Non-alcoholic fatty liver disease is an influencing factor for the association of SHBG with metabolic syndrome in diabetes patients

Metabolic syndrome (MS) and non-alcoholic fatty liver disease (NAFLD) have been identified as risk factors affecting serum sex hormone binding globulin (SHBG) levels. We conducted this cross-sectional study to delineate whether MS or NAFLD has more impact on circulating SHBG levels in type 2 diabetes (T2D) patients. Anthropometric and biochemical parameters including serums SHBG, testosterone (TT), liver enzymes, lipids, insulin, C-peptide and plasma glucose were measured. Regardless of the MS status, SHBG level was significantly lower in NAFLD patients than in non-NAFLD patients (P < 0.001). In the multiple linear regression analysis, lower serum SHBG level was strongly correlated with a higher incidence of NAFLD, but not MS components. In logistic regression analyses, after adjusted for age, sex, duration of diabetes, smoking status, and alcohol use, the ORs and 95%CI for presence of MS was 2.26 (95%CI 1.91–2.68) and for presence of NAFLD was 6.36 (95%CI 4.87–8.31) with per one SD decrease in serum SHBG (both P < 0.001). In conclusion, lower serum SHBG is associated with a higher prevalence of NAFLD, compared with MS and other metabolic disorders, in T2D patients. NAFLD might be an important influencing factor for the association of circulating SHBG with MS in T2D patients.


Diagnose of NAFLD.
Diagnose of NAFLD was based on abdominal ultrasound performed by experienced ultra-sonographer after excluding other causes of chronic liver disease and excessive alcohol consumption as detailed above. Hepatic steatosis was diagnosed according to the following ultra-sonographic criteria: abnormal intensity and high-level echoes from the hepatic parenchyma, depths of echo penetration into of the liver, liver-kidney differences in echo amplitude and blurred liver vessel structure and abnormal visualization of diaphragm 16 .
In order to determine the relationship of SHBG with the complexity of MS, we divided patients into five subgroups, MS1, MS2, MS3, MS4, and MS5, according to the number of MS components (MS score).
Anthropometric and biochemical measurements. Anthropometric parameters of height, weight and waist circumference (WC) were measured using standard equipment and techniques. Blood was drawn in the morning after a 12-h overnight fast and was analyzed for alanine aminotransferase (ALT), aspartate aminotransferase (AST), gamma-glutamyl transpeptidase (GGT), total cholesterol (TC), triglycerides (TG), high density lipoprotein cholesterol (HDL-C), low density lipoprotein cholesterol (LDL-C), fasting plasma glucose (FPG, glucose oxidize technique method), glycosylated haemoglobin (HbA1c, high-performance liquid chromatography), fasting serum insulin (F-INS), fasting C-peptide (F-CP), total testosterone (TT), and serum SHBG (Siemens healthcare Diagnostics Products limited, Bad Nauheim, Germany) at the central clinical laboratory of the hospital. Body mass index (BMI) was calculated as weight (in kilograms) divided by the square of height (in meters). The insulin resistance was calculated using the homeostasis model assessment method (HOMA-IR = fasting insulin*fasting glucose/22.5) 18 .

Statistical analysis.
Continuous data were expressed as means ± SD or standard error of the mean (SEM).
Categorical data were expressed as percentages. Variables not normally distributed were log-transformed before analyses. Continuous data were tested with independent samples t-test or Mann-Whitney U test, as appropriate. Categorical variables were tested with Pearson χ 2 -test. One-way analysis of variance (ANOVA) followed by Bonferroni post hoc test was used to compare the difference of SHBG between NAFLD and non-NAFLD groups and MS score subgroups. Relationships between SHBG and metabolic parameters were performed using partial correlation analysis adjusted for age and sex. Multiple linear regression models were conducted for the independent variables which significantly related to SHBG in univariate analysis before. The binary logistic regression analyses were used to determine the effects of decreasing serum SHBG levels on prevalence of MS and NAFLD. Potential confounding factors such as age, sex, duration of diabetes, smoking status, alcohol use, BMI, WC, GGT, TG, HDL-C, HOMA-IR and TT, which have been reported in researches to influence MS and NAFLD incidence were taken into consideration in different adjustment models. The results were expressed as odds ratios (OR) and 95% confidence intervals (CI). All statistical analyses were performed using SPSS 20.0 software (SPSS, Chicago, IL, USA). P-value of <0.05 was considered statistically significant. Data availability. The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.

Results
Baseline characteristics of T2D patients with and without NAFLD. According to ultrasonic diagnosis, all T2D patients were divided into NAFLD group (n = 453, age: 54.60 ± 12.92 years) and non-NAFLD group (n = 548, age: 61.75 ± 13.29 years). 71.9% (n = 326) patients were diagnosed with MS in NAFLD group, and 40.1% (n = 220) patients were diagnosed with MS in non-NAFLD group. Variables of BMI, WC, ALT, AST, GGT, TG, FPG, F-INS, fasting C-peptide, and HOMA-IR in the NAFLD group were significantly higher, compared with the non-NAFLD group (all P < 0.001). On the contrary, HDL-C in the NAFLD group was significantly lower than that in the non-NAFLD group (P < 0.001) ( Table 1).

Association of SHBG levels with MS components in patients with and without NAFLD.
To determine the association of SHBG with MS status, we divided all study subjects into groups with or without MS and divided them into five subgroups from M1 to M5 according to the number of metabolic components. As shown in Fig. 1A, SHBG levels in NAFLD group were significantly lower than that in the non-NAFLD group, with ~45.8% and ~39.9% decline in non-MS and MS subgroups [non-MS group: 44.38 ± 0.80 vs. 24.04 ± 1.29 (mean ± SEM) nmol/L; MS group: 39.79 ± 0.98 vs. 23.91 ± 0.83 nmol/L; both P < 0.001]. However, the difference of serum SHBG levels between MS and non-MS subgroups was only statically significant in non-NAFLD group (P = 0.002), but not in NAFLD group (P > 0.05) (Fig. 1A). This analysis shows that the NAFLD condition contributed more to the SHBG reduction (~43% decrease) in both non-MS and MS patients, whereas the MS status contributed to a less degree of SHBG reduction (~10.3%) in non-NAFLD patients (Fig. 1A).
With the number of MS components increasing, serum SHBG levels decreased, with the linear trend statically significant in non-NAFLD group (P for linear trend <0.001, one-way ANOVA), but not in NAFLD group (P for linear trend = 0.059, one-way ANOVA). Mean levels of SHBG in each MS score subgroups of non-NAFLD were significantly higher than those in NAFLD (P < 0.001) (Fig. 1B).

Correlation of serum SHBG with metabolic variables and sex hormones. Correlation analyses
showed that, after adjusted for age and sex, serum SHBG levels were negatively associated with BMI, WC, ALT, GGT, TG, FPG, F-INS, fasting C-peptide, HOMA-IR, and positively associated with HDL-C (P < 0.05), whereas serum SHBG levels were not related to AST, TC, LDL-C or HbA1c (P > 0.05) (Supplementary Table S1).
Factors found to be significantly associated with SHBG in the above correlation analysis were further analyzed for relative significance in multiple linear regression models. As a result, after adjustment for age, sex, BMI, WC, GGT, TG, HDL-C, HOMA-IR, and TT, the association between serum SHBG levels and MS scores reached a statistical significance with standardization coefficient (β) and 95% confidence interval (CI) was −0.165 (95%CI −0.241, −0.088) (P = 0.007). When NAFLD being added into this regression model as an independent variable, a strong correlation was found between lower serum SHBG levels and higher NAFLD incidence with β and 95%CI was −0.295 (95%CI −0.352, −0.239) (P < 0.001). Nevertheless, in this model, no independent relationship was non-NAFLD (n = 548) NAFLD (n = 453) P value found between the serum SHBG level and MS scores with β and 95%CI was −0.081 (95%CI −0.183, 0.021) (P = 0.057) (Fig. 2).

Independent relationship between low serum SHBG and presence of MS/NAFLD. Binary logis-
tic regression analyses were used to determine the effects of decreased serum SHBG levels on the incidence states of either MS or NAFLD. To illustrate the independent relationship between serum SHBG and MS/NAFLD, we conducted the following adjustment models to eliminate the potential confounders. In model a, after adjusted for age, sex, duration of diabetes (quartiles), smoking status, and alcohol use, the ORs and 95%CI for presence of MS   (Fig. 3). Thus, the decrease of SHBG exerts a larger impact on the presence of NAFLD than MS.

Discussion
In the present study, we provide evidence that NAFLD is an influencing factor for the association of serum SHBG with MS in T2D patients. Firstly, low circulating SHBG is prevalent in NAFLD than in non-NAFLD patients, regardless of the MS status. Secondly, although circulating SHBG levels are associated with scores of MS in T2D patients, the association is no longer significant when NAFLD is included in the study. Thirdly, the decrease of circulating SHBG has a larger impact on the presence of NAFLD than MS. Our finding is in agreement with a population-based cohort study by Flechtner-Mors et al. 9 . In their study, ultrasound diagnosed fatty liver was shown to exert a significant influence on serum SHBG concentrations in men and in premenopausal women, while no association between serum SHBG and metabolic syndrome was found. Besides, our data support that fatty liver is perhaps an influencing factor for the association of serum SHBG level with MS in that when the parameter of NAFLD was added into our linear regression model, the association turned to be no longer statistically significant. This clinical finding is also supported by the animal study by Selva et al. 19 that SHBG production is down-regulated in fatty liver due to the reduced expression of HNF-4a.
We also provide evidence that SHBG is more closely associated with NAFLD than MS. Although the result showed a correlation between SHBG and MS status, we found that the NAFLD condition contributed more to the lower SHBG, for that even in patients with all the five MS components, serum SHBG level was still higher in non-NAFLD group than that in NAFLD group (Fig. 1). Furthermore, with per SD decrease of serum SHBG, the ORs for presence of NAFLD increased 5.1-6.3 folds, compared to 1.8-2.2 folds for MS. Association of SHBG with MS has been observed in many studies where NAFLD is not separated from MS 5,6 . Our study provides an explanation for the association in that large percentage of MS patients coexist with NAFLD and that NAFLD and MS share with impaired insulin sensitivity and lipid dyslipidemia 10,20 . Since the association is independent of many known confounding factors such as obesity, insulin resistance, and lipid disorders, it indicates that the association of SHBG with those metabolic traits is probably a consequence of NAFLD on SHBG levels (Fig. 2).
Several studies have highlighted the significance of testosterone deficiency in pathogenesis of hepatic steatosis 21,22 , but these studies have not assessed the effect of SHBG on this association. Our studies indicated that serum SHBG, but not testosterone, is associated with fatty liver or other metabolic disorders. The mechanism for the link between NAFLD and SHBG level remains to be further illustrated 23 . Saez-Lopez et al. 14 recently showed that SHBG plays a suppressive role in hepatic lipogenesis in that SHBG overexpression in NAFLD mouse model ameliorates liver fat accumulation. Hence, there appears a vicious feedback between NAFLD and SHBG in the liver: fat accumulation suppresses SHBG production and the resulting lower SHBG results in fat accumulation. More studies in this respect may lead to new therapeutic approach for NAFLD.

Conclusions
The current study shows that serum SHBG is significantly associated with metabolic disorders of abdominal obesity, liver enzyme, blood lipids, and insulin resistance. However, SHBG levels are significantly lower in NAFLD patients than in non-NAFLD patients regardless of the MS status. Lower serum SHBG is more associated with a higher prevalence of NAFLD, compared with MS and other metabolic disorders. Hence, the reduction of SHBG level in MS and T2D subjects is probably driven by fatty liver.