Objective Increased fasting blood ethanol levels, suggested to stem from an increased endogenous ethanol synthesis in the GI tract, are discussed to be critical in the development of non-alcoholic fatty liver disease (NAFLD). The aim of the present study was to further delineate the mechanisms involved in the elevated blood ethanol levels found in patients with NAFLD.
Design In 20 nutritionally and metabolically screened children displaying early signs of NAFLD and 29 controls (aged 5–8 years), ethanol plasma levels were assessed. Ethanol levels along the GI tract, in vena cava and portal vein, intestinal and faecal microbiota, and activity of alcohol dehydrogenase (ADH) and cytochrome P450 2E1 (CYP2E1) were measured in wild-type, ob/ob and anti-TNFα antibody (aT) treated ob/ob mice.
Results Despite not differing in dietary pattern or prevalence of intestinal overgrowth, fasting ethanol levels being positively associated with measures of insulin resistance were significantly higher in children with NAFLD than in controls. Ethanol levels were similar in portal vein and chyme obtained from different parts of the GI tract between groups while ethanol levels in vena cava plasma were significantly higher in ob/ob mice. ADH activity was significantly lower in liver tissue obtained from ob/ob mice in comparison to wild-type controls and ob/ob mice treated with aT.
Conclusions Taken together, our data of animal experiments suggest that increased blood ethanol levels in patients with NAFLD may result from insulin-dependent impairments of ADH activity in liver tissue rather than from an increased endogenous ethanol synthesis.
Trial registration number NCT01306396.
- BACTERIAL OVERGROWTH
- FATTY LIVER
Statistics from Altmetric.com
Significance of this study
What is already known on this subject?
Overweight and obesity are key risk factors for insulin resistance and non-alcoholic fatty liver disease (NAFLD).
Patients with NAFLD and ob/ob mice with NAFLD have higher blood ethanol levels than controls, even in the absence of alcohol consumption.
What are the new findings?
Increased blood ethanol levels in patients with NAFLD correlate with markers of insulin resistance.
Whereas ethanol levels in vena cava plasma were significantly higher in ob/ob mice, ethanol levels were similar in portal vein plasma and chyme obtained from different parts of the GI tract of ob/ob and wild-type mice.
Alcohol dehydrogenase (ADH) activity being significantly lower in liver tissue obtained from ob/ob mice in comparison to wild-type controls was modulated by tumour necrosis factor α (TNFα).
Intestinal microbiota composition in small intestine and faeces was similar in ob/ob and wild-type mice.
How might it impact on clinical practice in the foreseeable future?
Ethanol metabolism may be impaired in patients with insulin resistance and/or NAFLD, and plasma ethanol levels may be a marker for insulin resistance and NAFLD, even in early stages of the disease.
In adults, children and adolescents, overweight and obesity are among the key risk factors for the development of insulin resistance and many chronic diseases related to insulin resistance, including non-alcoholic fatty liver disease (NAFLD).1 ,2 However, an increased body weight seems not to be the only risk factor for the development of NAFLD.3 Indeed, besides a general overnutrition, certain factors in diet, like an elevated intake of sugar and/or fat and also genetic factors are being discussed to be critical in the development and progression of NAFLD.3 Furthermore, in recent years, changes in gut microbiota and barrier function associated with an induction of toll-like receptors (TLRs) in the liver and an increased endogenous ethanol synthesis in the gut are also regarded as being critical in the development of NAFLD;4–9 however, to date, the role of the latter in the development of NAFLD has not been fully understood.
Results of several studies suggest that, in adult and paediatric patients with NAFLD, blood ethanol levels in fasting blood are markedly higher than those of normal-weight controls, despite no marked ethanol intake before blood samples were taken.7 ,10 ,11 In line with these findings, ethanol levels in the breath of genetically obese ob/ob mice suffering from NAFLD due to hyperphagy, induced by a lack of functioning leptin signalling, were also reported to be markedly higher than those of lean littermates.12 Results of a recent study suggest that in stool obtained from paediatric patients with non-alcoholic steatohepatitis, the abundance of alcohol-producing bacteria is markedly higher than in obese or normal-weight children and adolescents, further lending support to the hypothesis that an increased endogenous formation of ethanol may be critical in the development of NAFLD.7 Indeed, acute and chronic exposure to ethanol has been shown to be associated with alterations of the gut barrier function and subsequently leading to an increased translocation of bacterial endotoxins, induction of TLR-4-dependent signalling cascades and increased formation of reactive oxygen species in the liver as well as of tumour necrosis factor α (TNFα) (for overview see ref. 13), all being also discussed as being critical in the development of NAFLD. Starting from this background, the aim of the present study was to further delineate mechanisms involved in the elevated blood ethanol levels found in patients with NAFLD and their relation to the development of NAFLD.
Children were recruited as part of the so-called ‘Hohenheimer Fructose Intervention study’ (HoFI study). The study was approved by the ethics committee of the ‘Landesärztekammer Baden-Württemberg’ (Stuttgart, Germany), and was performed in accordance with the ethical standards laid down in the Declaration of Helsinki of 1975 as revised in 1983. The study is registered at http://www.clinicaltrials.gov (NCT01306396). Written informed consent was obtained from all subjects and their guardians before the study. For the present analysis, participants of the HoFI study were selected according to the following criteria: control group, no signs of health impairments as defined by Weiss et al,2 normal weight (body mass index (BMI) <90 percentile, according to reference data for German children14), age 5–8 years; NAFLD group, presence of steatosis as assessed by ultrasound by an experienced paediatrician using the following system for grading: grade 0, no steatosis; grade 1, mild steatosis; grade 2, moderate steatosis; and grade 3, severe steatosis.15 BMI percentile cut-offs for overweight were used according to the reference data for German children14 to avoid confusion and uncertainties of participants and guardians, and also healthcare providers involved in recruitment of subjects. For further details on health assessment, also see Maier et al.16
Six weeks old C57BL/6J and ob/ob mice (both Janvier SAS, Le-Genest-St-Isle, France) were housed in a pathogen-free barrier facility accredited by the Association for Assessment and Accreditation of Laboratory Animal Care. All procedures were approved by the local Institutional Animal Care and Use Committee. Food (ssniff M-Z Ereich V1184-0, Germany) and tap water were allowed ad libitum prior to experimentation for 2 weeks, and faecal samples were collected in week 2. Some animals were treated with intraperitoneal (i.p.) injections of 10 mg/kg body weight infliximab (anti-TNFα antibody (aT), REMICADE) or vehicle (0.9% NaCl) 4 days and 2 h before sacrifice. Mice were anaesthetised with 80 mg ketamine and 6 mg xylazine/kg body weight (i.p.). Blood was drawn simultaneously from the vena cava and the portal vein. Portions of tissue from liver and GI tract were snap-frozen immediately, fixed in neutral-buffered formalin or transferred immediately to ice-cold phosphate-buffered saline to determine ethanol levels.
All further methods used to determine dietary intake, physical activity and health status of children as well as methods used to analyse mouse tissue and blood samples and cell culture samples are described in detail in online supplementary materials.
Characteristics, health status and nutritional intake of normal-weight healthy children and children with early signs of NAFLD
While age, ethnicity and dietary pattern did not differ between healthy normal-weight children and those with early signs of NAFLD (eg, grade 1 steatosis, slightly higher transaminases than controls), BMI, BMI-SD score and waist circumference were significantly higher in children with early signs of NAFLD when compared with controls (see table 1). Time of physical inactivity per week was longer in children with early signs of NAFLD than in healthy normal-weight controls, and time spent physically active was shorter; however, as activity pattern varied considerably between children, differences did not reach the level of significance (see online supplementary table S2). Furthermore, insulin, alanine amino transferase (ALT) and triglyceride (TG) as well as leptin and TNFα levels determined in fasting blood samples and also homeostasis model assessment for insulin resistance (HOMA-IR) were significantly lower in healthy normal-weight children than in children with early signs of NAFLD (see table 1). Prevalence of small intestinal bacterial overgrowth (SIBO) did not differ between groups (see table 1).
Blood ethanol levels in healthy children and children with early signs of NAFLD
Ethanol levels in fasting plasma samples were significantly higher in children with early signs of NAFLD than in control children (see figure 1). Furthermore, correlation analysis revealed that insulin, leptin and TG levels determined in blood and also HOMA-IR and BMI, BMI- SD score as well as waist circumference were significantly positively related to plasma ethanol levels in children (see table 2). No correlations were found between dietary pattern, physical activity/inactivity, the prevalence of SIBO or TNFα and plasma ethanol levels (data not shown and table 2).
Ethanol concentrations in plasma obtained from vena cava and portal vein of genetically obese ob/ob mice and lean control mice
For ethical reasons, it was neither possible to obtain blood samples from portal vein nor gastric or intestinal juice of children. Therefore, we used ob/ob mice, shown before, to develop NAFLD and insulin resistance (also see online supplementary table S3 for markers of insulin resistance and figure 2A for representative pictures of liver histology of animals used in the present study) and also to have elevated breath ethanol levels,12 to further delineate the origin of the elevated plasma ethanol levels found in children. As our results further suggested that the increased blood alcohol levels found in children with early signs of NAFLD were not related to alterations in diet, all mice were fed the same standard chow. Ethanol levels determined in plasma samples obtained from vena cava of ob/ob mice were significantly higher by approximately 2.2-fold in comparison to normal-weight control mice (p<0.05); however, when comparing plasma ethanol levels in samples from portal vein, no differences between lean and obese mice were found. Plasma levels measured in portal vein of control mice were significantly higher by approximately 2.7-fold when compared with those determined in vena cava (see figure 2B). A similar difference was not found in ob/ob mice.
Ethanol concentrations in chyme obtained from stomach, small intestine and caecum as well as analysis of intestinal microbiota composition of genetically obese ob/ob mice and lean control mice
In line with the findings in plasma samples obtained from the portal vein of mice, ethanol concentrations, being much higher in the more distal parts of the intestine, did not differ between lean and obese ob/ob mice (see figure 3). Interestingly, in colon, ethanol levels were below the level of detection (data not shown). Neither the prevalence of 16S ribosomal RNA gene copy numbers of total eubacteria, Lactobacillus spp., Bifidobacterium spp., Bacteroides–Prevotella–Porphyromonas, nor Enterobacteriaceae, respectively, differed between lean and ob/ob mice in the small intestine. However, in faeces of ob/ob mice, prevalence of 16S ribosomal RNA gene copy numbers of Lactobacillus spp., Bifidobacterium spp., Bacteroides–Prevotella–Porphyromonas and Enterobacteriaceae, but not total eubacteria, was significantly lower than in faeces of lean controls (see table 3). Furthermore, mRNA expression levels of alcohol dehydrogenase (ADH) in duodenum and ileum of mice also did not differ between groups (see figure 4A, B).
Expression and activity of ADH and CYP2E1 in livers of lean control and ob/ob mice
Expression of cytochrome P450 2E1 (CYP2E1) mRNA and activity were significantly lower in livers of ob/ob mice than in lean control animals (see figure 4G, H). Neither mRNA nor protein levels of ADH1 differed between groups (see figure 4C, D). In contrast, total ADH activity was significantly lower in livers of ob/ob mice than in lean control mice (see figure 4E).
ADH activity in AML-12 cells challenged with TNFα
To determine if TNFα, shown repeatedly to be involved in the development of NAFLD and insulin resistance, is involved in the regulation of ADH activity, we challenged alpha mouse liver 12 (AML-12) cells, a mouse hepatocyte cell line, with TNFα for 6 h. Activity of ADH was markedly decreased when cells were challenged with 5 ng TNFα and even more so in cells treated with 10 ng of TNFα (p<0.05) (see figure 4F).
ADH activity and plasma ethanol levels in ob/ob mice after treatment with aT
ADH activity was significantly lower in livers of ob/ob mice treated with vehicle when compared with the lean controls. In livers of ob/ob mice treated with aT, ADH activity was significantly higher than in ob/ob mice treated with vehicle only (+35%); however, hepatic ADH activity was still significantly lower than that of lean controls (see figure 5A). A similar effect was not found for blood ethanol levels, being significantly higher in both ob/ob groups when compared with lean controls (see figure 5B).
Worldwide, the number of adults and children suffering from NAFLD has drastically increased during the past decades.17 Changes in the intestinal microbiota associated with alterations of the intestinal barrier function (for overview see ref. 9) and an elevation of endogenously synthesised ethanol are being discussed to contribute to the development and progression of NAFLD.7 Results of animal and human studies suggest that chronic intake of ethanol is associated with an impaired intestinal barrier function, increased permeation of bacterial endotoxins and subsequently an induction of TLR-dependent signalling cascades in the liver, most of which has also been described in patients with NAFLD.8 Indeed, it has been shown that acute and even more so chronic intake of ethanol can cause endotoxemia,18 and that rodents treated with antibiotics are almost completely protected from the development of alcohol-induced liver damage.19 Alcohol is constantly produced by the intestinal microbiota even in the absence of oral alcohol ingestion.20 ,21 For instance, it has been shown that a diet rich in sugar may lead to increased blood alcohol levels in rodents with a jejunal self-filling diverticulum,22 and that endogenously synthesised ethanol is eliminated by the ADH pathway.21 Furthermore, it has been shown that juvenile and adult patients with manifested NAFLD and mice with NAFLD have markedly higher blood and breath alcohol, and also acetaldehyde levels, even in the absence of alcohol consumption.7 ,11 ,12 ,23 Recent data further suggest that the increased fasting blood alcohol levels found in children and adolescents with non-alcoholic steatohepatitis are associated with an altered faecal microbiota composition;7 however, mechanisms involved in the elevated blood alcohol levels found in patients with NAFLD and their implications for the development of NAFLD have not yet been fully understood. In the present study, we were able to show that fasting plasma ethanol levels are already markedly higher in children suffering from very early signs of NAFLD (eg, steatosis grade 1, as diagnosed by ultrasound in the absence or with only slightly higher transaminases) than in healthy controls. Furthermore, we found that fasting plasma ethanol levels were strongly associated with indices of overweight and obesity, but even more so with markers of insulin resistance; however, no associations with dietary pattern, physical activity or the presence of SIBO were found. Taken together, these data support the hypothesis that increased fasting blood ethanol levels are associated with the development of NAFLD, and also suggest that these alterations already emerge during the very early onset of the disease. Our data also suggest that dietary pattern, physical activity or SIBO may not be key mediators of the increase of blood ethanol levels found in paediatric patients with early signs of NAFLD. Rather, our data imply that alterations in insulin signalling and/or insulin resistance might be critical. However, our data by no means preclude that in other settings and during later phases of the disease, changes in nutritional composition, SIBO and intestinal or faecal microbiota composition might contribute to the increase in blood ethanol levels found in patients with NAFLD. Indeed, due to ethical limitations, we only assessed SIBO by determining hydrogen exhalation after challenging subjects with glucose, allowing for a rough estimation of the presence of SIBO in the small intestine (for overview also see ref. 24). Small-bowel aspiration and quantitative culture, widely called ‘gold standard’, can often not be used due to the invasiveness of this method; however, besides combining different breath and exhalation tests, non-invasive and safe methods allowing for a more precise determination of SIBO are still lacking.24–27 In the present human study, intestinal microbiota composition was not determined for ethical reasons. As to our knowledge, to date, no non-invasive technique is available to obtain intestinal microbiota from upper parts of the GI tract, and faecal microbiota may not always reflect intestinal microbiota in its whole.28
The increase in blood ethanol levels in peripheral blood is not associated with an increased production of ethanol in the GI tract
To further investigate mechanisms involved in the increased fasting plasma ethanol levels found in children with early signs of NAFLD, a mouse model of genetically obese ob/ob mice was used. In line with the findings of Nair et al,10 determining breath ethanol levels, plasma ethanol levels in blood obtained from vena cava were significantly higher in ob/ob mice than in lean control mice. However, ethanol concentrations in blood obtained from portal vein and also in chyme obtained from stomach, duodenum and ileum as well as from caecum did not differ between ob/ob and control mice. Also, expression of ADH1 did not differ between the two groups in duodenum or ileum of mice. Interestingly, in samples obtained from colon, no ethanol was observed. The reason for the lack of ethanol production in this part of the GI tract may be that substances needed for ethanol synthesis, for example, sugars, are normally readily absorbed in the small intestine, and therefore, are not available in colon for ethanol synthesis. These data suggest that the differences found in blood ethanol levels in peripheral plasma of patients and also mice with NAFLD may not result from an altered intestinal microbiota and subsequently an increased synthesis of ethanol in the GI tract of patients with NAFLD, but rather that other mechanisms may be involved in the elevated levels of ethanol found in peripheral blood of these patients. In support, intestinal microbiota composition in small intestine, where marked amounts of ethanol were detected, was similar between groups. However, in line with the findings of others determining intestinal microbiota in further distal parts of the GI tract, for example, caecum,29 in faeces we found marked differences in regard to prevalence of different bacterial populations. Indeed, Ley et al29 also found a statistically significant reduction in Bacteroidetes in ob/ob in comparison to lean mice along with a significantly greater proportion of Firmicutes when sequencing intestinal microbiota obtained from the caecum of animals. Differences in the extent of changes between our study and those of others29 might have resulted from differences in analytical methods used (sequencing vs real-time PCR of selected bacteria), samples analysed (faeces vs caecum) and also differences in diet (ssniff M-Z Ereich V1184-0) versus polysaccharide-rich rodent chow (Purina). Furthermore, differences between our study and that of Zhu et al7 might have resulted from marked differences in study design. Specifically, Zhu et al7 determined microbiota composition in faeces of humans, and not in small intestine or faeces of mice, all fed with the same standardised diet as we did in the present study. Also, in the present study, ethanol was determined in different regions of the body, for example, along the GI tract and in blood samples obtained from different vessels of mice, whereas in our own human studies and the studies of other groups, ethanol levels were either determined in peripheral blood or breath.7 ,10–12 Indeed, as the liver is the major organ involved in ethanol metabolism and clearance, blood ethanol levels determined in peripheral blood may not always reflect blood ethanol levels in the GI tract or portal vein.
The increase in blood ethanol levels in peripheral blood is associated with a decreased activity of ADH in the liver
Earlier studies in rats suggested that activity of ADH, being the key enzyme of alcohol metabolism, is regulated gender-specific, age-specific and weight-specific.30 Indeed, it was shown that in rodents, an extended caloric restriction is associated with a decreased activity of hepatic ADH.31 It further was shown in in-vitro studies that the activity of ADH may, at least in part, be regulated through growth hormones like insulin-like growth factor I and insulin-dependent signalling pathways.32–34 In the studies of Lakshman et al,34 it was further shown that feeding a high carbohydrate, fat-free diet compared with a normal chow diet caused a marked decrease in ADH activity. Furthermore, in diabetic rats, ADH activity was found to be reduced by approximately 53% when compared with control animals.34 In the present study, fasting blood ethanol levels of children were strongly associated with markers of insulin resistance (eg, HOMA-IR, leptin, TG levels) and liver damage (eg, ALT levels) suggesting that alterations in insulin signalling (eg, beginning insulin resistance) may contribute to the increased ethanol levels found in fasting plasma of children suffering from early signs of NAFLD. In line with these findings, in livers of ob/ob mice displaying marked signs of hepatic insulin resistance like a lower expression of insulin receptor and also insulin-receptor substrate and increased levels of TNFα, activity of ADH was significantly lower than in livers of lean controls. Interestingly, neither ADH protein nor mRNA expression levels were altered in liver tissue and CYP2E1 mRNA expression, and activity was even lower in livers of ob/ob mice than in controls. The latter data are in line with those of other groups who also found CYP2E1 activity and mRNA expression to be lower in livers of ob/ob mice in comparison to lean control animals.35 The apparent discrepancy to the findings of others36 might have resulted from differences in dietary regimen (standard chow vs high-fat diet). Indeed, fatty acids have been shown to induce expression of CYPs in vitro.37 ,38 Furthermore, when ADH activity in AML-12 cells challenged with TNFα, at doses shown before to induce insulin resistance, was determined,39 we found a marked reduction in ADH activity. In support of the hypothesis that TNFα and subsequently insulin resistance might be involved in the regulation of ADH activity in settings of NAFLD, ADH activity in livers of ob/ob mice treated with an anti-TNFα antibody twice (4 days and 2 h prior to sacrifice) was markedly higher than in ob/ob mice treated with vehicle (+35%); however, probably due to the fact that hepatic ADH activity did not reach the level of lean controls and also due to the timing of treatment and measurements, blood alcohol levels of ob/ob mice treated with aT were still at the level of vehicle-treated ob/ob mice. Taken together, our mouse and cell-culture data suggest that impairments in insulin signalling may alter ADH activity in the liver, subsequently leading to an impaired ethanol metabolism and elevated blood ethanol levels in patients with NAFLD. Implications of these findings in regard to alcohol clearance after ingestion of alcoholic beverages remain to be determined. However, our animal data suggest that alcohol clearance might be impaired in settings of NAFLD/insulin resistance, thereby leading to (1) an extended direct effect of ethanol on the liver and maybe on the periphery (eg, brain, muscle and adipose tissue) and (2) the synergistic effects of obesity and alcohol consumption reported by others before in regard to the development of liver damage.40 ,41
Taken together, the results of the present study suggest that even very early stages of NAFLD are associated with elevated blood alcohol levels and that this is associated with impairments of insulin signalling in children. Our results from animal studies further suggest that alterations of intestinal or faecal microbiota, resulting in an increased ethanol synthesis in the intestine, and an altered activity of ADH in the liver may be critical factors adding to the elevated blood ethanol levels found in mice and, maybe, also in patients with NAFLD. Furthermore, our data also suggest that ethanol and maybe also metabolites of ethanol, such as acetaldehyde and acetate already shown before to be non-invasive measures in breath, could be used in future for the diagnosis of NAFLD.10 ,23 ,42 However, future studies will have to determine (1) if ADH activity is also lower in livers of patients with NAFLD, (2) if this is a direct effect of an impaired insulin signalling as well as (3) the implications of this for the development of the disease.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
- Data supplement 1 - Online supplement
Contributors AJE conducted research, analysed data and wrote paper. TA, CD and MD conducted research and analysed data. JMS, CS and CJJ conducted research. IBM carried out nutritional interviews and analysed data. EW designed and supervised experiments. IB designed research, wrote paper and had primary responsibility for final content. All authors read and approved the final manuscript.
Funding German Ministry of Education and Science (BMBF), FKZ: 01EA1305 and FKZ: 01KU1214A.
Competing interests None declared.
Ethics approval Landesärztekammer Baden-Württemberg (Stuttgart, Germany).
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.