Relationship of Liver Enzymes to Insulin Sensitivity and Intra-Abdominal Fat

The data presented are baseline measurements from 177 subjects (75 men and 102 women) from a study population of 234 subjects in whom data on insulin sensitivity, body fat distribution, and plasma liver enzyme concentrations were available. There were no significant differences in subject characteristics between all 234 subjects and the 177 who form the basis of the current analysis. The subjects, who had been recruited by advertisement to participate in a study of the effect of egg consumption on plasma lipids in people with various degrees of insulin sensitivity, were aged 31–75 years and were apparently healthy, had no history of diabetes, dyslipidemia, or uncontrolled hypertension, and had no known liver disease (13). Specific testing for liver disease was not performed at the time of the study. Subjects with fasting plasma glucose ≥6.4 mmol/l (≥115 mg/dl), biochemical evidence of renal disease, uncontrolled thyroid disease, coronary or other vascular disease, or anemia were excluded, but formal oral glucose tolerance tests were not performed. The subjects were predominantly Caucasian: Caucasian (n = 161), Asian (n = 5), African American (n = 7), Native American (n = 2), and Hispanic (n = 2). The study was approved by the Human Subjects Review Committee of the University of Washington, and subjects provided written informed consent.

Subjects were divided a priori into three groups on the basis of BMI and insulin sensitivity index (S_I) to analyze the relationship between liver enzyme concentrations, obesity, and insulin sensitivity. These three groups were lean insulin sensitive (LIS) (BMI <27.5 kg/m² and S_I ≥7.0 ×10⁻⁵ min/[pmol/l]), lean insulin resistant (LIR) (BMI <27.5 kg/m² and S_I <7.0 ×10⁻⁵ min/[pmol/l]), and obese insulin resistant (OIR) (BMI ≥27.5 kg/m² and S_I <7.0 ×10⁻⁵ min/[pmol/l]). The cutoff of 27.5 kg/m² was based on the criteria in place before the more recent definition of the criteria for overweight and obesity. The cutoff of 7.0 ×10⁻⁵ min/(pmol/l) for S_I represents the highest value for this parameter among a group of apparently healthy obese subjects studied in Seattle (14). Obese insulin-sensitive subjects were excluded from this analysis because of their small number (n = 8).

The averages of two weight and height measurements were used to calculate BMI as weight in kilograms divided by the square of height in meters. Waist and hip circumferences were calculated as the average of two measurements. Waist circumference was measured at the smallest circumference of the waist, and hip circumference was measured at the widest level of the buttocks, using a previously described protocol (National Health and Nutrition Examination Survey III Anthropometric Measurements. Videotape, National Center for Health Statistics).

A computed tomography scan of the abdomen was performed at the level of the umbilicus to quantify subcutaneous fat (SCF) area and intra-abdominal fat (IAF) area as described previously (15). Fat area was computed as the area with an attenuation range of −250 to −50 Hounsfield units. IAF and SCF areas were quantified by delineating the border of the peritoneal cavity. These measurements were performed by a single observer using standard GE 8800 computer software. The variability of these measures made by a single observer was 1.5% (15).

Subjects underwent a tolbutamide-modified, frequently sampled intravenous glucose tolerance test (FSIGT) to quantify insulin sensitivity as the S_I using Bergman's minimal model of glucose kinetics (16). Three basal blood samples were drawn at 15, 5, and 1 min before the intravenous administration of glucose at time 0. Glucose (11.4 g/m² body surface area) was infused over 1 min, and tolbutamide (125 mg/m² body surface area) was injected intravenously over 30 s at time 20 min. Blood samples were taken at 32 time points over 240 min after commencement of the glucose injection. Fasting glucose and insulin concentrations were calculated as the average of the three basal samples. Liver function tests were performed on the 3-min sample obtained during the FSIGT.

Alcohol intake was assessed using a standardized questionnaire and quantified as self-reported number of drinks per week.

Glucose was measured in duplicate using the glucose oxidase method. Immunoreactive insulin was measured in duplicate by radioimmunoassay using a modification of the double antibody technique. Samples for liver enzymes were assayed between 5 and 7 years after sampling. GGT was measured using an enzymatic colorimetric method (Modular P; Roche Diagnostics, Indianapolis, IN). AST and ALT were measured using the standardized kinetic method (Modular P). Samples were stored at −70°C before assay.

Statistical analyses were performed using SPSS 12.0 (SPSS, Chicago, IL). For regression analysis, dependent variables were logarithmically transformed where appropriate to satisfy the statistical assumptions of linear regression. Multiple regression analysis was used to determine whether associations between the dependent (liver transaminase levels) and independent variables of interest remained significant after adjustments for other potentially confounding independent variables. Model 1 contained S_I, IAF area, BMI, and age for each sex. Model 2 contained S_I, WHR, BMI, and age for each sex. Comparisons between groups were assessed by ANOVA with Tukey post hoc analysis, Kruskal-Wallis test, t test, or Mann-Whitney U test as appropriate. Data are presented as means ± SD unless specified. Non-normally distributed data with kurtosis were log transformed before parametric statistical tests were applied. P < 0.05 was considered significant.

Characteristics for all subjects are shown in Table 1 (n = 177) and subdivided into LIS (n = 53), LIR (n = 60), and OIR (n = 56) subjects and into men (n = 75) and women (n = 102). In this apparently healthy group of nondiabetic subjects, 66% were insulin resistant (defined as S_I <7.0 ×10⁻⁵ min/[pmol/l]) and 32% were obese (defined as BMI >27.5 kg/m²). In accordance with the a priori classification, the BMI of the obese group was significantly higher than that of both of the lean groups (P < 0.0001) (Table 1). S_I values were 2.3- and 2.8-fold higher in the LIS group than in the LIR and OIR groups, respectively (P ≤ 0.0001). The mean age of the LIS subjects was slightly lower than that of the insulin-resistant subjects.

LIR subjects were more centrally obese than LIS subjects, as evidenced by higher WHR (P = 0.009) and IAF area (P < 0.0001), despite a similar BMI in the two groups. LIR subjects were significantly less centrally obese (WHR P = 0.0005; IAF area P < 0.0001) and more insulin sensitive (P = 0.0001) than OIR subjects.

As listed in Table 1, fasting glycemia increased with increasing obesity and insulin resistance (LIS vs. LIR and LIR vs. OIR, P < 0.03; LIS vs. OIR, P < 0.0001), and a similar pattern was seen for triglycerides (LIS vs. LIR P < 0.006; LIR vs. OIR, P < 0.05; LIS vs. OIR, P < 0.0001). Systolic blood pressure was significantly higher in OIR subjects than in LIR and LIS subjects. There was no significant difference in alcohol intake, reported as median number of drinks per week (interquartile range [IQR]) between groups.

There was no sex-based difference in age or BMI (Table 1). As expected, men had higher WHR (P < 0.0001) and IAF area (P < 0.001) than women, whereas women had more SCF area (P < 0.005) than men (Table 1). Fasting glucose was higher in men (P < 0.001), but S_I did not differ between men and women (P = 0.1) (Table 1). All liver transferase levels, reported as median (IQR) were significantly higher in men compared with women: GGT 17 (14) vs. 10 (6) IU/l, ALT 16 (10) vs. 11 (6) IU/l, and AST 21 (7) vs. 17.5 (5) IU/l (P < 0.0001 for each).

Because transaminase levels were significantly higher in men than in women, the effect of obesity and insulin sensitivity on transaminase levels was analyzed separately for each sex. In men, GGT levels were significantly higher in insulin-resistant subjects (LIR and OIR) compared with LIS subjects (Fig. 1A). GGT levels did not differ between LIR and OIR subjects (P = 0.6). In women, GGT levels were also significantly higher in the OIR group than in the LIS group and tended to be higher in the LIR than in the LIS group (P = 0.09) (Fig. 1A). ALT and AST levels did not differ significantly among the LIS, LIR, and OIR groups in either men or women (Figs. 1B and C).

GGT was negatively associated with S_I and positively associated with IAF area, SCF area, WHR, and BMI (Table 2) after adjustment for age and sex. Waist circumference and alcohol consumption were not associated with GGT levels. ALT and AST were not associated with any of the variables and were thus not included in the multiple regression models.

Multiple linear regression analyses stratified by sex were performed with GGT as the dependent variable. In men, only S_I remained significantly associated with GGT levels independent of IAF area and WHR (models 1 and 2 in Table 3), age, and BMI. In contrast, in women, IAF area and WHR (models 1 and 2 in table 3) were significantly associated with GGT levels, but S_I was not.

We examined the relationship between body fat distribution, insulin sensitivity, and liver enzymes in a cohort of 177 nondiabetic subjects of whom >97% had GGT levels within the normal range. It is well recognized that body fat distribution and insulin sensitivity are associated (17,18), and in this cohort of apparently healthy individuals, we found that GGT was negatively associated with insulin sensitivity in men, whereas in women GGT was associated with central obesity. In common with other studies (19), we found that men had higher GGT levels and increased central adiposity than women, and these differences may explain the different results in men and women. ALT and AST were not associated with insulin sensitivity or body fat measures in our study.

The association between elevated liver transaminase levels and insulin resistance in the context of NAFLD is well established (20). In the Tübingen Family Study, GGT was associated with insulin sensitivity and glucose tolerance in both men and women. In addition, in this same study GGT was positively correlated with hepatic lipid content measured by magnetic resonance spectroscopy (21). ALT has previously been shown to be inversely related to insulin sensitivity, determined by the euglycemic clamp, and it has also been shown to have this same relationship with endothelial function in subjects with type 2 diabetes (22). Recently, the role of liver transaminases in predicting the development of type 2 diabetes has been examined in two large studies. In a study of 906 subjects, Hanley et al. (6) found that ALT and, to a lesser extent, AST were associated with the development of diabetes; however, they did not examine whether GGT predicted the development of hyperglycemia. In another study of 5,974 nondiabetic subjects, Sattar et al. (23) found that ALT levels within the normal range predicted incident diabetes. In the Mexico City Diabetes Study, GGT was shown to be an independent risk factor for the development of impaired glucose tolerance and diabetes (24), whereas Vozarova et al. (25) found that only ALT predicted progression to diabetes in Pima Indians.

Although GGT has been widely used as a marker of alcohol consumption, Lee et al. (9) have shown that GGT levels are also associated with an increased risk of development of type 2 diabetes independent of alcohol intake. In another study of >4,000 subjects, although an association between the incidence of diabetes and ALT levels was found, this was most strongly observed in the abnormal range of ALT and was weaker than the association with GGT levels (26). Others have found a strong, independent, and graded association between GGT levels and type 2 diabetes but not ALT or AST levels (7,8,27). However, to our knowledge, no previous study has examined the relationship between GGT, IAF area, and insulin sensitivity in nondiabetic subjects.

The recent emergence of the potential protective role of GGT against oxidative stress may explain the inverse association between GGT levels and insulin sensitivity we found here. The basis of the proposed link between GGT and oxidative stress is that glutathione is a major intracellular defense against free radicals and peroxides. However, as intact glutathione cannot be taken up by cells, the intracellular synthesis of glutathione depends on the metabolism of extracellular glutathione by GGT to release cysteine, which is then transported into the cell and used as a substrate for the de novo intracellular synthesis of glutathione (28). In vitro studies have demonstrated a protective effect of GGT against oxidative stress and cell death (29). Thus, increased GGT expression may initially represent an adaptive protective response to persistent oxidative stress. This would be consistent with the recent in vivo finding of a positive association between GGT and C-reactive protein levels (30). GGT levels have also been shown to predict future levels of inflammatory markers including C-reactive protein, fibrinogen, and F2-isoprostanes (a biomarker of lipid peroxidation) (10).

Yki-Jarvinen's group has shown that fatty liver is associated with fasting insulin as a surrogate measure of insulin sensitivity independently of IAF and SCF areas. In their study ALT was more strongly correlated with liver fat than GGT (31). We found that in men, GGT but not ALT or AST was associated with insulin sensitivity independently of body fat measures. As we quantified insulin sensitivity directly, we believe that our data raise the possibility that GGT may be a more sensitive marker of the liver's response to insulin sensitivity than ALT and AST. The finding that, even across the normal range, GGT levels are related to insulin sensitivity is of clinical relevance in the light of the emerging possible therapeutic role of the peroxisome proliferator–activated receptor-γ agonists in the treatment of NASH. Promrat et al. (32) demonstrated an improvement in transaminases and amelioration of insulin resistance in subjects with NASH after 48 weeks of treatment with pioglitazone. Lifestyle changes with weight loss and increased exercise have also been shown to improve liver enzymes and histological findings in subjects with NAFLD (4). Our data raise the possibility that increasing GGT levels (even within the normal range) in the context of insulin resistance may be an indication for lifestyle changes with the aim of weight loss or treatment with peroxisome proliferator–activated receptor-γ agonists.

The advantages of our analysis are that we examined a large number of subjects in whom insulin sensitivity had been determined by the FSIGT, and all of whom had fat distribution measured using computed tomography scans. However, the lack of any direct measure of hepatic fat is a drawback. Another potential limitation is that because alcohol intake was assessed by self-reported questionnaire, consumption may have been underestimated. Although liver enzyme measurements were made on a sample taken just after glucose administration, we doubt this affected our findings, as nutrient intake has been shown not to affect liver enzyme levels (33). The transferase levels were uniformly lower than would be expected in a normal population, which may be due to the fact that transaminase levels tend to decrease slightly (about 8%) with time, even when stored at −80°C (34,35). However, all samples were handled in the same manner. Although the absolute levels may have been affected, all samples should have been affected to the same degree, and therefore it is likely that although the absolute values may be lower, relative differences would have been robust and maintained.

In summary, GGT but not ALT or AST levels are inversely related to insulin sensitivity independently of central obesity in nondiabetic men. In contrast, in women GGT levels were positively associated with IAF area and WHR but were not associated with insulin sensitivity. If GGT is a marker of hepatic fat accumulation, this sex difference suggests that body fat distribution may be a more important player in the development of hepatic steatosis in women than in men. This finding suggests that GGT is a more sensitive marker of insulin resistance, at least in men, but whether this liver enzyme will prove useful to guide treatment decisions related to insulin resistance awaits further research.

This work was supported by the Medical Research Service of the Veterans Affairs; the American Egg Board; National Institutes of Health Grants DK-02654, DK-17047, DK-35747, DK-35816, HL-30086, HL-07028, RR-37, and RR-16066; the U.S. Department of Agriculture; the McMillen Family Trust; and an American Diabetes Association Distinguished Clinical Scientist Award to S.E.K.

We thank the subjects, Diane Collins, and the nursing staff of the General Clinical Research Center at University of Washington.