, 2002). Arterial hypertension was defined as systolic and diastolic blood pressures ≥140 mm Hg and ≥90 mm Hg, respectively. Liver steatosis was defined as FLI ≥ 60. Written informed consent was obtained from all participants. The study was approved by Hospital Ethics Committee (Pere Virgili Health Research Institute; ethical approval code: 200/2018) and was performed in full compliance with the Declaration of Helsinki.

Demographic and clinical characteristics including age, gender, both systolic and diastolic blood pressure, weight and waist circumference were obtained at the point of study inclusion. BMI was calculated as follows = weight (kg)/height (m)².

Venous blood sample were obtained from each patient in the study cohort after overnight fasting. Fasting blood was immediately centrifuged at 1500 g for 15 min at 4°C, and serum samples were aliquoted and stored at −80°C in the BioBank at our center until analysis. Biochemical parameters [glucose, glycated hemoglobin A1c (HbA1c), ultra-sensitive C-reactive protein (usCRP), creatinin], lipids [total cholesterol, triglycerides and high-density lipoprotein cholesterol (HDLc)], apolipoproteins [apolipoprotein A1 (apoA1) and apolipoprotein B100 (apoB100)], transaminases [aspartate aminotransferase (AST) and alanine aminotransferase (ALT)] and gamma-glutamyl transpeptidase (GGT) were measured using colorimetric, enzymatic and immunoturbidimetric assays, respectively (Spinreact, SA, Spain; Wako Chemicals GmbH, Germany; and Polymedco, New York, NY, United States; CV < 4%), which were adapted to the Cobas Mira Plus Autoanalyser (Roche Diagnostics, Spain). The low-density lipoprotein cholesterol (LDL-C) levels were calculated with the use of the Friedewald equation: LDL-C = total cholesterol − (HDL-C + [triglycerides ÷ 5]).

FLI was calculated using an algorithm based on BMI, waist circumference, triglycerides and GGT, which shows an accuracy of 0.84 (95% CI 0.81–0.87) detecting fatty liver. Specifically, FLI was calculated according to the following formula: FLI = (e^{0.953 × loge (triglycerides) + 0.139 × BMI + 0.718 × loge (GGT) + 0.053 × waist circumference − 15.745}) / (1 + e^{0.953 × loge (triglycerides) + 0.139 × BMI + 0.718 × loge (GGT) + 0.053 × waist circumference − 15.745}) × 100. Liver steatosis was defined as a FLI ≥ 60, as previously described (Bedogni et al., 2006).

The serum FABP4 levels were assessed using a commercial sandwich enzyme-linked immunosorbent assay kits (Biovendor, Brno, Czechia) with intra- and inter-assay coefficients of variation estimated at 5%. Briefly, serum samples were 10-fold diluted in the sample dilution buffer and then, the diluted samples, the standards and the quality controls were incubated with in duplicates in microplate wells pre-coated with polyclonal anti-human FABP4 antibody for 60 min on an orbital shaker. After washing with the wash solution, wells were incubated with the biotin labeled polyclonal anti-human FABP4 antibody for 60 min, washed and incubated with streptavidin-horseradish peroxidase (HRP) conjugate for 30 min. After a last washing, the substrate solution was added and the microtiter plate were incubated for 10 min protected from the direct sunlight. Then reaction was stopped by addition of acidic solution and absorbance of the resulting product was measured at 450 nm. The serum FABP4 concentrations were determined using a standard curve constructed with the kit’s standards.

The normality of continuous variables was determined by the Kolmogorov–Smirnov test. Continuous variables are expressed as median and interquartile range, unless otherwise indicated. Categorical variables are expressed as numbers with percentages. Differences in the FLI and FABP4 between patients with diabetes, obesity and metabolic syndrome were analyzed by the Mann–Whitney test. Correlations between two variables were performed by using the Spearman’s Rho coefficient (ρ). Univariate and multivariate linear regression models were constructed to search for independent relationships between FLI (dependent variable) and serum FABP4. Data are expressed as standardized beta (β). The study population was stratified in those with FLI ≥ 60 and those with FLI < 60 (Bedogni et al., 2006). Univariate and multivariate logistic binary regression models were performed for dichotomous variables to assess the risk of liver steatosis based on the serum FABP4 levels. The results are presented as odds ratios (OR) and 95% confidence interval (CI). Statistical analyses were performed using SPSS software (IBM SPSS Statistics, version 22.0). Differences were considered statistically significant at P < 0.05.

The clinical, anthropometric and biochemical characteristics of the study population are showed in Supplementary Table 1. Out of 389 subjects included, 191 were men and 197 women. The median age of the study population was 62.0 (25.0–85.0) years. 17.2% of the subjects were usual smokers and 20.8% ex-smokers. A total of 49.1% had a clinical history of arterial hypertension. Diabetes, obesity, and metabolic syndrome were present in 67.9, 51.9, and 74.8%, respectively. A total of 63.8% of the patients showed liver steatosis, with a median FLI of 76.6 (42.7–94.3). The median serum FABP4 levels were 25.3 (16.7–38.3) ng/mL.

Obese individuals showed higher weight, waist circumference and BMI than diabetic and metabolic syndrome individuals. Additionally, obese patients showed higher FLI than the other study groups. Diabetic patients had higher HbA1c and lower triglycerides than the metabolic syndrome individuals. No significant differences were found among the other studied variables in the three study groups (Supplementary Table 2).

We explored the potential correlations between the serum levels of FABP4 and hallmarks of liver injury, including AST, ALT, and GGT (Supplementary Table 3). No significant correlations were found between serum FABP4 and any transaminase, neither in the whole population nor in metabolic patients. Nevertheless, serum FABP4 was found positively correlated with GGT in all studied groups after adjusting for age, gender, glucose, triglycerides, apoA1, and apoB100. Additionally, serum FABP4 was strongly associated with the inflammatory hallmark usCRP in both the whole population and the metabolic patients (P < 0.001 for all comparisons) (Supplementary Table 3).

We next investigated whether FABP4 was associated with liver fat content assessed using the FLI. As it is shown in Figure 1, FLI was increased in diabetic (non-DM: 50.27 ± 3.10%; DM: 74.61 ± 1.50%, P < 0.001), obese (non-OB: 44.58 ± 2.01%; OB: 87.36 ± 0.95%, P < 0.001) and metabolic syndrome patients (non-MS: 26.97 ± 2.38%; MS: 79.15 ± 1.20%, P < 0.001) compared with subjects without the abovementioned metabolic disturbances (Figure 1A). Similarly, serum FABP4 levels were increased in patients with diabetes (non-DM: 25.20 ± 1.26 ng/mL; DM: 36.39 ± 1.82 ng/mL, P < 0.001), obesity (non-OB: 25.51 ± 1.40 ng/mL; OB: 41.24 ± 2.32 ng/mL, P < 0.01) and metabolic syndrome (non-MS: 20.39 ± 1.09 ng/mL; MS: 35.90 ± 1.59 ng/mL, P < 0.001) (Figure 1B).

Serum FABP4 levels positively correlated to FLI in the whole population (ρ = 0.387, P < 0.001), diabetic (ρ = 0.444, P < 0.001), obese (ρ = 0.392, P < 0.001), and metabolic syndrome (ρ = 0.300, P < 0.001) patients. After adjusting for age and gender (^*), or for age, gender, glucose, triglycerides, apoA1, apoB100, and AST (^#), the relationships remained statistically significant (P < 0.001 for all comparisons) (Table 1). Additionally, multivariate linear regression models performed with FLI designated as the dependent variable and with FABP4, age and gender (Model 1), or with FABP4, age, gender, glucose, triglycerides, apoA1, apoB100, and AST (Model 2), showed the independent relationships between FLI and serum FABP4 levels in metabolic subjects (Table 2). In the Model 1, FLI was independently associated with serum FABP4 in the whole population (R² = 0.127, standardized β = 0.323, P < 0.001), as well as in diabetic (R² = 0.165, standardized β = 0.267, P < 0.001), obese (R² = 0.149, standardized β = 0.303, P < 0.001) and metabolic syndrome (R² = 0.170, standardized β = 0.259, P < 0.001) patients. The positive relationships between serum FLI and serum FABP4 were improved in the Model 2 (All: R² = 0.424, standardized β = 0.255, P < 0.001; DM: R² = 0.306, standardized β = 0.251, P < 0.001; OB: R² = 0.290, standardized β = 0.308, P = 0.001; MS: R² = 0.286, standardized β = 0.256, P < 0.001).

Next, the population were stratified in those with FLI ≥ 60 and those with FLI < 60, according to the Bedogni criteria (Bedogni et al., 2006). The serum FABP4 levels were significantly increased in patients with FLI ≥ 60 (36.86 ± 2.13 ng/mL) compared with those with FLI < 60 (23.88 ± 1.17 ng/mL) (~ 1.5-fold, P < 0.01; vs FLI < 60) (Figure 2).

Then, univariate and multivariate logistic binary regression models were performed to assess the risk of liver steatosis (i.e., FLI ≥ 60) based on the serum FABP4 levels (Table 3). Serum FABP4 was positively related with liver steatosis in the whole population, in diabetic and metabolic syndrome, but not in obese patients in the crude model. After adjustment for age and gender (Model 1) and for age, gender, glucose, triglycerides, apoA1, apoB100, and AST (Model 2), the associations between FABP4 and liver steatosis were significant in all the studied groups, including obese patients.