Serum bile acid and unsaturated fatty acid profiles of non-alcoholic fatty liver disease in type 2 diabetic patients

Figure 4 Logistic regression risk prediction model about non-alcoholic fatty liver disease in type 2 diabetes mellitus. A: Odds ratio (OR) was represented in restricted cubic splines (RCS) nested showing association of palmitoleic acid (C16:1) level on a continuous scale and non-alcoholic fatty liver disease (NAFLD) in type 2 diabetes mellitus (T2DM); B: OR was represented in RCS nested showing association of alpha-octadecatrienoic acid (α-C18:3) level on a continuous scale and NAFLD in T2DM. Shaded areas represent 95% confidence interval (95%CI). 12.9 and 9.21 μmol/L were set as the reference values for C16:1 and α-C18:3 respectively; C: C16:1 was analyzed for their relationship with Triglyceride glucose index (TyG) by threshold; D: α-C18:3 was analyzed for their relationship with TyG by threshold. The statistical comparison was performed by the Wilcoxon test (statistically significant: P < 0.05); E: Comparison of three NAFLD predictive model assessments of the 95%CI of their area under the curve values; F: Comparison of model performance from Decision Curve analysis. Model 1 consists of C16:1, α-C18:3, eicosapentaenoic acid, ω-3 docosapentaenoic acid, γ-linolenic acid. Model 2 consists of sex, age, body mass index, systolic blood pressure, duration of diabetes, TyG, high-density lipoprotein cholesterol, glycosylated hemoglobin, fasting c-peptide, and c-peptide 2 h postprandial. Model 3 was constructed by the factors of model 2 and model 1 together. C16:1: Association of palmitoleic acid; α-C18:3: Alpha-octadecatrienoic acid; 95%CI: 95% confidence interval; AUC: Area under the curve.