Journal of Viral Hepatitis, 2011
Metabonomic analysis of hepatitis E patients shows deregulated
metabolic cycles and abnormalities in amino acid metabolism
S. U. Munshi,1 S. Taneja,1 N. S. Bhavesh,2 J. Shastri,3 R. Aggarwal4 and S. Jameel1 1Virology Group
and 2Structural and Computational Biology Group, International Centre for Genetic Engineering and Biotechnology, New Delhi; 3Department of
Microbiology, Kasturba Hospital for Infectious Diseases, Mumbai; and 4Department of Gastroenterology, Sanjay Gandhi Postgraduate Institute of Medical
Sciences, Lucknow, India
Received February 2011; accpeted for publication April 2011
SUMMARY. Hepatitis E, which is endemic to resource-poor
regions of the world, is largely an acute and self-limiting
disease, but some patients have an increased susceptibility to
develop fulminant hepatitis. The pathogenesis of hepatitis E
in humans is poorly characterized. To understand the met-
abolic pathways involved in the pathophysiology of hepatitis
E, we have used 1H nuclear magnetic resonance spectros-
copy to quantify various metabolites in the plasma and urine
of the patients with hepatitis E. These were compared with
specimens from patients with acute hepatitis B as disease
controls and healthy volunteers. Data were analysed using
chemometric statistical methods and metabolite databases.
The main metabonomic changes found in patients with
hepatitis E, but not in those with hepatitis B, included
increased plasma levels of L-isoleucine, acetone, and glycerol,
reduced plasma levels of glycine, and reduced urinary levels
of imidazole, 3-aminoisobutanoic acid, 1-methylnicotina-
INTRODUCTION
Hepatitis E is an acute and largely self-limiting form of viral
hepatitis that is caused by hepatitis E virus (HEV) [1]. The
Abbreviations: AAA, aromatic amino acid; BCAA, branched chain
amino acid; CPMG, Carr–Purcell–Meiboom–Gill; EBAM, Empirical
Bayes Analysis of Microarrays; FID, free induction decay; GC-MS,
Gas Chromatography–Mass Spectrometry; HBV, hepatitis B virus;
HCV, hepatitis C virus; HEV, hepatitis E virus; HIV, human immu-
nodeficiency virus; HMDB, Human Metabolome Data Bank; MetPA,
Metabolomic Pathway Analysis; NMR, nuclear magnetic resonance;
OTC, ornithine transcarbamoylase; PCA, Principal Component
Analysis; PLS, partial least squares; PLS-DA, Partial Least Squares
Discriminant Analysis; SAM, Significance Analysis of Microarrays
(and Metabolites); SER, smooth endoplasmic reticulum; VIP, vari-
able importance in projection.
Correspondence: Dr. Shahid Jameel, Virology Group, International
Centre for Genetic Engineering and Biotechnology, Aruna Asaf Ali
Marg, New Delhi–110067, India.
E-mail: shahid@icgeb.res.in
! 2011 Blackwell Publishing Ltd
doi:10.1111/j.1365-2893.2011.01488.
mide, biopterin, adenosine, 1-methylhistidine, and salicylu-
ric acid. Patients with hepatitis E or B both showed increased
levels of plasma and urinary l-proline and decreased levels of
various other metabolites. Pathway analysis tools suggest
the involvement of glycolysis, tricarboxylic acid cycle, urea
cycle, and amino acid metabolism in patients with acute
hepatitis E. These findings may help better understand the
clinical and biochemical manifestations in this disease and
the underlying pathophysiologic processes. Based on our
findings, it would be worthwhile determining whether
patients with hepatitis E are more prone to develop lactic
acidosis and ketosis compared with other forms of viral
hepatitis.
Keywords: hepatitis E, hepatitis E virus, metabonomics,
nuclear magnetic resonance.
virus is transmitted by the feco-oral route, primarily through
contaminated drinking water and uncooked meat, causes
both sporadic disease and large outbreaks, and is endemic in
many developing countries [1]. Although children are also
infected, the highest attack rates of the disease are observed
in young adults, aged 15–40 years [2]. Some infected per-
sons, especially pregnant women, show the propensity to
develop acute hepatic failure with high mortality [3,4].
As the liver is the main metabolic, biosynthetic, storage
and detoxifying organ, its infection by HEV causes a broad
range of clinical manifestations, most of which are related to
hepatocellular damage [2]. We have recently observed that
HEV modulates the in vitro gene expression pattern of hepatic
cells (unpublished data). Patients with hepatitis E also show
the differential expression of various acute-phase proteins
that are synthesized in the liver, in their plasma, and urine
[5]. Liver being a major metabolic organ, its infection is also
expected to result in measurable changes in metabolite levels.
Metabonomics includes quantitative measurements of
the dynamic multiparametric metabolic response of living
2 S. U. Munshi et al.
systems to genetic modification or pathophysiological stimuli
including dietary disease processes and alterations in gut
microflora. Comprehensive analysis of changes in the profiles
of metabolites in an organic biofluid has emerged as an
important area of investigation to differentiate between
healthy and diseased persons, to gain insights into disease
pathogenesis, and to develop prognostic biomarkers [6–8].
Nuclear magnetic resonance (NMR) spectroscopy is a pop-
ular tool for analysing metabonomic changes in biofluids
because of its noninvasiveness, rapidity, reproducibility,
need for small sample volumes, and high accuracy in iden-
tifying and quantifying metabolites [8].
The metabolite fingerprints in biofluids from patients
chronically infected with hepatitis B virus (HBV) [9], hepa-
titis C virus (HCV) [10], and HIV [11] have previously been
generated using high-resolution 1H NMR and Gas Chroma-
tography–Mass Spectrometry. This study was aimed at
identifying metabonomic changes during hepatitis E. To
examine whether these changes are unique to hepatitis E,
we also studied the patients with acute hepatitis B as disease
controls. We profiled the plasma and urine from patients
with acute hepatitis E and B and healthy controls using
1
H-NMR and carried out multivariate statistical analyses.
This has allowed us to observe metabonomic changes, de-
velop a working model of hepatitis E pathogenesis, and
identify potential biomarkers for this disease.
MATERIALS AND METHODS
Subjects
Blood and urine were collected from patients with acute viral
hepatitis who were inpatients or outpatients at one of the
collaborating hospitals, namely Army Base Hospital (New
Delhi, India), Sanjay Gandhi Postgraduate Institute of Med-
ical Sciences (Lucknow, India), or Kasturba Hospital for
Infectious Diseases (Mumbai, India). Ethics Committees of all
these institutions approved the study, and all patients and
volunteers provided the informed consent.
Clinical specimens
Blood (5–6 mL) was collected in EDTA-coated vacutainers
(Becton Dickinson, Franklin Lakes, NJ, USA). The tubes were
centrifuged at 800xg for 5 min at 4 "C, and the plasma was
kept frozen in aliquots at )70 "C. First morning, mid-stream
urine (50–150 mL) was collected in an autoclaved bottle
and kept on ice till freezing in aliquots at )70 "C. Specimens
were similarly obtained from healthy volunteers. Plasma
was tested for viral markers using commercial enzyme
immunoassays, according to the manufacturers! instructions
as follows: anti-HEV IgM (MP Biomedicals Asia Pacific Pvt.
Ltd, Singapore), HBsAg (Hepalisa; J Mitra and Co. Pvt. Ltd,
New Delhi, India), anti-HBc IgM (Anticorase MB-96; General
Biologicals Corp., Taipei, Taiwan), and anti-HCV (Hep-Chex-
C; XCyton Diagnostic Pvt. Ltd, Bangalore, India). Patients
testing positive only for anti-HEV IgM were diagnosed as
having acute hepatitis E. Those testing positive for HBsAg
were tested for anti-HBc IgM; those found positive were
diagnosed as having acute hepatitis B. Patients testing
positive for both HBsAg and anti-HEV IgM were considered
as having dual infection and were excluded. Patients with
positive anti-HCV test were also excluded.
NMR spectroscopy
For 1H NMR spectroscopy, aliquots of the plasma and urine
were lyophilized to reduce the water signal and reconstituted
in 400 lL of D2O (Deuterium oxide, 99.8% atom D) and
100 lL Tetramethylsilane (TMS, NMR grade, 99.9+%) (both
from Sigma Aldrich, St. Louis, MO, USA). All NMR experi-
ments were carried out at 298K on a Bruker Avance III
spectrometer equipped with cryogenic triple-resonance
probes, operating at the proton field strength of 500 MHz.
Temperature calibration was performed using 100%
2
H-Methanol [12]. The 1H spectra of urine were measured
using normal acquisition with presaturation. For plasma,
the Carr–Purcell–Meiboom–Gill (CPMG) echo sequence was
used to partially suppress the protein signals. Based on the
experimental optimization, a total CPMG delay of 300 ms
was used with an echo time of 200 ls. One-dimensional
1
H NMR spectra were measured using 32 scans with a
relaxation delay of 4 s. For each sample, FIDs were collected
with a spectral width of 6009.62 Hz and acquisition time of
2.73 s (t1max). The spectra were referenced to TMS at 0 ppm
and were processed using Topspin 2.1 (Bruker AG, Fal-
laenden, Switzerland). The spectral region 6.0–4.5 ppm was
set to zero integral to remove interference from urea and
water signals.
Data analysis
Several statistical tools allow simple, rapid, and reliable
analysis of complex NMR spectra and permit extraction of
hidden useful patterns [7,8]. Statistical analyses and spectral
annotations were carried out with "MetaboAnalyst! (http://
www.metaboanalyst.ca/MetaboAnalyst/faces/Home.jsp).
This module contains several statistical and machine
learning algorithms, including chemometric analysis [Prin-
cipal Component Analysis (PCA), Partial Least Squares Dis-
criminant Analysis (PLS-DA)] and feature selection path
[Significance Analysis of Microarrays (and Metabolites;
SAM), Empirical Bayes Analysis of Microarrays, etc.). We
analysed the 1H NMR spectra using PCA, PLS-DA, and SAM.
For NMR peaks found through different statistical analyses,
corresponding metabolites were identified from a library in
Human Metabolome Data Bank (HMDB; http://www.
hmdb.ca) and MetaboMiner, linked with MetaboAnalyst.
Each metabolite was also confirmed by comparing its 1H and
13
C chemical shifts and coupling patterns with values pre-
! 2011 Blackwell Publishing Ltd

viously published [13,14] or available from HMDB. The
pathways most involved in acute hepatitis E and B were then
identified using MetPA, a pathway enrichment analysis
software (http://metpa.metabolomics.ca).
RESULTS
Specimens from patients with acute hepatitis E (26 plasma,
24 urine), those with acute hepatitis B (six plasma, five
urine), and healthy controls (18 plasma, 10 urine) were
studied. All the patients were men with clinical and bio-
chemical findings typical of acute viral hepatitis (Table 1).
The healthy controls were age- and sex-matched, tested
negative for the viral hepatitis markers, and had normal
liver function tests.
The chemical shifts and signal intensities in 1H NMR
spectra of the plasma and urine from patients with acute
hepatitis E, acute hepatitis B, and healthy controls were
compared (Figs 1a,b). Pattern recognition and multivariate
data analysis were carried out using PCA, which is an
unsupervised method. It can analyse a data set in which
each individual measurement contains several data points,
such as a series of NMR spectra. In this analysis, the first
principal component (PC1) represents the largest source of
variation in the data set, PC2 represents the next largest
source of remaining variation, independent of PC1, and so
on with higher PCs. In effect, the first three PCs explain the
vast majority of variation present in the data. Using PCA
with the first three PCs, it was possible to distinguish
between plasma and urine NMR spectra from patients with
hepatitis E and controls with the first three PCs. For patients
with hepatitis E, PC1 alone explained 99.1% and 92.9%
variations in the NMR spectra of plasma and urine, respec-
tively. For patients with hepatitis B, PC1 explained 74.8%
and 48.5% variations in the NMR spectra of plasma and
urine, respectively (Figs 2a,b).
Unlike PCA, PLS-DA describes maximum separation
between predefined classes of data. We classified each spec-
Table 1 Clinical characteristics of study
subjects Characteristics
Mean age, years (range) 30.08 (24–43)
Time between onset of
illness and
specimen collection
(days)
Bilirubin (mg/dL)
SGOT (IU/L)
SGPT (IU/L)
Alkaline phosphatase
(IU/L)
Hb (g/dL)
Total leucocyte count
(·1000/lL)
! 2011 Blackwell Publishing Ltd
Hepatitis E metabonomics 3
imen into one of two groups, i.e. patients with hepatitis E and
controls, or patients with hepatitis B and controls, and per-
formed a regression analysis of the original data set against
these groups. In the PLS-DA scores plot, the first three PLS
components, i.e. PLS1, PLS2 and PLS3, showed improved
separation between patients with acute hepatitis E or B and
controls, compared with the PCA analysis (Figs 3a,b). Being
a supervised method, PLS-DA classifies and selects important
features depending on the variable importance in projection
(VIP) score, which is a weighted sum of squares of the PLS
loadings, with more important variables (metabolites) hav-
ing higher VIP scores. We used a more stringent VIP score
cut-off of ‡1.0 than the recommended 0.8 [15].
The levels of selected metabolites in the plasma and urine
data sets were determined using SAM (Figs 4a,b). It recog-
nizes the chemical shifts (metabolites) with statistically sig-
nificant differences between experimental groups by
integrating data from a set of metabolite-specific "t!-tests
wherein each chemical shift is given a score (di), calculated
on the basis of changes in its level relative to the standard
deviation of repeated measurements for those chemical
shifts. The chemical shifts with scores above a threshold are
considered as potentially significant. These were attributable
to 34 metabolites whose levels were significantly different in
either plasma or urine of patients with hepatitis E and/or B
compared with healthy controls (Table 2). These profiles
indicated significant modulation of several metabolic path-
ways during hepatitis E or B (Table 3) and also identified
some differences between the two. These changes were
especially prominent in glycolysis, TCA and urea cycles, and
the amino acid biosynthesis pathways. Compared with
healthy controls, there were reduced levels of l-tryptophan,
l-phenylalanine, l-histidine, l-tyrosine, l-ornithine, hippuric
acid, fumaric acid, ethanol, inosine, and l-valine and
increased levels of l-proline in both hepatitis E and B. The
levels of imidazole, 3-aminoisobutanoic acid, 1-methylnico-
tinamide, biopterin, adenosine, 1-methylhistidine, salicyluric
acid, and glycine were reduced only in patients with
Acute hepatitis E Acute hepatitis B Healthy controls
35.16 (23–56) 28.76 (22–47)
7. 5 (1–24) 13.3 (1–52) –
7.5 (2–25) 16.0 (1–38) 0.8 ± 0.3
309 (46–1684) 626.7 (20–2303) 31 ± 8
477 (83–1612) 647.4 (27–2368) 26 ± 9
267 (96–453) 106 (46–183) 93 ± 17
13.9 (9.8–16.0) 12.6 (9.2–16.2) 13.6 ± 0.2
7.9 (4.6–10.2) 7.6 (6.0–8.6) 6.3 ± 0.3
4 S. U. Munshi et al.
Fig. 1 Representative one-dimensional 1H NMR spectra of (a) plasma and (b) urine obtained from a healthy control, patient
with hepatitis E, and patient with hepatitis B.
hepatitis E. In contrast, methylsuccinic acid, N-acetyl-l-
aspartic acid, l-acetylcarnitine, acetoacetic acid, l-carnitine,
and succinic acid were reduced only in patients with hepa-
titis B. We also found the levels of 2-hydroxybutyric acid,
pantothenic acid, citric acid, and l-lactic acid to increase
during hepatitis E but to decrease in hepatitis B. Increased
levels of l-isoleucine, acetone, and glycerol were observed in
hepatitis E, whereas levels of betaine and l-glutamine were
higher in hepatitis B.
The aromatic amino acid (AAAs) such as l-tryptophan,
l-phenylalanine, and l-tyrosine, the branched chain amino
acid (BCAA) valine, the glucogenic amino acid glycine, and
the basic amino acid l-ornithine all decreased in patients
with hepatitis E. In contrast, l-proline and l-isoleucine
(another BCAA) were higher in these patients.
For pathway analyses, we found some of these to be sig-
nificantly (P < 0.05) modulated in both hepatitis B and E.
However, the nature of the metabolites and their fold
changes in a given pathway were different for the two dis-
eases. Besides these common pathways, the metabolism of
alanine, aspartate, glutamate, tyrosine, and butanoate was
modulated during hepatitis B, whereas thiamine metabolism
was affected during hepatitis E.
DISCUSSION
Hepatitis caused by different viruses has different outcomes.
Whereas infection with HAV or HEV is almost always acute
and self-limiting, that with HBV or HCV may become
chronic following an acute phase. Therefore, the pathobio-
logical processes in these infections are expected to be dif-
ferent. Here, we studied the effects of HEV on host
metabolism by identifying metabolites whose levels are sig-
nificantly altered in the body fluids of patients with hepatitis
! 2011 Blackwell Publishing Ltd

Fig. 2 Three-Dimensional PCA score plots of plasma and urine samples of patients with hepatitis E (top panels) and patients
with hepatitis B (bottom panels) compared with healthy controls. The explained variances of the selected PCs are shown
in brackets.
E. These were compared with patients having acute hepatitis
B to delineate pathogen/disease-specific host metabolite
signatures.
Liver biopsies in patients with hepatitis E and nonhuman
primates with experimental HEV infection show necro-
inflammatory changes with inflammation of the bile duct
[16]. HEV infects hepatocytes, [2] causes swelling of mito-
chondria, and the dilation of smooth endoplasmic reticulum
(SER) in these cells [17]. The SER is involved in several
metabolic processes including lipid and steroid synthesis,
carbohydrate metabolism, etc. The mitochondria play a
dominant role in ATP production through the TCA cycle,
regulation of cellular metabolism and proliferation, synthesis
of haem and steroids, apoptosis, and storage of calcium ions.
We have observed dysregulation of some of these pathways
in patients with hepatitis E. Figure 5 shows a schematic of
the most important metabolic changes we found. These
pathways and their relationship to clinical and molecular
features of hepatitis E and B are discussed later.
! 2011 Blackwell Publishing Ltd
Hepatitis E metabonomics 5
Deregulated metabolic cycles
The catabolism of carbohydrates, fats, and proteins produces
a common product, acetyl-CoA, which enters the TCA cycle
to generate ATP under aerobic conditions. In glycolysis,
glucose produces pyruvate that is converted to acetyl-CoA.
The ORF3 protein of HEV activates the transcription factor
hypoxia-inducible factor 1 (HIF-1), leading to the increased
expression of lactate dehydrogenase (LDH) and glycolytic
enzymes including pyruvate kinase (PK) [18]. Higher levels
of plasma lactate and urinary citrate, lower levels of fuma-
rate, and increased levels of plasma 2-hydroxybutyric acid
are indicative of metabolic stress, poor energy metabolism,
or lactic acidosis in patients with hepatitis E [19]. Anaerobic
glycolysis, wherein glucose metabolism leads to elevated
plasma lactic acid, appears to be accentuated in patients
with hepatitis E. Increased levels of acetone in these patients
also suggest the involvement of acetyl-CoA in a pathway
leading to ketoacidosis [20], as in diabetes and in inborn
6 S. U. Munshi et al.

Fig. 3 Three-dimensional PLS-DA score plots of plasma and urine samples of patients with hepatitis E (top panels) and
patients with hepatitis B (bottom panels) compared with healthy controls. The explained variances of the selected PLSs are as
follows: hepatitis E – plasma – R2 = 0.067, Q2 = 0.45; urine – R2 = 0.84, Q2 = 0.35; hepatitis B – plasma – R2 = 0.99,
Q2 = 0.99; urine R2 = 0.97 Q2 = 0.67.

errors of metabolism [21–23]. Lactic acidosis and ketosis
may in turn contribute to hepatic encephalopathy observed
in some patients with hepatitis E.
Ammonia is a harmful product of protein and amino acid
catabolism, which is detoxified through the urea cycle in
hepatocytes (Fig. 5). The urea so produced is transported to
the plasma for excretion through urine and faeces. We found
decreased levels of ornithine and fumarate in patients with
hepatitis E, suggesting anomalies in ammonia detoxification.
This could be due to decreased expression of ornithine
transcarbamoylase (OTC), the first enzyme of the urea cycle;
transcription of the OTC gene is regulated by HNF4-alpha,
the nuclear levels of which are reduced in HEV-infected cells
(V. Chandra, P. Holla, D. Ghosh, D. Chakrabarti, M. Padi-
garu and S. Jameel, unpublished data).
Abnormalities in amino acid metabolism
There are reduced levels of AAAs – l-tryptophan, l-phenyl-
alanine, and l-tyrosine in patients with hepatitis E (Fig. 5).
Decreased fumarate levels in these patients also support
lower levels of l-phenylalanine and l-tyrosine. Because of the
increased expression of PK and LDH in HEV-infected cells
[18], there would be reduced levels of phosphoenolpyruvate,
which together with erythrose-4-phosphate forms choris-
mate, an AAA precursor.
The BCAAs – leucine, isoleucine, and valine are essential
amino acids that animals cannot synthesize and have to
acquire through diet. In patients with hepatitis E, plasma
isoleucine levels are higher, valine levels lower, and leucine
levels similar to healthy controls. The main BCAA catabo-
lism enzymes, branched-chain aminotransferase and mito-
chondrial branched-chain ketoacid dehydrogenase [24], are
not limited to the liver, but are also distributed extensively
and mainly functional in extra-hepatic tissues [25–27]. The
BCAA levels may differ because of dietary intake and not
HEV-induced hepatic injury.
Reduced plasma AAA levels are likely to lead to lower
concentrations in the brain, compromising the synthesis of
serotonin and dopamine [28,29]. In patients with hepatitis E,
we also found lower levels of biopterin, an oxidative product
of tetrahydrobiopterin, which is a cofactor for dopamine,

! 2011 Blackwell Publishing Ltd

 Hepatitis E metabonomics 7

Plasma Urine

0.5 1.0 1.5 2.0 2.5 3.0

Cutlow: ––1.396
Cutlow: ––2.246 Cutup: 2.185

––5 ––4 ––3 ––2 ––1 0 1 2 3 4 5 6 7

Cutup: 2.232 p0: 0.344
p0: 0.252 Significant: 42
Significant: 79 False: 3.46
False: 2.91 FDR: 0.028
FDR: 0.009

Observed d(i) values

Observed d(i) values
Hepatitis E

––0.5
––1.5
––2.5
––3.5
––5 ––4 ––3 ––2 ––1 0 1 2 3 4 5 6 7 ––3.5 ––2.5 ––1.5 ––0.5 0.5 1.0 1.5 2.0 2.5 3.0

––8 ––7 ––6 ––5 ––4 ––3 ––2 ––1 0 1 2 3 4 5

Cutlow: ––2.006 Cutlow: ––2.31
3

Cutup: 2.152 Cutup: 2.523

p0: 0.042 p0: 0.128
Significant: 64
2

Significant: 87
False: 0 False: 1.8
FDR: 0 FDR: 0.003

Observed d(i) values

Observed d(i) values
1
Hepatitis B

––2 ––1 0 ––3

––4

––4 ––3 ––2 ––1 0 1 2 3 ––8 ––7 ––6 ––5 ––4 ––3 ––2 ––1 0 1 2 3 4 5
Expected d(i) values Expected d(i) values

Fig. 4 Significance Analysis of Microarrays (and Metabolites) (SAM) plots of plasma and urine from patients with hepatitis E
(top panels) and hepatitis B (bottom panels) as compared to healthy controls. Spots above the upper and below the lower
horizontal dotted line indicate up- and down-regulated metabolites, respectively.

serotonin, epinephrine, and nitric oxide synthesis [30]. Lower

Metabolic differences between hepatitis E and B
levels of biopterin would reflect decreased neurotransmitter
synthesis [31], and this might account for the neurological Based on the contrasting changes in metabolite levels, dif-
symptoms related to sleep regulation, circadian rhythmicity, ferent pathways appear to be modulated during acute hep-
and motor functions in patients with fulminant hepatitis. atitis B and E. In the TCA cycle, fumaric acid was reduced in
Plasma levels of glycine were also reduced in patients with both infections, whereas citric acid was increased in HEV,
hepatitis E. Glycine is involved in glutathione (GSH) but not in HBV infection. In glycolysis and gluconeogenesis,
metabolism and bile synthesis [32,33] and may become a ethanol was reduced and lactic acid was increased in hep-
limiting factor in GSH synthesis in the c-glutamyl cycle. atitis E, but both were decreased in hepatitis B. In propan-
Oxidative stress or detoxification demands can dramatically oate metabolism, 2-hydroxybutyric acid, l-lactic acid, and
increase the formation of hepatic GSH [34]. Limiting the acetone were elevated in hepatitis E, but declined in hepatitis
levels of glycine would reduce the GSH levels, thus dimin- B. Altered alanine, aspartate, glutamate, tyrosine, and but-
ishing the ability of patients with hepatitis E to overcome anoate metabolism was observed in hepatitis B but not E.
oxidative stress. A marked reduction of GSH concentration Lactic acid, which is involved in glycolysis, gluconeogenesis,
in erythrocytes, but not in plasma, was also reported in and pyruvate metabolism, and two other metabolites, ace-
hepatitis A, B, and C [35,36]. Lower levels of hippuric acid in tone and 2-hydroxybutyric acid, which are associated with
the urine of patients could also result from reduced glycine ketoacidosis were elevated significantly in patients with
levels as the former is a product of glycine and benzoic acid hepatitis E, but not in patients with hepatitis B. This appears
[37]. Bile acids are synthesized in vertebrate liver from to be a major difference between acute hepatitis B and E.
cholesterol and conjugated with taurine or glycine before There are no published metabonomic data from patients
their secretion via bile into the intestine. Reduced glycine with acute hepatitis B, but some information is available on
levels would decrease the levels of conjugated intestinal bile those with chronic hepatitis B [9,39]. Compared with heal-
acids, impair micelle formation, and consequently intestinal thy controls, patients with hepatitis B-related chronic hep-
fat absorption, leading to steatorrhea [38]. atitis and hepatocellular carcinoma had reduced plasma

! 2011 Blackwell Publishing Ltd

8 S. U. Munshi et al.

Table 2 Differential metabolites in plasma and/or urine of patients with acute hepatitis E or hepatitis B, and the chemical shifts
corresponding to these metabolites

Chemical shift (ppm)*/Source#

Metabolite Acute hepatitis E Acute hepatitis B

Decreased in both acute hepatitis E and hepatitis B

l-tryptophan (Trp) 7.73 (d), 7.72 (t)/U 7.69 (s)/U
7.21 (t)/U, P
l-phenylalanine (Phe) 7.43 (m), 7.38 (m)/U 7.32 (d)/U
l-histidine (His) 7.08 (d)/U 7.11 (d), 7.88 (d)/U
l-tyrosine (Tyr) 7.18 (m)/P 6.83 (m), 6.90 (m)/P
6.91 (m), 6.83 (m)/U
l-ornithine (Orn) 1.80 (m) 1.80 (m)
Hippuric acid 7.87 (dd), 7.54/U 7.64 (tt)/U
Fumaric acid 6.51 (s), 6.55 (s)/P 6.51 (s), 6.55 (s), 6.46 (s)/P
6.48 (s)/U 6.48 (s)/U
Ethanol 1.14 (t)/P 1.14 (t)/P, 1.12 (t)/U
Inosine 8.35 (s)/U 8.35 (s)/U
l-valine 1.05 (d)/P 1.05 (d)/P, 0.99 (d)/U
Increased in both acute hepatitis E and hepatitis B
l-proline 2.02 (m)/P 3.38 (dt)/U
Decreased only in acute hepatitis E
Imidazole 8.22 (s)/U
3-Aminoisobutanoic acid 2.94 (m)/U
1-Methylnicotinamide 9.02 (d)/U
Biopterin 8.73 (s)/U
Adenosine 8.32 (S)/U
1-Methylhistidine 7.66 (s)/U
Salicyluric acid 7.02 (dd), 7.46 (ddd)/U
Glycine 3.55 (s)/P
Decreased only in acute hepatitis B
Methylsuccinic acid 1.09 (d)/U
N-Acetyl-l-aspartic acid 2.01 (s)/U
l-acetylcarnitine 2.13 (s)/U
Acetoacetic acid 2.26 (s)/U
l-carnitine 2.39 (s), 2.43 (s)/P
Succinic acid 2.39 (s)/U
Increased only in acute hepatitis E
l-isoleucine 1.27 (m)/P
Acetone 2.23 (s)/P
Glycerol 3.56 (m)/P
Increased only in acute hepatitis B
Betaine 3.24 (s)/ U
Increased in acute hepatitis E and decreased in acute hepatitis B
2-Hydroxybutyric acid (2HB) 0.87 (t)/P 0.87 (t)/P
Pantothenic acid 0.85 (s)/U 0.85 (s), 0.91 (s)/U
Citric acid 2.67 (d)/U 2.53 (d), 2.62 (dd)/U
l-lactic acid 1.33 (d)/P 1.33 (d)/P

*Multiplicity: d, doublet; t, triplet; m, multiplet; s, singlet; dd, doublet of doublet; dt, doublet of triplet. #U, urine, P, Plasma.

levels of acetone [39], and those with hepatitis B-related AAAs and methionine are found in patients with chronic
cirrhosis had higher plasma lactic acid levels [40]. In hepatitis B or C [41]. The reduction of BCAAs in chronic
patients with hepatitis B, we found lower levels of lactic acid infections is likely to be due to the reduced lean muscle mass
and acetone. Reduced levels of BCAAs and higher levels of [42]. This is unlikely in acute hepatitis, and reduced dietary

! 2011 Blackwell Publishing Ltd

 Hepatitis E metabonomics 9

Table 3 Combined list of metabolites, present differentially in plasma and/or urine of patients with acute hepatitis E or B, and
their involvement in different metabolic pathways. The levels of metabolites shown in bold were elevated in patients with
hepatitis E and/or B; the others were reduced in patients compared with healthy controls

Hepatitis E Hepatitis B
Total
Pathway name metabolites Metabolites P value Metabolites P value

Aminoacyl-tRNA 75 l-histidine, l-phenylalanine, 6.07E)07 l-histidine, l-phenylalanine, 2.05E)05

biosynthesis l-glycine, l-valine, l-valine, l-tryptophan,
l-tryptophan, l-tyrosine, l-tyrosine, l-proline
l-proline, l-isoleucine
Propanoate 35 2-Hydroxybutyric acid, 4.47E)04 2-Hydroxybutyric acid, 6.42E)06
metabolism l-lactic acid, l-valine, l-valine, l-lactic acid,
Acetone Succinic acid, Acetoacetic
acid
Nitrogen metabolism 39 l-histidine, l-phenylalanine, 4.40E)05 l-histidine, l-phenylalanine, 2.37E)04
l-tryptophan, l-tyrosine, l-tryptophan, l-tyrosine
l-glycine
Phenylalanine 45 Hippuric acid, Fumaric acid, 0.001 Hippuric acid, Succinic acid, 2.29E)05
metabolism l-tyrosine, l- Fumaric acid, l-tyrosine,
Phenylalanine l-phenylalanine
Alanine, aspartate 24 – – Succinic acid, Fumaric acid, 8.89E)04
and glutamate N-Acetyl-l-aspartic acid
metabolism
Citrate cycle 20 Fumaric acid, Citric Acid 0.018 Hippuric acid, Succinic acid, 0.011
(TCA cycle) Fumaric acid
Phenylalanine, 27 l-phenylalanine, l-tyrosine, 0.002 l-phenylalanine, l-tyrosine, 0.001
tyrosine and l-tryptophan l-tryptophan
tryptophan
biosynthesis
Tyrosine metabolism 76 – Succinic acid, Fumaric acid, 0.003
Acetoacetic acid, l-tyrosine
Arginine and proline 77 l-proline, Ornithine, 0.04 l-proline, Ornithine, 0.024
metabolism Fumaric acid Fumaric acid
Butanoate 40 – – Succinic acid, Fumaric acid, 0.003
metabolism Acetoacetic acid
Pantothenate and 27 Pantothenic acid, l-valine 0.03 Pantothenic acid, l-valine 0.02
CoA biosynthesis
Glycolysis or 31 Ethanol, l-lactic acid 0.04 Ethanol, l-lactic acid 0.02
Gluconeogenesis
b-Alanine 28 l-histidine, Pantothenic acid 0.03 l-histidine, Pantothenic acid 0.021
metabolism
Valine, leucine and 40 l-valine, l-isoleucine 0.03 l-valine, Acetoacetic acid 0.04
isoleucine
degradation
Thiamine metabolism 24 l-Glycine, l-tyrosine 0.02 l-tyrosine 0.18

intake during symptomatic disease may instead be respon-
sible. Increased plasma levels of proline were found in
patients with acute hepatitis B and E. In hepatitis E, this could
be due to reduced nuclear levels of HNF4a, a transcription
factor required for the expression of proline oxidase, which
catalyses the first step in proline catabolism. Mutations in the
proline oxidase gene are also associated with hyperprolin-
emia [43], and increased proline levels are also reported in
patients with HBV-mediated liver failure [44].
An in vivo proton magnetic resonance spectroscopy study
suggested that cerebral metabolite changes during hepatic
encephalopathy differ from changes because of acute-on-
chronic liver failure and chronic liver disease [45]. This may
be due to differences in the pathogenesis of these two
somewhat overlapping disease conditions. Besides metabo-
lites, the levels of plasma acute-phase proteins in acute
hepatitis A and E were different from that in chronic hepa-
titis B and C [5,46,47].

! 2011 Blackwell Publishing Ltd

10 S. U. Munshi et al.

Fig. 5 Schematic representation of the metabolites and metabolic pathways affected in hepatitis E. The modulated metabolites
found in this study are shown in boxes; up or down arrows indicate their relative increase and decrease, respectively, in
patients with hepatitis E compared with healthy controls. Shaded boxes represent findings from other studies.

In this metabonomic study, we have identified changes
in several metabolites during hepatitis E and have com-
pared these with acute hepatitis B. The main limitation of
this study is that all our subjects were men. Hepatitis E
results in significant mortality among pregnant women in
whom it can also take a fulminant course. It will thus be
important to also study pregnant and nonpregnant women
with hepatitis E. Nevertheless, this approach has allowed
us to better understand the hepatitis E pathogenesis in
terms of deregulated metabolic pathways, the primary ones
being TCA and urea cycles, and amino acid metabolism.
Our results also suggest that patients with hepatitis E (and
not hepatitis B) may be prone to lactic acidosis and
ketosis.
Metabonomics is a comparatively new field, and its
coverage is not as wide as that of genomics and proteo-
mics. While currently available tools cover the entire
genome and almost the entire proteome, only <20% of the
metabolome is covered in the available databases [48,49].
Yet, its simplicity and noninvasive nature makes it an
attractive tool to understand the disease conditions and
pathogenesis.
ACKNOWLEDGEMENTS AND DISCLOSURES
This work was supported by a grant from the National
Institutes of Health, USA (5R01AI076192) to SJ. SUM is a
recipient of the ICGEB Predoctoral Fellowship; ST received a
fellowship from CSIR, India. The Department of Biotechnol-
ogy, Government of India provided financial support for the
NMR facility at the ICGEB, New Delhi. None of the authors
have any competing interests to declare.

REFERENCES
1 Emerson SU, Purcell RH. Hepatitis E
virus. In: Knipe DM, Howley PM, eds.
Fields Virology, 5th edn. Philadel-
phia: Wolters Kluwer Lippincott
Williams & Wilkins, 2007: 3047–
3058.
2 Aggarwal R, Naik S. Epidemiology of
hepatitis E: current status. J Gastro-
enterol Hepatol 2009; 24: 1484–
1493.
3 Khuroo MS, Kamili S. Aetiology,
clinical course and outcome of
sporadic acute viral hepatitis in preg-
nancy. J Viral Hepat 2003; 10: 61–69.
4 Hamid SS, Atiq M, Shehzad F et al.
Hepatitis E virus superinfection in
patients with chronic liver disease.
Hepatology 2002; 36: 474–478.
5 Taneja S, Sen S, Gupta VK, Aggarwal
R, Jameel S. Plasma and urine bio-
markers in acute viral hepatitis E.
Proteome Sci 2009; 7: 39.
6 Dumas M-E, Barton RH, Toye A et al.
Metabolic profiling reveals a contri-
bution of gut microbiota to fatty liver
phenotype in insulin-resistant mice.
Proc Natl Acad Sci USA 2006; 103:
12511–12516.
7 Nicholson JK, Lindon JC, Holmes E.
"Metabonomics!: understanding the
metabolic responses of living systems
to pathophysiological stimuli via
multivariate statistical analysis of
biological NMR spectroscopic data.
Xenobiotica 1999; 29: 1181–1189.

! 2011 Blackwell Publishing Ltd


8 Lindon JC, Nicholson JK, Holmes E,
Everett JR. Metabonomics: metabolic
processes studied by NMR spectros-
copy of biofluids combined with pat-
tern recognition. Concepts Magn
Reson 2000; 12: 289–320.
9 Mao Y, Huang X, Yu K, Qu H, Liu C,
Chang Y. Metabonomic analysis of
hepatitis B virus-induced liver fail-
ure: identification of potential diag-
nostic biomarkers by fuzzy support
vector machine. J Zhejiang Univ Sci B
2008; 9: 474–481.
10 Godoy MM, Lopes EP, Silva RO et al.
Hepatitis C virus infection diagnosis
using metabonomics. J Viral Hepat
2010; 17: 854–858.
11 Hewer R, Vorster J, Steffens FE,
Meyer D. Applying biofluid 1H NMR-
based metabonomic techniques to
distinguish between HIV-1 positive/
AIDS patients on antiretroviral
treatment and HIV-1 negative indi-
viduals. J Pharm Biomed Anal 2006;
41: 1442–1446.
12 Findeisen M, Brand T, Berger S. A
1
H-NMR thermometer suitable for
cryoprobes. Magn Reson Chem 2006;
45: 175–178.
13 Lindon JC, Nicholson JK, Everette JR.
NMR spectroscopy of biofluids. Ann
Rep NMR Spectros 1999; 38: 1–88.
14 Nicholson JK, Foxall PJ, Farrant RD,
Lindon JC. 750 MHz 1H and 1H-13C
NMR spectroscopy of human blood
plasma. Anal Chem 1996; 67: 793–
811.
15 Tan J, Li QQ, Loganath A, Chong YS,
Xiao M, Obbard JP. Multivariate data
analyses of persistent organic pollu-
tants in maternal adipose tissue in
Singapore. Environ Sci Technol 2008;
42: 2681–2687.
16 Halbur PG, Kasorndorkbua C, Gilbert
C et al. Comparative pathogenesis of
infection of pigs with hepatitis E
viruses recovered from a pig and a
human. J Clin Microbiol 2001; 39:
918–923.
17 Gupta H, Iyenger B, Tandon BN.
Localization of a new enteric non-A,
non-B [HEV] virus in target organ
liver. Gastroenterol Jpn 1993; 28: 46–
50.
18 Moin SM, Chandra V, Arya R, Jameel
S. The hepatitis E virus ORF3 protein
stabilizes HIF-1a and enhances HIF-
1-mediated transcriptional activity
! 2011 Blackwell Publishing Ltd
through p300/CBP. Cell Microbiol
2009; 11: 1409–1421.
19 Pettersena JE, Landaas S, Eldjarn L.
The occurrence of 2-hydroxybutyric
acid in urine from patients with lac-
tic acidosis. Clin Chim Acta 1973; 48:
213–219.
20 McGarry JD, Foster DW. Regulation
of hepatic fatty acid oxidation and
ketone body production. Ann Rev
Biochem 1980; 49: 395–420.
21 Landaas S. The formation of 2-
hydroxybutyric acid in experimental
animals. Clin Chim Acta 1975; 58:
23–32.
22 Landaas S, Pettersen JE. Clinical
conditions associated with urinary
excretion of 2-hydroxybutyric acid.
Scand J Clin Lab Invest 1975; 35:
259–266.
23 Yang W, Roth KS. Defect in alpha-
ketobutyrate metabolism: a new
inborn error. Clin Chim Acta 1985;
145: 173–182.
24 Hutson SM, Fenstermacher D, Mahar
C. Role of mitochondrial transami-
nation in branched chain amino acid
metabolism. J Biol Chem 1988; 263:
3618–3625.
25 Hutson SM, Cree TC, Harper AE.
Regulation of leucine and ketoi-
socaproate metabolism in skeletal
muscle. J Biol Chem 1978; 253:
8126–8133.
26 Shinnick FL, Harper AE. Branched-
chain amino acid oxidation by iso-
lated rat tissue preparations. Biochim
Biophys Acta 1976; 437: 477–486.
27 Hutson SM, Sweatt AJ, LaNoue KF.
Branched-chain amino acid metabo-
lism: implications for establishing
safe intakes. J Nutr 2005; 135:
1557S–1564S.
28 Young SN, Leyton M. The role of
serotonin in human mood and social
interaction: insight from altered
tryptophan levels. Pharmacol Bio-
chem Behav 2002; 71: 857–865.
29 Hyland K. Inherited disorders affect-
ing dopamine and serotonin: critical
neurotransmitters derived from aro-
matic amino acids. J Nutr 2007;
137: 1568S–1572S.
30 Kaufman S. The structure of the
phenylalanine hydroxylation cofac-
tor. Proc Natl Acad Sci USA 1963; 50:
1085–1090.
31 Weiss G, Thuma PE, Biemba G, Ma-
beza G, Werner ER, Gordeuk VR.
Hepatitis E metabonomics 11
Cerebrospinal fluid levels of biopterin,
nitric oxide metabolites, and immune
activation markers and the clinical
course of human cerebral malaria.
J Infect Dis 1998; 177: 1064–1068.
32 Hernandes MS, Troncone LR. Gly-
cine as a neurotransmitter in the
forebrain: a short review. J Neural
Transm 2009; 116: 1551–1560.
33 Lu SC. Regulation of hepatic gluta-
thione synthesis: current concepts
and controversies. FASEB J 1999;
13: 1169–1183.
34 Bhatia V, Bhardwaj P, Elikkottil J,
Batra J, Saraya A. A 7-day profile of
oxidative stress and antioxidant sta-
tus in patients with acute liver fail-
ure. Hepatol Int 2008; 2: 465–470.
35 Swietek K, Juszczyk J. Reduced glu-
tathione concentration in erythro-
cytes of patients with acute and
chronic viral hepatitis. J Viral Hepat
1997; 4: 139–141.
36 Kulinsky V, Leonova Z, Kol-
esnichenko L, Malov I, Danilov Yu.
Glutathione system in erythocytes
and plasma in viral hepatitis. Bio-
chem (Moscow) Suppl B: Biomed Chem
2007; 1: 270–275.
37 Quick AJ. The conjugation of benzoic
acid in man. J Biol Chem 1931; 92:
65–85.
38 Binder HJ. Disorders of Absorption.
In: Kasper DL, Braunwalr E, Fauci
AS, Hauser SL, Longo DL, Jameson
JL, eds. Harrison!s Principles of
Internal Medicine, 16th edn. New
York: McGraw-Hill, 2005: 1764–
1776.
39 Shariff MI, Ladep NG, Cox IJ et al.
Characterization of urinary biomar-
kers of hepatocellular carcinoma
using magnetic resonance spectros-
copy in a Nigerian population.
J Proteome Res 2010; 9: 1096–2103.
40 Xue R, Dong L, Wu H, Liu T, Wang J,
Shen X. Gas chromatography/mass
spectrometry screening of serum
metabolomic biomarkers in hepatitis
B virus infected cirrhosis patients. Clin
Chem Lab Med 2009; 47: 305–310.
41 Blonde-Cynober F, Aussel C, Cynober
L. Abnormalities in branched-chain
amino acid metabolism in cirrhosis:
influence of hormonal and nutritional
factors and directions for future re-
search. Clin Nutr 1999; 18: 5–13.
42 Khanna S, Gopalan S. Role of bran-
ched-chain amino acids in liver dis-
12 S. U. Munshi et al.
ease: the evidence for and against.
Curr Opin Nutr Metab Care 2007; 10:
297–303.
43 Bender HU, Almashanu S, Steel G
et al. Functional consequences of
PRODH missense mutations. Am
J Hum Genet 2005; 76: 409–420.
44 Yu K, Sheng G, Sheng J et al. A me-
tabonomic investigation on the bio-
chemical perturbation in liver failure
patients caused by hepatitis B virus.
J Proteome Res 2007; 6: 2413–2419.
45 Verma A, Saraswat VA, Krishna YR,
Nath K, Thomas MA, Gupta RK. In
vivo 1H magnetic resonance spec-
troscopy-derived metabolite varia-
tions between acute-on-chronic liver
failure and acute liver failure. Liver
Int 2008; 28: 1095–2103.
46 Noeman SA, Sharada K, EL-Bena H,
Dardery SE, Sawaf KA. Validity of
acute phase proteins and trace met-
als as adjuncts in viral hepatitis A
and B. Indian J Clin Biochem 1994; 9:
40–42.
47 Mas VR, Maluf DG, Archer KJ, Yanek
K, Bornstein K, Fisher RA. Proteomic
analysis of HCV cirrhosis and HCV-
induced HCC: identifying biomarkers
for monitoring HCV-cirrhotic patients
awaiting liver transplantation.
Transplantation 2009; 87: 143–152.
48 Wishart DS. Quantitative metabolo-
mics using NMR. Trends Anal Chem
2008; 27: 228–237.
49 Hollywood K, Brison DR, Goodacre
R. Metabolomics: current technolo-
gies and future trends. Proteomics
2006; 6: 4716–4723.
! 2011 Blackwell Publishing Ltd