Proceedings
oftheNutritionSo
ciety
The Nutrition Society Irish Section 25th Annual Postgraduate Meeting, hosted by the University of Cork and held at the RadissonBlu Hotel on 11–12 February 2016
Irish Section Postgraduate Meeting
Metabolomics as a tool in the identification of dietary biomarkers
Helena Gibbons1,2 and Lorraine Brennan1,2*1Institute of Food and Health, UCD School of Agriculture and Food Science, University College Dublin (UCD),
Belfield, Dublin, Ireland2UCD Conway Institute of Biomolecular Research, UCD, Belfield, Dublin, Ireland
Current dietary assessment methods including FFQ, 24-h recalls and weighed food diariesare associated with many measurement errors. In an attempt to overcome some of theseerrors, dietary biomarkers have emerged as a complementary approach to these traditionalmethods. Metabolomics has developed as a key technology for the identification of new diet-ary biomarkers and to date, metabolomic-based approaches have led to the identification ofa number of putative biomarkers. The three approaches generally employed when usingmetabolomics in dietary biomarker discovery are: (i) acute interventions where participantsconsume specific amounts of a test food, (ii) cohort studies where metabolic profiles are com-pared between consumers and non-consumers of a specific food and (iii) the analysis of diet-ary patterns and metabolic profiles to identify nutritypes and biomarkers. The present reviewcritiques the current literature in terms of the approaches used for dietary biomarker discov-ery and gives a detailed overview of the currently proposed biomarkers, highlighting stepsneeded for their full validation. Furthermore, the present review also evaluates areas suchas current databases and software tools, which are needed to advance the interpretationof results and therefore enhance the utility of dietary biomarkers in nutrition research.
Metabolomics: Dietary biomarkers: Diet and nutrition: Dietary assessment
Dietary biomarkers and the concept of metabolomics
The contribution of diet to the increasing burdens ofCVD, diabetes, obesity and cancers has been recognisedsince the 1970s(1). Selected foods and nutrients as well asdietary patterns are now known to interact with variousmetabolic processes contributing to a reduction or an in-crease in the risk of disease(2). For example, it is wellestablished that high salt consumption raises blood pres-sure(3) and high consumption of red meat has been asso-ciated with increased incidence of type 2 diabetes(4,5),CVD(6) and cancers(7). In contrast, dietary patternssuch as the dietary approaches to stop hypertensiondiet, which emphasises consumption of fruit and vegeta-bles, low-fat dairy foods and whole grains and reducedintake of red meats and sugars has been shown to de-crease blood pressure and CVD risk(8,9). Similarly, theMediterranean diet which emphasises high fruit, vege-table and olive oil consumption has been shown to
reduce CVD and type-2 diabetes risk(10,11). As diet is akey environmental risk factor, the identification and tar-geting of dietary factors with the greatest prospective forreducing or increasing disease risk is of major scientificand public health importance(12). It is therefore essentialthat dietary assessment methods are reliable and accuratefor the advancement of our understanding of the linksbetween diet and health.
Diet is traditionally measured via self-reporting meth-ods such as FFQ, 24-h recalls and weighed food diaries.There is however a number of methodological issuesassociated with each of these assessment methods, in-cluding energy underreporting, recall errors and diffi-culty in assessment of portion sizes(2,13,14). Such errorscan lead to reduced power, underestimated associationsand false findings which may contribute to inconsisten-cies in the field of nutritional epidemiology(14,15). In aneffort to address some of these measurement issues, theuse of dietary biomarkers, which are found in biological
*Corresponding author: Professor L. Brennan, email [email protected]: SSB, sugar-sweetened beverage.
Proceedings of the Nutrition Society (2017), 76, 42–53 doi:10.1017/S002966511600032X© The Authors 2016 First published online 25 May 2016
https://doi.org/10.1017/S002966511600032XDownloaded from https://www.cambridge.org/core. IP address: 65.21.228.167, on 30 Jan 2022 at 22:50:00, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
Proceedings
oftheNutritionSo
ciety
samples and are related to ingestion of a specific food orfood group, have emerged(16). Currently dietary biomar-kers exist for salt, protein, sucrose/fructose intake (so-dium/nitrogen/sucrose and fructose measured in 24 hurine samples) and energy expenditure (the doubly labelledwater technique)(2,17). These dietary biomarkers can beused in conjunction with traditional dietary assessmentmethods to improve the accuracy of dietary intake meas-urement and can also be used to more accurately associatedietary intake with disease risk and nutritional status(18).
In recent years, metabolomics has developed as a keytechnology for the identification of new dietary biomarkers.Metabolomics provides a powerful approach for the com-prehensive description of all low molecular weight mole-cules present in biological samples(16). In metabolomicsresearch, the analytical platforms predominantly used areNMR spectroscopy and MS coupled with a chromato-graphic step, for example, GC or liquid chromatography.Each of these techniques is associated with a number ofadvantages and disadvantages, for example MS-basedtechniques have high sensitivity and therefore may detectmetabolites below the detection limit of NMR spectros-copy; however, sample treatment is necessary beforeMS-based analysis, while little or no pre-treatment isrequired for NMR(19).While in the past many articlesdetailed the advantages and disadvantages of differentapproaches there has now been a realisation that usingone platform alone will not give complete coverage of themetabolite profile; therefore, a combination of technologiesand approaches is usually recommended for optimal cover-age. Analysis of metabolomic data is commonly performedusing multivariate statistics and there are an increasing se-lection of databases and tools available to assist in the in-terpretation of these multivariate results(20).
Examination of the literature reveals that there are threeapproaches generally employed for dietary biomarker dis-covery. These can be summarised as: (i) acute or mediuminterventionswhere participants consume specific amountsof a test food and biological samples are collected post con-sumption, (ii) cohort studies where metabolic profiles arecompared between consumers and non-consumers of aspecific food and (iii) the analysis of dietary patterns andmetabolic profiles to identify nutritypes and biomarkers.Although these study designs have led to the identificationof a number of biomarkers in the literature in recent years,each of these approaches has a number of limitationsassociated with it. Awareness of these is important in theinterpretation and potential use of such biomarkers.Therefore the objective of the present review is to give anoverview of currently proposed biomarkers and secondlythe present review aims to critique the current literaturein terms of approaches for dietary biomarker discovery,highlighting steps needed for their full validation.
Dietary biomarker discovery using intervention studies
Dietary intervention studies involve participants consum-ing specific amounts of a test food in a single meal (acuteintervention) or for a short-to-medium term interventionthe test food is consumed in repeated meals. In this
approach, baseline and postprandial biofluids are col-lected and following analysis, potential biomarkers areidentified. This approach has led to the identification ofa number of putative biomarkers of specific foods andbeverages as summarised in Table 1. An excellent ex-ample of a biomarker successfully identified using thisapproach is proline betaine, a robust biomarker of citrusfruit intake. Proline betaine was originally identified byAtkinson et al.(21) and following this Heinzmann et al.performed an acute intervention study with a mixed-fruitmeal, which consisted of apples, grapes, oranges andgrapefruit(22). Eight participants consumed standardisedmeals over 3 d and on the second day the mixed-fruitmeal was consumed(22). Urine samples were collectedand analysed using NMR spectroscopy. Followingmultivariate analysis proline betaine was identified as apotential biomarker. To assign the origin of urinary pro-line betaine excretion after the mixed-fruit meal, concen-trations of proline betaine in fruits and fruit juices weremeasured. Concentrations of proline betaine were higherin citrus fruit compared with other commonly availablefruit and fruit juices tested. The urinary excretionprofile of proline betaine was then measured in six indi-viduals after consumption of orange juice. This biomark-er was confirmed using data from participants in theINTERMAP UK cohort and demonstrated a high sensi-tivity and specificity for citrus fruit consumption (90·6and 86·3 %, respectively)(22). Lloyd et al. also identifiedproline betaine and a number of biotransformed pro-ducts in postprandial urine samples after consumptionof 200 ml orange juice as part of a standardised testbreakfast(23). Subsequent biomarker validation demon-strated sensitivities and specificities of 80·8–92·2 and74·2–94·1 %, respectively, for elevated proline betainein high reporters of citrus fruit consumption(23).Following on from these acute studies, a medium-termintervention study used MS to profile the urinary meta-bolomes of twelve volunteers that consumed orangejuice regularly for 1 month as part of their habitualdiet. Proline betaine was again identified as a potentialmarker of citrus fruit(24). Considering the range of studiesthat consistently report proline betaine as a marker of cit-rus fruit intake the evidence base is strong to support itsuse.
A number of research groups have also used dietaryinterventions to investigate biomarkers of cruciferous vege-tables(25–27). Andersen et al. performed a controlled cross-over meal study with nine brassica-containing New NordicDiet meals in seventeen subjects(26). The 24 h urine sampleswere collected and analysed by ultra-performance liquidchromatography–quadruple time-of-flight–MS. To investi-gate the food sources of the biomarkers found in the mealstudy, a range of small single food studies were performedwith three to four participants in each. Using a sensitivityand specificity analyses to select the most promising bio-markers, a range of conjugated isothiocyanates were iden-tified as biomarkers of brassica intake(26). Furtherbiomarkers of other foods, including fish were also iden-tified(26). To validate the biomarkers from this study,Andersen et al.(27) carried out a 6-month parallel dietaryintervention study where 107 participants were randomised
Metabolomics in dietary biomarker discovery 43
https://doi.org/10.1017/S002966511600032XDownloaded from https://www.cambridge.org/core. IP address: 65.21.228.167, on 30 Jan 2022 at 22:50:00, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
Proceedings
oftheNutritionSo
ciety
Table
1.Sum
maryof
putativebiomarke
rsiden
tified
usingametab
olom
icsap
proac
hin
interven
tionstud
ies
Dietary
factor
Study
duration
No.
ofsu
bjects
Sam
ple
Metab
olom
ictech
nique
Biomarke
rReferen
ce
Citrus
fruit
Acu
teinterven
tion
8Fa
stingan
dpos
tprand
ial
urine
NMR
Prolinebetaine
Heinz
man
net
al.(2
2)
Citrus
fruit
Acu
teinterven
tion
424
hurine
LC–ESI–qTO
F,LT
Q-O
rbitrap
Prolinebetaine
,hyd
roxy
prolinebetaine
,hes
peretin
3′-O
-glucu
ronide,
narin
genin7-O-glucu
ronide,
limon
ene8,9-diolg
lucu
ronide,
nootka
tone
13,14-diolg
lucu
ronide,
N-M
ethy
ltyraminesu
lpha
te
Pujos
-Guillo
tet
al.(2
4)
Citrus
fruit
4-wee
kinterven
tion
1224
hurine
LC–ESI–qTO
F,LT
Q-O
rbitrap
Prolinebetaine
,hyd
roxy
prolinebetaine
,hes
peretin
3′-O
-glucu
ronide,
narin
genin7-O-glucu
ronide,
limon
ene8,9-diolg
lucu
ronide,
nootka
tone
13,14-diolg
lucu
ronide,
N-M
ethy
ltyraminesu
lpha
te
Pujos
-Guillo
tet
al.(2
4)
Citrus
fruit
Acu
teinterven
tion
12Fa
stingan
dpos
tprand
ial
urine
FIE–FT
ICR–MS
Prolinebetaine
,hyd
roxy
prolinebetaine
Lloy
det
al.(2
3)
Citrus
fruit
6-mon
thinterven
tion
107
24hurine
LC–qTO
FProlinebetaine
,hes
peretin-3-glucu
ronide
And
erse
net
al.(2
7)
Red
cabbag
e6-mon
thinterven
tion
107
24hurine
LC–qTO
F3-Hyd
roxy
-3-(methy
lsulfiny
l)propa
noic
acid,3
-hyd
roxy
hippuric
acid-sulpha
te,3
-hyd
roxy
hippuric
acid,ibe
rinN-ace
tyl-cy
steine
,N-ace
tyl-S-(N-3-m
ethy
lthiopropyl)cysteine,
N-ace
tyl-S-(N-
lylth
ioca
rbam
oyl)c
ysteine,
sulfo
rapha
neN-ace
tylcysteine
And
erse
net
al.(2
7)
Bee
troo
t6-mon
thinterven
tion
107
24hurine
LC–qTO
F4-Ethyl-5-aminop
yroc
atec
hols
ulpha
te,
4-ethy
l-5-methy
laminop
yroc
atec
hols
ulpha
te,
4-ethy
lpyridine-2-ca
rbox
ylic
acid
glyc
ineco
njug
ate
And
erse
net
al.(2
7)
Walnu
ts6-mon
thinterven
tion
107
24hurine
LC–qTO
F5-Hyd
roxy
indole-3-ac
etic
acid
And
erse
net
al.(2
7)
Straw
berrie
s6-mon
thinterven
tion
107
24hurine
LC–qTO
F2,5-Dim
ethy
l-4-metho
xy-3(2H)-furano
ne-sulpha
teAnd
erse
net
al.(
27)
Cho
colate
6-mon
thinterven
tion
107
24hurine
LC–qTO
F6-Amino-5-(N-m
ethy
lform
ylam
ino)-1-m
ethy
luracil,theo
brom
ine,
7-methy
luric
acid
And
erse
net
al.(2
7)
Ras
pberrie
sAcu
teinterven
tion
24Fa
stingan
dpos
tprand
ial
urine
FIE–FT
ICR–MS,
GC–TO
F–MS
Caffeic
acid-sulpha
te,m
ethy
lepicatec
hin-su
lpha
teLloy
det
al.(2
8)
Cruciferous
vege
tables
2-wee
kinterven
tion
20Fa
stingan
dpos
tprand
ial
urine
NMR
S-m
ethy
l-L-cy
steine
sulfo
xide
Edman
dset
al.(2
5)
Cruciferous
vege
tables
Acu
teinterven
tion
17Fa
stingan
dpos
tprand
ial
urine
UPLC
–qTO
F–MS
N-ace
tyl-S-(N-3-m
ethy
lthiopropy
l)cy
steine
,N-ace
tyl-S-(Nallylth
ioca
rbam
oyl)cy
steine
,Iberin
N-ace
tyl-cy
steine
,N-ace
tyl-cy
steine
conjug
ate,
4-im
inop
entylisothioc
yana
te,S
ulforapha
neN-ace
tyl-cy
steine
,ErucinN-ace
tyl-cy
steine
,N-A
cetyl-(N
′ -ben
zylth
ioca
rbam
oyl)-cy
steine
,Sulforapha
neN-cysteine
And
erse
net
al.(2
6)
Brocc
olli
Acu
teinterven
tion
24Fa
stingan
dpos
tprand
ial
urine
FIE–FT
ICR–MS
Tetron
icac
id,x
ylon
ate/lyxo
nate,threitol/e
rythritol
Lloy
det
al.(2
8)
Coffee
Acu
teinterven
tion
5Fa
stingan
dpos
tprand
ial
urine
NMR
2-furoylglyc
ine
Heinz
man
net
al.(4
1)
H. Gibbons et al.44
https://doi.org/10.1017/S002966511600032XDownloaded from https://www.cambridge.org/core. IP address: 65.21.228.167, on 30 Jan 2022 at 22:50:00, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
Proceedings
oftheNutritionSo
ciety
Coffee
Acu
teinterven
tion
9Fa
sting,
morning
spot,
24hurine
HILIC–MS/M
SN-M
ethy
lpyridinium,trig
onelline
Lang
etal.(4
6)
Black
tea
Acu
teinterven
tion
20Fa
stingan
dpos
tprand
ial
urine
NMR
Hippuric
acid,4
-hyd
roxy
hippuric
acid,1
,3-dihyd
rophe
nyl-2-O-sulph
ate,
allic
acid,4
-O-m
ethy
lgallic
acid
Van
Velze
net
al.(4
7)
Black
tea
Acu
teinterven
tion
324
hurine
NMR
Hippuric
acid,g
allic
acid,1
,3-dihyd
roxy
phe
nyl-2-O-sulph
ate
Day
kinet
al.(4
8)
Black
andgree
ntea
2d interven
tion
1724
hurine
NMR
Hippuric
acid,1
,3-dihyd
roph
enyl-2-O
-sulph
ate
vanDorsten
etal.(4
9)
Cha
mom
iletea
2-wee
kinterven
tion
14Spot
urine
NMR
Hippuric
acid
Wan
get
al.(5
0)
Mixed
nuts
12-w
eek
interven
tion
4224
hurine
LC–qTO
F,LT
Q-O
rbitrap
10-H
ydroxy
dec
ene-4,6-diyno
icac
id-sulpha
te,trid
ecad
ieno
ic/trid
ecyn
oic
acidgluc
uron
ide,
dod
ecan
edioic
acid,1
,3-dihyd
roxy
phe
nyl-2-O-
sulpha
te,p
-cou
maroy
lalcoh
ol-glucu
ronidean
d-sulph
ate,
N-ace
tylseroton
ine-su
lpha
te,5
-hyd
roxy
indolea
cetic
acid,u
rolitin
A-glucu
ronide,
sulpha
te,s
ulpha
tegluc
uron
ide
Tulip
anie
tal.(5
1)
Bee
fAcu
teinterven
tion
17Pos
tprand
ial
plasm
aGC–MS
2-am
inoa
dipic
acid,β
-alanine
,4-hyd
roxy
proline
Ros
set
al.(3
0)
Herrin
gAcu
teinterven
tion
17Pos
tprand
ial
plasm
aGC–MS
Cetoleicac
id,d
ocos
ahex
aeno
icac
idRos
set
al.(3
0)
Salmon
Acu
teinterven
tion
24Fa
stingan
dpos
tprand
ial
urine
FIE–FT
ICR–MS
Ans
erine,
methy
lhistid
ine,
TMAO
Lloy
det
al.(2
8)
Red
mea
t15
dinterven
tion
1724
hurine
Ionex
chan
gech
romatog
raph
y1an
d3methy
lhistid
ine
Cross
etal.(5
2)
Red
mea
t15
dinterven
tion
1224
hurine
NMR
Carnitin
e,crea
tinine,
TMAO,a
cetyl-ca
rnitine
,tau
rine,
1an
d3
methy
lhistid
ine
Stella
etal.(2
9)
Cruciferous
vege
tables
,citrus
andso
ya
2-wee
kinterven
tion
10Fa
stingurine
LTQ–FT
LC–MS/
MS
Prolinebetaine
,sulforapha
ne,h
ippuric
acid,g
enistein,d
aidze
in,e
quol,
glyc
itein,O
-des
methy
lang
olen
sin,
trigon
elline,
(iso)va
lerlg
lycine
,hy
droxy
phe
nylace
tyl-glyc
ine,
nico
tinuric
acid
May
etal.(4
0)
Ling
onberrie
sAcu
teinterven
tion
14Pos
tprand
ial
urine
NMR
Hippuric
acid,4
-hyd
roxy
hippuric
acid
Lehton
enet
al.(5
3)
Wine
28d
interven
tion
6124
hurine
NMR
Tartrate,4
-hyd
roxy
phe
nylace
tate,m
annitol,etha
nol
Vaz
eque
z-Fres
noet
al.(5
4)
Mixed
redwine/
grap
ejuiceex
trac
ts4-wee
kinterven
tion
5824
hurine
NMR,G
C–TO
F–MS
Syringicac
id,3
-hyd
roxy
hippuric
acid,4
-hyd
roxy
hippuric
acid,
3-hy
droxy
phe
nylace
ticac
id,4
-hyd
roxy
man
delic
acid,v
anilm
andelic
acid,h
ippuric
acid,3
-hyd
roxy
phe
nylpropionicac
id,
1,2,3-trihyd
roxy
ben
zene
,4-hyd
roxy
ben
zoic
acid,h
omov
anillic
acid,
dihyd
roferulic
acid,p
heny
lace
tylglutamine
vanDorsten
etal.(5
5)
Mixed
redwine/
grap
ejuiceex
trac
ts4d interven
tion
3524
hurine
GC–MS,L
C–MS
Syringicac
id,3
-hyd
roxy
hippuric
acid,p
yrog
allol,3-hy
droxy
phe
nylace
ticac
id,3
-hyd
roxy
phe
nylpropionicac
id,3
,4-dihyd
roxy
phe
nylpropion
icac
id,
indo
le-3-lac
ticac
id,h
ippuric
acid,c
atec
hol,4-hy
droxy
hippuric
acid,
3,4-dihyd
roxy
phe
nylace
ticac
id,v
anillic
acid
Jaco
bset
al.(5
6)
Dietary
fibres(oat
bran,
ryebran,
and
suga
rbee
tfibres)
5-wee
kinterven
tion
25Fa
stingplasm
aLC
–qTO
F–MS
2-am
inop
heno
lsulpha
te,2
,6-dihyd
roxy
ben
zoic
acid,h
ydroxy
latedan
dgluc
uron
idated
nuatigen
inJo
hans
son-Persson
etal.(5
7)
Dietary
fibre
6-mon
thinterven
tion
7724
hurine
NMR
Hippuric
acid
Ras
mus
senet
al.(5
8)
Metabolomics in dietary biomarker discovery 45
https://doi.org/10.1017/S002966511600032XDownloaded from https://www.cambridge.org/core. IP address: 65.21.228.167, on 30 Jan 2022 at 22:50:00, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
Proceedings
oftheNutritionSo
ciety
into two distinct dietary patterns. Combining liquid chro-matography–MS data from 24 h urine samples and datafrom 3-d weighed dietary data the present study againidentified conjugates of isothiocyanates as brassica biomar-kers. However, using this approach it was only possible toverify 23 % of potential biomarkers observed in theprevious-meal studies(27). As this was a less controlledintervention that included a wider selection of foods withvaried amounts of intake and different preparation meth-ods, it highlights the need for the validation of biomarkersin different subjects and study settings(27).
A number of red meat and fish biomarkers have beenidentified using this intervention approach(7,28,29). Mostrecently, metabolomics has been applied to comparethe different effects of meat and fish on the plasma meta-bolome(30). Ross et al.(30) carried out an interventionstudy analysing the differences in the postprandialplasma metabolic response to meals containing bakedbeef, baked herring and pickled herring. Seventeenmales consumed three test meals in a crossover designwith 1 week washout between the meals. Postprandialblood plasma samples were taken over 7 h and analysedby GC–MS. Concentrations of 2-aminoadipic acid,β-alanine and 4-hydroxyproline were significantly higherfollowing the beef meal compared with the baked herringmeal. Herring intake led to a greater plasma postprandialresponse from DHA and cetoleic acid compared withbeef(30). However, further studies are needed to confirmthese dietary biomarkers and decipher their specificity.
Dietary biomarker discovery using cohort studies
Searching for new dietary biomarkers in cohort studiesrequires the use of self-reported dietary data to identifylow and high consumers of a specific food. Followingthis, the metabolomic profiles are compared betweenlow and high consumers and potential biomarkers areidentified. Putative biomarkers of foods, identifiedusing this approach, are presented in Table 2. Work inour laboratory combined this approach with an acuteintervention to identify and confirm a panel of biomar-kers indicative of sugar-sweetened beverage (SSB) in-take(31). Heat map analysis was performed to identifycorrelations between NMR spectral regions and SSBintakes in the cohort study. A panel of four biomarkers:formate, citrulline, taurine and isocitrate were identifiedas markers of SSB intake. Following the acute consump-tion of the SSB all four metabolites were shown to in-crease in the urine and the panel of biomarkers weresuccessfully identified in the SSB(31). Another studyusing this cohort study approach, analysed the correla-tions between serum profiles and dietary data collectedusing FFQ in participants from the Prostate, Lung,Colorectal and Ovarian Cancer Screening Trial(32). Theapplication of untargeted metabolomics to this epidemio-logic data set detected thirty-nine metabolites of knownidentity that were correlated with a total of thirteen diet-ary groups, for example citrus intake was associated withstachydrine, chiro-inositol, scyllo-inositol and N-methylproline, fish with 3-carboxy-4-methyl-5-propyl-2-Ta
ble1.
(Con
t.)
Dietary
factor
Study
duration
No.
ofsu
bjects
Sam
ple
Metab
olom
ictech
nique
Biomarke
rReferen
ce
Who
le-grain
rye
bread
4-wee
kinterven
tion
2024
hurine
LC–qTO
F3-(3,5-D
ihyd
roxy
phe
nyl)-1-prop
anoicac
id-sulph
atean
d-glucu
ronide,
enterolacton
e-gluc
uron
ide,
azelaicac
id,2
-aminop
heno
l-su
lpha
te,
2,4-dihyd
roxy
-1,4-ben
zoxa
zin-3-on
e,2-am
inop
heno
l-su
lpha
te,2
–
4-dihyd
roxy
-1,4-ben
zoxa
zin-3-on
e-su
lpha
te,ind
olylac
ryloylglyc
ine,
ferulic
acid-sulpha
te,3
,5-dihyd
roxy
phe
nylethan
ol-sulpha
te,
3,5-dihyd
roxy
cinn
amic
acid-sulpha
te
Bon
dia-P
onset
al.(5
9)
Who
le-grain
sourdou
ghrye
bread
8-wee
kinterven
tion
2824
hurine
FIE–FT
ICR–MS
HHPAAgluc
uron
ide,HPAAsu
lpha
te,H
BOAgluc
uron
ide,
N-feruloy
lglycine
sulpha
te,H
HPAAsu
lpha
teBec
kman
net
al.(6
0)
Che
ese
6-wee
kinterven
tion
2324
hurine
UPLC
–ESI–qTO
FTy
ramine,
sulpha
te,iso
butyryl
glyc
ine(and
othe
rac
ylglyc
ines
),xa
nthu
renicac
id,4
-hyd
roxy
phe
nylace
ticac
idHjerpsted
etal.(6
1)
Milk
andch
eese
14d
interven
tion
15Fa
eces
,24h
urine
NMR
Milk;c
itrate,
crea
tine,
crea
tinine,
urea
,che
ese;
prolinebetaine
,tyros
ine,
hipp
urate
Zhe
nget
al.(6
2)
LC,liquidch
romatog
raphy
;ESI,elec
tros
prayionisa
tion;
qTO
F,qua
drupoletim
e-of-flight;LT
Q,linea
rtrap
qua
drupole;
FIE,flow
infusion
elec
tros
prayionisa
tion;
FTICR,Fo
uriertran
sform-ion
cyclotronreso
nanc
e;qTO
F,qua
drupoletim
e-of-flight;U
PLC
,ultra-perform
ance
liquidch
romatog
raphy
;HILIC,h
ydrophilic
liquidinteractionch
romatog
raphy
;TMAO,trim
ethy
lamine-N-oxide;
LTQ-FT.
linea
riontrap
-Fou
riertran
sform
mas
ssp
ectrom
eter;HHPAA,2
-hyd
roxy
-N-(2-hy
droxy
phe
nyl)a
cetamide;
HPAA,N
-(2-hy
droxy
phe
nyl)a
cetamide;
HBOA,2
-hyd
roxy
-1,4-ben
zoxa
zin-3-one
;HHPAA,2
-hyd
roxy
-N-(2-hy
droxy
phe
nyl)a
cetamide.
H. Gibbons et al.46
https://doi.org/10.1017/S002966511600032XDownloaded from https://www.cambridge.org/core. IP address: 65.21.228.167, on 30 Jan 2022 at 22:50:00, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
Proceedings
oftheNutritionSo
ciety
Table
2.Sum
maryof
putativebiomarke
rsiden
tified
usingametab
olom
icsap
proac
hin
coho
rtstud
ies
Dietary
factor
Dietary
asse
ssmen
ttool
No.
ofsu
bjects
Sam
ple
Metab
olom
ictech
nique
Biomarke
rsReferen
ce
Oily
fish
FFQ
68Fa
sting,
morning
spot,2
4hurine
FIE–FT
ICR–MS
Methy
lhistid
ine
Lloy
det
al.(6
3)
Citrus
fruit
24-h
dietary
reco
rdEighty
Fastingurine
LC–ESI–qTO
F,LT
Q-O
rbitrap
Prolinebetaine
,hyd
roxy
prolinebetaine
,hes
peretin
3′-O
-glucu
ronide,
narin
genin7-O-glucu
ronide,
limon
ene8,9-diol
gluc
uron
ide,
nootka
tone
13,14-diolg
lucu
ronide,
N-M
ethy
ltyramine
sulpha
te
Pujos
-Guillo
tet
al.(2
4)
Sug
ar-swee
tene
dbev
erag
es4-dfood
diary
565
Fastingurine
NMR
Form
ate,
isoc
itrate,
citrulline,
taurine
Gibbon
set
al.(3
1)
Citrus
,green
vege
tables
,redmea
t,sh
ellfish
,fish
,pea
nuts,c
offee,
etc.
FFQ
502
Fastingse
rum
UHPLC
–MS/M
S,
GC–MS
Citrus
;Scy
llo-&ch
iro-ino
sitol,Green
s;CMPF,
Red
mea
t;indo
lepropiona
te,S
hellfish
;CMPF,
Pea
nuts;T
ryptopha
nbetaine
,4-Vinylphe
nols
ulpha
te,C
offee;
trigon
elline-N-m
ethy
lnicotinate
andquina
te
Gue
rtin
etal.(3
2)
Coffee
24-h
dietary
reco
rd,F
FQ39
Morning
spot
urine
UPLC
–qTO
F–MS
Atrac
tylig
enin
gluc
uron
ide,
Cyc
lo(isoleu
cyl-prolyl),
1-Methy
lxan
thine,
1,7-dim
ethy
luric
acid,k
ahweo
loxide
gluc
uron
ide,
1-methy
luric
acid,trig
onelline,
dim
ethy
lxan
thine
gluc
uron
ide,
5-ac
etylam
ino-6-form
ylam
ino-3-methy
luracil(AMFU
),hipp
uric
acid,trim
ethy
luric
acid,p
arax
anthine,
3-hy
droxy
hippuric
acid,1
,3or
3,7-dim
ethy
luric
acid,c
affeine
Rothw
ell
etal.(6
4)
Coffee
FFQ
68Fa
sting,
morning
spot,2
4hurine
FIE–FT
ICR–MS
Dihyd
roca
ffeicac
idLloy
det
al.(6
3)
Red
mea
t24
-hdietary
reco
rd,F
FQ12
6Fa
stingurine
Ionex
chan
gech
romatog
raph
y1-Methy
lhistid
ine
Myint
etal.(6
5)
Red
mea
tFF
Q20
47Serum
FIA–MS/M
SPCaa
36:0
,PCaa
36:4
,PCaa
38:0
,PCaa
38:4
,PCae
34:2
,PC
ae34
:3,P
Cae
36:3
,PCae
36:4
,PCae
36:5
,PCae
38:4
,PCae
38:5
,PCae
38:6
,PCae
40:4
,Lyso-PC20
:4,S
M24
:1,F
erritin
Witten
bech
eret
al.(3
3)
White
bread
and
who
le-grain
bread
FFQ
155
Fastingsp
oturine
HPLC
–qTO
F–MS
2-Aminop
heno
lsulpha
te,H
PAAgluc
uron
ide,
HHPAA,H
MBOA
gluc
uron
ide,
HBOAglyc
oside,
HPPA,H
MBOA,D
HPPA
gluc
uron
ide,
3,5-dihyd
roxy
phe
nylethan
olsu
lpha
te,D
HPPTA
sulpha
te,h
ydroxy
ben
zoic
acid
gluc
uron
ide,
dihyd
roferulic
acid
sulpha
te,e
nterolac
tone
gluc
uron
ide,
pyrraline,
3-indolec
arbox
ylic
acid
gluc
uron
ide,
ribofl
avin,2
,8-dihyd
roxy
quino
linegluc
uron
ide
Garcia-Aloy
etal.(3
4)
Cruciferous
vege
tables,
citrus
andso
ya3-dfood
reco
rds,
FFQ
93Fa
stingurine
LTQ–FT
LC–MS/
MS
Prolinebetaine
May
etal.(4
0)
Polyp
heno
l-ric
hfood
s24
-hdietary
reco
rd,F
FQ48
124
hurine
UHPLC
–qTO
F–MS
Coffee;
dihyd
roferulic
acid
sulpha
te.R
edwine;
gallicac
idethy
les
ter.Citrus
fruit;na
ringe
ningluc
uron
ide.
Tea;
4-O-m
ethy
lgallic
acid.A
pplesan
dpea
rs;p
hloretingluc
uron
ide.
Cho
colate
produ
cts;
methy
l(epi)catec
hinsu
lpha
te
Edman
ds
etal.(6
6)
Walnu
tsFF
Q38
1Fa
stingsp
oturine
HPLC
–qTo
F–MS
3-indolec
arbox
ylic
acid
gluc
uron
ide,
hydroxy
indolea
cetic
acid
sulpha
te,N
-ace
tylseroton
insu
lpha
te,
10-hyd
roxy
-dec
ene-4,6-diyno
icac
idsu
lpha
te,trid
ecad
ieno
ic/
tridec
ynoicac
idgluc
uron
ide,
enterolacton
egluc
uron
ide,
urolith
ins
Garcia-Aloy
etal.(6
7)
FIE,fl
owinfusion
elec
tros
prayionisa
tion;
FTICR,Fo
uriertran
sform-ion
cyclotronreso
nanc
e;LC
,liquidch
romatog
raphy
;ESI,elec
tros
prayionisa
tion;
qTO
F,qua
drupoletim
e-of-flight;L
TQ,linea
rtrap
qua
drupole;
UHPLC
,ultra-HPLC
;CMPF,
3-ca
rbox
y-4-methy
l-5-propyl-2-furan
propan
oicac
id;F
IA,fl
owinjectionan
alysis;P
Caa
,diacy
lpho
spha
tidylch
olines
;PC
ae,a
cylalkyl
pho
spha
tidylch
olines
;Ly
so-P
C,
lyso
pho
spha
tidylch
olines
;SM,s
phing
omye
lin;HPAA,N-(2-hy
droxy
phe
nyl)ac
etam
ide;
HHPAA,2-hy
droxy
-N-(2-hy
droxy
phe
nyl)ac
etam
ide;
HMBOA,2
-hyd
roxy
-7-m
etho
xy-2H-1,4
-ben
zoxa
zin-3-on
e;HBOA,
2-hy
droxy
-1,4-ben
zoxa
zin-3-on
e;HPPA,2
-hyd
roxy
-N-(2-hy
droxy
phe
nyl)ac
etam
ide;
DHPPA,3-(3,5-dihyd
roxy
phe
nyl)propan
oicac
id;D
HPPTA
,5-(3,5-dihyd
roxy
phe
nyl)pen
tano
icac
id.
Metabolomics in dietary biomarker discovery 47
https://doi.org/10.1017/S002966511600032XDownloaded from https://www.cambridge.org/core. IP address: 65.21.228.167, on 30 Jan 2022 at 22:50:00, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
Proceedings
oftheNutritionSo
ciety
furanpropanoic acid, DHA and EPA, peanut intake withtryptophan betaine and 4-vinylphenol sulphate and coffeeintake was associated with trigonelline-N-methylnicotinateand quinate(32). To complicate interpretation further, the in-take of foods is highly correlated making identification ofspecific biomarkers difficult and this highlights the needfor the validation of biomarkers. The majority of biomar-kers identified using cohort studies have been predominantlyidentified in urine, this study demonstrates the potential useof serum samples in dietary biomarker discovery. However,the proposed biomarkers identified are only based on asso-ciations and some biomarkers were not food specific, for ex-ample DHA was correlated with fish and rice intake.Further validation in intervention studies is therefore neces-sary to demonstrate responsiveness to intake.
Wittenbecher et al. also demonstrated the use of serumsamples when identifying biomarkers of red meat intakein a subset of participants from the European ProspectiveInvestigation into Cancer and Nutrition-Potsdam cohort(n 2047)(33). Total red meat consumption was assessedusing FFQ and serum samples were analysed using a tar-geted metabolomics approach. Ferritin, glycine, four diacylphosphatidylcholines, eleven acylalkyl phosphatidylcho-lines, two lysophosphatidylcholines and two sphingomye-lins were associated with total red meat consumption andsix of these biomarkers were also found to be associatedwith type-2 diabetes risk(33). This is the first study evaluat-ing a large set of metabolites as potential mediators of theassociation between red meat intake and diabetes risk,however, dietary information relied on estimates of habit-ual consumption over the past year by FFQ and metabo-lites were measured at a single time point. Furthermore,total red meat was defined as processed and unprocessedmeat and therefore did not identify biomarkers of specifictypes of meat. Additional study is essential to validatethe biomarkers identified and to further dissect such rela-tionships with disease risk.
Biomarkers of bread intake have also been investi-gated in 155 subjects from the PERIMED study(34). A137-item FFQ was used to stratify subjects into threegroups: non-consumers of bread (n 56), white-bread con-sumers (n 48) and whole-grain bread consumers (n 51).Fasting urine samples, analysed by untargeted HPLC–quadruple time-of-flight–MS, identified higher concen-trations of compounds, including benzoxazinoids andalkylresorcinol metabolites and compounds producedby gut microbiota (enterolactones, hydroxybenzoic anddihydroferulic acid metabolites) in bread consumers. 2,8-dihydroxyquinoline glucuronide was also found to bemore abundant in whole-grain bread consumers(34).The biomarkers identified are based on a FFQ; thereforefurther validation is essential to demonstrate a direct re-lationship with bread consumption.
Dietary biomarker discovery using dietary patterns
The third approach, analysing dietary patterns and meta-bolomic profiles to identify nutritypes (i.e. metabolicprofiles that reflect dietary intake) and biomarkers havebeen demonstrated by a number of research groups (see
Table 3). One of the first examples emerged from our la-boratory when a k-means cluster analysis was performedon self-reporting dietary data and three distinct dietary pat-terns, which were associated with unique food intakes wereidentified(35). Dietary clusters were reflected in the urinarymetabolomic profiles of the 125 participants and a numberof metabolites were identified and linked to the intake ofspecific food groups(35). These nutritypes have the potentialto aid dietary assessment by unobjectively classifying peo-ple into certain dietary patterns. Further work within ourresearch group, applying the concept of using biomarkersto reflect dietary patterns, has focused on lipidomics, asubfield of metabolomics that concentrates on the globalstudy of lipids(36). Dietary data, measured by FFQ andlipid profiles measured from serum samples, in thirty-fourMetabolic Challenge Study participants were used for thisanalysis. Principal component analysis reduced lipidprofiles into lipid patterns and these were regressed againstdietary data to identify biomarkers related to the intake ofcertain foods and nutrients. Six lipid patterns were iden-tified including lipid pattern 1 which was found to be high-ly predictive of dietary fat intake (AUC= 0·82), lipidpattern 4 which was highly predictive of alcohol intake(AUC= 0·81) and lipid pattern 6 which had a reasonablygood ability to predict dietary fish intake (AUC= 0·76).Lysophosphatidylcholine alkyl C18 : 0 (LPCeC18 : 0)was identified as a potential biomarker of alcohol con-sumption and lysophosphatidylethanolamine acyl C18 : 2(LPEaC18 : 2) and phoshatidylethanolamine diacyl C38 :4 (PEaaC38 : 4) were identified as potential biomarkersof fish intake(36). This approach demonstrates the utilityof serum in the identification of key dietary factors thatinfluence the lipidomic profile. However, again validationof the biomarkers through use of intervention studies isneeded.
Most recently, Andersen et al. used an untargeted meta-bolomics approach to distinguish between two dietary pat-terns with the purpose of developing a compliancemeasure(37). In a parallel intervention study, 181 partici-pants were randomly assigned to follow a New NordicDiet or an Average Danish Diet. The 24 h urine sampleswere collected, analysed by ultra-performance liquid chro-matography–quadruple time-of-flight–MS and partialleast-squares discriminant analysis was applied to developa compliance model for Average Danish Diet and NewNordic Diet based on the most discriminative featuresdetected in urine. This resulted in a robust model with amisclassification rate of 19 %(37). Metabolites characteris-ing the Average Danish Diet and the New Nordic Dietare listed in Table 3. The present study demonstrates thepotential of metabolomics in discovering biomarkers indi-cative of dietary patterns, but furthermore it highlights apromising approach that may be used to develop compli-ance measures that cover the most important discriminantmetabolites of complex diets.
Limitations of current approaches/study designs
In general, metabolomic-based approaches have pro-duced reasonably robust models for dietary biomarker
H. Gibbons et al.48
https://doi.org/10.1017/S002966511600032XDownloaded from https://www.cambridge.org/core. IP address: 65.21.228.167, on 30 Jan 2022 at 22:50:00, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
Proceedings
oftheNutritionSo
ciety
Table
3.Sum
maryof
putativebiomarke
rsiden
tified
usingdietary
patternsan
dmetab
olom
icprofiles
Dietary
patterns
Dietary
pattern
approac
hSam
ple
Metab
olom
ictech
nique
Biomarke
rsReferen
ce
Pruden
tan
dWes
tern
dietary
patterns
PCA
Fastingplasm
aESI–MS/M
SWes
tern
dietary
pattern;inc
reas
edam
inoac
idsan
dsh
ort-ch
ainac
ylca
rnitine
sBou
chard-M
ercier
etal.(6
8)
Hea
lthy,
unhe
althy,
trad
ition
alIrish
dietary
pattern
k-mea
nsclus
ter
analysis
Fastingurine
NMR
Hea
lthy;
glyc
ine,
phe
nylace
tylglutaminean
dac
tetoac
etateTrad
ition
alIrish
;TMAO,O
-ace
tylcarnitin
ean
dnn
dimethy
lglycine
O’Sullivan
etal.(3
5)
Sev
endietary
patterns(e.g.h
ealth
ydiet,trad
ition
alBav
arian)
PCA
Fastingplasm
aESI–MS/M
SHea
lthydiet;dec
reas
ein
thedeg
reeof
saturatio
nof
the
fattyac
idmoietiesof
differen
tglyc
erol-
pho
spha
tidylch
olines
Altm
aier
etal.(6
9)
Sev
endietary
patterns(e.g.d
ietary
fat
lipid
pattern,a
lcoh
ollip
idpattern)
PCA
Fastingse
rum
ESI–MS/M
SAlcoh
olco
nsum
ptio
n;LP
CeC
18:0
Fish
cons
umptio
n;LP
EaC
18:2
,PEaa
C38
:4O’G
orman
etal.(3
6)
Five
dietary
patterns(e.g.e
nergy
intake
,plant
v.an
imal-bas
eddiet)
PCA
Fastingplasm
aNMR
Ene
rgyintake
;greater
conc
entrations
oflip
id-related
high
-ene
rgyintake
,highe
rcirculating
pho
spha
tidyc
holinerelatedto
lower
energy
intake
.Animal-bas
eddiet;high
erco
ncen
trations
oflysine
,arginine
,glutamine/glutam
ate,
threon
ine,
aspartate/
asparag
ine,
citratean
dpolyo
lcom
poun
ds
Peré-Trep
atet
al.(7
0)
NNDan
dan
ADD
24hurine
UPLC
–qTO
F–MS
NNDdiet;TM
AO,h
ippuric
acid,h
ydroquino
ne-
gluc
uron
ide,
(2-oxo
-2,3-dihyd
ro-1H-ind
ol-3-yl)a
cetic
acid
and3,4,5,6-tetrah
ydrohippu
rate.A
DDdiet;
pyrraline,
gluc
uron
ideco
njug
ated
produc
ts,
theo
brom
ine,
7-methy
luric
acid,3
,7-dim
ethy
luric
acid,
7-methy
lxan
thine,
6-am
ino-5-[N-m
ethy
lform
ylam
ino]-1-
methy
luracil,prolinebetaine
andgluc
uron
ides
ofperillic
acid
And
erse
net
al.(3
7)
Dietary
patternse.g.
high
intake
ofbutter/low
intake
ofmarga
rine,
high
intake
ofredmea
tan
dfish
/low
intake
ofwho
le-grain
bread
,tea
andco
ffee
RRR
Fastingse
rum
FIA–MS/M
SHighintake
ofbutteran
dlow
intake
ofmarga
rine;
acylca
rnitine
s,ac
yl-alkyl-pho
spha
tidylch
olines
,lyso-
pho
spha
tidylch
olines
andhy
droxy
-sphing
omye
lins.
Highintake
ofredmea
tand
fish
andlowintake
ofwho
le-
grainbread
andtea;
hexo
sean
dpho
spha
tidylch
olines
Floe
gale
tal.(7
1)
ADD,A
verage
Dan
ishDiet;PCA,prin
cipal
compon
entan
alysis;E
SI,elec
tros
prayionisa
tion;
LPCeC
18:0
,lysop
hosp
hatid
ylch
olinealky
lC18
:0.L
PEaC
18:2
,lysop
hosp
hatid
yletha
nolamineac
ylC18
:2.N
ND,N
ewNordic
Diet;PEaa
C38
:4,p
hosh
atidyletha
nolaminediaclyl
C38
:4.T
MAO,trim
ethy
lamine-N-oxide;
UPLC
,ultra-perform
ance
liquidch
romatog
raphy
;qTO
F,qua
drupoletim
e-of-flight;R
RR,red
uced
rank
regres
sion
;FIA,flow
injectionan
alysis.
Metabolomics in dietary biomarker discovery 49
https://doi.org/10.1017/S002966511600032XDownloaded from https://www.cambridge.org/core. IP address: 65.21.228.167, on 30 Jan 2022 at 22:50:00, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
Proceedings
oftheNutritionSo
ciety
identification. However, following the discovery of anybiomarker, validation in an independent study is criticalto enable the generalisability of the results. This valid-ation step is essential because factors which may not bepresent in traditional dietary assessment methods, includ-ing genetic factors, lifestyle and physiological factors,dietary factors, the biological sample or the analyticmethodology could skew biomarker measures of dietaryintake(38). For many of the study designs discussed, val-idation of the biomarker is often absent, making it diffi-cult for the translation of these biomarkers into practice.
It has been proposed that the confirmation of dietarybiomarkers should occur in two stages, firstly the dose–response effect should be included in intervention studiesand secondly the suitability of the candidate biomarkerin a free-living population should be investigated usinga (controlled) habitual diet(39). Evaluation of the dose–re-sponse relationship is critical as it allows for the assess-ment of the suitability of the biomarker over a range ofintakes(20). Unfortunately, in many studies, this import-ant step is often absent. Biomarkers identified using sam-ples from cohort studies do not assess the directrelationships of food amounts consumed and levels ofbiomarkers and do not demonstrate responsiveness tointakes, therefore the relationship is only an associ-ation(16). Such studies should ideally be combined withintervention studies to demonstrate direct relationshipsand dose–response relationships. Conversely, dietarybiomarkers identified within acute intervention studiesadvantageously allow for the examination of dose–response relationships; however, to date few studieshave incorporated such designs.
When using self-reporting dietary data from cohortstudies in the biomarker discovery process, one shouldbe aware of reporting errors and the potential for missingimportant correlations and attenuation of results. Mayet al. investigated the metabolomic profiles of partici-pants consuming a high-phytochemical diet comparedwith a diet without fruit and vegetables in a randomisedcontrolled trial and also investigated the metabolomicprofiles of participants in a cross-sectional study, wherehigh and low fruit and vegetable diets were identifiedbased on 3-d food records and FFQ. The interventionstudy found forty-six putatively annotated ions, withMS/MS fragment ion support that were differentiallyabundant between the two intervention diets; however,within the cross-sectional study only one compound anno-tated with MS/MS support was identified using the 3-dfood records and there were no metabolites that signifi-cantly separated groups based on FFQ data(40). This there-fore demonstrates the drawbacks of using self-reporteddata in dietary biomarker discovery. Furthermore, whenusing cohort studies to identify or confirm biomarkers itis imperative that it is acknowledged that many of thefoods consumed are highly correlated and therefore bio-markers identified may not be specific to the particularfood of interest(20). Following identification of putativebiomarkers from cohort studies we recommend that the re-lationship is confirmed using an intervention study in adose–response manner where the sensitivity and specificityof the biomarkers can also be assessed. The importance of
such a step is key to the validation of the biomarkers andimportant to support their use.
Use of acute and medium-term interventions is notwithout limitations in terms of dietary biomarker iden-tification: many of the biomarkers identified using thisapproach are markers of acute intake. For example, pro-line betaine is excreted rapidly in urine and excretion isalmost complete ≤24 h(22). These acute biomarkers maytherefore only be valid for people that regularly and fre-quently consume the particular foods. The identificationof dietary biomarkers that reflect habitual intake requireslonger-term studies. Furthermore, it must also be notedthat the majority of the acute and medium-term interven-tion study designs involve only a small number of partici-pants(22,24,41). The proposed dietary biomarkers identifiedusing these approaches therefore cannot always be extra-polated to population studies in free-living individuals.However, this can be in part be dealt with by confirmationin cohort studies with a diverse range of characteristics.
While considering the earlier described limitations instudy designs, there is also the need for development ofdatabases and software tools to advance the interpret-ation of metabolomics results and therefore enhancethe utility of dietary biomarkers in nutrition research.Current databases such as the Human MetabolomeDatabase provides access to an online database containingdetailed information about small molecule metabolites(>40 000) found in the human body(42). Since it was firstdescribed in 2007, it is constantly being expanded andupdated and has become a valuable resource that containsspectroscopic, quantitative, analytic and physiological in-formation about human metabolites(42). The FoodMetabolome Database is another database of >28 000food constituents that contains information about foodsources and food concentrations(43). This resource pro-vides an aid for the identification of new metabolitesthat are reflective of food intake. While this resource isvaluable, the identification of metabolites originatingfrom food remains difficult and there is a need for sharingof databases to aid identification. Most recently, a com-prehensive and electronically accessible human urinemetabolome database, which includes quantitative con-centrations of metabolites in urine samples was estab-lished(44). This database also represents a significantdevelopment and resource for biomarker identificationand quantification. Other new software tools includeBAYESIL; this system provides fully automated andfully quantitative NMR-based metabolomics of complexmixtures(45). This will have a significant impact onNMR spectroscopy and NMR-based metabolomics.
Conclusion
The use of dietary biomarkers in nutrition research holdsgreat promise. However, prior to having a suite of reli-able dietary biomarkers that could be used in nutritionresearch a number of validation steps need to considered.Furthermore, the challenges identified in this reviewneed to be acknowledged and addressed. Appropriatevalidation steps are essential, otherwise the robustness
H. Gibbons et al.50
https://doi.org/10.1017/S002966511600032XDownloaded from https://www.cambridge.org/core. IP address: 65.21.228.167, on 30 Jan 2022 at 22:50:00, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
Proceedings
oftheNutritionSo
ciety
of biomarkers will remain uncertain and the translationof these biomarkers into practice will be challenging.Longer-term studies are also needed for the identificationof dietary biomarkers reflective of habitual dietary in-take. Until well-validated biomarkers are identified it isunlikely we will see uptake by the research communityof the emerging biomarkers. The challenge for theresearchers working in this field, in the coming years,will be to develop a suite of well-validated biomarkers.To this end the Joint Programming Initiatives fundedprogramme FoodBall will address some of these issuesand pave the way forward (http://foodmetabolome.org/).They may also have the potential for the assessment ofcompliance to dietary interventions in both a clinicaland a research setting. Ultimately these dietary biomar-kers will be used to further elucidate the proposed linksbetween certain foods and disease.
Financial Support
The authors would like to acknowledge the followingfunding: FP7 Project NutriTech (289511), SFI (14/JP-HDHL/B3075) and ERC (647783).
Conflict of Interest
None.
Authorship
H. G. drafted the outline of the manuscript, conductedthe literature search and drafted the manuscript. L. B.was responsible for critically reviewing the manuscript.Both authors read and approved the final manuscript be-fore submission.
References
1. Mozaffarian D & Ludwig DS (2010) Dietary guidelines inthe 21st century – a time for food. JAMA 304, 681–682.
2. Bingham SA (2002) Biomarkers in nutritional epidemi-ology. Public Health Nutr 5, 821–827.
3. Kotchen TA, Cowley AW & Frohlich ED (2013) Salt inhealth and disease – a delicate balance. N Engl J Med368, 1229–1237.
4. Feskens EJ, Sluik D, Fau - van Woudenbergh GJ et al.(2013) Meat consumption, diabetes, and its complications.Curr Diab Rep 13, 298–306.
5. Pan A, Sun Q, Bernstein AM et al. (2011) Red meat con-sumption and risk of type 2 diabetes: 3 cohorts of USadults and an updated meta-analysis. Am J Clin Nutr 94,1088–1096.
6. Hu FB, Rimm EB, Stampfer MJ et al. (2000) Prospectivestudy of major dietary patterns and risk of coronaryheart disease in men. Am J Clin Nutr 72, 912–921.
7. Cross AJ, Leitzmann MF, Gail MH et al. (2007) A pro-spective study of red and processed meat intake in relationto cancer risk. PLoS Med 4, e325.
8. Appel LJ, Moore TJ, Obarzanek E et al. (1997) A clinicaltrial of the effects of dietary patterns on blood pressure. NEngl J Med 336, 1117–1124.
9. Sacks FM, Svetkey LP, Vollmer WM et al. (2001) Effectson blood pressure of reduced dietary sodium and the diet-ary approaches to stop hypertension (DASH) diet. N EnglJ Med 344, 3–10.
10. Estruch R, Ros E, Salas-Salvado J et al. (2013) Primary pre-vention of cardiovascular disease with a Mediterraneandiet. N Engl J Med 368, 1279–1290.
11. Salas-Salvadó J, Bulló M, Babio N et al. (2011) Reductionin the incidence of Type 2 diabetes with the mediterraneandiet: results of the PREDIMED-Reus nutrition interven-tion randomized trial. Diab Care 34, 14–19.
12. Mozaffarian D, Appel LJ & Van Horn L (2011)Components of a cardioprotective diet: new insights.Circulation 123, 2870–2891.
13. Kipnis V, Midthune D, Freedman L et al. (2002) Bias indietary-report instruments and its implications for nutri-tional epidemiology. Public Health Nutr 5, 915–923.
14. Dhurandhar NV, Schoeller D, Brown AW et al. (2015)Energy balance measurement: when something is not betterthan nothing. Int J Obes 39, 1109–1113.
15. Marshall JR & Chen Z (1999) Diet and health risk: riskpatterns and disease-specific associations. Am J Clin Nutr69, 1351s–1356s.
16. Brennan L, Gibbons H & O’Gorman A (2015) An over-view of the role of metabolomics in the identification ofdietary biomarkers. Curr Nutr Rep 4, 304–312.
17. Tasevska N, Runswick SA, McTaggart A et al. (2005) urin-ary sucrose and fructose as biomarkers for sugar consump-tion. Cancer Epidemiol Biomarkers Prev 14, 1287–1294.
18. Potischman N (2003) Biologic and methodologic issues fornutritional biomarkers. J Nutr 133, 875s–880s.
19. Wishart DS (2008)Metabolomics: applications to food scienceand nutrition research. Trends Food Sci Technol 19, 482–493.
20. Scalbert A, Brennan L, Manach C et al. (2014) The foodmetabolome: a window over dietary exposure. Am J ClinNutr 99, 1286–1308.
21. Atkinson W, Downer P, Lever M et al. (2007) Effects of or-ange juice and proline betaine on glycine betaine and homo-cysteine in healthy male subjects. Eur J Nutr 46, 446–452.
22. Heinzmann SS, Brown IJ, Chan Q et al. (2010) Metabolicprofiling strategy for discovery of nutritional biomarkers:proline betaine as a marker of citrus consumption. Am JClin Nutr 92, 436–443.
23. Lloyd AJ, Beckmann M, Favé G et al. (2011) Proline beta-ine and its biotransformation products in fasting urine sam-ples are potential biomarkers of habitual citrus fruitconsumption. Br J Nutr 106, 812–824.
24. Pujos-Guillot E, Hubert J, Martin JF et al. (2013) Massspectrometry-based metabolomics for the discovery of bio-markers of fruit and vegetable intake: citrus fruit as a casestudy. J Proteome Res 12, 1645–1659.
25. Edmands WMB, Beckonert OP, Stella C et al. (2011)Identification of human urinary biomarkers of cruciferousvegetable consumption by metabonomic profiling. JProteome Res 10, 4513–4521.
26. Andersen M-BS, Reinbach HC, Rinnan Å et al. (2013)Discovery of exposure markers in urine for Brassica-containing meals served with different protein sources byUPLC–qTOF-MS untargeted metabolomics. Metabolomics9, 984–997.
27. Andersen MB, Kristensen M, Manach C et al. (2014)Discovery and validation of urinary exposure markers fordifferent plant foods by untargeted metabolomics. AnalBioanal Chem 406, 1829–1844.
Metabolomics in dietary biomarker discovery 51
https://doi.org/10.1017/S002966511600032XDownloaded from https://www.cambridge.org/core. IP address: 65.21.228.167, on 30 Jan 2022 at 22:50:00, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
Proceedings
oftheNutritionSo
ciety
28. Lloyd AJ, Fave G, Beckmann M et al. (2011) Use of massspectrometry fingerprinting to identify urinary metabolitesafter consumption of specific foods. Am J Clin Nutr 94,981–991.
29. Stella C, Beckwith-Hall B, Cloarec O et al. (2006)Susceptibility of human metabolic phenotypes to dietarymodulation. J Proteome Res 5, 2780–2788.
30. Ross AB, Svelander C, Undeland I et al. (2015) Herringand beef meals lead to differences in plasma 2-aminoadipicacid, β-alanine, 4-hydroxyproline, cetoleic acid, and doco-sahexaenoic acid concentrations in overweight men. JNutr 11, 2456–2463.
31. Gibbons H, McNulty BA, Nugent AP et al. (2015) A meta-bolomics approach to the identification of biomarkers ofsugar-sweetened beverage intake. Am J Clin Nutr 101,471–477.
32. Guertin KA, Moore SC, Sampson JN et al. (2014)Metabolomics in nutritional epidemiology: identifyingmetabolites associated with diet and quantifying their po-tential to uncover diet-disease relations in populations.Am J Clin Nutr 101, 1000–1011.
33. Wittenbecher C, Mühlenbruch K, Kröger J et al. (2015)Amino acids, lipid metabolites, and ferritin as potentialmediators linking red meat consumption to type 2 diabetes.Am J Clin Nutr 101, 1241–1250.
34. Garcia-Aloy M, Llorach R, Urpi-Sarda M et al. (2014)Nutrimetabolomics fingerprinting to identify biomarkersof bread exposure in a free-living population from thePREDIMED study cohort. Metabolomics 11, 155–165.
35. O’Sullivan A, Gibney MJ & Brennan L (2011) Dietary in-take patterns are reflected in metabolomic profiles: poten-tial role in dietary assessment studies. Am J Clin Nutr 93,314–321.
36. O’Gorman A, Morris C, Ryan M et al. (2014) Habitualdietary intake impacts on the lipidomic profile. JChromatogr B, Anal Technol Biomed Life Sci 966, 140–146.
37. Andersen M-BS, Rinnan Å, Manach C et al. (2014)Untargeted metabolomics as a screening tool for estimatingcompliance to a dietary pattern. J Proteome Res 13, 1405–1418.
38. Jenab M, Slimani N, Bictash M et al. (2009) Biomarkers innutritional epidemiology: applications, needs and new hor-izons. Hum Genet 125, 507–525.
39. Kuhnle GG (2012) Nutritional biomarkers for objectivedietary assessment. J Sci Food Agric 92, 1145–1149.
40. May DH, Navarro SL, Ruczinski I et al. (2013)Metabolomic profiling of urine: response to a randomised,controlled feeding study of select fruits and vegetables, andapplication to an observational study. Br J Nutr 110, 1760–1770.
41. Heinzmann SS, Holmes E, Kochhar S et al. (2015)2-Furoylglycine as a candidate biomarker of coffee con-sumption. J Agric Food Chem 63, 8615–8621.
42. Wishart DS, Jewison T, Guo AC et al. (2013) HMDB 3.0 –the human metabolome database in 2013. Nucleic AcidsRes 41, D801–D807.
43. Wishart D (2012) Systems biology resources arising fromthe human metabolome project. In Genetics MeetsMetabolomics: from Experiment to Systems Biology, pp.157–175 [K Suhre, editor]. New York, NY: SpringerNew York.
44. Bouatra S, Aziat F, Mandal R et al. (2013) The humanurine metabolome. PLoS ONE 8, e73076.
45. Ravanbakhsh S, Liu P, Bjordahl TC et al. (2015) Accurate,fully-automated nmr spectral profiling for metabolomics.PLoS ONE 10, e0124219.
46. Lang R, Wahl A, Stark T et al. (2011) UrinaryN-methylpyridinium and trigonelline as candidate dietarybiomarkers of coffee consumption. Mol Nutr Food Res55, 1613–1623.
47. van Velzen EJJ, Westerhuis JA, van Duynhoven JPM et al.(2009) Phenotyping tea consumers by nutrikinetic analysisof polyphenolic end-metabolites. J Proteome Res 8, 3317–3330.
48. Daykin CA, Duynhoven JPMV, Groenewegen A et al.(2005) Nuclear magnetic resonance spectroscopic basedstudies of the metabolism of black tea polyphenols inhumans. J Agric Food Chem 53, 1428–1434.
49. Van Dorsten FA, Daykin CA, Mulder TPJ et al. (2006)Metabonomics approach to determine metabolic differ-ences between green tea and black tea consumption. JAgric Food Chem 54, 6929–6938.
50. Wang Y, Tang H, Nicholson JK et al. (2005) AMetabonomic strategy for the detection of the metaboliceffects of chamomile (Matricaria recutita L.) ingestion. JAgric Food Chem 53, 191–196.
51. Tulipani S, Llorach R, Jáuregui O et al. (2011)Metabolomics unveils urinary changes in subjects withmetabolic syndrome following 12-week nut consumption.J Proteome Res 10, 5047–5058.
52. Cross AJ, Major JM & Sinha R (2011) Urinary biomarkersof meat consumption. Cancer Epidemiol Biomarkers Prev20, 1107–1111.
53. Lehtonen H-M, Lindstedt A, Järvinen R et al. (2013) 1HNMR-based metabolic fingerprinting of urine metabolitesafter consumption of lingonberries (Vaccinium vitis-idaea)with a high-fat meal. Food Chem 138, 982–990.
54. Vázquez-Fresno R, Llorach R, Alcaro F et al. (2012)1H-NMR-based metabolomic analysis of the effect ofmoderate wine consumption on subjects with cardiovascu-lar risk factors. Electrophoresis 33, 2345–2354.
55. van Dorsten FA, Grün CH, van Velzen EJJ et al. (2010)The metabolic fate of red wine and grape juice polyphenolsin humans assessed by metabolomics. Mol Nutr Food Res54, 897–908.
56. Jacobs DM, Fuhrmann JC, van Dorsten FA et al. (2012)Impact of short-term intake of red wine and grape polyphe-nol extract on the human metabolome. J Agric Food Chem60, 3078–3085.
57. Johansson-Persson A, Barri T, Ulmius M et al. (2013) LC–QTOF/MS metabolomic profiles in human plasma after a5-week high dietary fiber intake. Anal Bioanal Chem 405,4799–4809.
58. Rasmussen LG, Winning H, Savorani F et al. (2012)Assessment of dietary exposure related to dietary GI andfibre intake in a nutritional metabolomic study of humanurine. Genes Nutr 7, 281–293.
59. Bondia-Pons I, Barri T, Hanhineva K et al. (2013) UPLC–QTOF/MS metabolic profiling unveils urinary changes inhumans after a whole grain rye versus refined wheatbread intervention. Mol Nutr Food Res 57, 412–422.
60. Beckmann M, Lloyd AJ, Haldar S et al. (2013)Hydroxylated phenylacetamides derived from bioactivebenzoxazinoids are bioavailable in humans after habitualconsumption of whole grain sourdough rye bread. MolNutr Food Res 57, 1859–1873.
61. Hjerpsted JB, Ritz C, Schou SS et al. (2014) Effect ofcheese and butter intake on metabolites in urine using anuntargeted metabolomics approach. Metabolomics 10,1176–1185.
62. Zheng H, Yde CC, Clausen MR et al. (2015)Metabolomics investigation to shed light on cheese as a
H. Gibbons et al.52
https://doi.org/10.1017/S002966511600032XDownloaded from https://www.cambridge.org/core. IP address: 65.21.228.167, on 30 Jan 2022 at 22:50:00, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
Proceedings
oftheNutritionSo
ciety
possible piece in the french paradox puzzle. J Agric FoodChem 63, 2830–2839.
63. Lloyd AJ, BeckmannM, Haldar S et al. (2013) Data-drivenstrategy for the discovery of potential urinary biomarkers ofhabitual dietary exposure. Am J Clin Nutr 97, 377–389.
64. Rothwell JA, Fillâtre Y, Martin JF et al. (2014) New bio-markers of coffee consumption identified by the non-targeted metabolomic profiling of cohort study subjects.PLoS ONE 9, e93474.
65. Myint T, Fraser GE, Lindsted KD et al. (2000) Urinary1-methylhistidine is a marker of meat consumption inblack and in white California seventh-day adventists. AmJ Epidemiol 152, 752–755.
66. Edmands WMB, Ferrari P, Rothwell JA et al. (2015)Polyphenol metabolome in human urine and its associationwith intake of polyphenol-rich foods across Europeancountries. Am J Clin Nutr 102, 905–913.
67. Garcia-Aloy M, Llorach R, Urpi-Sarda M et al. (2014)Novel multimetabolite prediction of walnut consumption
by a urinary biomarker model in a free-living population:the PREDIMED study. J Proteome Res 13, 3476–3483.
68. Bouchard-Mercier A, Paradis A-M, Rudkowska I et al.(2013) Associations between dietary patterns and gene ex-pression profiles of healthy men and women: a cross-sectional study. Nutr J 12, 1–13.
69. Altmaier E, Kastenmüller G, Römisch-Margl W et al.(2010) Questionnaire-based self-reported nutritionhabits associate with serum metabolism as revealed byquantitative targeted metabolomics. Eur J Epidemiol 26,145–156.
70. Peré-Trepat E, Ross AB, Martin F-P et al. (2010)Chemometric strategies to assess metabonomic imprintingof food habits in epidemiological studies. ChemometricsIntell Lab Syst 104, 95–100.
71. Floegel A, von Ruesten A, Drogan D et al. (2013)Variation of serum metabolites related to habitual diet: atargeted metabolomic approach in EPIC-Potsdam. Eur JClin Nutr 67, 1100–1108.
Metabolomics in dietary biomarker discovery 53
https://doi.org/10.1017/S002966511600032XDownloaded from https://www.cambridge.org/core. IP address: 65.21.228.167, on 30 Jan 2022 at 22:50:00, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.