HIGH ADIPONECTIN AND TNF-α LEVELS IN MODERATE DRINKERS SUFFERING FROM LIVER STEATOSIS: COMPARISON WITH NON-DRINKERS SUFFERING FROM SIMILAR HEPATOPATHY

Moderate alcohol consumption is associated with increased insulin sensitivity and a reduced risk for type 2 diabetes. An important endogenous mediator of insulin sensitivity is adiponectin (AN), an adipokine that displays numerous antiatherogenic, antidiabetogenic and antiinflammatory effects. Recently, acute increase in alcohol consumption has been shown to be associated with increase in plasma adiponectin and, concomitantly, insulin sensitivity. Whether chronic alcohol consumption predicts an increase in plasma AN and whether this is independent of adiposity, markers of liver dysfunction, and plasma adipokines such as tumor necrosis factor (TNF)-alpha is not known. We, therefore, investigated these relationships in 75 men who were diagnosed with liver steatosis using ultrasound/liver biopsy. We examined 75 men, who were diagnosed for having liver steatosis (ultrasound/liver biopsy). Each filled in a questionnaire on alcohol intake. Subjects were divided into two subgroups according to alcohol history and CDT concentrations--drinkers and non-drinkers. All individuals were examined for serum concentrations of AN, glucose, triglycerides, alanine aminotransferase (ALT), aspartate aminotransferase (AST) and glutamate transferase (GMT) activity; carbohydrate-deficient transferrin (CDT%) a marker of chronic alcohol consumption, insulin and TNF-alpha. The Quicki insulin sensitivity index was calculated. Forty-eight individuals were found to be moderate drinkers and 27 subjects non-drinkers. Moderate drinkers had significantly higher concentrations of AN (13.8 +/- 3,7 versus 9.1 +/- 5.4 mg/l, means +/- SD, p = 0.012) compared with non-drinkers, independent of adiposity. Plasma AN concentrations in the whole group were positively correlated with TNF-alpha concentrations (r = 0.6; p = 0.0001), CDT (r = 0.26; p = 0.0084), AST/ALT index (r = 0.3, p = 0.009), AST (r = 0.29; p = 0.011) and GMT (r = 0.29; p = 0.011) and negatively with BMI (r = -0.48; p = 0.0002) and glycemia (r = -0.22; p = 0.049). The positive associations of AN with TNF-alpha (0.8; p = 0.001), CDT (0.55; p = 0.017), AST/ALT index (0.55; p = 0.019) and the negative correlation with glycemia (-0.35; p = 0.0158) were independent of BMI. Stratified according to alcohol intake, in moderate drinkers, a positive correlation was found between AN and TNF-alpha concentrations (r = 0.6, p = 0.0001, AST/ALT index (r = 0.34, p = 0.0295) whereas in non-drinkers no such correlations were found. The concentration of AN and BMI displayed a negative correlation in both drinker and nondrinker patients (r = -0.42, p = 0.01 and -0.61; p = 0.012, respectively). We concluded that plasma AN is higher in moderate drinkers compared to non-drinkers, even after correction for BMI. Drinkers suffering from liver steatosis were found to have a positive correlation between AN concentrations, laboratory markers of liver disease and TNF-alpha. Such correlation was absent in non-drinkers suffering from liver steatosis. This suggests that alcohol may modulate the inhibitory effect of TNF-alpha on AN production, and thus, increase its plasma concentrations.

Experimental and clinical studies have repeatedly confirmed that AN concentration shows a positive correlation with insulin sensitivity and a negative correlation with amount of adipose tissue.Low AN values have been associated with a high basal and reduced insulin-induced phosphorylation of tyroxin receptor for tyrosine kinase in muscle, resulting in progression of insulin resistance 8 .
It has been confirmed that administration of recombinant AN to mice corrected hyperglycemia (reduction of liver gluconeogenesis), decreased insulin resistance and reduced plasma atherogenicity 3,4,9,18 .AN also inhibits cell adhesion and neointimal formation in vessels, thus inhibiting the progression of atherosclerotic alterations 19 .
Moderate alcohol consumption influences many metabolic parameters positively.It is known to be associated with higher HDL concentrations, lower atherogenic risk, lower oxidative stress and higher insulin sensitivity 20 .
The effect of alcohol consumption on AN production and concentration in blood has not yet been investigated.At the time this paper was being prepared, two communications have been published.One was based on a mice experiment in which mice had been administered ethanol as a substantial component of caloric supply.This resulted in a significant decrease of AN in blood 21 .Another investigation made use of human volunteers who consumed alcohol for several weeks which resulted, in contrast to the mice experiment, in increased AN concentrations 20 .
Alcohol consumption may cause liver steatosis.Both alcohol-induced (ASH) and non-alcoholic liver steatosis (NASH) are chronic liver diseases.Their incidence is increasing and is becoming a significant cause of mortality in developed countries. 21NASH occurs mainly in indi-viduals with metabolic syndrome; almost 20 % of affected individuals develop liver cirrhosis 22,23 .
The aim of the present study was to find relations between AN and long-term alcohol consumption with regard to TNF-α concentrations, insulin sensitivity and laboratory markers of liver injury in subjects with liver steatosis.

METHODS
We examined 75 men with liver steatosis diagnosed by sonography, increased laboratory markers and liver biopsy.The subjects filled in a questionnaire on frequency and quantity of alcohol intake.The study was approved by the ethical commission of the Hospital Šternberk.
The group under study was divided into two subgroups -drinkers and non-drinkers according to the following two criteria defined: a) mean alcohol intake > 40 g of alcohol daily for over 4 weeks versus absolute non-drinking for the same period b) CDT concentration (cut-off > 2.75 %) If CDT values discriminated patients into the other group than the one they specified in the questionnaire (non-drinkers with CDT > 2.75 %, drinkers with CDT < 2.75 %), such individuals were excluded from the study.
Blood samples was drawn under aseptic precautions from vena cubiti, samples were centrifuged 10 min in Body weight of patients was measured in their underwear, without any belongings (keys, shoes etc.).

RESULTS
Forty eight probands were classified as moderate drinkers, twenty seven as non-drinkers.Four subjects were excluded due to differences between anamnestic information on alcohol consumption and corresponding CDT % values (see Methods).
Normal distribution of all parameters was found in both subgroups (drinkers and non-drinkers).
Moderate drinkers had significantly higher serum adiponectin concentrations (mean 13.8 vs 9.1 mg/l) and adiponectin after correction to BMI compared to nondrinkers (0.5 vs 0.26).They also showed higher activities of AST, GMT as well as higher medium erythrocyte volume (MCV) compared to non-drinkers.
The stepwise variable regression showed that out of all measured parameters only the adiponectin value could be used to calculate the regression formula to assess drinkers (F 8.44; p = 0.0103, R 2 0.35) (of course, except for CDT % which was used for defining this subgroup).
Anti-inflammatory cytokines have hepatoprotective effect 27 .Decreased concentration of Interleukin-10 in mice model after alcohol intake or its absence in Interleukin-10 knockout mice leads to a markedly enhanced incidence of hepatic necrosis 30 .High adiponectin and TNF-α levels in moderate drinkers suffering from liver steatosis: comparison with non-drinkers suffering from similar hepatopathy The alcohol-induced hepatic involvement is also markedly reduced by administration of antibodies against TNF-α 31 and completely eliminated in TNF-α receptor 1 knock-out mice model 32 .In addition, therapy with thalidomid (inhibitor of TNF-α production in Kupffer cells) inhibits alcohol-induced liver injury 11 .
High concentrations of TNF-α lead to high basal phosphorylation of muscle insulin receptor and is manifested as insulin resistance 34,35 .
One experimental study showed that administration of AN resulted in decreased production of TNF-α by the liver and reduced concentration of TNF-α in circulation 39 .These findings were also confirmed in adiponectin-knockout mice models, where high expression and production of TNF-α were demonstrated 4 .
Enhanced expression of cytokines and chemokines, such as TNF-α, Interleukin-6, Interleukin-18, MCP-1 and resistin is also typical of obese individuals.Obese adipose tissue is infiltrated by cells of monocyte origin producing such peptides 20 This suggests that the conditions leading to enhanced production and activity of cytokines and chemokines (e.g.liver steatosis or obesity) are associated with low levels of adiponectin.However, the very opposite was recorded in our group of alcohol drinkers.
These findings lead us to the hypothesis that alcohol blocks the inhibitory effect of one or more anti-inflammatory factors (e.g.TNF-α) on adiponectin expression and secretion.
Extensive literature documents that chronic alcohol consumption in humans is related to enhanced insulin sensitivity.Many prospective studies proved have shown regular alcohol intake (so-called moderate drinking) is associated with augmented type 2 diabetes and cardiovascular mortality [40][41][42][43][44] .We found three experimental studies dealing with a direct impact of chronic alcohol abuse on insulin sensitivity.One of them reported a significant increase in insulin sensitivity after moderate alcohol intake in 51 postmenopausal women 44 , two other studies proved that alcohol had no significant effect on the parameters of insulin sensitivity 12,[42][43] .
The absence of an effect of chronic alcohol intake on insulin sensitivity in these studies might be due to an insufficient daily alcohol dose, short follow-up and many other factors.
A recent study employing healthy volunteers 20 revealed that regular alcohol intake did not only lead to a decrease but to an increase of AN concentrations and enhanced insulin sensitivity (by 11% after a 4-week consumption of 40 g alcohol/day).TNF-α levels did not change significantly due to drinking.This makes us conclude that similar TNF-α levels in both drinkers and non-drinkers lead to attenuated inhibition of adiponectin expression and/or secretion in drinkers and, subsequently, to higher adiponectin levels measured in circulation.
If alcohol intake results in reduction of the inhibitory effect of TNF-α on adiponectin expression and/or secretion, it may be speculated that alcohol also eliminates or blocks other effects of TNF-α which are causal or supportive on the emergence and propagation of metabolic syndrome and atherosclerosis.
To clarify all these relations, extensive experiments should be carried out, particularly at the molecular level.

Fig. 3 .
Fig. 3. Scatter, adiponectin and CDT %.The results are represented by a whisker box plot where the upper and the lower boundary of the box and the line within the box indicate the 75 th and 25 th percentiles and This study was supported by grantes VVZ MSM 6198959216 from the Ministry of Education, Youth and Sports of the Czech Republic, IGA 1A/8250-2/2004 and 1A/8250-2/2004 from the Ministry of Health of the Czech Republic and IGA.

Table 1 .
Principal statistic in whole group, subgroups of drinkers and non-drinkers.
Normality assessed by Komolgorov Smirnov test High adiponectin and TNF-α levels in moderate drinkers suffering from liver steatosis: comparison with non-drinkers suffering from similar hepatopathy