Metabolomic profiling reveals novel biomarkers of alcohol intake and alcohol-induced liver injury in community-dwelling men
Metabolomic profiling reveals novel biomarkers of alcohol intake and alcohol-induced liver injury in community-dwelling men
Sei Harada 0 1 2 3
Toru Takebayashi 0 1 2 3
Ayako Kurihara 0 1 2 3
Miki Akiyama 0 1 2 3
Asako Suzuki 0 1 2 3
Yoko Hatakeyama 0 1 2 3
Daisuke Sugiyama 0 1 2 3
Kazuyo Kuwabara 0 1 2 3
Ayano Takeuchi 0 1 2 3
Tomonori Okamura 0 1 2 3
Yuji Nishiwaki 0 1 2 3
Taichiro Tanaka 0 1 2 3
Akiyoshi Hirayama 0 1 2 3
Masahiro Sugimoto 0 1 2 3
Tomoyoshi Soga 0 1 2 3
Masaru Tomita 0 1 2 3
0 Faculty of Environment and Information Studies, Keio University , Fujisawa 252-0882 , Japan
1 Institute for Advanced Biosciences, Keio University , Tsuruoka 997-0052 , Japan
2 Department of Preventive Medicine and Public Health, School of Medicine, Keio University , 35 Shinanomachi, Shinjuku-ku, Tokyo 160-8582 , Japan
3 Division of Environmental and Occupational Health, Department of Social Medicine, Faculty of Medicine, Toho University , Ota-ku, Tokyo 143-0015 , Japan
Objective Metabolomics is a promising approach to the identification of biomarkers in plasma. Here, we performed a population-based, cross-sectional study to identify potential biomarkers of alcohol intake and alcohol-induced liver injury by metabolomic profiling using capillary electrophoresis-mass spectrometry (CE-MS). Methods Fasting plasma samples were collected from 896 Japanese men who participated in the baseline survey of the Tsuruoka Metabolomics Cohort Study, and 115 polar metabolites were identified and absolutely quantified by CE-MS. Information on daily ethanol intake was collected through a standardized, self-administered questionnaire. The associations between ethanol intake and plasma concentration of metabolites were examined. Relationships between metabolite concentrations or their ratios and serum liver enzyme levels in the highest ethanol intake group ([46.0 g/day) were then examined by linear regression analysis. Replication analysis was conducted in 193 samples collected from independent population of this cohort. Results Nineteen metabolites were identified to have an association with daily alcohol consumption both in the original and replication population. Three of these metabolites (threonine, glutamine, and guanidinosuccinate) were found to associate well with elevated levels of serum liver enzymes in the highest ethanol intake group, but not in the non-drinker group. We also found that the glutamate/ glutamine ratio had a much stronger relation to serum cglutamyltransferase, aspartate transaminase, and alanine transaminase than glutamate or glutamine alone (standardized beta = 0.678, 0.558, 0.498, respectively). Conclusions We found 19 metabolites associated with alcohol intake, and three biomarker candidates (threonine, guanidinosuccinate and glutamine) of alcohol-induced liver injury. Glutamate/glutamine ratio might also be good biomarker.
Metabolomics; Alcohol disease; Biomarker; Amino acids; Alcoholic liver
Alcoholic liver disease is a worldwide burden, with
493,300 deaths and 14,544,000 disability-adjusted life
years (DALYs) attributed to alcoholic liver disease
worldwide in 2010, accounting for 0.9 % of all deaths and
0.6 % of all DALYs in that year . Liver diseases have
the highest alcohol-attributable fractions of any disease
besides alcohol disorders and fetal alcohol syndrome, and
alcohol consumption contributes to 50 % of the disease
burden of liver cirrhosis . Liver enzymes such as
cglutamyltransferase (c-GTP) and aminotransferase were
known as biomarkers of alcohol consumption and alcoholic
liver injury. However, further study is needed to clarify
pathological or metabolic alterations leading to liver injury
induced by alcohol intake specifically. Identification of
novel biomarker candidates for monitoring of alcohol
consumption or detecting signs of alcoholic liver disease
could aid in the design of effective prevention programs.
Global metabolomic profiling [3, 4] in human plasma is
a powerful tool for the comprehensive identification of
metabolic alterations caused by chronic alcohol
consumption and ensuing alcoholic liver diseases. The metabolomic
profile reflects cellular activities affected by any and all
possible genetic or environmental factors and facilitates
identification of potential biomarkers of alcohol intake and
alcohol-induced liver diseases [5, 6]. To our knowledge,
the only study of alcohol-induced metabolomic differences
in humans to date focused primarily on lipid metabolism
. However, development of alcohol-induced liver injury
is also accompanied by alterations in carbohydrate and
amino acid metabolism, reflecting the metabolic
consequences of alcohol-induced oxidative stress [8–10]. Among
several methods of metabolomic profiling, the capillary
electrophoresis-mass spectrometry (CE-MS) method has
high separation efficiency and compound identification
capability, which allows the absolute quantification of
global polar metabolites, including carbohydrates and
amino acids [11–14], while being unsuitable for non-polar
metabolites including most of lipid metabolites. We thus
expected to add new insights into alcohol-induced
metabolomic alterations with polar metabolites.
We aimed to find metabolomic biomarker candidates of
alcohol intake and alcohol-induced liver injury based on
the effect on plasma profiles of global polar metabolites.
Here, we performed a cross-sectional study with
CE-MSbased metabolomic profiling in a large, community-based
Materials and methods
The study base was 1017 men and 1109 women aged
35–74 years participating in the ongoing Tsuruoka
Metabolomic Cohort Study, initiated in April 2012 in Tsuruoka
City (Yamagata Prefecture, Japan). The baseline period is
3 years, during which we aim to enroll 10,000 subjects
aged 35–74 years from among participants in annual
municipal or worksite health check-up programs in the city.
The total population of Tsuruoka City at initiation was
136,623, of whom 71,868 were aged 35–74 years. Plasma
metabolomic profiling in the 1017 men and 1109 women
who consented to participate (participation rate 90 %
among health check-up program attendees) in the first
3 months (from April to June 2012) was completed by the
end of 2013. Proportion of regular drinkers was lower in
women than in men (23.8 vs. 74.2 %), as was daily ethanol
intake (median 8.0 vs. 35.7 g/day). Due to the small
number of female participants who were regular drinkers,
only male participants were ultimately included in analysis.
To confirm associations observed in the original dataset,
replication analysis was performed using independent 215
plasma samples. These replication samples were collected
from subsequent male participants of our cohort study
between July and August 2012.
The study was approved by the Medical Ethics
Committee of the School of Medicine, Keio University, Tokyo,
Japan (Approval No 20110264). Informed consent was
obtained from all individual participants included in the
study in written form.
Data and sample collection
All data and samples were obtained during the annual
health check-up, including blood and urine specimens.
Information on drinking, smoking, diet, stress, and physical
activity was collected through a standardized
Alcohol intake per week was calculated from the
frequency of alcohol consumption during a typical week and
the total alcohol intake on each occasion, and then divided
by seven to obtain average alcohol intake per day .
Subjects were then classified as never drinkers, ex-drinkers,
or current drinkers. Current drinkers were defined as
subjects consuming one or more grams of ethanol per day on
To avoid variation due to fasting state and circadian
rhythm, blood samples were collected in the morning
between 8:30 am and 10:30 am after overnight fasting.
Plasma samples were collected with
ethylenediaminetetraacetic acid-2Na as an anticoagulant and kept at 4 C
immediately after collection. The plasma samples were
centrifuged for 15 min (1500 g at 4 C) within 1 h of
collection, divided into aliquots, and kept for a maximum
of 6 h at 4 C until extraction of metabolites. Serum
samples were collected with serum-separating medium and
kept at room temperature after collection. Levels of c-GTP,
serum aspartate transaminase (AST), and alanine
transaminase (ALT) were measured via the Japan Society
of Clinical Chemistry transferable national standardized
method, with c-GTP levels measured via a colorimetric
method and AST and ALT via an ultraviolet
Non-targeted mass spectrometry-based metabolomic
profiling was performed with fasting plasma samples via
capillary electrophoresis time-of-flight mass spectrometry
(CE-TOFMS). Metabolite extraction from plasma was
completed within 6 h after collection to minimize the
effect of metabolic change in plasma. The extraction
method has been described in detail elsewhere .
CE-TOFMS analysis of cationic metabolites and anionic
metabolites was performed as described previously [12,
13]. The raw data were processed using our proprietary
software (MasterHands) [12, 17]. As a preliminary study,
we identified 290 metabolite peaks (131 cations and 159
anions) in plasma; 154 known with standard compounds
and 136 unknown. We decided to routinely measure
absolute concentrations of 115 metabolites (63 cations and
52 anions) a priori that were expected to be stably observed
in most human plasma samples and had matched standards.
To eliminate the possibility of liver damage caused by
diseases other than alcohol liver diseases, we excluded 99
subjects who had any self-reported history of cancer,
positive results on examination for hepatitis B virus surface
antigen, or hepatitis C virus antibody from the original
population. We also excluded 16 subjects without correctly
evaluated alcohol consumption, two whose metabolome
were outliers in the primary component analysis, and four
without overnight fasting. We excluded 22 subjects from
the replication population for the same reasons. The final
dataset included 896 men in the original population and
193 men in the replication population with complete
fasting plasma metabolomics measurement and estimated daily
alcohol intake data.
For metabolomics data, six of 115 metabolites detected
in less than 1 % of subjects were excluded from further
analysis. The remaining 107 metabolite concentrations
were treated as continuous variables, and 49 metabolites of
them were log-transformed according to the shape of
We classified the subjects into four groups by tertile
according to current amount of daily alcohol intake:
nondrinkers and low, middle, and high alcohol intake groups
(low group: 1.0–24.9 g/day, middle group: 25.0–46.0 g/day,
high group: 46.1–205.1 g/day). To examine the association
between daily alcohol intake and 107 metabolite
concentrations as continuous variables (i.e., biomarkers of alcohol
intake), we performed linear regression analysis between
alcohol intake groups (1: non-drinkers, 2: low, 3: middle 4,
high alcohol intake) and each metabolite concentration. We
calculated difference between non-drinkers and high alcohol
intake group for metabolites treated as normal variables, and
calculated fold change from non-drinkers to high alcohol
intake group for metabolites treated as log-transformed
variables, using beta of the linear regression analysis. To
adjust for multiple comparison, we presented p values using
Benjamini and Hochberg’s false discovery rate (BH-FDR)
method (a = 0.05). We also examined the association
between alcohol intake and high-density lipoprotein (HDL)
cholesterol, which is well-known traditional alcohol related
biomarker [18–22], to compare with the metabolite
associations discovered in our study. Low-density lipoprotein
(LDL) -cholesterol and triglyceride were also examined for
the same reason, though associations of these and alcohol
intake vary by studies . In order to confirm whether
found associations (FDR p \ 0.05) of alcohol intake and the
metabolites in the original dataset could be replicated in a
different dataset, linear regression analysis was performed in
the replication population. As a sensitivity analysis, we also
performed the linear regression analysis between each
metabolite concentration and alcohol intake (g/day) as a
continuous variable, excluding top 5 percent drinkers
([92.0 g/day alcohol intake) as outliers.
To further explore whether the identified alcohol intake
biomarkers, whose associations were replicated, were
associated with alcoholic liver injury or not after controlling for
age, we performed linear regression analyses in the highest
tertile group of daily alcohol consumption between alcohol
intake-related metabolites and serum liver enzymes,
including c-GTP, AST, and ALT. To adjust for multiple
comparison, we calculated p values adjusted using the
BHFDR method (a = 0.05). The same analyses were done in
the non-drinker group to examine the possibility that some
metabolites were related to an increase in liver enzymes due
to non-alcoholic liver diseases. We also examined the
association between glutamate/glutamine ratio on serum
cGTP, AST, and ALT via linear regression analysis because
this ratio was reported to correlate well with c-GTP in
patients of alcoholic liver diseases in the previous clinical
study . The metabolites associated with alcoholic liver
injury in the original dataset (FDR p \ 0.05) were also
examined in the replication population.
Sensitivity analysis was also performed in each analysis
by incorporating possible confounders into the constructed
models as continuous variables: age, BMI, smoking
numbers per year, systolic blood pressure, HDL-cholesterol,
and hemoglobin A1c. Daily dietary energy intake, and
daily physical activity were also adjusted in the full model.
We used SAS 9.3 (SAS Institute Inc., Cary, NC) for all
Table 1 Characteristics of original population
High daily activityd, Yes
High dietary intaked, Yes
Non-drinker (n = 231)
Low (n = 220)
Middle (n = 219)
High (n = 226)
ALT alanine aminotransferase, AST aspartate aminotransferase, c-GTP gamma-glutamyl transpeptidase, DBP diastolic blood pressure, FPG
fasting plasma glucose; HDL high-density lipoprotein, IGT impaired glucose tolerance, LDL low-density lipoprotein, SBP systolic blood pressure
a Reported as median (range)
b Reported as mean (standard deviation)
c Reported as geometric mean (range)
d Percent and numbers of the highest quantile are shown
1 Hypertension: Systolic blood pressure C 140 mmHg, diastolic blood pressure C90 mmHg or on medication
2 Impaired glucose tolerance: Glucose C 110 mg/dL, hemoglobinA1c C 6.5 % or on medication
3 Dyslipidemia: Triglyceride C 150 mg/dL, LDL cholesterol C 140 mg/dL, HDL cholesterol B 40 mg/dL or on medication
Table 1 shows the characteristics of the four groups by
alcohol intake in the original population. No marked
differences between groups were observed in body mass index,
fasting glucose, or hemoglobin A1c. Further, no obvious
difference was noted in ALT levels. In contrast, c-GTP and
AST levels were elevated with increasing alcohol intake, as
were systolic blood pressure, diastolic blood pressure,
triglycerides, HDL-cholesterol, and the percentage of
participants with high daily physical activity, while
LDLcholesterol and the percentage of participants with high
dietary energy intake decreased with increasing alcohol
intake. The characteristics of the replication population were
similar to the original population, as shown in eTable 1.
Association between alcohol intake and metabolome
In total, 36 polar metabolites related to alcohol consumption
even after adjusted for age. (FDR p \ 0.05) (Fig. 1; results
of all metabolites in eTables 2, 3). Twenty-seven
metabolites still showed p values less than 0.05 in full-adjusted
Fig. 1 The associations between plasma metabolites and alcohol
intake groups. The associations between plasma metabolites and
alcohol intake groups (1: non-drinkers, 2: low 3: middle 4: high
alcohol intake groups) in the original (a, c) and the replication
population (b, d). Linear regression analysis between each metabolite
and alcohol intake group was performed (p values are shown); then
difference between non-drinkers and the high alcohol intake group for
normal variables (a, b) and fold change for log-transformed variables
(c, d) were calculated using beta of the linear regression analysis. The
metabolites with less than 0.05 FDR p values in the original
model. Exclusion of ex-drinkers (n = 62) from the
nondrinker group also did not change the results substantially.
Of the 27 metabolites associated with alcohol
consumption after full adjustment, 19 associations (70.4 %) were
confirmed in the replication set (p \ 0.05). Thirteen
metabolites were involved in amino acid metabolism, three
in carbohydrate metabolism, two in lipid metabolism, and
one in cofactor and vitamin metabolism. Various amino acid
population were shown in this figure (a, c). Replication analyses were
performed for only these metabolites (b, d). CI Confidence interval,
CSSG Cysteine-glutathione disulfide, FDR False discovery rate, HDL
High-density lipoprotein, LDL Low-density lipoprotein. *Adjusted for
age, BMI, smoking numbers per year, systolic blood pressure,
HDLcholesterol, hemoglobin A1c, daily dietary energy intake, and daily
physical activity. #Adjusted for age, BMI, smoking numbers per year,
systolic blood pressure, hemoglobin A1c, daily dietary energy intake,
and daily physical activity
pathways were associated with alcohol consumption,
including branched chain amino acids, arginine, threonine,
and glutamine. These results were similar in the sensitivity
analysis to examine the association between each metabolite
concentration and alcohol intake (g/day) as a continuous
variable, instead of the alcohol intake group score.
As common lipid biomarkers, the results of
HDL-cholesterol, LDL-cholesterol, and triglyceride were also
Fig. 2 The associations between alcohol-related plasma metabolites
and serum c-GTP, AST and ALT. The associations between
alcoholrelated plasma metabolites and serum c-GTP, AST and ALT in the
high alcohol intake group and non-drinkers were shown. Linear
regression analysis between each alcohol-related metabolite
(logtransformed if necessary) and c-GTP, AST, and ALT
(log-transformed) was performed (p values are shown); then we calculated fold
change and 95 % confidence interval of serum c-GTP, AST, and ALT
per one standard deviation increase in each metabolite using
standardized beta of the linear regression analysis. The metabolites
analyzed. HDL-cholesterol was associated with alcohol
intake as well known. LDL-cholesterol also was associated,
but this association was not replicated. Triglyceride had
nearly one fold change. The magnitude of difference
seemed to be stronger than HDL-C or LDL-C in some of
metabolites such as hippurate and pipecolate, while being
similar in most metabolites.
Association between alcohol-related metabolites
and serum c-GTP, AST, and ALT elevation
Figure 2 and eTables 4–6 indicate the results of the linear
regression analyses of the association between alcohol-related
metabolites and c-GTP, AST, and ALT levels among the high
alcohol intake group. Eight metabolites were associated with
serum c-GTP in the original set, and six metabolites’
[threonine, guanidinosuccinate, glutamine, choline, carnitine, and
cysteine-glutathione disulfide (CSSG)] associations were also
observed in the replication set. These results remained mostly
unchanged even after adjustment for possible confounders.
Threonine, guanidinosuccinate, glutamine, and CSSG also
related to serum AST in both the original set and the
replication set. Because elevation of serum c-GTP and AST in
drinkers often reflects alcoholic liver injury  threonine,
with less than 0.05 false discovery rate p-values in the original
population and glutamine/glutamine ratio were shown in this figure.
Replication analyses were performed for only these variables. ALT
alanine aminotransferase, AST aspartate aminotransferase, CSSG
Cysteine-glutathione disulfide, c-GTP gamma-glutamyl
transpeptidase, SD standard deviation. #log-transformed variables. *Adjusted
for age. **Adjusted for age, BMI, smoking numbers per year, systolic
blood pressure, HDL-cholesterol, hemoglobin A1c, daily dietary
energy intake, and daily physical activity
guanidinosuccinate, glutamine, choline, carnitine, and CSSG
metabolites could be considered to have an association with
alcoholic liver injury.
To determine whether similar associations were
observable or not under ‘non-drinking’ conditions, we also
performed linear regression analyses in the non-drinker
group (Fig. 2; eTables 4–6). Threonine and
2-aminobutyrate (2AB) levels clearly differed from those in the
highintake group. Threonine had negative associations with
serum c-GTP and AST levels among non-drinkers but
positive associations in the high-intake group. These
observations were also found in the replication set. Further,
there was no association between plasma 2AB level and
cGTP level in the non-drinker group while a negative
association was observed in the high-intake group,
although it was not replicated. Guanidinosuccinate,
glutamine, and choline which showed associations (FDR
p \ 0.05) in the high-intake group also had no association
in the non-drinkers. In contrast to these metabolites, CSSG
had strong associations with serum liver enzymes in both
the non-drinker and high-intake groups, indicating that the
observed associations were likely due to factors other than
alcohol intake. These results were consistent even after
excluding ex-drinkers from non-drinkers.
In the present study, we revealed the metabolomic
differences induced by alcohol intake and found new biomarker
candidates of alcohol-induced liver injury in human plasma
using CE-MS-based global metabolomic profiling among
community-dwelling men. Plasma concentrations of 27
metabolites were associated with alcohol consumption
after being adjusted for possible confounders (FDR
p \ 0.05), and 19 metabolites associations were
reconfirmed in the replication set. Among the 19 metabolites,
three of which (threonine, guanidinosuccinate and
glutamine) were simultaneously associated with liver injury
manifested by elevated serum liver enzymes among regular
drinkers. Metabolomic profiling studies for alcohol-related
health effects in humans are scarce. One study of
alcoholinduced metabolomic difference has been previously
reported, focused mainly on lipid metabolome and
suggested that metabolomic profiles based on
phosphatidylcholines, lysophosphatidylcholines, ether lipids, and
sphingolipids form a new class of biomarkers for alcohol
consumption . Our results presented new insights to
understand alcohol-induced alterations in broad
metabolome including both of polar and non-polar metabolites,
adding the findings in polar metabolites such as amino
acids and carbohydrates by using CE-MS platform to
known findings in lipids. To our knowledge, this is the first
epidemiological study to use metabolomics to investigate
potential biomarkers associated with alcohol-related liver
Among the 107 polar metabolites we examined, 19 were
associated with daily alcohol consumption in both the
original and the replication set. In particular, changes in
polar metabolite concentrations in plasma related to
methionine metabolism and glutathione pathway, such as
CSSG, 2AB, and choline, may reflect reactions to oxidative
stress induced by high daily alcohol consumption. Plasma
levels of CSSG, a biomarker of oxidative stress that is
strongly related to hepatic glutathione level , decreased
with increasing alcohol intake. Given that hepatic
glutathione depletion after chronic alcohol consumption has
been demonstrated in both experimental animals as well as
humans  and that most glutathione is rapidly converted
to CSSG in human plasma because of poor stability ,
low plasma CSSG levels suggest that chronic alcohol
consumption may induce an oxidizing state in the liver and
cause subsequent depletion of glutathione in the organ.
Plasma levels of 2AB were also strongly related to
alcohol intake in our results. Acceleration of glutathione
turnover due to alcohol administration might play an
important role in the increase in plasma 2AB because
acceleration of glutathione enhances the conversion of
homocysteine to cysteine but inhibits that of homocysteine
to methionine [8, 10], which results in increased production
of 2AB. Our observation of higher choline and lower
N,Ndimethylglycine concentrations in the drinkers’ plasma
than in nondrinkers’ plasma may support the assumption
that conversion of homocysteine to methionine was
inhibited, as has been previously reported .
2AB has been extensively investigated as a biomarker of
alcohol-related chronic metabolic change. In a healthy
population, active drinkers without liver disease tended to
have a higher serum 2AB concentration than non-drinkers
. In contrast, patients in a clinical study with severely
developed alcohol-induced liver disease had lower serum
2AB concentration than non-drinkers . Consistent with
these findings, we observed that plasma 2AB level
increased with increasing alcohol intake and decreased
with increasing serum c-GTP level only in the high alcohol
intake group, though this finding was not confirmed in the
replication set. These findings may indicate that plasma
2AB levels increase due to oxidative stress with alcohol
intake in healthy people but decrease once hepatic
pathological change is observed.
Of note, plasma threonine had strong positive
associations with c-GTP and AST levels only in the high alcohol
intake group, while a weak negative association was
observed among non-drinkers as well as low and middle
alcohol intake groups. In the high alcohol intake group,
geometric mean concentrations of plasma threonine were
obviously different between subjects with increased AST
and normal serum AST [172.9 lM for subjects with
increased AST (C40 IU/L) and 134.4 lM for those with
normal serum AST (\40 IU/L); p \ 0.0001]. On the other
hand, this association was not observed in the other groups,
including non-drinkers (117.4 lM for the AST C 40 IU/L
group and 127.9 lM for the AST \ 40 IU/L group).
Further, plasma threonine level exceeded 250 lM when the
analysis was restricted to the high alcohol intake group
with AST C 100 IU/L, although this group consisted of
only 2 subjects. These observations suggest that threonine
may be a specific biomarker of alcohol-induced liver
The mechanism by which plasma threonine is elevated
among heavy drinkers with accompanying liver damage is
unclear, with one possible explanation being that the main
metabolic pathway of threonine in humans, catabolism to
2-oxobutyrate via serine dehydratase , might be
inhibited, as plasma 2AB concentration, a good biomarker
of alcohol drinking and one of the end products of
2-oxobutyrate, is decreased only when alcohol-induced
liver injury exists. A second possibility is that intake of
threonine, an essential amino acid, is increased in the high
alcohol intake group; this seems unlikely, however,
because essential amino acid concentrations other than
threonine were lower in the high alcohol intake group than
in the other groups (shown in eTables 2, 3), indicating that
heavy drinkers tended to eat less. A previous study
suggested that levels of both threonine and 2AB in plasma
should increase when threonine intake is high . Thus,
threonine might be an important biomarker candidate of
alcohol-induced liver injury, reflecting metabolic
alterations in the liver.
Our findings that plasma glutamine concentration
decreased in an alcohol intake-dependent manner, while a
decrease in plasma glutamine was associated with
elevation of serum liver enzymes in the high alcohol intake
group only, suggest that glutamine might play an important
role in protecting the liver from alcohol. These results are
consistent with a previous clinical study showing that
drinking patients with alcohol-induced liver diseases had
low plasma glutamine levels compared to healthy people or
patients with non-alcoholic liver diseases, which was
alleviated after one month of alcohol abstinence . This
putative protective role of glutamine is supported by the
finding that pretreatment with glutamine prevented
ethanol-induced liver injury in mice by improving
ethanol-induced inflammatory response . Our observation of
lower p values in glutamate/glutamine ratio suggests that
alteration of glutamate and glutamine pathway is strongly
related to alcoholic liver injury.
Guanidinosuccinate concentration also decreased in an
alcohol intake-dependent manner. In addition,
guanidinosuccinate concentration had highly negative association
with elevation of serum liver enzymes only in the high
alcohol intake group. This finding suggests that
guanidinosuccinate declines specifically in alcohol-induced liver
injury. Considering the previous report which showed that
alcoholic cirrhotic patients had much lower serum
guanidinosuccinate than non-alcoholic cirrhotic patients or
controls , guanidinosuccinate could be a biomarker
candidate to discriminate alcohol-induced liver injury from
other liver injuries.
Although this study was performed using carefully
designed epidemiological protocols to minimize
measurement errors, some limitations warrant mention. First, the
study was conducted under a cross-sectional design, and
the temporality of cause-effect relationships is not assured.
Reverse causality is likely to occur if a subject who
experienced liver enzyme abnormalities changed his
drinking habit. Careful follow-up and intervention studies
with proper assessment of liver functions are further
needed to overcome these limitations. Second, various factors
might have influenced measurement variability in the
metabolomic analysis, in turn potentially causing both
random and biased misclassification of metabolomic data.
Sampson et al. indicated that within-subject variability in
assay accounted for the majority of variability in more than
half of metabolites they measured using LC–MS . To
minimize variability in our study, we set a uniform fasting
condition on study participants and standardized the quality
control procedures for metabolomic analysis. Third,
information was obtained as average alcohol intake for a
typical day only. However, metabolic pathways might be
affected by unusual alcohol intake, such as binging.
In conclusion, we found 19 metabolites different for
alcohol intake, and three biomarker candidates (threonine,
guanidinosuccinate and glutamine) of alcohol-induced
liver injury. The glutamate/glutamine ratio might also be a
good biomarker. Follow-up study of the subjects to
elucidate causality is now on-going.
Acknowledgments We thank the study participants and members
of the Tsuruoka Metabolomics Cohort Study team, especially Mitsu
Narutomi, Yuko Ando, Sayaka Togashi, and Miki Kudo.
Compliance with ethical standards
Funding sources This work was supported in part by research funds
from the Yamagata Prefectural Government and the city of Tsuruoka
and by a Grant-in-Aid for Scientific Research (Grant Number
24390168) from the Japan Society for the Promotion of Science.
Open Access This article is distributed under the terms of the
Creative Commons Attribution 4.0 International License (http://crea
tivecommons.org/licenses/by/4.0/), which permits unrestricted use,
distribution, and reproduction in any medium, provided you give
appropriate credit to the original author(s) and the source, provide a
link to the Creative Commons license, and indicate if changes were
1. Rehm J , Samokhvalov AV , Shield KD . Global burden of alcoholic liver diseases . J Hepatol . 2013 ; 59 : 160 - 8 .
2. World Health Organization. Global status report on alcohol and health 2014 . Geneva: World Health Organization; 2014 (cited 2015 July 17) . http://www.who. int/substance_abuse/publications/ global_alcohol_report/msb_gsr_ 2014 _1.pdf?ua=1.
3. Lutz NW , Sweedler JV , Wevers RA . Methodologies for metabolomics: experimental strategies and techniques . Cambridge : Cambridge University Press ; 2013 .
4. Issaq HJ , Van QN , Waybright TJ , Muschik GM , Veenstra TD . Analytical and statistical approaches to metabolomics research . J Sep Sci . 2009 ; 32 : 2183 - 99 .
5. Suhre K , Shin S , Petersen A. Human metabolic individuality in biomedical and pharmaceutical research . Nature . 2011 ; 477 : 54 - 60 .
6. Warren KR , Murray MM . Alcoholic liver disease and pancreatitis: global health problems being addressed by the US National Institute on Alcohol Abuse and Alcoholism . J Gastroenterol Hepatol . 2013 ; 28 (Suppl 1): 4 - 6 .
7. Jaremek M , Yu Z , Mangino M , Mittelstrass K , Prehn C , Singmann P , et al. Alcohol-induced metabolomic differences in humans . Transl Psychiatry . 2013 ; 3 : e276 .
8. Lieber CS . Role of oxidative stress and antioxidant therapy in alcoholic and nonalcoholic liver diseases . Adv Pharmacol . 1997 ; 38 : 601 - 28 .
9. Zakhari S , Li TK . Determinants of alcohol use and abuse: impact of quantity and frequency patterns on liver disease . Hepatology . 2007 ; 46 : 2032 - 9 .
10. Kharbanda KK . Methionine metabolic pathway in alcoholic liver injury . Curr Opin Clin Nutr Metab Care . 2013 ; 16 : 89 - 95 .
11. van der Werf MJ , Overkamp KM , Muilwijk B , Coulier L , Hankemeier T. Microbial metabolomics: toward a platform with full metabolome coverage . Anal Biochem . 2007 ; 370 : 17 - 25 .
12. Sugimoto M , Wong DT , Hirayama A , Soga T , Tomita M. Capillary electrophoresis mass spectrometry-based saliva metabolomics identified oral, breast and pancreatic cancer-specific profiles . Metabolomics . 2010 ; 6 : 78 - 95 .
13. Soga T , Igarashi K , Ito C , Mizobuchi K , Zimmermann H-P , Tomita M. Metabolomic profiling of anionic metabolites by capillary electrophoresis mass spectrometry . Anal Chem . 2009 ; 81 : 6165 - 74 .
14. Ramautar R , Somsen GW , de Jong GJ . CE-MS in metabolomics . Electrophoresis . 2009 ; 30 : 276 - 91 .
15. Okamura T , Tanaka T , Yoshita K , Chiba N , Takebayashi T , Kikuchi Y , et al. Specific alcoholic beverage and blood pressure in a middle-aged Japanese population: the High-risk and Population Strategy for Occupational Health Promotion (HIPOP-OHP) Study . J Hum Hypertens . 2004 ; 18 : 9 - 16 .
16. Hirayama A , Nakashima E , Sugimoto M , Akiyama S , Sato W , Maruyama S , et al. Metabolic profiling reveals new serum biomarkers for differentiating diabetic nephropathy . Anal Bioanal Chem . 2012 ; 404 : 3101 - 9 .
17. Hirayama A , Kami K , Sugimoto M , Sugawara M , Toki N , Onozuka H , et al. Quantitative metabolome profiling of colon and stomach cancer microenvironment by capillary electrophoresis time-of-flight mass spectrometry . Cancer Res . 2009 ; 69 : 4918 - 25 .
18. Gardner CD , Tribble DL , Young DR , Ahn D , Fortmann SP . Associations of HDL, HDL( 2 ), and HDL (3) cholesterol and apolipoproteins A-I and B with lifestyle factors in healthy women and men: the Stanford Five City Project . Prev Med . 2000 ; 31 : 346 - 56 .
19. Brinton EA . Effects of ethanol intake on lipoproteins . Curr. Atheroscler Rep . 2012 ; 14 : 108 - 14 .
20. Tanaka H , Ikai E , Yamada Y. Genetic polymorphisms in alcohol metabolizing enzymes as related to sensitivity to alcohol-induced health effects . Environ Health Prev Med . 1997 ; 1 : 193 - 200 .
21. Yamada Y , Noborisaka Y , Ishizaki M , Tsuritani I , Honda R. Low fasting serum insulin in Japanese alcohol consumers does not imply improved coronary risk factors . Environ Health Prev Med . 2004 ; 9 : 243 - 50 .
22. Nakamura S , Ito Y , Suzuki K , Hashimoto S. Blood pressure, levels of serum lipids, liver enzymes and blood glucose by aldehyde dehydrogenase 2 and drinking habit in Japanese men . Environ Health Prev Med . 2006 ; 11 : 82 - 8 .
23. Tominaga T , Suzuki H , Mizuno H , Kouno M , Suzuki M , Kato Y , et al. Clinical significance of measuring plasma concentrations of glutamine and glutamate in alcoholic liver diseases . Alcohol Alcohol Suppl . 1993 ;1A: 103 - 9 .
24. Reichling JJ , Kaplan MM . Clinical use of serum enzymes in liver disease . Dig Dis Sci . 1988 ; 33 : 1601 - 14 .
25. Jones DP , Carlson JL , Mody VC , Cai J , Lynn MJ , Sternberg P. Redox state of glutathione in human plasma . Free Radic Biol Med . 2000 ; 28 : 625 - 35 .
26. Kleinman WA , Richie JP . Status of glutathione and other thiols and disulfides in human plasma . Biochem Pharmacol . 2000 ; 60 : 19 - 29 .
27. Medici V , Peerson JM , Stabler SP , French SW , Gregory JF , Virata MC , et al. Impaired homocysteine transsulfuration is an indicator of alcoholic liver disease . J Hepatol . 2010 ; 53 : 551 - 7 .
28. Musshoff F , Daldrup T. Determination of biological markers for alcohol abuse . J Chromatogr B Biomed Sci Appl . 1998 ; 713 : 245 - 64 .
29. Darling PB , Grunow J , Rafii M , Brookes S , Ball RO , Pencharz PB . Threonine dehydrogenase is a minor degradative pathway of threonine catabolism in adult humans . Am J Physiol Endocrinol Metab . 2000 ; 278 : E877 - 84 .
30. Peng H , Chen Y , Chen J. Effects of glutamine administration on inflammatory responses in chronic ethanol-fed rats . J Nutr Biochem . 2011 ; 22 : 282 - 8 .
31. Marescau B , De Deyn PP , Holvoet J , Possemiers I , Nagels G , Saxena V , et al. Guanidino compounds in serum and urine of cirrhotic patients . Metabolism . 1995 ; 44 : 584 - 8 .
32. Sampson JN , Boca SM , Shu XO , Stolzenberg-Solomon RZ , Matthews CE , Hsing AW , et al. Metabolomics in epidemiology: sources of variability in metabolite measurements and implications . Cancer Epidemiol Biomark Prev . 2013 ; 22 : 631 - 40 .