To investigate the value of high density lipoprotein-cholesterol (HDL-C) in the diagnosis and risk assessment of non-alcoholic fatty liver disease (NAFLD).

A cross-sectional study and multistage stratified random sampling method were performed in epidemiological survey. According to inclusion and exclusion criteria, a total of 3 312 individuals were enrolled and divided into NAFLD group (913 cases) and non-NAFLD group (2 399 cases). The serum lipid levels were compared between the two groups. Receiver operating characteristic (ROC) curve was performed to evaluate the value of HDL-C in the diagnosis of NAFLD. The binary logistic regression models were established based on HDL-C level.The differences in liver function indexes were compared among the research objects with different HDL-C levels.T test and Mann-Whitney U test were performed for statistical analysis.

The serum levels of total cholesterol, triglyceride and low density lipoprotein-cholesterol (LDL-C) of NAFLD group were all higher than those of non-NAFLD group ((5.24±0.92) mmol/L vs.(4.98±0.92) mmol/L, (1.95±1.41) mmol/L vs.(1.13±0.68) mmol/L, (3.31±0.84) mmol/L vs.(3.09±0.84) mmol/L), and the differences were statistically significant (t=-7.29, -22.38 and -6.84, all P<0.01). However the serum HDL-C level of NAFLD group was lower than that of non-NAFLD group((1.30±0.33) mmol/L vs.(1.64±0.40) mmol/L), and the difference was statistically significant (t=24.93, P<0.01). The incidence of hypercholesterolemia, hypertriglyceridemia, hypo-high-density lipoprotein cholesterolemia and hyper-low-density lipoprotein cholesterolemia of NAFLD group was 48.0%(438/913), 44.8%(409/913), 31.0%(283/913) and 82.8%(756/913), respectively, which were significantly higher than that of non-NAFLD group (36.8%, 882/2 399; 13.2%, 317/2 399; 10.5%, 251/2 399; 71.8%, 1 723/2 399), and the differences were statistically significant (χ²=34.65, 385.43, 206.18 and 42.37, all P<0.01). Using the cut-off values of HDL-C≤1.66 mmol/L in female and≤1.33 mmol/L in male, the area under curve (AUC) values for NAFLD diagnosis were 0.720(95% confidence interval (CI) 0.693 to 0.747) and 0.708 (95%CI 0.679 to 0.737), respectively, the sensitivity was 79.1% and 76.6%, and the specificity was 55.0%and 54.6%. The results of binary logistic regression models based on HDL-C level indicated that prevalence of NAFLD in female with low HDL-C was 4.584 times (95%CI 3.530 to 5.940, P<0.01) higher than that in female with high HDL-C; the prevalence of NAFLD in male with low HDL-C was 3.898 times (95%CI 3.020 to 5.030, P<0.01) higher than that of male with high HDL-C. The alanine aminotransferase (ALT), aspartate transaminase (AST), gamma glutamyl transpeptidase (GGT) and alkaline phosphatase (ALP) levels of low HDL-C group were all higher than those of high HDL-C group (20.10 U/L, 14.40 U/L to 29.40 U/L vs. 16.80 U/L, 12.70 U/L to 23.00 U/L; 19.20 U/L, 16.00 U/L to 23.70 U/L vs. 19.00 U/L, 16.00 U/L to 22.17 U/L; 22.00 U/L, 14.00 U/L to 34.00 U/L vs. 15.00 U/L, 11.00 U/L to 23.00 U/L and 71.00 U/L, 59.00 U/L to 85.00 U/L vs. 66.00 U/L, 55.00 U/L to 82.00 U/L), and the differences were statistically significant (Z=-10.53, -2.20, -14.19 and -5.87, all P<0.05).

The serum HDL-C level is negatively correlated with the risk of NAFLD level, and the NAFLD risk of individuals with low HDL-C level is significantly higher than individuals with high HDL-C level.

Cholesterol, HDL; Non-alcoholic fatty liver disease; Risk factors; Anti-inflammation effects