Characteristics of the subjects

No statistical difference in subject age [32.00 (25.50,41.00) vs. 35.00 (28.00,41.50), p = 0.235], sex [56/35 vs. 59/32, p = 0.645], and serum levels of ALB [48.60 (47.50,50.65) vs. 49.20 (47.85,51.00), p = 0.199] was found between NFs and controls. Significant disparities in the anthropometric and laboratory data were shown in Table 1. As predicted, NF patients exhibited higher BMI than HCs (p < 0.05). Serum concentrations of AST, ALT, FBG, INS, CpS, HbA1c, PA, urea, CREA, TRIG, CHOL, and LDL-C were obviously higher in NFs than HCs (all p < 0.05). PLT and 25(OH) D were lower in NFs than in HCs (all p < 0.05). NFs exhibited higher levels of WBC, NEU, IL-6, and CRP than HCs (all p < 0.05). The two indicators of insulin resistance, HOMA-IR and QUICK, showed notable differences between the two groups. NFS, an internationally recognised biomarker for liver fibrosis degree, was also elevated in NFs.

Table 1 Demographic, anthropometric and laboratory variables in all subjectsIL-38 levels in NAFLD patients

To characterise the status of IL-38 in NFALD cases, we detected the level of serum IL-38 using ELISA. Quantification analysis revealed circulating IL-38 in NFs was dramatically higher than HCs (p < 0.001, Fig. 1a). Based on the ultrasonographic degrees, NFs patients were categorized into three groups: mild group (n = 33), moderate group (n = 31), and severe group (n = 27). We further performed subgroup analysis within these grading groups, and the results showed that cases in the severe group had the highest levels of IL-38, followed by the moderate and mild groups (Fig. 1b). Furthermore, the IL-38 levels of patients with abnormal liver enzyme levels (ALT and/or AST) were higher than in those with normal liver enzyme levels (Fig. 1c). This result suggested that circulating IL-38 might involve in NAFLD pathogenesis.

Fig. 1

The level of serum IL-38 in NAFLD patients. a Comparison of serum IL-38 level in NAFLD cases and healthy control. b Comparison of serum IL-38 level in NAFLD patients with different ultrasonographic degrees. c Comparison of serum IL-38 level in NAFLD patients with abnormal liver enzymes or not

Characteristics of participants according to serum IL-38 tertiles

Participants were classified into tertiles (low, medium, or high) based on serum IL-38 levels to further investigate the correlation these and NAFLD. Table 2 displayed the baseline characteristics of all subjects. Compared with individuals in the lower tertile of IL-38 (T1), those expressing higher levels of IL-38 had higher ALT, AST, FBG, HOMA-IR, CRP, INS, CpS, WBC, NEU, PA, TRIG, and BMI, but lower baseline levels of QUICK, HDL-C, and 25(OH)D. Furthermore, with increasing IL-38, the proportion of NAFLD patients in T1, T2, and T3 was 13.33, 48.39, and 88.33%, respectively, exhibiting a clear increasing trend (p < 0.0001 for the trend).

Table 2 Clinical and laboratory characteristics by tertile distribution of serum IL-38 levelSerum IL-38 and anthropometric/biochemical parameters correlations in NAFLD patients

In NFs, the correlations between serum IL-38 and demographic or anthropometric parameters (age and BMI); liver enzymes (ALT and AST); lipids (CHOL, HDL-C, LDL-C, and TRIG); glucose metabolism (FBG, HbA1c, INS, CpS, HOMA-IR, and QUICK); inflammation indices (WBC, NEU, CRP, and IL-6); indicators of kidney function (urea and CREA); and 25(OH) D were further explored (Fig. 2). Significant positive associations of serum IL-38 levels with HOMA-IR, INS, and CpS were found, whereas QUICK was negatively associated with IL-38 level in glucose metabolism. Positive associations were observed between IL-38 and ALT or AST levels. When inflammation parameters were evaluated, a positive association between serum IL-38 and IL-6 was also observed. IL-38 and 25(OH) D were found to have a negative correlation.

Fig. 2

Correlation analyses between serum IL-38 levels with liver status, glucose metabolism and inflammation-related markers

Odd ratios (ORs) of NAFLD by IL-38 tertiles

The association between serum IL-38 and ORs of NAFLD was evaluated with a multivariable logistic regression analysis. The ORs for NAFLD in T2 or T3 vs. T1 for serum IL-38 were shown in Table 3. The OR of T2 vs. T1 was statistically significant after adjusting age, sex, BMI, ALT, and AST, but it was not significant after adjustment for HDL-C and LDL-C. In the multivariate analysis with adjustment for all the above explanatory variables, the OR of T3 vs. T1 remained a significantly positive determinant of NAFLD. These results indicated that higher serum IL-38 was linked to a higher risk of NAFLD.

Table 3 The association of serum IL-38 levels with risk of NAFLD in different modelsROC curve of IL-38

The ROC curve of serum IL-38 was plotted to predict NAFLD (Fig. 3.). The area under the ROC curve (AUC) was 0.859 (95% CI, 0.807–0.911, p < 0.0001). And at the cut-off value of 132.53 pg/mL for serum IL-38, the sensitivity and specificity were 79.12 and 74.73%, severally.

Fig. 3

ROC curve of serum IL-38 levels in diagnosing NAFLD

Variables related to serum IL-38

In all subjects, Spearman’s analysis revealed that serum IL-38 level was correlated with ALT, AST, PLT, FBG, HOMA-IR, QUICK, INS, CpS, HbA1c, WBC, PA, CREA, TRIG, HDL-C, BMI, IL-6, and 25(OH)D. Based on these results, we performed linear regression analysis using serum IL-38 level as the dependent variable and these above indexes as independent variables. AST, HOMA-IR, and TRIG were all independently linked to serum IL-38 in a multivariate linear regression model (Table 4).

Table 4 Multivariate linear regression analysis of variables associated with serum IL-38