Sei sulla pagina 1di 19

Food Chemistry 278 (2019) 144–162

Contents lists available at ScienceDirect

Food Chemistry
journal homepage: www.elsevier.com/locate/foodchem

Review

Food fingerprints – A valuable tool to monitor food authenticity and safety T


Sonia Medina , Jorge A. Pereira, Pedro Silva, Rosa Perestrelo, José S. Câmara
⁎ ⁎

CQM – Centro de Química da Madeira, Universidade da Madeira, Campus da Penteada, 9020-105 Funchal, Portugal

ARTICLE INFO ABSTRACT

Keywords: In recent years, food frauds and adulterations have increased significantly. This practice is motivated by fast
Food fingerprints economical gains and has an enormous impact on public health, representing an important issue in food science.
Authenticity In this context, this review has been designed to be a useful guide of potential biomarkers of food authenticity
Food safety and safety. In terms of food authenticity, we focused our attention on biomarkers reported to specify different
Adulteration
botanical or geographical origins, genetic diversity or production systems, while at the food safety level, mo-
Biomarkers
lecular evidences of food adulteration or spoilage will be highlighted. This report is the first to combine results
from recent studies in a format that allows a ready overview of metabolites (< 1200 Da) and potentially mo-
lecular routes to monitor food authentication and safety. This review has therefore the potential to unveil im-
portant aspects in food adulteration and safety, contributing to improve the current regulatory frameworks.

1. Introduction adulteration is carried out to increase volume and weight, mask inferior
quality and replace authentic substances with cheaper ones, with ben-
Food fingerprints can be defined as molecular markers that re- efits for sellers and possible harms to consumers. It is therefore a major
present a characteristic food state or condition, allowing more effective concern for consumers, food producers, companies, and regulatory
product discrimination. Essentially, it is a marker or set of markers that agencies. The food frauds most often detected are adulteration (e.g.,
allow us to answer many questions about food authenticity, as “Are addition of water), sophistication processes (e.g., use of sugars, ad-
those organic carrots truly organic? Does this saffron really originate ditives, or flavours), and counterfeiting (e.g., use of prohibited additives
from Spain? Can we discriminate between orange juice and pulp wash? and colorants, incorrect botanical and geographical declarations, reuse
(Cubero-Leon, De Rudder, & Maquet, 2018; Le Gall, Puaud, & of expired products, or incorrectly naming something an organic pro-
Colquhoun, 2001; Rubert, Lacina, Zachariasova, & Hajslova, 2016). duct) (Avula, Smillie, Wang, Zweigenbaum, & Khan, 2015; Granato,
Unfortunately, the search for these fingerprints is not only related with Koot, Schnitzler, & van Ruth, 2015; Tette, Guidi, Bastos, Fernandes, &
the quality of the products, but also with their safety to human health. Gloria, 2017; van Ruth, et al., 2011; Xue et al., 2013). Therefore, when
Although the deliberate adulteration of food and beverages to deceive there is a food suspected of adulteration, rapid and robust methods, as
consumers is illegal in any part of the world, the promise of fast eco- well as specific and reliable markers, must be available to support the
nomical incomes is becoming very prevalent and affects a broad range withdraw of such products from the food chain and act against the
of products, from dietary food products to beverages. Food products facilitators of that frauds (Llano, Muñoz-Jiménez, Jiménez-Cartagena,

Abbreviations: ABA, abscisic acid; CINs, cytoplasmic invertases; CLA, conjugated linoleic acids; CQA, caffeoylquinic acids; DART, direct analysis in real time; DART-
q-TOF-MS, direct analysis in real time-quadrupole-time of flight-mass spectrometry; DUFA, diunsaturated fatty acids; DXP, 1-deoxy-D-xylulose-5- phosphate; EFSA,
European Food Safety Authority; FAME, fatty acyl methyl ester; FDA, Food and Drug Administration; FSMA, Food Safety Modernization Act; FT-ICR-MS, Fourier
transform-ion cyclotron resonance-mass spectrometry; GC-O, gas chromatography-olfactometry; GLVs, green leaf volatiles; HPLC-ESI-MS/MS, high-performance
liquid chromatography-electrospray ionization-tandem mass spectrometry; HPLC-q-TOF-MS, high-performance liquid chromatography-quadrupole-time of flight-
mass spectrometry; HS-SPME-GC–MS, headspace solid phase microextraction gas chromatography-mass spectrometry; LC, liquid chromatography; LOOCV, leave-
one-out cross validation; MDGC, multidimensional gas chromatography; MS, mass spectrometry; MUFA, monounsaturated fatty acids; MVA, mevalonic acid; NIRS,
near-infrared spectroscopy; NMR, nuclear magnetic resonance; O/C, organic and conventional; OPLS, orthogonal partial least square; PCA, principal component
analysis; PDO, protected designation of origin; PLOT, porous layer open tubular; PTR-MS, proton transfer reaction mass spectrometry; PUFA, polyunsaturated fatty
acids; SBSE, stir bar sorptive extraction; SIMCA, soft independent modelling of class analogy; SIFT-MS, selected ion flow tube mass spectrometry; SLDA, stepwise
linear discriminant analysis; TMA-N, trimethylamine nitrogen; TVB-N, total volatile bases nitrogen; UPLC-HR-MS, ultra-performance liquid chromatography-high
resolution-mass spectrometry; VIN, vacuolar invertases; VOCs, volatile organic compounds

Corresponding authors.
E-mail addresses: sonia.escudero@staff.uma.pt (S. Medina), jsc@staff.uma.pt (J.S. Câmara).

https://doi.org/10.1016/j.foodchem.2018.11.046
Received 13 July 2018; Received in revised form 2 November 2018; Accepted 8 November 2018
Available online 09 November 2018
0308-8146/ © 2018 Elsevier Ltd. All rights reserved.
S. Medina et al. Food Chemistry 278 (2019) 144–162

Londoño-Londoño, & Medina, 2018). This is particularly relevant when, current regulatory frameworks.
besides the economic loss, the adulteration involves health risks for the
consumers. Hence, the ability to trace and authenticate food products is 2. Potential biomarkers to monitor food authenticity and safety
a great concern for the food industry, but also for the public authorities.
For this reason, there is extensive legislation regarding food safety and Food chemistry research is a very dynamic field and multiple studies
authenticity all over the world, being USA and EU responsible for the aiming food products authenticity have performed, producing many
most comprehensive and strict guidelines. In USA, the Food Safety high quality and robust classification models. However, in many of
Modernization Act (FSMA) (https://www.fda.gov/Food/ these reports the discriminant metabolites were not unequivocally
GuidanceRegulation/FSMA/, accessed October 19, 2018) defined by identified, because this is a very time-consuming step, still constituting
the Food and Drug Administration (FDA) establishes several rules to nowadays the major bottleneck in the metabolomics workflow. For this
prevent foodstuffs adulteration. In turn, the European Food Safety reason, high-throughput approaches have been developed to detect
Authority (EFSA) (https://www.efsa.europa.eu/en/aboutefsa, accessed specific metabolites contributing for samples discrimination. These
October 19, 2018) includes a set of laws and guidelines for the food metabolites may be used as promising biomarkers of authentication,
quality and safety assurance, designed as General Food Laws. but also monitor biochemical changes in many foodstuffs. The main
In food authentication field, there are targeted and untargeted biomarkers and biological pathways involved in food authenticity and
methods of analysis. The first approach focuses on the analysis of a safety according to different varieties/cultivars, production systems,
specific metabolite or group of metabolites, while in the untargeted geographical/botanical origin, adulteration practices, as well as spoi-
approach, the major goal is discrimination of patterns of metabolites lage/freshness indicators are detailed below. As already referred, due to
that may change in response to environmental, genetic or human al- the extensive scientific literature publish on this subject, we only focus
terations (adulteration) (Llano et al., 2018). In this context, application on studies that were able to identify discriminant metabolites (bio-
of metabolomics to food science for assessing safety and authenticity markers) using consistent statistical models (that are also referred in
has gained much interest (Arbona, Iglesias, & Gómez-Cadenas, 2015; Fig. 1 and in the respective Tables 1–5).
Bonte, et al., 2014; Eisenmann, et al., 2016; Souard et al., 2018). Me-
tabolomics studies are mainly discriminative and predictive, aiming to 2.1. Species, varieties or cultivars
find differences between samples (e.g., for the oranges origin dis-
crimination, for the variety and vintage differentiation of wines or for There are serious economic and quality reasons to certify the au-
the civet coffees authentication) and to create statistical models to thenticity of the varieties used in different foodstuffs. However, as food
predict class memberships (e.g., for the classification of green coffee processing progresses, it becomes extremely difficult to distinguish
beans or wines according to botanical and geographical origins). Food among species, varieties or cultivars and this fact is often used to carry
fingerprints may be obtained for food quality, safety and authenticity out food adulterations. This problem may be solved by the detection
purposes. To achieve this, an experimental design should be carefully and eventually quantification of specific metabolites able to dis-
established, being the sample preparation and the analytical platforms criminate the adulteration. In Table 1 are referred relevant examples of
selected critical parameters to evaluate. A wide range of instrumental potential biomarkers of food adulteration in different foodstuffs, being
techniques, as chromatography (gas and liquid), mass spectrometry or the respective matrices, methodology and statistical analyses identified
spectroscopy is used to monitor food authenticity and safety. To support for the reader guidance. Grapes (Vitis vinifera) have a critical im-
this review, it was queried PubMed and Scopus databases with the portance in wine properties and quality and different studies have in-
keywords “biomarkers” and “food authentication” to obtain the re- vestigated a putative discrimination between cultivars. Different me-
search published in the last five years in food authenticity. This has tabolites, as organic acids, amino acids, carbohydrates,
retrieved us hundreds of reports, essentially involving mass spectro- phenylpropanoids, and flavonoids, have been previously pointed as
metry (MS)-based studies (mainly gas chromatography coupled with potential biomarkers for such discrimination. In this context, Ali et al.
mass spectrometry (GC–MS) and liquid chromatography mass spectro- (2009) reported the use of 1H NMR technique and multivariate analyses
metry with quadrupole time of flight technology (LC-q-TOF-MS)) for the differential characterization of plant samples based on their
(Arbona, et al., 2015; Bonte, et al., 2014; Klockmann, Reiner, metabolic composition. This study also demonstrates that the sample
Bachmann, Hackl, & Fischer, 2016), spectroscopy (nuclear magnetic storage conditions and the solvent extraction may produce artifacts,
resonance (NMR), Raman, UV–Visible, etc.) and spectrofluorimetry leading to the detection of outliers in the statistical analysis. Never-
methods (Farag, Labib, Noleto, Porzel, & Wessjohann, 2018; Li, Zhang, theless, small variations as the age of the plant samples (young versus
Jin, Liu, & Wang, 2016; Longobardi, et al., 2017; Nekvapil, et al., old leaves of the same plant) may result in a different metabolic profile
2018). Nevertheless, more than 75% of the publications cited in this and so the interpretation of the results should consider these inter-
review were performed by MS technology (Tables 1–5). The popularity ferences (Ali, et al., 2009). Related with this work, another study pre-
of MS approaches is due to ability to distinguish small differences be- sented C6-alcohols (1-hexenol, (E)-3-hexenol and (Z)-3-hexenol) and
tween samples in complex matrices. In contrast, the development of the ratios among them as varietal markers in grapes. It allowed the
spectroscopy and spectrofluorimetry methods may be related with their discrimination in six different white grapes varieties (Alvarinho, Arinto,
non-destructive nature, which is very relevant for expensive samples. Avesso, Azal, Loureiro, Trajadura) and three red grapes varieties
Other analytical strategies, as stable isotope and DNA analyses are (Amaral, Borraçal and Vinhão) from Portugal (Oliveira, Faria, Sá,
being also applied routinely for food authenticity purposes. However, in Barros, & Araújo, 2006). In the same context, five monovarietal red
this review, we will focus mainly on studies involving small molecules wines made from Negramoll, Listán negro, Tintilla, Vijariego negro and
(< 1200 Da) identified by chromatographic methodologies and labelled Baboso negro were analysed with gas chromatography-olfactometry
as biomarkers given their ability to discriminate different samples. (GC-O) and multidimensional gas chromatography (MDGC). In this
Moreover, we will only consider metabolites certified through robust report, several compounds were identified as potential markers and
statistical models. Hence, the main objective of this work is to assign some of them were detected as important odorants in only one variety.
high significant biomarkers to food authenticity and safety monitoring This was the case of 2,3-pentanedione and 3-ethylphenol (present only
according to their geographical and botanical origin, genetic diversity, in Negramoll), ethyl valerate and 3-isopropyl-2-methoxypyrazine (de-
production systems, adulteration and spoilage or freshness indication. tected in Tintilla) and ethyl 4-methylpentanoate (Vijariego negro). This
This practical guide will allow an easy search of metabolites, con- fact indicates that there is strong variability among wine varietals and
tributing to eventually identify new molecular routes for food authen- GC-O and MDGC is a powerful technology to detect these differences
tication and safety assessment, fostering the improvement of the (Culleré, Escudero, Pérez-Trujillo, Cacho, & Ferreira, 2008). A very

145
Table 1
Potential biomarkers for the differentiation among species, varieties and cultivars of different foodstuffs.
Biomarkers Matrix (number of Methodology Statistical analysis References
S. Medina et al.

samples)

1
Adenine, alanine, caffeic acid, fructose, inositol, linolenic acid, proline, quercetin-3-O-glucoside, succinic Grape (6) H NMR HCA; PCA; PLS-DA Ali et al. (2009)
acid
2-Methoxyphenol, 2, 3-pentadione, 3-isopropyl-2-methoxypyrazine, β-damascenone, ethyl cinnamate, ethyl Wine (5) GC-O, MDGC ANOVA Culleré et al. (2008)
4-methylpentanoate, 2-furfural, octanal
1-Hexenol, (E)-3-hexenol, (Z)-3-hexenol Wine (43) GC–MS ANOVA; Mann–Whitney test Oliveira et al. (2006)
1,1,6-Trimethyl-1,2-dihydronaphtalene, α-terpineol, β-damascenone, (E,E)-farnesal, geraniol, linalool, Wine (36) GC–MS PCA; LDA; LOOCV Câmara et al. (2007)
nerolidol, trans-furan linalool oxide, vitispirane
1
Alanine, citramalic acid, citrate, epicatechin, galactose, isoleucine, myo-inositol, quinic acid, xylose Apple (140) H NMR PCA Eisenmann et al. (2016)
1-Hexanol, 2-methylbutyl acetate, 6-methyl-5-hepten-2-ol, butyl 2-methylbutanoate, ethyl 2- Apple juice (50) GC–MS LDA; PCA; SLDA; LOOCV Guo et al. (2012)
methylbutanoate, ethyl hexanoate, hexyl acetate, hexyl butanoate, hexyl propanoate, pentyl 2-
methylbutanoate
1-Butanol, 1-hexanol, 2-methylbutyl acetate; 2-methylpropyl acetate, acetaldehyde, butyl acetate, (E)-3- Apple juice (60) APCI-MS, GC–MS LDA; PCA; SIMCA; LOOCV Gan et al. (2014)
hexenol, ethanol, hexanal, hexyl acetate, (Z)-2-hexenal
Abscisic acid, ferulic acid hexoside, linonoate A-ring lactone, naringin, neohesperidin, nomilin glucoside, Citrus juice (36) LC-ESI-q-TOF-MS HCA; PLS-DA; ANOVA Arbona et al. (2015)
obacunone, tryptophan
3-O-Trans-feruloyl tormentic acid Shrub chaste, Agnus castus LC-MS, NMR OPLS-DA Yahagi et al. (2016)
fruits (8)
C16:0, C18:0, C18:1, C20:0, C20:2, α-tocopherol, β-sitosterol, ɣ-aminobutyric acid, glucose, isoleucine, Rice (23) GC-FID, GC–MS HCA; PCA Frank et al. (2012)
methionine, myo-inositol, phosphoric acid, raffinose, sucrose
4-Guanidinobutanoate, agmatine, alanine, arginine, asparragine, ɣ-aminobutyric acid, glutamate, glycine, Rice (100) UHPLC-MS/MS, GC–MS ANOVA; PCA; ICV Hu et al. (2014)
pipecolate, serine, trans-4-hydroxyproline, tyrosine
Alanine, ɣ-aminobutyric acid, leucine, oxalic acid, valine Rice (48) GC–MS PCA Danial et al. (2013)
2-Methoxybenzoic acid, 2-methylbenzofuran, 2,6,6-trimethyl-2-cyclohexene-1,4-dione, 2′- Honey (9) UHPLC-PDA-MS/MS, PCA Beitlich et al. (2014)
hydroxyacetophenone, 2′- methoxyacetophenone, 3-hydroxy-1-(2-methoxyphenyl)-penta-1,4-dion, GC–MS

146
3,4,5-trimethylphenol, 4-methoxyphenyllactic acid, 5-methyl-3-furancarboxylic acid, acetyl-2-hydroxy-
4-(2-methoxyphenyl)-4 oxobutanal, cis-linalool, kojic acid, leptosin, lumichrome, methyl syringate, p-
anisaldehyde, p-anisic acid, phenethyl alcohol
2,3-Butanediol, cinnamaldehyde, hexanal, hexanol Date (13) GC–MS HCA; PCA; OPLS-DA Khalil et al. (2017)
Chrysoeriol rhamnosyl hexoside, fructose, glucose, isoquercetin, luteolin rhamnosyl dihexoside, rutin, Date (21) UPLC-PDA-ESI-q-TOF-MS HCA; PCA Farag et al. (2014)
sucrose
Caffeoyl shikimic acid, chrysoeriol glucoside, fructose, glucose, luteolin glycoside, quercetin conjugates Date (18) UPLC-q-TOF-MS PCA; OPLS-DA Farag et al. (2016)
13
Caffeine, caffeoylquinic acids, choline, citrate, malate, sucrose, trigonelline Coffee (12) C NMR PCA; OPLS-DA; ICV Wei et al. (2012)
Caffeoylquinic acid, dicaffeoylquinic acid, feruloylquinic acid, linoleic acid, palmitic acid, sucrose Coffee (23) ESI-FT-ICR-MS PCA; PLS-DA; LOOCV Garrett et al. (2013)
1
16-O-mehylcafestol, kahweol Coffee (101) H NMR ANOVA; PCA; ICV Monakhova et al. (2015)
Caffeine, ent-kaurane-diterpenoid derivatives (C20) Coffee (150) LC-q-TOF-MS PCA; PLS-DA Souard et al. (2018)
Caffeine, inositol, pyroglutamic acid Kopi Luwak (21) GC–MS PCA; OPLS-DA; ICV Jumhawan et al. (2013)
1-Octen-3-ol, 1-octen-3-one, 3-octanol, 3-octanone, octanal Truffle (6) GC–MS, GC-O – Culleré et al. (2013)
1,2-Dimethoxybenzene, 2-phenyl-2-buten-1-al, 5-methyl-2-phenyl-2-hexenal Truffle (25) GC–MS CA; Mann–Whitney test Splivallo et al. (2007)
5-(12-Nonadecenyl)-resorcinol, 5-heptadecylresorcinol, 5-(heneicosenyl)-resorcinol, 5-heneicosylresorcinol, Wheat (52) UHPLC-q-TOF-MS PCA; OPLS-DA; ROC curve Righetti et al. (2018)
5-nonadecanylresorcinol, 5-pentacosylresorcinol, 5-tricosylresorcinol, digalactosyl diglyceride,
monogalactosyl diglyceride, triacylglycerols
2-Methoxyhydroxytyrosol glucoside, dimethyl oleuropein, hydroxytyrosol glucoside, methoxytyrosol Olive tree (70) HPLC-ESI-MS/MS LDA; SIMCA; KNN; ICV Di Donna et al. (2010)
glucoside
2-Methyl-6-methyleneoctan-2-ol, 3- octanone, 4-octadecylmorpholine, 2,2,6-Trimetylcyclohexanone, (Z)- Tomato (5) GC–MS ANOVA; LSD; PCA; LOOCV Figueira et al. (2014)
methyl-3-hexenoate
2-Methyl-2-methylpropyl butyrate, 2-pentyl acetate, 3-methylhexyl butyrate, heptan-2-ol, isopentyl Banana (5) GC–MS PCA; SLDA; ICV Pontes et al. (2012)
hexanoate, n-butyl butyrate, (Z)-2-hexenyl butyrate
1-Octanol, 1-pentanol, 3-octanol, 3-octanone, (E)-2-octen-1-ol, (E)-2-octenal, hexanal, hexanol, linalool, ρ- Mushroom (18) GC–MS PCA Malheiro et al. (2013)
anisaldehyde
2-Heptanone, 2-methyl-4-propyl-1,3-oxathiane, α-terpinolene, benzyl pentanoate, butyl hexanoate, ethyl Passion fruits (9) GC–MS PCA; PLS-DA; LOOCV Porto-Figueira, Freitas, Cruz, Figueira,
crotonate, ethyl hexanoate, ethyl hexyl carbonate, furaneol, hexyl acetate, isopentyl acetate, methyl and Câmara, (2015)
butanoate, methyl hexanoate, methyl-(E)-2-hexenoate, methyl pentanoate, propyl hexanoate
(continued on next page)
Food Chemistry 278 (2019) 144–162
S. Medina et al. Food Chemistry 278 (2019) 144–162

spectrometry; ESI-FT-ICR-MS: electrospray ionization Fourier transform ion cyclotron resonance mass spectrometry; GC-FID: gas chromatography-flame ionization detector; GC-O: gas chromatography-olfactometry; HCA;
hierarchical cluster analysis; ICV: internal cross validation; GC–MS: gas chromatography mass spectrometry; GC-O: gas chromatography-olfactometry; KNN: K-nearest neighbours; MDGC: multidimensional gas-chro-
matography; MS/MS: tandem mass spectrometry; NMR: nuclear magnetic resonance; LC-ESI-q-TOF-MS: liquid chromatography-electrospray ionization-quadrupole-time of flight-mass spectrometry; LC-MS: liquid

analysis; PCA: principal component analysis; PDA: photodiode array; PLS-DA: partial least squares-discriminant analysis; ROC: receiver operating characteristics; SIMCA: soft independent modeling of class analogy; SLDA:
ANOVA: analysis of variance; APCI-MS: atmospheric-pressure chemical ionization-mass spectrometry; CA: cluster analysis; DA: discriminant analysis; DART-TOF-MS: direct analysis in real time- time of flight mass

chromatography mass spectrometry; LDA: linear discriminant analysis; LOOCV: leave-one-out cross validation; LSD: least square difference; OPLS-DA: orthogonal projections to latent structures modeling discriminant
recent study carried out in two white wines (Feteasca Regala and Sau-
vignon Blanc), indicates tannins as discriminant markers between these
wines. This was a little surprising because tannins are usually asso-
ciated to red wines. However, Feteasca Regala is a wine characterized by
a higher content of tannins when compared with other white wines
References

(Magdas, Guyon, Feher, & Pinzaru, 2018). Moving to fruit juices, Guo
Ferreira et al. (2009)
Avula et al. (2015)

Farag et al. (2018)

and colleagues performed a method for an efficient discrimination of


apple juice according to apple varieties used in their production in
China (Starkrimson, Qinguan, Gala, Jonagold, Golden Delicious, and Fuji).
The authors used headspace solid phase microextraction gas chroma-
tography-mass spectrometry (HS-SPME-GC–MS) and processed the
chromatographic data obtained using different chemometrics tools. In
ANOVA; PCA; OPLS; LOOCV

this work, stepwise linear discriminant analysis (SLDA) revealed sa-


tisfactory discrimination of apple juices and provided 100% success
Statistical analysis

rate in terms of prediction ability. Hence, this discriminative model is a


valuable tool to control the quality of the apple juices produced in
ANOVA; PCA

China and detect eventual adulterations (Guo, Yue, & Yuan, 2012).
Several citrus juices obtained from 12 commercial varieties (lemon
(Eureka and Fino), grapefruit (Marsh and Star Ruby), mandarin (Fortune,
PCA

Nadorcott, Pixie, and Hernandina), blonde orange (Sucrenya, Midknight)


and navel orange (Washington and Lane Late)) were also evaluated by
non-targeted metabolite profiling (Arbona et al., 2015). The results
obtained showed that several metabolites (abscisic acid (ABA), limo-
Methodology

DART-q-TOF-MS

noids, flavonoids and amino acids) were able to discriminate the citrus
varieties used in the study. Accordingly, Sucrenya variety showed a
specific accumulation of obacunone, nomilin glucoside and limonoate
GC–MS
NMR

A-ring lactone, as well as the highest ABA levels among all varieties,
suggesting a different ABA metabolite fingerprinting for each variety. In
turn, naringin and neohesperidin were exclusively present in grape-
fruits, while tryptophan was found in two grapefruits cultivars with
Matrix (number of

values four-fold higher than the average levels. In contrast, ferulic acid
samples)

hexoside was scarce in the juice of Sucrenya but it was found in sig-
Cherimoya (12)
Cinnamon (35)

Cinnamon (10)

nificantly higher levels in Fortune and Midknight varieties. These com-


pounds have gained therefore scientific and commercial interest as
important phylogenetic markets, but also due to their beneficial effects
on human health (Arbona et al., 2015).
Rice is a widely grown crop, endowed with a rich genetic diversity,
furan-2-one, α-terpeneol, benzoic acid, butyl propanoate, furaneol, linalool, methyl 2-furoate, methyl 3-

but the harvested seeds are often adulterated with others of lower
3-Methoxybutan-1-ol, 3-methylbutanoate, 3-methylbutan-1-ol, 3-methylbutyl butanoate, 5-methyl-(3H)-

stepwise linear discriminant analysis; UHPLC: ultra-high performance liquid chromatography.


Cinnamaldehyde, coumarin, dehydro-sesquiterpene, dehydro-sesquiterpene-oxide, methyl cinnamate,

Acetic acid, (E)-cinnamic acid; (Z)-cinnamic acid, eugenol, (E)-cinnamaldehyde, (E)-cinnamaldehyde

quality. So, it is very complicated to differentiate seeds of various rice


varieties based on visual observation. In this context, the metabolite
hydroxy-3-methylbutanoate, methyl-3-hydroxybutanoate, (Z)-2-penten-1-ol, (Z)-3-hexen-1-ol

profiling of rice (Oryza sativa L.) revealed an increase in the content of


fatty acid methyl ester (FAME), free fatty acids, organic acids and
amino acids and allowed the discrimination of red, black and non-co-
loured indica and japonica rice varieties (Frank, Reichardt, Shu, & Engel,
2012). These results were in agreement with another study that also
reported stigmasterol and ɣ-aminobutyrate as biomarkers to distinguish
between indica and japonica rice cultivars using a non-targeted meta-
bolomics approach (Hu et al., 2014). Moreover, in this recent work, 92
metabolites including 28 amino acids, 23 carbohydrates, 22 lipids, 12
CPGECs (cofactors, prosthetic groups and electron carriers), five nu-
Biomarkers

dimethyl acetal, glycerol, o-hydroxy-cinnamaldehyde

cleotides and two secondary metabolites exhibited statistically sig-


nificant differences between the referred cultivars. They observed that
metabolites, particularly amino acids, can be used to differentiate be-
tween different rice cultivars, with asparagine ranked the highest (Hu
et al., 2014). Using principal component analysis (PCA), it was also
possible to observe differences among rice cultivars (MR 219, MR220,
MR232 and Intani-2). So, high levels of leucine and oxalic acid appear
to be associated with a good tillering ability (Danial, et al., 2013). These
results highlighted the potential role of metabolites as early predictor
sesquiterpene oxide

biomarkers that may be very useful in crop breeding programmes and


Table 1 (continued)

can serve as a basis for the future improvement of rice quality via
metabolic engineering. Honey is a food product facing a great pressure
by the consumers. As it production shortages, honey price increases as
well as the number of adulterations of this product in the international
market. Moreover, counterfeiting honey is analytically very difficult to
discriminate. In this context, a recent work reported that 18 metabolites

147
Table 2
Metabolites as potential biomarkers for food authenticity according to production systems.
Production systems Biomarkers Matrix (number of Methodology Statistical analysis References
S. Medina et al.

samples)

Farming practices (organic vs. Glutamylphenylalanine, N-malonyltryptophan Tomato (15) HPLC-q-TOF-MS ANOVA Vallverdú-Queralt et al.
conventional) (2011)
β1-Tomatine, dehydrotomatine, tomatidine Tomato (11) HPLC-LTQ- – Caprioli et al. (2014)
Orbitrap-MS
2-Furanone, α-ketobutyric acid, alanine, ascorbic acid, catechol, citramalic acid, citric Tomato, pepper (40, DART-TOF-MS PCA; LDA; LOOCV Novotná et al. (2012)
acid, dehydroascorbic acid, ferulic acid, fructosyl glutamate, furfural, glucoheptonic 24)
acid, gluconolactone, glutaric acid, glycerate, hydroxymethylfural, itaconic acid,
lactate, lactic acid, leucine, malic acid, proline, phenylalanine, piruvic acid,
pyroglutamic acid, quinic acid, succinic acid, threonine, valine;
Alanine, β-alanine, glycerate, hydroxyglutarate, myo-inositol, panthotenic acid, urea, Wheat (20) GC–MS Tukey test Zörb et al. (2006)
valine
4-Aminobutanoate, alanine, adenosine, arabinose, asparagine, aspartate, Wheat (22) GC–MS Studentś t-test; PCA Bonte et al. (2014)
ethanolamine, malate, myo-inositol, tryptophan, urea
Protocatechuic acid Wheat (36) LC-HRMS ANOVA; PCA Weesepoel, et al. (2016)
2-Piperidinecarboxylic acid, 2,3-dihydroxypropanoic acid, α-solanine, allantoin, Potato (48) GC–MS ANOVA; PCA Shepherd et al. (2014)
arginine; asparagine, aspartic acid, β-alanine, ɣ-aminobutyric acid, fumaric acid, LC-MS
galactinol, galactose, glucose, glutamine, glutathione, glutamic acid, glycerol,
glycine, histidine; inositol; isoleucine, leucine, L-alanine; lysine; methionine, oxalic
acid, oxoproline, phenylalanine; proline, quinic acid, serine; solasonine, sucrose,
thiamine, threonine, trihydroxypentanoic acid, tryptophan, tyrosine, valine
Phosphate, malic acid, myo-inositol Maize (54) GC–MS ANOVA; PCA Röhlig and Engel (2010)
Arabinofuranosyl-[a-L-arabinofuranosyl]-L-arabinose, chlorogenic acid, citric acid, L- Carrot (140) UHPLC-MS PCA; OPLS-DA; ROC curve; ICV; Cubero-Leon, et al. (2018)
arginine, sedoheptulose ECV
1-Acetyl-3,14,20-trihydroxywitha-5,24-dienolide 3-glucoside, butyl (S)-3- Goldenberry (30) UHPLC-q-TOF-MS Studentś t-test; PCA Llano et al. (2018)

148
hydroxybutyrate [arabinosyl-(1-greater than6)-glucoside], physagulin D
2-Oxoglutarate semialdehyde, 4-oxoglutarate semialdehyde, β-myrcene, Orange juices (19) UPLC-HR-MS; PCA; PLS-DA; HCA; LOOCV; PT; RT Cuevas et al. (2017)
dehydroascorbic acid, ethyl butanoate, ɣ-terpinene, hexanal, limonene, (Z)-3-hexanal GC–MS
(+)-Catechin, chlorogenic acid, (−)-epicatechin, gallic acid, myricetin, p-coumaric Grape juices (96) HPLC-DAD Mann-Whitney test; ANOVA; PCA; Granato et al. (2015)
acid quercetin HPLC-FLD PLS-DA; SIMCA; KNN; LOOCV;
ECV
Delphinidin 3-arabinoside, delphinidin 3-galactoside, delphinidin 3-glucoside, Blueberry (10) HPLC-DAD ANOVA Wang et al. (2008)
malvidin 3-arabinoside, myricetin 3-arabinoside, petunidin 3-galactoside, petunidin
3-glucoside, quercetin 3-glucoside
Caffeine, chlorogenic acid, trigonelline Coffee HPLC-DAD Newman-Keuls test; ANOVA do Carmo Carvalho et al.
(2011)
β-apo-carotenal, β-carotene, β-cryptoxanthin, canthaxanthin, luteoin/zeaxanthin Egg (84) HPLC-DAD ANOVA; PCA; LSD van Ruth et al. (2011)
α-carotene, β-carotene, chlorogenic acid, cis-β-carotene, gallic acid, kaempferol, Pepper (9) HPLC-DAD ANOVA Hallmann and
quercetin, quercetin-D-glucoside Rembiałkowska (2012)
1-Pentenol, 1-penten-3-ol, 2-heptyl butanoate, 2-hexanol, 3-hydroxy-2-butanone, 3- Banana (120) GC–MS t-test Dong et al. (2014)
methoxy-1-propanol acetate, 4-methyl-2-pentyl acetate, decanal, eugenol,
hexadecane, isobutyl isovalerate, isopropyl myristate, (Z)-3-hexenyl acetate
Phytanic acid, pristanic acid Cheese, milk, cream, GC–MS ANOVA Vetter and Schröder (2010)
butter (32)

Rearing practices (wild vs. Cholesterol, choline, C22:6 n-3, C20:5 n-3, fumaric acid, glutamine, malic acid, Sea bass (7) NMR ANOVA; PCA Mannina et al. (2008)
cultured) phosphatidylethanolamine, trimethylamine oxide
C14:0, C15:0, C16:0, C17:0, C18:0, C16:1 n-7, C18:1 n-9, C20:1 n-9, C18:2 n-6, C20:4 Sea bass (27) HPLC; GC–MS ANOVA; LSD Fuentes et al. (2010)
n-6, C20:5 n-3, C22:6 n-3, arginine, alanine, asparagine, glutamic acid, glycine,
histidine, methionine, phenylalanine, serine, taurine, tyrosine

(continued on next page)


Food Chemistry 278 (2019) 144–162
S. Medina et al. Food Chemistry 278 (2019) 144–162

liquid chromatography-fluorescence detector; HPLC-q-TOF-MS: high performance liquid chromatography-quadrupole- time of flight mass spectrometry; ICV: internal cross validation; LC-HRMS: liquid chromatography
high resolution mass spectrometry; LC-MS: liquid chromatography mass spectrometry LDA: linear discriminant analysis; LOOCV: leave-one-out cross validation; LSD: least square difference; LTQ: linear trap quadrupole;

permutation test; ROC: receiver operating characteristic; RT: randomization test; SLDA: stepwise linear discriminant analysis; SWDA: stepwise discriminant analysis; UHPLC-MS: ultra-high performance liquid chro-
ANOVA: analysis of variance; CANDISC: canonical discriminant analysis; CLA: conjugated linoleic acids; DA: discriminant analysis; DART-TOF-MS: direct analysis in real time- time of flight mass spectrometry; ECV:
external cross-validationGC-MS: gas chromatography mass spectrometry; HCA; hierarchical cluster analysis; HPLC-DAD: high performance liquid chromatography -diode array detector; HPLC-FLD: high performance

NMR: nuclear magnetic resonance; OPLS-DA: orthogonal projections to latent structures modeling discriminant analysis; PCA: principal component analysis: PLS-DA: partial least squares-discriminant analysis; PT:
(8 volatiles and 10 non-volatiles compounds) allowed a clear differ-
entiation among three kinds of honey derived from monocultures
(Manuka (8), Hanuka (7), and Jelly bush (1)) harvested in 2012.

Marseglia et al. (2013)


Caligiani et al. (2016)
Povolo et al. (2009)
Segato et al. (2017)
References

Nevertheless, in order to achieve a definitive classification of these

Vasta et al. (2011)


kinds of honey, further research is still necessary, particularly with a
larger number of samples, as well as with samples from different years
(Beitlich, Koelling-Speer, Oelschlaegel, & Speer, 2014). Date fruits have
been also subjected to an interesting discriminant analysis. Initially,
Farag, Mohsen, Heinke, and Wessjohann, (2014) reported differences
among 21 date varieties from Egypt, being flavonols glycosides (rutin
and isoquercetin), flavones glycoside (luteolin rhamnosyl dihexoside
ANOVA; SWDA; CANDISC; DA

and chrysoeriol rhamnosyl hexoside) and sugars (fructose, glucose, and


Kolmogorov-Smirnov’s test
Statistical analysis

sucrose) the principal contributors for the discrimination of the vari-


eties studied. Then, the same authors analysed 18 date cultivars from
ANOVA; SWDA; ICV

PCA; CA; LOOCV

Saudi Arabia using a metabolomics approach. The results obtained


using PCA and orthogonal partial least square (OPLS) analyses revealed
that the ratios flavones versus flavonol and sugars levels, were the main
ANOVA

contributors for the cultivar differentiation (Farag, Handoussa, Fekry, &


Wessjohann, 2016). Finally, using SPME-GC–MS, Khalil, Fekry, and
Farag, (2017) analysed the volatile fraction of date fruits and identified
a total of 89 VOCs in 13 Egyptian date varieties. Among these, lipid-
GC–MS; 1H NMR
Methodology

derived volatiles and phenylpropanoid derivatives were the major


HPLC; GC–MS

components of date fruit aroma and volatile markers as 2,3-butanediol,


GC–MS
GC–MS

cinnamaldehyde, hexanal, and hexanol, are among the ones with the
higher contribution for the discrimination of the date fruits varieties
GC

analysed. Coffee is globally one of the most widely used food products
and in the past two decades, there has been continuous growth in the
Matrix (number of

Asiago PDO Cheese

Milk, Cheese (220)

global consumption. Consequently, a huge number of articles have been


Milk, Cheese (43)
samples)

published concerning the coffee adulteration and some of them have


Cheese (304)

focused on identifying biomarkers for different species or varieties. In


Meat (100)

this context, Monakhova et al. (2015) reported that kahweol and 16-O-
(81)

methylcafestol may be used to authenticate arabica and robusta coffees,


respectively. Furthermore, arabica coffee was also reported to contain
significantly higher levels of sucrose, citrate, trigonelline, and malate,
whereas robusta coffee was characterized by higher levels of caffeine,
choline and caffeoylquinic acids (CQA) (Table 1) (Wei, et al., 2012). In
2-methyl-1-Butanol, 3-undecanone, cuminic alcohol, germacrene D, skatole
Anteiso-C15:0, C9:0, C18:3 n-6, C20:3 n-6, C20:4 n-6, cis-9, trans-11 CLA

another report, phenolic compounds, fatty acids, and sucrose were


cyclopropane fatty acids, Σ trans/trans CLA, trans-11, cis-13 CLA, retinol

described as cultivars biomarkers for eight arabica coffee cultivars,


what constitutes a valuable tool for the certification processes and
traceability of coffee beans (Garrett, et al., 2013). Very recently, coffees
leaves from nine species grown in the same environmental conditions
were subjected to discriminant analysis and caffeine and ent-kaurane
diterpenoid (C20) derivatives were found to be key metabolites in the
Biomarkers

discrimination of the selected coffee species. Caffeine concentration


Cyclohexyl tridecanoic acid, lactobacillic acid

was found approximately 800 times higher in C. arabica leaves com-


pared to C. canephora, what makes it a good marker to authenticate
Dihydrosterculic acid, lactobacillic acid
1-Phytene, 2-phytene, neophytadiene

coffee species and ent-kaurane diterpenoid (C20) derivative was found


in three different species (C. arabica, C. canephora and C anthonyi),
exhibiting similar metabolic profiles in the analytical conditions de-
scribed (Souard et al., 2018). Finally, it is worthwhile to refer the pe-
culiar arabica kopi luwak, an exotic Indonesian coffee and eventually the
world́s most expensive. The coffee berries from kopi luwak are eaten by
Asian palm civet, being the coffee pericarp completely digested and the
beans excreted. Then, they are collected, cleaned, wet-fermented, sun-
dried and roasted. A recent study allowed the discrimination between
robusta and arabica kopi luwak, being caffeine, inositol and pyr-
oglutamic acid the discriminant markers. Moreover, this was the first
matography mass spectrometry.
Feed practices (pasture vs. silage)

work contributing for the authentication of this expensive coffee and in


fact citric acid, malic acid, and the ratio inositol/pyroglutamic acids
Production systems

were able to differentiate among authentic and fake kopi luwak samples
Table 2 (continued)

(Jumhawan et al., 2013). Another very expensive matrix highly sus-


ceptible to frauds are the truffles, which can be morphologically very
similar, although having very different aromas. The characterization of
the volatile composition of different truffle species constitutes there-
fore, a very relevant opportunity to define biomarkers to detect frauds.
This has been reported by Splivallo, Bossi, Maffei, and Bonfante,

149
Table 3
Potential biomarkers for the food authentication according to geographical and botanical origin.
Biomarkers Matrix (number of samples) Methodology Statistical analysis References
S. Medina et al.

Lepteridine Manuka honey (27) (New Zealand) HPLC Studentś t test Lin et al. (2017)
Synephrine Honeys (34) Monofloral citrus, ‘assa-peixe’, eucalyptus LC-MS/MS ANOVA Tette et al. (2017)
and ‘aroeira’ honey. Wildflower (polyfloral) honeys (São
Paulo, Minas Gerais)
Heptanal, hexanal, octanal, p-cymene Honeys (374) (Corsican, non-Corsican-France, Italy, GC–MS DPLS; LDA; SIMCA; SVM; RMSCV; Stanimirova et al. (2010)
Austria, Ireland, and Germany) LOOCV
1,5,8-p-Menthatriene, 1,8-nonadiyne, 2-furanmethanol, 2-hydroxy-3,5,5-trimethyl- Honeys (55) (Coffee, longan, lychee, sunflower and wild GC–MS ANOVA Pattamayutanon et al. (2017)
2-cyclohexen-1,4-dione, 2,2-dimethyl butanal, 2,6,6-trimethyl-2-cyclohexen-1- honeys of five provinces of Thailand)
ol, 3,4,5-trimethyl-phenol, 5-hydroxymethylfurfural, 10-oxodecanoate,
anisaldehyde, anise alcohol, benzaldehyde, butyryl lactone, diethyl
decanedioate, diethyl hexanedioate, ethyl heptanoate; hexadecane, lilac
alcohol B, lilac alcohol C, lilac alcohol D, methyl caproate, methyl oleate,
pentadecane, phenylmethanol, terpineol, (Z)-5,6-dimethyl-1,3-cyclohexadiene
4-Terpineol, α-terpineol, hotrienol, lilac aldehydes, linalool Honeys (25) (Acacia, sunflower, linden, rapeseed, GC–MS – Spanik et al. (2014)
chestnut and orange honeys from EU countries)
Kynurenic acid Honeys (32) (Botanical origins (multiflora, acacia, SEC-UV-DAD HCA; PCA Beretta et al. (2012)
chestnut, dog rose, mango and honeydew. Geographical
origins (Italy, Kosovo, Cameroon, Kenya and Brazil))
Perseitol Honeys (140) (Avocado nectar) NIRS PCR; PLS; SECV Dvash et al. (2002)
5-Hydroxymethylfurfural, citric acid, ethanol, lactic acid, phenylalanine, sucrose, Honeys (46) (Brazilian wildtype-, citrus-, eucalyptus, assa- NMR HCA; KNN; PCA; PLS-DA; SIMCA; Boffo et al. (2012)
tyrosine peixe, sugar-cane) SECV
4-(1-Hydroxy-1-methylethyl)benzoic acid, 4-(1-Hydroxy-1-methylethyl)- Honeys (305) (Linden, chestnut, orange and eucalyptus NMR 02PLS-DA Schievano et al. (2013)
cyclohexa-1,3-dienecarboxylic acid, 4-(1-Hydroxy-1-methylethyl)cyclohexa- from Italy)
1,3-dienecarboxylic acid methyl ester, 4-(1-methylethenyl)-cyclohexa-1,3-
dienecarboxylic acid, 2-quinolone, 3-oxo-α-ionone, 4-quinolone, 8-

150
hydroxylinalool, caffeine, deoxyvasicinone, dehydrovomifoliol
Acetyl-dicaffeoylglycerol, acetyl-feruloyl-caffeoylglycerol acetyl-p-coumaroyl- Propolis (13) (Germany) TLC, TLC-MS – Betrams et al. (2013)
caffeoylglycerol, acetyl-p-coumaroyl-feruloylglycerol, acetyl-di-p-
coumaroylglycerol, apigenin, caffeic acid, caffeic acid phenethyl ester, chrysin,
galangin, kaempferol, naringenin, pinocembrin, quercetin
Kaempferol-dimethylether Propolis (40) (Portugal) LC-DAD-ESI-MS(n) – Falcao et al. (2013)
4-(4′-Hydroxyphenyl)-2-butanone Arabica coffees (27) (Ethiopia, Tanzania, Guatemala) GC–MS, GC-O DA; PCA Akiyama et al. (2008)
2-Methylpyrazine, 2,3-diethylpyrazine, 2,3-dimethylpyrazine, 5-methylfurfural Arabica coffees (47) (Brazil, Colombia, Costa Rica, GC-TOF-MS PCA Risticevic, Carasek, and
Guatemala, Ethiopia, Indonesia) Pawliszyn, (2008)
(−)-Epicatechin, caffeine cinnamtannin A2, cyanidin-3-O-arabinoside, cyanidin-3- Cocoa (31) (31 different geographical origins) HPTLC – Pedan et al. (2017)
O-galactoside, proanthocyanidin B2 and C1, theobromine
1-Hexanol, (E)-2-hexenal, (E)-2-hexen-1-ol, hexanal, (Z)-3-hexenyl acetate Olive oils (36) (Southern Italian regions) GC–MS PCA Benincasa et al. (2003)
Olive oils (37) (Calabrian and Tunisian areas) GC–MS ANOVA; LDA; Wald–Wolfowitz Cavaliere et al. (2007)
test
3-Methylbutanal, ethanol tetrahydro-2-(3-pentynyloxy)-2H-pyran, hexanal Tomatoes (80) (Italian regions) GC–MS LDA; PCA; SIMCA kCV Feudo et al. (2011)
(E)-2-hexenal, (E,E)-2,4-heptadienal, ethyl hexanoate, hexyl 2-methylbutyrate, p- Apples (9) (Madeira Islands regions) GC–MS LDA; PCA; LOOCV Ferreira et al. (2009)
ethyl styrene
1-Butanol, 1-hexanol, (E)-3-hexenol, hexanal, (Z)-2-hexenal Apple juices (210) (New Zealand, South Africa, Chile) APCI-MS, GC–MS ANOVA, PCA; PLS-DA; LOOCV; Gan et al. (2014)
ECV
Citrusin D Oranges (65) (Spain, Brazil, Argentina, South Africa) UHPLC-q-TOF MS PLS; OPLS-DA; ECV Diaz et al. (2014)
4-Aminobutyrate, O-phosphocholine, acetate, asparagine, isoleucine, leucine, Cabbages (322) (Korea, China) NMR PCA; ICV Kim et al. (2013)
phenylacetate, phenylalanine, succinate, sucrose, tyrosine, valine
Diacylglycerols, γ-tocopherol, phosphatidylcholines, phosphatidylethanolamines, Hazelnuts (196) (Germany, France, Italy, Turkey, UPLC-q-TOF-MS ANOVA; Studentś t-test; PCA; Klockmann et al. (2016)
triacylglycerols Georgia) LDA; PLS-DA; SIMCA; SVM; ICV
1-Hexanol, 1-pentanol, 1-penten-3-ol, hexanal, pentanal, trimethylbenzenes Walnuts (China, Ukraine, Chile) GC–MS ANOVA Elmore, Nisyrios, and Mottram,
(2005)
2-Furaldehyde, 5-(hydroxymethyl)-2-furaldehyde, vanillin, syringaldehyde, Sherry vinegars (Aging in acacia or cherry wood barrels) GC–MS LDA; BPANN García-Parrilla et al. (2011)
coniferaldehyde
(continued on next page)
Food Chemistry 278 (2019) 144–162
Table 3 (continued)

Biomarkers Matrix (number of samples) Methodology Statistical analysis References


S. Medina et al.

(+)-Catechin; (−)-epicatechin, cis-piceid, p-hydroxybenzoic acid, caftaric acid, Wines (28) (Czech Republic regions) HPLC ANOVA Lampir and Pavlousek (2013)
protocatechuic acid
(+)-Catechin, (−)-epicatechin, (−)-epicatechingallate, (−)-epigallocatechin, p- Wines (90) (China regions) HPLC ANOVA; CDA; ICV Sun et al. (2015)
coumaric acid, p-hydroxy benzoic acid, caffeic acid, chlorogenic acid, ferulic
acid, gallic acid, gentisic acid, epigallocatechin gallate, protocatechuic acid,
sinapic acid, syringic acid, vanillic acid
Malvidin-3-O-glucoside, malvidin-3-O-glucoside-4-vinylguaiacol, malvidin-3-(6-O- Wines (27) (Argentina regions) LC-MS PLS; LOOCV Pisano et al. (2015)
acetylglucoside)
1-Octen-3-ol, 2-ethylhexan-1-ol, 3-ethylbutanoic acid, 3-ethoxypropan-1-ol, 3,5,5- Wines (34) (Azores, Canary, Madeira Islands) GC–MS PLS-DA Perestrelo et al. (2014)
trimethylhexan-1-ol, 4-methylphenol, 4-(methylthio)-1-butanol, decanoic acid,
(E)-whiskey lactone, ethyl 2-methylbutanoate, ethyl DL-2-hydroxycaproate,
ethyl 3-hydroxybutanoate, γ-octalactone, isoamyl lactate, (Z)-3-hexenyl
butanoate
(E)-β-damascenone, α-bisabolene, α-terpineol, α-ylangene, β-bourbonene, β- Grapes (22) (Madeira Island regions) GC–MS PCA; HCA Perestrelo, Barros, Rocha, and
ionene, geraniol, hotrienol, linalool Câmara, (2014)
1
Citrate, malate, proline, threonine, trigonelline Grapes (20) (South Korea regions) H NMR PCA; PLS-DA Son et al. (2009)
1
Acetate, alanine, aspartate, choline, choline phosphate, citrate, citrulline, Lentils (85) (Italy, Canada) H NMR PCA-LDA; PLS-DA; SIMCA; MCCV Longobardi et al. (2017)
isoleucine, galactose, glucose, malate
2-Hydroxy-4-isopropylnaphthalene, 2-heptenal, 2-methylbutanal, 2- Truffles (Italy regions) PTR-TOF-MS PCA; HCPC Vita et al. (2015)
methylbutanoate, 2-methyl-1-propanol, 2-pentylfuran, 2,3-octanedione, 2,4-
dimethylfuran, acetaldehyde, acrylic acid, carveol, cyclopentyl-1-thiaethane,
diethanol sulfide, ethenone (akyl fragment), ethyl;, pyridine
2-Carene, 3-carene, α-phellandrene, α-pinene, β-pinene, γ-terpinene, camphene, Cheeses (41) (Five Slovenian PDO) GC–MS CA Tompa et al. (2013)
limonene, tricyclene
β-Caryophyllene Cheeses (35) (PDO Asiago Mountain) GC–MS — Favaro, Magno, Boaretto, Bailoni,
and Mantovani, (2005)

151
Pachymic acid Mushroom (60) (Wolfiporia extensa from five China UFLC, UV ANOVA; PLS-DA; HCA Li et al. (2016)
regions) spectroscopy
Glycerophospholipids and their oxidized lipids Saffron (44) (Spanish PDO La Mancha-Aragón) UHPLC-q-TOF-MS/ PCA; OPLS-DA; ICV Rubert et al. (2016)
MS

ANOVA: analysis of variance; APCI-MS: atmospheric-pressure chemical ionization-mass spectrometry; BPANN: back propagation artificial neural network; CA: cluster analysis; CDA: canonical discriminant analysis; CE:
capillary electrophoresis; DA: discriminant analysis; DPLS: dispersive partial least squares; ECV: external cross validation; GC–MS: gas chromatography mass spectrometry; GC-O: gas chromatography-olfactometry; HCA;
hierarchical cluster analysis; HCPC: hierarchical clustering on principal components; HPLC: high performance liquid chromatography; HPTLC: high performance thin layer chromatography; ICV: internal cross validation;
kCV: k-fold cross validation; KNN: K-nearest neighbour; LC-DAD-ESI-Ms(n): liquid chromatography- diode array detector-electrospray ionization- tandem mass spectrometry; LC-MS/MS: liquid chromatography tandem
mass spectrometry; LDA: linear discriminant analysis; LOOCV: leave-one-out cross validation; MCCV: monte carlo cross validation; NIRS: near-infrared spectroscopy; NMR: nuclear magnetic resonance; OPLS-DA:
orthogonal projections to latent structures modeling discriminant analysis; PCA: principal component analysis; PCR: principal component regressions; PLS-DA: partial least squares-discriminant analysis; PDO: protected
designation of origin PTR-TOF-MS: proton transfer reaction time of flight mass spectrometry; RMSCV: root mean square error of cross validation; SEC-UV-DAD: size exclusion chromatography ultraviolet diode array
detection; SECV: standard error of cross validation; SIMCA: soft independent modeling of class analogy; SVM: support vector machines; TLC: thin layer chromatography; UFLC: ultra-fast liquid chromatography; UHPLC-q-
TOF-MS: ultra-high performance liquid chromatography-quadrupole-time of flight-mass spectrometry; UV: ultraviolet.
Food Chemistry 278 (2019) 144–162
S. Medina et al.

Table 4
Biomarkers candidates to detect food adulteration.
Biomarkers Matrix (number of samples) Methodology Statistical analysis References

Filbertone, lupeol Olive oil (Adulteration with hazelnut oil) GC–MS – Azadmard-Damirchi (2010)
Lactose semicarbazone, semicarbazide Milk (Adulteration with nitrogen-containing LC-MS – Abernethy and Higgs (2013)
compounds)
Linoleic acid, oleic acid, saturated fatty acids (total concentration), triacylglycerol, cholesterol Milk (Adulteration with vegetable oils and GC–MS t-test Kim et al. (2015)
contents animal fats)
Hexanal, heptanal, methyl-butanal, nonanal, pentanal Maté tea (6) (Adulteration of Ilex GC–MS – Dallago, Valduga, Luccio, Benin,
paraguariensis with other Ilex species) and Tres, (2011)
3-Oxohexadecanoic acid glycerides, arabitol, CE(22:5), Cer(d18:1/24:1), citric acid, creatinine, Meat (85) (Adulteration of beef mince with GC–MS; UHPLC-MS ANOVA; PCA; PLS-DA Trivedi et al. (2016)
decanoylcholine, ethanolamide, glucose 6-phosphate, glycine, glycyl-lysine, heptadecane, pork mince)
hexano-dibutyrin, malic acid, myo-inositol, N-carboxyethyl-gaminobutyric acid, oleic acid,
pentadecane, PG(36:4), phosphate, prostaglandin D2, pyroglutamic acid, TG (16:0/15:0/18:4),
xi-2-ethyl-1-hexanol
Caffeoylphenylalanine, p-coumaroyltyrosine Coffee (Adulteration arabica with robusta) QTOF-MS; FT-ICR PCA; PLS-DA; LOOCV Garrett et al. (2013)

152
2,5-Dimethylfuran, 4-ethyl-2-methoxyphenol, indole, maltol, phenol Coffee (Roast defects) GC–MS ANOVA; PCA Yang et al. (2016)
Catechol Propolis (22) (Adulteration with poplar HPLC-MS; 1H NMR; 13C – Huang et al. (2014)
extract) NMR
2-acetylfuran-3-glucopyranoside Honey (186) (Adulteration by rice syrup) HPLC-DAD; LC-q-TOF-MS; – Xue et al. (2013)
NMR
1
Dimethylproline Orange juice (313) (Adulteration with pulp H NMR ANOVA; LDA; PLS; Le Gall et al. (2001)
wash) SLDA; CVA
Arbutin Apple juice (132) (Adulteration with pear HPLC-PDA – Thavarajah and Low (2006)
juice)
4-O-p-Coumarylquinic acid, abscisic acid, arbutin, isorhamnetin-3-O-rutinoside Apple juice (58) (Adulteration with pear HPLC-PDA; LC-MS/MS; – Willems and Low (2018)
juice) NMR

ANOVA: analysis of variance; BPANN: back propagation artificial neural network; CVA: canonical variates analysis; FT-ICR: Fourier transform ion cyclotron resonance; GC–MS: gas chromatography mass spectrometry;
HPLC-DAD: high performance liquid chromatography-diode array detector; HPLC-PDA: high performance liquid chromatography-photodiode array; LC-MS/MS: liquid chromatography-tandem mass spectrometry; LC-q-
TOF-MS: liquid chromatography quadrupole time of flight-mass spectrometry; LDA: linear discriminant analysis; LOOCV: leave-one-out cross validation; NMR: nuclear magnetic resonance; PCA: principal component
analysis; PLS-DA: partial least squares-discriminant analysis; SLDA: stepwise linear discriminant analysis; UHPLC-MS: ultra-high performance liquid chromatography mass spectrometry.
Food Chemistry 278 (2019) 144–162
S. Medina et al.

Table 5
Metabolites as potential biomarker compounds for the detection of food spoilage and freshness.
Biomarkers Matrix (number of Methodology Statistical analysis References
samples)

Spoilage 2-Ethyl-1-hexanol, 2-methyl-1-butanol, 3-methylbutanal, 3-methyl-1-butanol, ethanol, ethyl acetate, Sea bass (8) GC–MS PCA, Tukeýs test Parlapani et al. (2015)
ethyl isobutyrate, ethyl propionate
Dimethyl sulfide, trimethylamine Oyster GC–MS Bioelectronic – Lee et al. (2018)
nose
3-Methyl butanoic acid, acetic acid, acetoin, ethyl benzene, propyl benzene, styrene Salmon (12) GC–MS ANOVA, REML Wierda et al. (2006)
2,3-Butanediol, 3-methyl-1-butanol, acetic acid, dimethyl sulfide, ethyl acetate, isobutyl alcohol, Brown shrimp Atlantic SIFT-MS HCA, PCA, PLS; LOOCV Kuuliala, Abatih, et al.
trimethylamine cod (18) (2018)
2,3-Butanediol, 3-methyl-1-butanol, ethanol, ethyl acetate, trimethylamine Atlantic cod (31) SIFT-MS Friedmańs test Kuuliala, Al Hage, et al.
(2018)
2,4,6-Trimethypyridine, allyl methyl sulfide, dimethyl disulfide, dimethyl trisulfide, methyl Chicken (12) GC–MS – Lovestead and Bruno
thiolacetate, phenyl sulfide (2010)
2-Butanol, 3-methyl-1-butanol, acetic acid, ethanol, dimethyl disulfide, dimethyl trisulfide, proponic Chicken (168) GC–MS/FASST ANOVA, CA, HCA, PCA Miks-Krajnik et al. (2016)
acid, sulfides methanethiol
Acetone, alanine, arginine, asparagine, carbon disulfide, dimethyl sulfide, ethanol, ethyl acetate, Chicken (128) HPLC; GC–MS – Alexandrakis et al. (2012)
glucose, glutamine, glycine, hexanal, heptane, isoleucine, L-lactate, lysine, methionine, methyl

153
benzoate, methyl ethyl ketone, phenylalanine, serine, threonine, toluene, tyrosine
Acetoin, acetone, diacetyl, dimethyl sulfide ethanol Chicken (21) PTR-MS – Franke and Beauchamp
(2017)
3-Methylindole, acetoin, adenosin diphosphate, aspartic acid, betaine, butanoic acid, creatine, Beef (28) GC–MS HCA, PLS, Pearson correlation Ercolini et al. (2011)
1
glycogen, histidine, inosine monophosphate, lactate, leucine, proline; threonine H NMR
2,3,5,-Trimethyl pyrazine, 2,6-dimethyl pyrazine, cyclohexanone, ethanol 2,2 (ethoxy-ethoxy), Iberian ham (30) GC–MS ANOVA, Pearson correlation Martín et al. (2010)
tetramethyl pyrazine, trimethyl pyrazine
Ergosterol Grape (10) HPLC Pearson correlation Porep et al. (2014)
5,10-Diethoxy-2,3,7,8-tetrahydro-1H,6Hdipyrrolo[1,2-a:1′,2′-d]pyrazine, pyruvic acid, tryptophan, Bread (25) GC x GC-TOF-MS PCA, PLS-DA, Studentś test; Cheng et al. (2015)
2,3-butanediol, leucinic acid, malonic acid, Egg, (25) Jonckheere-Terpstra test; LOOCV
Nitrosotrimethylurea, putrescine, Cucumber (25)
Ethanolamine, fumaric acid, homoproline; pyroglutamic acid, threonic acid

Freshness 1-Penten-3-ol, 1-octen-3-ol, cyclopentanol, (E)-2-hexenal, (E)-2-pentenal, hexenal, octanal, (Z)-2- Salmon (12) GC–MS ANOVA, REML Wierda et al. (2006)
penten-3-ol
Carotenoids Citrus fruits (18) Raman spectroscopy – Nekvapil et al. (2018)

ANOVA: analysis of variance; CA: cluster analysis; FASST: fast automated scan/sim type; HCA; hierarchical cluster analysis; HPLC: high performance liquid chromatography; GC x GC-TOF-MS: two dimensional gas
chromatography time of flight mass spectrometry; GC–MS: gas chromatography mass spectrometry; LOOCV: leave-one-out cross validation; PCA: principal component analysis; PLS-DA: partial least squares-discriminant
analysis; PTR-MS: proton-transfer-reaction mass spectrometry; REML: residual maximum likelihood; SIFT-MS: selected-ion-flow-tube mass spectrometry.
Food Chemistry 278 (2019) 144–162
S. Medina et al. Food Chemistry 278 (2019) 144–162

Fig. 1. Potential candidates for biomarkers in food authenticity and safety.

(2007), that used headspace stir bar sorptive extraction-gas chromato- Cassanese, Coratina, Nocellara and Leccino) and reported that four
graphy-mass spectrometry (HS-SBSE-GC–MS) to characterize Tuber phenolic compounds, 2-methoxyhydroxytyrosol glucoside, dimethyl
melanosporum (black truffle) and Tuber indicum (Chinese truffle) and oleuropein, hydroxytyrosol glucoside and methoxytyrosol glucoside,
identified 1,2-dimethoxybenzene, 2-phenyl-2-buten-1-al and 5-methyl- were discriminant biomarkers of the selected cultivars. A similar work
2-phenyl-2-hexenal as potential biomarkers to differentiate the selected using HS-SPME-GC–MS was carried out with different tomato cultivars
truffles. In another study involving GC-O and HS-SPME-GC–MS tech- (Lycopersicum esculentum L., plum, campari, grape, cherry and regional)
nologies, the VOCs 1-octen-3-ol, 1-octen-3-one, 3-octanol, 3-octanone and identified five VOCs (2-methyl-6-methyleneoctan-2-ol; 3-octanone;
and octanal were also reported to distinguish between these species of 4-octadecylmorpholine; 2,2,6-trimetylcyclohexanone and (Z)-methyl-3-
truffles (Culleré, Ferreira, Venturini, Marco, & Blanco, 2013). A similar hexenoate) as discriminant markers (Figueira, Camara, Pereira, &
application involved the characterization of the volatile fraction of six Camara, 2014). The same strategy was applied to five banana cultivars
wild mushroom species (Clitocybe odora, Clitocybe fragrans, Hebeloma of the Musaceae family (dwarf cavendish, prata, maçã, ouro, and pla-
crustuliniforme, Lepista nuda, Tricholoma fracticum and Tricholoma ter- tano). In this case, the esters group was the largest chemical class
reum). Using HS-SPME-GC–MS followed by unsupervised analysis, as identified in all cultivars and 2-methyl-2-methylpropyl butyrate, 2-
PCA, 1-octanol, 1-pentanol, 3-octanol, 3-octanone, (E)-2-octen-1-ol, pentyl acetate; 3-methylhexyl butyrate, heptan-2-ol, isopentyl hex-
(E)-2-octenal, hexanal, hexanol, linalool, and ρ-anisaldehyde were de- anoate, n-butyl butyrate and (Z)-2-hexenyl butyrate were identified as
scribed as markers of the six mushrooms species analysed. This study metabolites that clearly classify the banana samples by cultivars
corroborated the importance of using the volatile metabolites as a (Pontes, Pereira, & Câmara, 2012). 1H NMR also constitutes a powerful
powerful taxonomic tool (Malheiro et al., 2013). metabolomics tool, as the work from Eisenmann, et al. (2016) showed.
On the other hand, a recent study described an untargeted lipi- Using an untargeted metabolomics approach, the authors applied
domics strategy to discriminate common (Blasco, Triticum aestivum) and multivariate analysis to the peel and pulp extracts of 14 apple cultivars
durum wheat (Odisseo, Triticum turgidum spp. durum) varieties. In this and observed a remarkable variation in the content of sugars, poly-
work, alkylresorcinols, in particular 5-heptadecylresorcinol (AR 17:0), phenols and acids among the selected cultivars. In turn, this allowed a
triacylglycerols (56:2, 56:3, 56:4 and 56:5), monogalactosyl diglyceride clear discrimination between the 14 apple cultivars analysed (Table 1).
(36:3, 36:5 and 38:6) and digalactosyl diglyceride (34:2, 36:3 and 36:4) Cinnamon is a common spice with multiple uses in cosmetic, food, and
were reported as potential biomarkers to differentiate common and pharmaceutical industries. It is nevertheless, very susceptible to adul-
durum wheat varieties. This result demonstrated that untargeted lipi- teration, with many species of plants being labelled as cinnamon;
domics is a powerful tool for the detection of wheat frauds and it might however, the “true cinnamon” refers only to the dried inner bark of
be applied in other lipid rich matrices (Righetti, et al., 2018). Olive oil Cinnamon verum J.S. Presl. Regarding this fraud, a study carried out a
is very susceptible to adulteration, particularly with oils from lower robust method to authenticate the “true cinnamon” and biomarkers as
quality and species. Some olive oil cultivars are recognized as being of coumarin, cinnamaldehyde, methyl cinnamate and various sesqui-
high quality, originating more expensive olive oils and so are legally terpenes were reported to differentiate C. verum from other species
protected by different certifications of origin. Using a high-performance (Avula, et al., 2015). Another recent report added that several meta-
liquid chromatography-electrospray ionization coupled with tandem bolites are capable of distinguishing between “cinnamon” species C.
mass spectrometry (HPLC-ESI-MS/MS) technique, Di Donna et al. verum and C. cassia, particularly eugenol (2-methoxy-4-prop-2-en-
(2010) analysed the leaves of five cultivars of Olea europaea (Carolea, ylphenol), that was found mostly in C. verum (see Table 1) (Farag, et al.,

154
S. Medina et al. Food Chemistry 278 (2019) 144–162

2018). All studies referred above emphasize that genetic diversity statistically different between the two crops (see Table 2). This fact may
(plant cultivars) modulate the intrinsic composition for several com- be due to N levels in the tubers, which are significantly higher in
pounds. These putative biomarkers have different physicochemical conventional fertilization regimes when compared to the organic ones
properties, mainly involving amino acids, phenolic compounds, sugars (Shepherd, et al., 2014). The application of a comprehensive metabolite
and organic acids, but also VOCs from a wide diversity of chemical approach using GC–MS to discriminate between O/C productions was
families. Therefore, it is important to consider different analytical also applied to maize samples (Zea mays). In this case, malic acid, myo-
platforms to identify biomarkers able to reliably discriminate foodstuffs inositol and phosphate were the metabolites responsible for such dif-
by the plant cultivar used in its production. ferentiation. However, in this work, other factors like the genotype,
farming location, and growing season may have contributed to the
2.2. Production systems different metabolite profile of maize kernels obtained (Röhlig & Engel,
2010). In contrast, the influence of these factors has been considered in
The systems behind the production of different foodstuffs poten- a recent untargeted metabolomics approach performed in carrot sam-
tially affect the levels of certain phytochemicals. This may result in the ples (Daucus carota L.). Cubero-Leon, et al. (2018) found a consistent
identification of food production biomarkers, similarly to the examples influence of the production systems on the crop’s metabolome of two
given in the previous section. In this context, several studies have as- carrots varieties (Nerac and Namur) produced by different fertilization
sayed the influence of farming, rearing or feed practices in the meta- approaches, in different regions of the same country, and over a period
bolites profile of different foodstuffs (food metabolome) (Cubero-Leon, of four years. In this case, markers related to carbohydrate metabolism
et al., 2018). This discussion is particularly relevant in the context of and plant defence mechanisms were identified as responsible for the
the growing consumer’s interest in organic foods, whose authentication differences between O/C carrot samples (Table 2) (Cubero-Leon, et al.,
requires robust analytical platforms and bioinformatics tools. Re- 2018). The influence of the farming system has been also assayed in the
garding this, the influence of farming practices (organic and conven- goldenberry fruits (Physalis peruviana). In a recent report, untargeted
tional; O/C) in the context of a single nutrient or class of compounds is metabolomics analysis identified two withanolides and one fatty acyl
still matter of debate in the scientific literature. Nevertheless, un- glycoside as putative markers to differentiate O/C goldenberry fruits.
targeted metabolomics approach has a great potential to certify the However, targeted metabolomics assays involving carotenoids and as-
authenticity of organic plant products (Llano, et al., 2018). In recent corbic acid did not find statistical differences between both crops.
years, several studies presented potential biomarkers to discriminate Therefore, untargeted metabolomics seems to be more suitable for the
samples from O/C farming (Table 2). Vallverdú-Queralt, Medina- authentication of organic products (Llano, et al., 2018). In addition,
Remón, Casals-Ribes, Amat, and Lamuela-Raventós, (2011), for in- other studies are based on carotenoids profiling to verify the differences
stance, compared tomato-derived products (O/C ketchup) using high- between farming systems. In this context, a previous report highlighted
performance liquid chromatography-quadrupole time of flight-mass the changes in carotenoids profile between O/C chicken eggs. The re-
spectrometry (HPLC-q-TOF-MS) and reported that glutamylphenylala- sults of this study indicate that food fingerprinting approach is a va-
nine and N-malonyltryptophan were biomarkers able to differentiate luable tool for analytical verification of the production systems of or-
organic ketchup from the conventional one. Also, in tomato products, a ganic eggs (van Ruth, et al., 2011). Related with this, another study
method developed to analyse tomatidine using LC coupled to LTQ-Or- showed that the volatile fraction of organic bananas was higher than
bitrap mass spectrometry, identified higher levels of this compound in conventional ones and these augmented VOCs could be used as tenta-
conventional samples. Additionally, the authors of this study also used a tive biomarkers to authenticate organic bananas. This shows that the
non-targeted metabolomics approach to identify another two metabo- method of cultivation also alters the fruit aromatic composition, pro-
lites, dehydrotomatine and β1-tomatine, that discriminate the two types viding an higher aromatic quality to the fruits grown under organic
of samples (Caprioli, Logrippo, Cahill, & James, 2014). A previous farming, which is certainly a valuable information for producers and
metabolomics fingerprinting assay employing direct analysis in real consumers (Dong, Chen, Wang, Huang, & Yi, 2014). Nevertheless, ap-
time (DART) coupled to TOF-MS for authentication of tomatoes and proaches involving complementary analytical techniques, as the ones
peppers from O/C farming was also reported. This approach identified combining metabolomics fingerprinting by ultra-performance liquid
several metabolites altered in both tomatoes and peppers (2-furanone, chromatography high resolution mass spectrometry (UPLC-HR-MS),
furfural, α-ketobutyric acid, citric acid, dehydroascorbic acid, glutaric volatile profile analysis by HS-SPME-GC–MS, and chemometrics
acid, hydroxymethylfuran, itaconic acid, malic acid, pyruvic acid, and models, certainly constitute a more powerful framework for the au-
succinic acid), showing the effect of agronomic systems on metabolome thentication of organic foodstuffs. This integrated approach involving
and specific metabolic pathways, irrespective of the matrix tested the characterization of both the volatile and non-volatile metabolites
(Novotná, et al., 2012). The potential of metabolomics for the au- was recently carried out in orange juices and allowed the differentiation
thentication of organic crops has been also explored in cereals. Zörb between O/C juices (Table 2) (Cuevas, Pereira-Caro, Moreno-Rojas,
and co-workers, for instance, determined the levels of 52 metabolites, Muñoz-Redondo, & Ruiz-Moreno, 2017). The phenolic compounds
including amino acids, organic acids, sugars, sugar phosphates, alco- content of organic products relative to their conventional counterparts
hols, and nucleotides in wheat (Triticum aestivum L.) grown under O/C has been extensively investigated. Although this is still matter of debate
conditions. The authors reported that only eight of them (alanine, β- the scientific literature, it appears that organic products have a higher
alanine, glycerate, hydroxyglutarate, myo-inositol, panthotenic acid, content of phenolic compounds than the traditional ones. This ob-
urea and valine) showed significant differences between two farming servation was reported, for instance, for polyphenols in grapes juices
systems (Table 2) (Zörb, Langenkämper, Betsche, Niehaus, & Barsch, (Granato, et al., 2015), anthocyanins in blueberries var. Bluecrop
2006). Another study involving wheat grains, detected 11 metabolites (Vaccinium corymbosum L.) (Wang, Chen, Sciarappa, Wang, & Camp,
with significant differences between the O/C farming practices (Bonte, 2008), and different antioxidant compounds (phenolic compounds,
et al., 2014). Moreover, among these metabolites, three of them, ala- carotenoids, vitamin C, among others) in sweet bell peppers (Hallmann
nine, myo-inositol and urea, were also identified in the work of Zörb & Rembiałkowska, 2012), suggesting that organically grown foods have
and co-workers (Zörb, et al., 2006) previously reported in this section. higher content of these compounds and this fact may have positive
Finally, Weesepoel, et al. (2016) also reported that protocatechuic acid effects on human health. Regarding this, an animal experiment showed
may serve as a single marker for discrimination between O/C produced that coffee (both O/C) had beneficial effects on rat health, although
wheat. Metabolomics assays were also performed in potato samples chlorogenic acids levels were higher in organic coffees than conven-
(Solanum tuberosum L.) grown under both O/C fertilizer systems and tional ones. More studies are therefore necessary to clarify whether
most of the metabolites identified, mainly amino acids, were there is an effective higher content in phenolics in the organic products

155
S. Medina et al. Food Chemistry 278 (2019) 144–162

relatively to conventional products and if this represents a real positive that required robust statistical methods in order to extract the relevant
effect in human health (do Carmo Carvalho et al., 2011). Food pro- features related with product origin and allow an eventual assessment
cessing adds an extra level of complexity to the authentication problem of the influence of the product origin on food metabolome (Klockmann
because its effects on food composition and eventually can mask the et al., 2016). As previous emphasized, the verification of the true
discrimination of frauds. Regarding this, two fatty acids, phytanic acid composition of different foodstuffs is very important to certify their
and pristanic acid, were found significantly higher concentration in authenticity and defend their economic value. This is particularly re-
organic products from German markets like cheese, milk, cream and levant for products whose specificity is based on its origin and orga-
butter than conventional ones. A level of 200 mg of phytanic acid per noleptic characteristics, as wines, honeys, coffees, cheeses or olive oils.
100 g of lipids that was then suggested as the value for control for or- Usually, this certification is achieved through the identification of
ganic products and this will constitute a valuable biomarker for the biomarkers of different botanical species used in the production of a
authentication of this organic product (Vetter & Schröder, 2010). given product. Honeys, for instance, were subject of extensive studies to
Discrimination of the rearing practices used to obtain certain food certify their authenticity and detect frauds and adulterated products.
product is certainly very relevant to the producers and consumers. The Overall, honeys can be discriminated by the type of flowers (unifloral or
work from Mannina, et al. (2008), used NMR to differentiate between multifloral) that different species of bees pollinate in different geo-
wild and cultured sea bass (Dicentrarchus labrax). The metabolite pro- graphies, resulting in almost unlimited number of honeys with different
filing in conjunction with statistical analysis identified fatty acids characteristics and flavours (Beretta, Fermo, & Maffei Facino, 2012;
(monounsaturated fatty acids (MUFA), diunsaturated fatty acids Boffo, Tavares, Tobias, Ferreira, & Ferreira, 2012; Spanik, Pazitna,
(DUFA), and polyunsaturated fatty acids (PUFA)) and other metabolites Siska, & Szolcsanyi, 2014). Originally, honeys were certificated by its
as compounds responsible for the discrimination between different botanical origin based in the microscopic analysis and quantification of
rearing systems. Fumaric and malic acids, which are metabolites of the pollens used in their production (melissopalynological analysis).
Krebs cycle, were found among the identified metabolites, suggesting Meanwhile, alternative methodologies were explored to certify the
differences at the level of energy metabolism in cultured fish versus wild authenticity of honeys and a huge number of studies can be found in the
type fish. Moreover, accumulation of glutamine was also reported in the literature (Pattamayutanon, et al., 2017). Here we will present selected
wild type fish, what can be correlated with the higher level of these examples of different methodologies (HPLC, GC, NMR, NIRS) and
organic acids, since glutamine metabolism is closely related to Krebs markers reported to differentiate specific honeys, as orange honeys
cycle (Mannina, et al., 2008). As can be seen in Table 2, these results (Tette, et al., 2017), manuka honey (Lin, Loomes, Prijic, Schlothauer, &
are in accordance with another work reporting fatty acids and free Stephens, 2017), chestnut honeys (Beretta, et al., 2012) or even honeys
amino acids as discriminative compounds between farmer and wild sea obtained with avocado nectar (Dvash, et al., 2002), as shown in Table 3.
bass (Fuentes, Fernández-Segovia, Serra, & Barat, 2010). Recently, the In addition, propolis, a beehive sealing material and one of the most
influence of feed practices in the production of different foodstuffs has heterogeneous natural cocktail built from plant sources, has been also
been investigated by different researchers. Segato, et al. (2017), for subjected to chemometric analysis and several studies successfully as-
instance, were able to distinguish among three cheese production sys- signed propolis type to the corresponding plant source (Betrams,
tems with milk from cows fed, pasture-based, hay-based and maize si- Muller, Kunz, Kammerer, & Stinzing, 2013; Falcao et al., 2013). On the
lage-based, which one was used in the production of cheese from other hand, coffee constitutes another commodity of great economic
Asiago, and Italian protected designation of origin (PDO). Using GC and value and so it has been extensively studied, particularly its different
the discriminant analysis, the authors found nine fatty acids and fat- aroma and flavours. Table 3 presents potential biomarkers for coffee
soluble vitamins as useful biomarkers to separate the three production discrimination by geographic origin using chemometric analysis of the
systems. Moreover, high contents of conjugated linoleic acids (CLA), volatile composition. In the same sense, wine is a matrix that has been
antesio-C15:0 and vitamin A were enough to discriminate pasture-based extensively analysed regarding the contribution of different factors to
cheese (Segato, et al., 2017). Also related with the cheese production, its typicity, namely grape varieties and winemaking processes used.
another study reported the discrimination of the simultaneously use of Wine composition in bioactive compounds, like phenolic compounds,
milk originating from two different production systems (animals constitute a reliable signature to classify wines according to their bo-
feeding on pasture and feeding with concentrates or silages) to produce tanical and geographical origin, as exemplified in Table 3 for Chinese
the same cheese type. This allowed an effective product protection and red wines (Sun, et al., 2015), Czech white wines (Lampir & Pavlousek,
a corresponding increase of economic value (Povolo, Pelizzola, Ravera, 2013) and Argentinean wines (Pisano, Silva, & Olivieri, 2015). The
& Contarini, 2009). Similarly, another work reported the presence of wine aroma, usually assessed using GC–MS, has a high potential for
lactobacillic acid and dihydrosterculic acid (components of bacterial wine discrimination, particularly in what concerns to complex wines,
membranes) in milk and dairy products from cows fed with silage. allowing a deeper characterization of their aging and quality. In Table 3
Therefore, these acids may be used as biomarkers of silage feedings for are presented volatile markers that were reported to discriminate dif-
milk-based products (Caligiani, Nocetti, Lolli, Marseglia, & Palla, ferent aspects of Madeira wine, one of the oldest fortified wines known,
2016). Lactobacillic acid was also detected in milk samples from cows according to the cultivars used in its production (Câmara, Alves, &
fed with forages containing maize silage, which is not allowed to pro- Marques, 2007; Perestrelo, Silva, & Câmara, 2014). In general, it can be
duce milk for some PDO cheeses as Parmigiano-Reggiano, constituting easily observed that the botanicals used in food production, like food
therefore a very useful biomarker to detect cheese adulteration supplements, are the main source of markers for their certification and
(Marseglia, et al., 2013). In the same context, but using volatile markers traceability. In this context, several classes of compounds have been
characterized through GC–MS analysis, skatole, 3-undecanone, cuminic explored as botanical markers, particularly phenolics, flavonoids, or-
alcohol, 1-butanol, and 2-methyl were able to discriminate between ganic acids, terpenes, sugars, and amino acids. Phenolic compounds, for
animals fed concentrate and animals fed non-concentrate diets and the instance, are secondary metabolites arising from shikimate/phenyl-
terpenoid germacrene D was reported as a prospective marker of grass propanoid pathway or ‘polyketide’ acetate/malonate pathway, or both,
feeding (Vasta et al., 2011). and are produced by plants, often in response to abiotic and abiotic
adverse conditions (García-Parrilla, Cerezo, Tesfaye, & Troncoso,
2.3. Geographical/botanical origin 2011). Furthermore, several health protective properties beyond their
antioxidant effect have been associated to phenolics and hundreds of
Ongoing consumer’s demands about greater labelling specifications such compounds have been reported so far in different plants. This
regarding the foods geographical origin have driven great research ef- constitutes a reliable fingerprint to discriminate different commodities
forts using advanced technologies. This has produced complex results according to their geographic and botanic origin, including cocoa

156
S. Medina et al. Food Chemistry 278 (2019) 144–162

(Pedan et al., 2017), wines (Lampir & Pavlousek, 2013; Pisano, et al., present on wild trees (Beretta, et al., 2012).
2015), and fruits, as apples or oranges (Diaz, Pozo, Sancho, &
Hernandez, 2014; Gan, Soukoulis, & Fisk, 2014). On the other side,
terpenoids (or isoprenoids), arising from mevalonic acid (MVA, or 2.4. Food adulteration
isoprenoid) pathway and 1-deoxy-D-xylulose 5-phosphate (DXP, or non-
mevalonate) pathway, constitute the largest class of known plant me- Nowadays, food adulteration represents a major concern because
tabolites, with over 40,000 structures identified. In this respect, foods are often contaminated with additives to increase bulk, disguise
monoterpenoids, sesquiterpenoids diterpenoids, triterpenoids, and spoilage and enhance the benefits. This challenge requires a continuous
carotenoids are among the most relevant terpenoids, and they have effort to improve the analytical power to uncover sophisticated pro-
important applications as pharmaceuticals, flavours, or fragrances. ducts adulterations, particularly when food safety, rather than food
Most of these terpenoids are volatile, conferring unique flavours and quality, is a real issue. In this context, olive oil, due to its commercial
fragrances and they may be used as biomarkers to authenticate the value, is the product more often adulterated with other cheaper oils
origin of foodstuffs in which they are present, such as cheeses (Tompa, from different species. Hazelnut oil is the oil most often used to adul-
Susic, Rogelj, & Pompe, 2013), honeys (Pattamayutanon, et al., 2017; terate olive oil inasmuch as it has a similar composition of triacylgly-
Schievano, Morelato, Facchin, & Mammi, 2013; Spanik, et al., 2014; cerols, fatty acids, and sterols (Azadmard-Damirchi, 2010). This fact
Stanimirova, et al., 2010), fruits (Ferreira, Perestrelo, & Camara, 2009) makes it very difficult to detect olive oil adulterated with hazelnut oil,
or wines (Perestrelo, et al., 2014). Several volatile carbonyl compounds but lupeol and filbertone, which are specific to hazelnut oil, could as-
are often reported as markers of different plant species and foods in sess the fraud down to an adulteration level of 2% and 10%, respec-
terms of geographical origin. This is the case of green leaf volatiles tively. Lupeol is a pentacyclic triterpene with promising anticancer
(GLVs), consisting of C6 and C9 aldehydes, alcohols and their esters, as activity and filbertone, (2E)-5-methyl-2-hepten-4-one, also known as
alkyl esters, carbonyl compounds (e.g., hexanal, trans-2-hexenal) and hazelnut ketone, is an unsaturated ketone often described as a highly
alcohols (e.g., 1-hexanol, 1-butanol, cis-3-hexenol) reported by Gan, potent plant flavour molecule (Azadmard-Damirchi, 2010) (Table 4).
et al. (2014) to discriminate apple juices from different geographic The adulteration of olive oil with cheaper olive oils from other regions
origins (New Zealand, South Africa and Chile; see Table 3). GLVs are and quality is also very frequent. However, a chemometric analysis is
obtained from the oxylipins pathway, through the peroxidation of reliable enough to obtain markers for the detection of these adultera-
PUFA and they are involved in plant defence, plant–plant interactions tions, as is shown in Table 3 with Italian olive oils (Benincasa, et al.,
and plant–insect interactions. Beyond the previous example, different 2003; Cavaliere, et al., 2007). Milk is the second product more often
GLVs have been pointed as markers of botanical or geographical origin adulterated. This is usually achieved with the addition of vegetable
in olive oils (Benincasa, et al., 2003; Cavaliere, et al., 2007) and in a proteins and fats (Kim, Kim, & Park, 2015), milk from different species,
variety of fruits and vegetables, as apples and tomatoes (Ferreira, and an addition of whey and watering are more and less harmless to
Perestrelo, Caldeira, & Camara, 2009; Feudo, Macchione, Naccarato, human health. However, there are milk adulterations which are not
Sindona, & Tagarelli, 2011). Sugars also present huge variations among safe, causing adverse health effects, as urea (Abernethy & Higgs, 2013),
different botanical and geographical origin, making them also excellent formalin, detergents, ammonium sulfate, boric acid, caustic soda,
biomarkers. In higher plants, sugar metabolism is essentially driven by benzoic acid, salicylic acid, hydrogen peroxide, sugars and melamine.
different invertases that hydrolyse sucrose into glucose and fructose. In Nevertheless, these adulterations, unlike the less harmful to human
turn, these are the key molecules that plants use as nutrients, energy health referred above, are easily detected. Overall, lipids diversity
sources, and signalling for growth, yield formation, and cope with stress constitutes a good marker of adulteration of different foodstuffs,
(Wan, Wu, Yang, Zhou, & Ruan, 2018). Broadly, plant invertases can be namely milk and derivatives. The lipid fraction of milk, for instance,
classified in two types; cytoplasmic invertases (CINs) and acid in- contains mono-, di- and triglycerides, glycolipids, cholesterol and re-
vertases, which in turn can be associated to the cell wall (CWIN) or exist spective esters derivatives, free fatty acids, phospholipids and sphin-
in a vacuolar soluble form (VIN) (Wan, et al., 2018). Thus, CINs pro- golipids, and the presence, diversity and amount of each of these group
duce sugars that are important to maintain cytosolic sugar homeostasis of lipids forms a specific lipid profile that allows an easy detection of
for cellular functions and are more conserved in phylogenetic terms. adulteration (Kim, et al., 2015). Many studies are proposing strategies
However, CWINs and VIN, are involved in plant response and adapta- to certify meat composition, a concern that is related not only with
tion to biotic and abiotic stresses, being able to process oligosaccharides meat quality but also with meat safety. This is the case, for example of
containing both sucrose and fructose, generating a myriad of different the adulteration of mince with meat from other species. In this context,
sugars that can be used as plant biomarkers. This is the case of complex it is noteworthy to refer the integrated approach followed by (Trivedi
sugar derivatives as perseitol (a sugar alcohol) detected in honey and it et al., 2016) that underwent a metabolite profiling and lipidomics ap-
may be used to certify that honeybees have harvested avocado nectar proach to detect the adulteration of beef with pork meat. By combining
(Dvash, et al., 2002). Also, citrusin D, a fatty acyl glycoside of mono- data from GC–MS and UHPLC-MS analyses using chemometric tools,
and disaccharides, was tentatively identified as optimal marker to dif- these authors reported over 20 metabolites from different classes that
ferentiate Valencian (Spain), Argentinean, Brazilian and South African allowed the detection of beef meat adulterated with pork meat. Fur-
oranges (Diaz, et al., 2014). Furthermore, amino acids not only are the thermore, pathways involved in this process may be researched to de-
backbone of proteins, but also of other molecules that plant use for tect significant differences in glutathione, inositol and sphingolipid
different functions. Many of these molecules constitute botanical mar- metabolism that justify the differences reported between the two types
kers, while other metabolites accumulate in response to certain abiotic of meat studied (Table 4). The fraudulent addiction of a juice of lesser
factors and they could be considered geographical markers. For in- economic value to a more expensive one, known as juice-to-juice de-
stance, synephrine, an aromatic amine formed in citrus by a pathway basing, is often detected. In this regard, a recent study reported that
involving tyramine and N-methyltyramine, which in turn derive from arbutin (4-hydroxyphenyl-β-D-glucopyranoside) was a specific marker
the amino acid tyrosine, has been recently reported as a potential for apple juice debasing with pear juice (Thavarajah & Low, 2006).
biomarker for orange honey authenticity (Tette, et al., 2017). Likewise, More recently, another work informed that 4-O-p-coumarylquinic acid,
kynurenic acid, arising from the L-tryptophan metabolism, was reported isorhamnetin-3-O-rutinoside and ABA were able to demonstrate the
as a marker of chestnut honey and was found in honey varieties of debasing procedure (Willems & Low, 2018). Thus, complex sugars as
arboreal origin, but it was absent in honeys obtained from herbal arbutin and isorhamnetin-3-O-rutinoside can be also reliable markers of
flowers. This distribution may be explained as a plant defence me- food fraud because they are specifically produced by some species that
chanism against fungal pathogens and herbivorous parasites, ever- often are not the ones used to obtain the original product.

157
S. Medina et al. Food Chemistry 278 (2019) 144–162

2.5. Spoilage and freshness markers metabolites, mainly sulfur compounds, were detected just after 4 days
of storage, while microbial populations were still below the accepted
Food spoilage is a complex process to which many factors, as tem- sensory level of spoilage. This fact is very significant and it may in-
perature, moisture, pH level or oxygen pressure, contribute. It has a crease our knowledge for the early detection of spoilage process
crucial importance in food safety as many metabolites produced during (Alexandrakis, Brunton, Downey, & Scannell, 2012). More recently,
spoilage are toxic to humans (Kuuliala, Al Hage, et al., 2018). There- proton-transfer-reaction mass spectrometry (PTR-MS), an on-line tech-
fore, the identification of key markers of food spoilage is very relevant. nique that enables the detection of compounds in real-time, has been
Likewise, food freshness certainly corresponds to the consumer pre- also applied to study chicken spoilage, identifying most the volatile
ference and simultaneously ensures food safety, at least in what con- biomarkers previously referred, as acetone, dimethyl sulfide and
cerns to the toxic effects caused by spoilage. Thus, rapid monitoring of ethanol (Franke & Beauchamp, 2017). The combination of advanced
food deterioration is considered essential, and apart from the tradi- analytical platforms, as GC–MS and 1H NMR, to analyse the same
tional chemical spoilage markers, such as total volatile bases nitrogen sample, allows a deeper and comprehensive analysis, but is simulta-
(TVB-N) and trimethylamine nitrogen (TMA-N), research for new neously very important to identify specific metabolites. In this context,
markers able to monitor food spoilage processes is essential. In this several markers able to monitor the progress of the spoilage process
context, Parlapani, Haroutounian, Nychas, and Boziaris, (2015) used were detected in beef meat. Among them, butanoic acid and acetoin (3-
SPME-GC–MS to analyse sea bass storage and reported changes in hydroxy-2-butanone) were related to the growth of different microbial
specific VOCs, mainly alcohols, aldehydes, and esters, which could be groups. Butanoic acid, commonly associated with meat spoilage, could
used as markers of deterioration (Table 5). In the same way, a recent be derived from microbial consumption of free amino acids via the
study studied freshness reduction in oyster (Crassostrea gigas) by the Stickland reaction. In addition, the accumulation of acetoin has been
detection of dimethyl sulfide by GC–MS upon 4 days of harvesting and associated not to spoiled meat, but rather to “not fresh” product. Re-
trimethylamine upon 2 days of harvesting, in this case using a bioe- frigeration is a common process of meat storage, but this procedure
lectronics nose. Hence, these biomarkers have a powerful potential to causes the dehydration of the meat surface and stimulates the pro-
monitor seafood spoilage for real-time and on-site analysis, thus con- duction of betaine, another relevant marker to monitor the deteriora-
tributing to prevent seafood-borne diseases (Lee, et al., 2018). Other tion process in beef that was identified in the study (Ercolini, et al.,
study carried out in salmon samples (Oncorhynchus tshawytscha) iden- 2011). Still related with meat products, a work carried out in Iberian
tified VOCs as biomarker candidates to supervise salmon spoilage, as ham identified many VOCs correlated with the deterioration of this
well as certain alcohols and aldehydes able to reveal salmon freshness product, particularly pyrazines, ethers and carbonyls (Table 5) (Martín
(Table 5) (Wierda, Fletcher, Xu, & Dufour, 2006). Two recent and et al., 2010). Furthermore, it has been also reported that fungi infec-
promising studies carried out in Atlantic cod (Gadus morhua) were tions, particularly the ones caused by Botrytis cinerea and different
performed using selected-ion-flow-tube mass spectrometry (SIFT-MS). species of Penicillium and Aspergillus destroy food flavours, affecting
This approach eliminated the need of sample preparation procedures very negatively food quality and safety. Regarding this, ergosterol was
while allowing fast and sensitive analysis of the VOCs profile over recently reported as a specific marker for the evaluation of fungi in-
storage time. Fish packaging and storage conditions affected the evo- fection and phytosanitary status of grapes (Porep, Walter, Kortekamp, &
lution of the VOCs profile and should be considered in the selection of Carle, 2014). Another study was carried out in different food matrices
spoilage indicators, which included VOCs as 2,3-butanediol; 3-methyl- (bread, chicken egg and cucumber), using an untargeted metabolic
1-butanol; ethyl acetate and trimethylamine. Thus, the evolution of profiling approach based on two-dimensional GC-TOF-MS, allowed the
these spoilage markers in conjunction with microbial growth assays and differential identification of 13 metabolites, including amino acids,
sensory rejection could be used for the development of intelligent biogenic amines, and organic acids. The major metabolic changes in the
packaging applications (Kuuliala, Abatih, et al., 2018; Kuuliala, Al spoilage process of the target foods were closely related to some me-
Hage, et al., 2018). Chicken meat is widely appreciated across the globe tabolic pathways, as amino acids metabolism, Krebs cycle and ammonia
and so the identification of volatile markers to certify its degree of recycling (Cheng et al., 2015). One of the most recent studies about
conservation is certainly highly relevant. In this context, Lovestead and food freshness performed in three different citrus varieties (C. clem-
Bruno (2010) developed an headspace analysis using cryo-adsorption entina, C. tangerina and C. reticulata), reported that the Raman signal
with a porous layer open tubular plot (PLOT) column to separate and due to the carotenoids from fruit peel was a good indicator of citrus
detect the VOCs emitted by the sample by GC–MS. Then, the authors freshness, therefore introducing an objective criterion to assess citrus
applied this methodology to identify markers of chicken spoilage in freshness. These results could have a strong impact in food industries
breast tissue samples. In another recent study performed with chicken and consumer satisfaction (Nekvapil, et al., 2018).
breast, a PCA analysis unveils some details about the spoilage process.
According to the results obtain, breast chicken spoilage seems to occur 3. Factors affecting the robustness of food fingerprints
in two steps or, alternatively, involve two different pathways in which
the glycolytic and proteolytic reactions occur simultaneously, but at There are many factors affecting the analysis of food fingerprints
different rates. This mechanism is supported by the observation of a and the robustness of the models obtained to certify food authenticity.
first increase in alcohols and free fatty acids levels (primary spoilage One of the main bottlenecks is the critical lack of standardization in the
markers), followed by a rise in the sulfide contents (secondary spoilage experimental layouts used. The sample size we find with this review, for
markers) caused by the nitrogenous compounds and amino acids cat- instance, presents a great heterogeneity, ranging from five to 374
abolism, mainly cysteine and methionine (Table 5). Overall, ethanol samples (Tables 1–5). This fact certainly represents a major problem in
was the predominant VOCs in all spoiled samples, a fact that may result the comparison of different studies for authentication purposes. Al-
from the metabolic activity of lactic acid bacteria, as Pseudomonas spp., though an optimal samples number may be difficult to establish, this
often present in spoilage food. In turn, 3-methyl-1-butanol and acetic number should be very large (Novotná, et al., 2012), preferentially
acid are usually generated by activity of Brochothrix thermosphacta and higher than the number of variables identified so the statistical models
lactic acid bacteria. Together, these three VOCs could be considered as can produce reliable classification models. Regarding the quality of the
helpful indicators to evaluate the deterioration of aerobically stored samples sets, the sampling period is also an important factor. In fact, in
chicken meat (Miks-Krajnik, Yoon, Ukuku, & Yuk, 2016). A third study previous studies, year-to-year variations in the samples set have been
on chicken breast using HPLC and GC–MS reported amino acids and reported for several food matrices, as wheat, potato, honey or orange
VOCs as putative biomarkers of chicken spoilage prior to organoleptic (Diaz, et al., 2014; Shepherd, et al., 2014; Stanimirova, et al., 2010;
changes in this foodstuff. A relevant aspect of this study is that certain Weesepoel, et al., 2016). Overall, most studies have investigated the

158
S. Medina et al. Food Chemistry 278 (2019) 144–162

differences on the chemical composition among two years of produc- acetoin, lactate and threonine have also been identified as spoilage
tion, with the exception of two works, involving carrot samples markers in both chicken and beef meat (Alexandrakis, et al., 2012;
(Cubero-Leon, et al., 2018) and wines samples (Perestrelo, et al., 2014), Ercolini, et al., 2011; Franke & Beauchamp, 2017). These metabolites
where the differences among four years of production were in- may be very useful as spoilage markers because they have been de-
vestigated. Nonetheless, the seasonal variability can also produce dif- tected in several foodstuffs, but they should be quantitatively evaluated
ferences in the fingerprints in the same year. Souard, et al. (2018), for to know their sensory contribution to the “gone-off” odour
instance, reported that most of the coffee samples harvested in No- (Alexandrakis, et al., 2012). Furthermore, several metabolites have
vember had lower intensities for most of the metabolites than the been detected in the same food matrix using different technologies. For
samples harvested in other periods. In other studies, however, there instance, caffeine, caffeoylquinic acid and sucrose have been reported
were not described differences among wine samples of different years of as significant metabolites to differentiate coffee varieties by 13C NMR
production, suggesting that, at least in these cases, the influence of the (Wei, et al., 2012) and MS (FT-ICR-MS and LC-q-TOF-MS) (Garrett,
genetic background may be higher than the environmental conditions et al., 2013; Souard, et al., 2018). Another example is the detection of
fluctuations (Câmara, et al., 2007; Righetti, et al., 2018). Considering cinnamaldehyde in cinnamon by both DART-q-TOF-MS and NMR
the high variability of the results reported, more efforts are needed to technologies (Avula, et al., 2015; Farag, et al., 2018). These technolo-
understand the conjugated effects of genetics and seasonal climate gies combined with chemometrics tools have a powerful potential for
variations on the chemical composition of food (Longobardi, et al., food authentication with traceability and certification purposes. How-
2017). Still related with this heterogeneity in the results, the influence ever, the selection and quality of statistical classification models are
of other aspects as natural contaminations should be considered. For essential for the robustness of the biomarkers identification and dis-
example, in honeys from diverse botanical origins, overlapping flow- crimination. In this context, the implementation of model validation
ering periods of different plants may occur, resulting in a cross con- through, for example, the leave-one-out cross validation (LOOCV, is a
tamination of the samples (Schievano, et al., 2013). Besides, storage particular case of leave-p-out cross-validation with p = 1) is required
conditions before analysis must be standardized because samples are not only to estimate the robustness of statistical analysis, but also to
often stored frozen, refrigerated, kept at room temperature or even on prevent overfitting (Farag, et al., 2018; Gan, et al., 2014; Garrett, et al.,
ice, and these variations may modulate metabolite levels. 2013). Furthermore, the models should allow a reliable sample dis-
In this review, hundreds of metabolites presented in Tables 1–5 are crimination with a correct classification (∼100%) upon external vali-
described as putative biomarkers and many of them have been char- dation (Gan, et al., 2014). It is noteworthy that the choice of statistical
acterized according to specific authentication issues. In addition, models may influence the outcomes obtained. In this context, in a re-
identical molecules have been described among different studies in the cent work carried out with two varieties of carrot samples during four
same matrix. This is the case of β-damascenone that was found as production years, PCA was not able to discriminate between O/C car-
discriminant metabolite in wine samples of different cultivars in two rots, but it was a successful model for the classification based on the
studies (Câmara, et al., 2007; Culleré, et al., 2008). Also, three VOCs, 1- production year. OPLS-DA, in turn, was able to distinguish among
hexanol, 2-methylbutyl acetate, and hexyl acetate, were reported in farming systems (Cubero-Leon, et al., 2018). In another study carried
different works as molecules able to distinguish different varieties/ out in grape juices (biodynamic, conventional and organic) from Brazil
cultivars of apple juices, with an assignment success rate of 100% in and Europe, PLS-DA allowed the classification between Brazilian and
classification and prediction test (with internal and external validation European grape juices (intraregional models with a 92% of efficiency),
set for both studies - inter-laboratory reproducibility) (Gan, et al., 2014; kNN algorithm differentiated between European conventional and or-
Guo, et al., 2012). In contrast, the amino acids alanine and γ-amino- ganic/biodynamic samples and SIMCA model enabled the discrimina-
butyric acid have been described in three reports as discriminant tion among juices under different farming systems. However, neither
markers of different rice cultivars/varieties (Danial, et al., 2013; Frank, ANOVA nor HCA had significant/clear statistical results (Granato, et al.,
et al., 2012; Hu, et al., 2014). The sample preparation, extraction, and 2015). More recently, the same author, in a detailed review about
storage (−18, −20 and −80 °C) were quite different among these chemometrics methods applied to food authentication, declared that
three works, so the detection of these specific amino acids was largely HCA is a highly arbitrary model and it should not be used with food
independent of experimental design. Regarding the farming systems, authenticity purposes. It is therefore a very difficult task to select the
three metabolites, alanine, myo-inositol, and urea, were described as “best” statistical model because of all of them present negative and
significant metabolites for the discrimination between O/C wheat positive characteristics (Granato et al., 2018). This is another major gap
samples in two different studies using GC–MS (Bonte, et al., 2014; Zörb, in food authentication protocols that, together with the lack of super-
et al., 2006). Similarly, lactobacillic acid was established as a molecular vision from government‘s authorities and the weak communications
marker for cheese authentication by two different studies using GC–MS between consumers and producers, is making food authenticity certi-
and confirmed by NMR (Caligiani, et al., 2016; Marseglia, et al., 2013). fication a demanding and very challenging process to food handlers and
Another two works allowed wine differentiation according to geo- food standardization authorities.
graphic origin from Czech Republic (Lampir & Pavlousek, 2013) and
China regions (Sun, et al., 2015) with identical molecules as dis- 4. Conclusions and future perspectives
criminant markers ((+)-catechin, (−)-epicatechin, and protocatechuic
acid). Although, more evidences are necessary, these metabolites are Biochemical changes in foodstuffs caused by environmental, genetic
strong candidates to authenticate PDO wines. diversity or human activity result in specific variations that can be used
Related to food adulteration issue, arbutin is a molecular marker as markers of those products. In this context, compounds as amino
able to detect juice-to-juice debasing between apple and pear juices acids, phenolic compounds, organic acids and VOCs of different che-
described in two articles (Thavarajah & Low, 2006; Willems & Low, mical classes, are the potential biomarkers for food authentication most
2018). Concerning to spoilage biomarkers, several reports have de- often reported. As these compounds possess different physicochemical
tected dimethyl sulfide and trimethylamine in different foods, as oyster, properties, several analytical platforms must be employed to obtain
cod and chicken (only dimethyl sulfide) (Kuuliala, Abatih, et al., 2018; reliable methodologies to detect and quantify biomarkers able to certify
Lee, et al., 2018; Lovestead & Bruno, 2010). Particularly for chicken foodstuffs origin, variety or production system, as well as to detect food
samples, it is very important to note that dimethyl disulfide, dimethyl adulteration or spoilage/freshness indicators. Regarding this, strategies
trisulfide, acetone and dimethyl sulfide ethanol have been reported in envisaging the characterization of the volatile and non-volatile frac-
several works (Alexandrakis, et al., 2012; Franke & Beauchamp, 2017; tions of food products samples, followed by robust data fusion and
Lovestead & Bruno, 2010; Miks-Krajnik, et al., 2016). Moreover, processing algorithms involving advanced statistical analysis are

159
S. Medina et al. Food Chemistry 278 (2019) 144–162

becoming crucial to cope with the complexity of foodstuffs authenti- Beretta, G., Fermo, P., & Maffei Facino, R. (2012). Simple and rapid simultaneous pro-
cation. In this review, several promising biomarkers for food authen- filing of minor components of honey by size exclusion chromatography (SEC) coupled
to ultraviolet diode array detection (UV-DAD), combined with chemometric methods.
ticity have been pointed, but most of them still require deeper and more Journal of Pharmaceutical and Biomedical Analysis, 58, 193–199.
extensive validation procedures. To accelerate this process, we antici- Betrams, J., Muller, M., Kunz, N., Kammerer, D., & Stinzing, F. (2013). Phenolic com-
pate that more technologies facilitating routine analysis will become pounds as marker compounds for botanical origin determination of German propolis
samples based on TLC and TLC-MS. Journal of Applied Botany and Food Quality, 86,
available for rapid monitoring of food authentication. In turn, in- 143–153.
creasingly sophisticated untargeted methods will improve the ability to Boffo, E. F., Tavares, L. A., Tobias, A. C. T., Ferreira, M. M. C., & Ferreira, A. G. (2012).
detect the minor differences between genuine and fraudulent foods and Identification of components of Brazilian honey by 1H NMR and classification of its
botanical origin by chemometric methods. LWT – Food Science and Technology, 49(1),
beverages. 55–63.
Moreover, better analytical performances will allow not only con- Bonte, A., Neuweger, H., Goesmann, A., Thonar, C., Mäder, P., Langenkämper, G., &
firming the existence of a given biomarker, but also accurately quan- Niehaus, K. (2014). Metabolite profiling on wheat grain to enable a distinction of
samples from organic and conventional farming systems. Journal of the Science of Food
tifying it, a feature that will be very relevant in the definition of spoi-
and Agriculture, 94(13), 2605–2612.
lage markers. Additionally, PTR, SIFT and NMR outputs will certainly Caligiani, A., Nocetti, M., Lolli, V., Marseglia, A., & Palla, G. (2016). Development of a
result in future non-destructive analytical techniques for on-site and quantitative GC-MS method for the detection of cyclopropane fatty acids in cheese as
real-time food authentication, to which “user-friendly” chemometrics new molecular markers for parmigiano Reggiano authentication. Journal of
Agricultural and Food Chemistry, 64(20), 4158–4164.
approaches and authentication biomarkers databases need to be gen- Câmara, J. S., Alves, M. A., & Marques, J. C. (2007). Classification of Boal, Malvazia,
erated. Future generations of consumers will certainly be even more Sercial and Verdelho wines based on terpenoid patterns. Food Chemistry, 101(2),
informed, demanding for food integrity (food traceability, safety, 475–484.
Caprioli, G., Logrippo, S., Cahill, M. G., & James, K. J. (2014). High-performance liquid
quality, and authenticity), assuring that food should be not only heal- chromatography LTQ-Orbitrap mass spectrometry method for tomatidine and non-
thier and safer, but also be obtained through sustainable procedures target metabolites quantification in organic and normal tomatoes. International
that respect animal welfare and are environment-friendly. Journal of Food Sciences and Nutrition, 65(8), 942–947.
Cavaliere, B., De Nino, A., Hayet, F., Lazez, A., Macchione, B., Moncef, C., ... Tagarelli, A.
(2007). A metabolomic approach to the evaluation of the origin of extra virgin olive
Acknowledgments oil: A convenient statistical treatment of mass spectrometric analytical data. Journal
of Agricultural and Food Chemistry, 55(4), 1454–1462.
Cheng, J., Gao, R., Li, H., Wu, S., Fang, J., Ma, K., ... Dong, F. (2015). Evaluating potential
This work was supported by FCT-Fundação para a Ciência e a markers of spoilage foods using a metabolic profiling approach. Food Analytical
Tecnologia (project PEst-OE/QUI/UI0674/2013, CQM, Portuguese Methods, 8(5), 1141–1149.
Government funds) and by ARDITI-Agência Regional para o Cubero-Leon, E., De Rudder, O., & Maquet, A. (2018). Metabolomics for organic food
authentication: Results from a long-term field study in carrots. Food Chemistry, 239,
Desenvolvimento da Investigação Tecnologia e Inovação through the
760–770.
project M1420-01-0145-FEDER-000005 – Centro de Química da Cuevas, F. J., Pereira-Caro, G., Moreno-Rojas, J. M., Muñoz-Redondo, J. M., & Ruiz-
Madeira – CQM+ (Madeira 14-20). SM was supported by the Post- Moreno, M. J. (2017). Assessment of premium organic orange juices authenticity
Doctoral fellowship granted by ARDITI – CQM+ project (ARDITI- using HPLC-HR-MS and HS-SPME-GC-MS combining data fusion and chemometrics.
Food Control, 82, 203–211.
CQM/2017/008-PDG). JAMP and PS were supported by project Culleré, L., Escudero, A., Pérez-Trujillo, J. P., Cacho, J., & Ferreira, V. (2008). 2-Methyl-3-
M1420- 09-5369-FSE-000001 through the Post-Doctoral and PhD fel- (methyldithio)furan: A new odorant identified in different monovarietal red wines
lowships, respectively, while RP was supported by FCT Post-Doctoral from the Canary Islands and aromatic profile of these wines. Journal of Food
Composition and Analysis, 21(8), 708–715.
grant (SFRH/BPD/97387/2013). Culleré, L., Ferreira, V., Venturini, M. E., Marco, P., & Blanco, D. (2013). Potential aro-
matic compounds as markers to differentiate between Tuber melanosporum and Tuber
Conflicts of interest indicum truffles. Food Chemistry, 141(1), 105–110.
Dallago, R. M., Valduga, A. T., Luccio, M. D., Benin, S., & Tres, M. V. (2011). Analysis of
volatile compounds of Ilex paraguariensis A. St. - Hil. and its main adulterating species
The authors declare no conflict of interest. Ilex theizans Mart. ex Reissek and Ilex dumosa Reissek. Ciência e Agrotecnologia, 35,
1166–1171.
Danial, A. D., Neoh, B. K., Ersad, M. A., Ruzlan, N., Win, W., Musa, N. A., ... Yusof, H. M.
References
(2013). Differentiation of rice varieties using metabolite markers. Journal of
Agricultural Technology, 9(3), 601–610.
Abernethy, G., & Higgs, K. (2013). Lactose semicarbazone as a marker for semicarbazide Di Donna, L., Mazzotti, F., Naccarato, A., Salerno, R., Tagarelli, A., Taverna, D., &
adulteration in milk. Journal of Chromatography A, 1295, 152–155. Sindona, G. (2010). Secondary metabolites of Olea europaea leaves as markers for the
Akiyama, M., Murakami, K., Hirano, Y., Ikeda, M., Iwatsuki, K., Wada, A., ... Iwabuchi, H. discrimination of cultivars and cultivation zones by multivariate analysis. Food
(2008). Characterization of headspace aroma compounds of freshly brewed arabica Chemistry, 121(2), 492–496.
coffees and studies on a characteristic aroma compound of Ethiopian coffee. Journal Diaz, R., Pozo, O. J., Sancho, J. V., & Hernandez, F. (2014). Metabolomic approaches for
of Food Science, 73(5), C335–346. orange origin discrimination by ultra-high performance liquid chromatography
Alexandrakis, D., Brunton, N. P., Downey, G., & Scannell, A. G. (2012). Identification of coupled to quadrupole time-of-flight mass spectrometry. Food Chemistry, 157, 84–93.
spoilage marker metabolites in Irish chicken breast muscle using HPLC, GC–MS do Carmo Carvalho, D., Brigagão, M. R. P. L., dos Santos, M. H., de Paula, F. B. A., Giusti-
coupled with SPME and traditional chemical techniques. Food and Bioprocess Paiva, A., & Azevedo, L. (2011). Organic and conventional Coffea arabica L.: A
Technology, 5(5), 1917–1923. comparative study of the chemical composition and physiological, biochemical and
Ali, K., Maltese, F., Zyprian, E., Rex, M., Choi, Y. H., & Verpoorte, R. (2009). NMR toxicological effects in wistar rats. Plant Foods for Human Nutrition, 66(2), 114–121.
Metabolic Fingerprinting Based Identification of Grapevine Metabolites Associated Dong, T., Chen, X. J., Wang, M., Huang, Y. H., & Yi, G. J. (2014). Comparison of volatile
with Downy Mildew Resistance. Journal of Agricultural and Food Chemistry, 57(20), aroma compounds in Dwarf Cavendish banana (Musa spp. AAA) grown under organic
9599–9606. or traditional cultivation. The Journal of Horticultural Science and Biotechnology, 89(4),
Arbona, V., Iglesias, D. J., & Gómez-Cadenas, A. (2015). Non-targeted metabolite pro- 441–447.
filing of citrus juices as a tool for variety discrimination and metabolite flow analysis. Dvash, L., Afik, O., Shafir, S., Schaffer, A., Yeselson, Y., Dag, A., & Landau, S. (2002).
BMC Plant Biology, 15(1), 38. Determination by near-infrared spectroscopy of perseitol used as a marker for the
Avula, B., Smillie, T. J., Wang, Y.-H., Zweigenbaum, J., & Khan, I. A. (2015). botanical origin of avocado (Persea americana Mill.) honey. Journal of Agricultural and
Authentication of true cinnamon (Cinnamon verum) utilising direct analysis in real Food Chemistry, 50(19), 5283–5287.
time (DART)-QToF-MS. Food Additives & Contaminants: Part A, 32(1), 1–8. Eisenmann, P., Ehlers, M., Weinert, C., Tzvetkova, P., Silber, M., Rist, M., ... Muhle-Goll,
Azadmard-Damirchi, S. (2010). Review of the use of phytosterols as a detection tool for C. (2016). Untargeted NMR spectroscopic analysis of the metabolic variety of new
adulteration of olive oil with hazelnut oil. Food Additives & Contaminants: Part A, apple cultivars. Metabolites, 6(3), 29.
27(1), 1–10. Elmore, J. S., Nisyrios, I., & Mottram, D. S. (2005). Analysis of the headspace aroma
Beitlich, N., Koelling-Speer, I., Oelschlaegel, S., & Speer, K. (2014). Differentiation of compounds of walnuts (Juglans regia L.). Flavour and Fragrance Journal, 20(5),
manuka honey from kanuka honey and from jelly bush honey using HS-SPME-GC/MS 501–506.
and UHPLC-PDA-MS/MS. Journal of Agricultural and Food Chemistry, 62(27), Ercolini, D., Ferrocino, I., Nasi, A., Ndagijimana, M., Vernocchi, P., La Storia, A., ...
6435–6444. Villani, F. (2011). Monitoring of microbial metabolites and bacterial diversity in beef
Benincasa, C., De Nino, A., Lombardo, N., Perri, E., Sindona, G., & Tagarelli, A. (2003). stored under different packaging conditions. Applied and Environmental Microbiology,
Assay of aroma active components of virgin olive oils from southern Italian regions 77(20), 7372–7381.
by SPME-GC/ion trap mass spectrometry. Journal of Agricultural and Food Chemistry, Falcao, S. I., Vale, N., Gomes, P., Domingues, M. R., Freire, C., Cardoso, S. M., & Vilas-
51(3), 733–741. Boas, M. (2013). Phenolic profiling of Portuguese propolis by LC-MS spectrometry:

160
S. Medina et al. Food Chemistry 278 (2019) 144–162

Uncommon propolis rich in flavonoid glycosides. Phytochemical Analysis, 24(4), Devlieghere, F. (2018). Multivariate statistical analysis for the identification of po-
309–318. tential seafood spoilage indicators. Food Control, 84, 49–60.
Farag, M. A., Handoussa, H., Fekry, M. I., & Wessjohann, L. A. (2016). Metabolite pro- Kuuliala, L., Al Hage, Y., Ioannidis, A. G., Sader, M., Kerckhof, F. M., Vanderroost, M., ...
filing in 18 Saudi date palm fruit cultivars and their antioxidant potential via UPLC- Devlieghere, F. (2018). Microbiological, chemical and sensory spoilage analysis of
qTOF-MS and multivariate data analyses. Food & Function, 7(2), 1077–1086. raw Atlantic cod (Gadus morhua) stored under modified atmospheres. Food
Farag, M. A., Labib, R. M., Noleto, C., Porzel, A., & Wessjohann, L. A. (2018). NMR ap- Microbiology, 70, 232–244.
proach for the authentication of 10 cinnamon spice accessions analyzed via chemo- Lampir, L., & Pavlousek, P. (2013). Influence of locality on content of phenolic com-
metric tools. LWT – Food Science and Technology, 90, 491–498. pounds in white wines. Czech Journal of Food Science, 31(6), 619–626.
Farag, M. A., Mohsen, M., Heinke, R., & Wessjohann, L. A. (2014). Metabolomic finger- Le Gall, G., Puaud, M., & Colquhoun, I. J. (2001). Discrimination between orange juice
prints of 21 date palm fruit varieties from Egypt using UPLC/PDA/ESI–qTOF-MS and and pulp wash by 1H nuclear magnetic resonance spectroscopy: Identification of
GC–MS analyzed by chemometrics. Food Research International, 64, 218–226. marker compounds. Journal of Agricultural and Food Chemistry, 49(2), 580–588.
Favaro, G., Magno, F., Boaretto, A., Bailoni, L., & Mantovani, R. (2005). Traceability of Lee, K. M., Son, M., Kang, J. H., Kim, D., Hong, S., Park, T. H., ... Choi, S. S. (2018). A
Asiago mountain cheese: A rapid, low-cost analytical procedure for its identification triangle study of human, instrument and bioelectronic nose for non-destructive sen-
based on solid-phase microextraction. Journal of Dairy Science, 88(10), 3426–3434. sing of seafood freshness. Scientific Reports, 8(1), 547.
Ferreira, L., Perestrelo, R., Caldeira, M., & Camara, J. S. (2009). Characterization of vo- Li, Y., Zhang, J., Jin, H., Liu, H., & Wang, Y. (2016). Ultraviolet spectroscopy combined
latile substances in apples from Rosaceae family by headspace solid-phase micro- with ultra-fast liquid chromatography and multivariate statistical analysis for quality
extraction followed by GC-qMS. Journal of Separation Science, 32(11), 1875–1888. assessment of wild Wolfiporia extensa from different geographical origins.
Ferreira, L., Perestrelo, R., & Camara, J. (2009). Comparative analysis of the volatile Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 165, 61–68.
fraction from Annona cherimola Mill. cultivars by solid-phase microextraction and gas Lin, B., Loomes, K. M., Prijic, G., Schlothauer, R., & Stephens, J. M. (2017). Lepteridine as
chromatography–quadrupole mass spectrometry detection. Talanta, 77(3), a unique fluorescent marker for the authentication of manuka honey. Food Chemistry,
1087–1096. 225, 175–180.
Feudo, G. L., Macchione, B., Naccarato, A., Sindona, G., & Tagarelli, A. (2011). The vo- Llano, S. M., Muñoz-Jiménez, A. M., Jiménez-Cartagena, C., Londoño-Londoño, J., &
latile fraction profiling of fresh tomatoes and triple concentrate tomato pastes as Medina, S. (2018). Untargeted metabolomics reveals specific withanolides and fatty
parameter for the determination of geographical origin. Food Research International, acyl glycoside as tentative metabolites to differentiate organic and conventional
44(3), 781–788. Physalis peruviana fruits. Food Chemistry, 244, 120–127.
Figueira, J., Camara, H., Pereira, J., & Camara, J. S. (2014). Evaluation of volatile me- Longobardi, F., Innamorato, V., Di Gioia, A., Ventrella, A., Lippolis, V., Logrieco, A. F., ...
tabolites as markers in Lycopersicon esculentum L. cultivars discrimination by multi- Agostiano, A. (2017). Geographical origin discrimination of lentils (Lens culinaris
variate analysis of headspace solid phase microextraction and mass spectrometry Medik.) using (1)H NMR fingerprinting and multivariate statistical analyses. Food
data. Food Chemistry, 145, 653–663. Chemistry, 237, 743–748.
Frank, T., Reichardt, B., Shu, Q., & Engel, K.-H. (2012). Metabolite profiling of colored Lovestead, T. M., & Bruno, T. J. (2010). Detection of poultry spoilage markers from
rice (Oryza sativa L.) grains. Journal of Cereal Science, 55(2), 112–119. headspace analysis with cryoadsorption on a short alumina PLOT column. Food
Franke, C., & Beauchamp, J. (2017). Real-time detection of volatiles released during meat Chemistry, 121(4), 1274–1282.
spoilage: A case study of modified atmosphere-packaged chicken breast fillets in- Magdas, D. A., Guyon, F., Feher, I., & Pinzaru, S. C. (2018). Wine discrimination based on
oculated with Br. thermosphacta. Food Analytical Methods, 10(2), 310–319. chemometric analysis of untargeted markers using FT-Raman spectroscopy. Food
Fuentes, A., Fernández-Segovia, I., Serra, J. A., & Barat, J. M. (2010). Comparison of wild Control, 85, 385–391.
and cultured sea bass (Dicentrarchus labrax) quality. Food Chemistry, 119(4), Malheiro, R., Guedes de Pinho, P., Soares, S., da Silva, César, Ferreira, A., & Baptista, P.
1514–1518. (2013). Volatile biomarkers for wild mushrooms species discrimination. Food
Gan, H. H., Soukoulis, C., & Fisk, I. (2014). Atmospheric pressure chemical ionisation Research International, 54(1), 186–194.
mass spectrometry analysis linked with chemometrics for food classification - a case Mannina, L., Sobolev, A. P., Capitani, D., Iaffaldano, N., Rosato, M. P., Ragni, P., ...
study: Geographical provenance and cultivar classification of monovarietal clarified Coppola, R. (2008). NMR metabolic profiling of organic and aqueous sea bass ex-
apple juices. Food Chemistry, 146, 149–156. tracts: Implications in the discrimination of wild and cultured sea bass. Talanta,
García-Parrilla, M. C., Cerezo, A. B., Tesfaye, W., & Troncoso, A. M. (2011). Phenolic 77(1), 433–444.
compounds as markers for the authentication of sherry vinegars: A foresight for high Marseglia, A., Caligiani, A., Comino, L., Righi, F., Quarantelli, A., & Palla, G. (2013).
quality vinegars characterization. ACS Symposium Series, 1081, 201–213. Cyclopropyl and omega-cyclohexyl fatty acids as quality markers of cow milk and
Garrett, R., Schmidt, E. M., Pereira, L. F. P., Kitzberger, C. S., Scholz, M. B. S., Eberlin, M. cheese. Food Chemistry, 140(4), 711–716.
N., & Rezende, C. M. (2013). Discrimination of arabica coffee cultivars by electro- Martín, A., Benito, M. J., Aranda, E., Ruiz-Moyano, S., Córdoba, J. J., & Córdoba, M. G.
spray ionization Fourier transform ion cyclotron resonance mass spectrometry and (2010). Characterization by volatile compounds of microbial deep spoilage in iberian
chemometrics. LWT – Food Science and Technology, 50(2), 496–502. dry-cured ham. Journal of Food Science, 75(6), 1–10.
Granato, D., Koot, A., Schnitzler, E., & van Ruth, S. M. (2015). Authentication of geo- Miks-Krajnik, M., Yoon, Y. J., Ukuku, D. O., & Yuk, H. G. (2016). Identification and
graphical origin and crop system of grape juices by phenolic compounds and anti- quantification of volatile chemical spoilage indexes associated with bacterial growth
oxidant activity using chemometrics. Journal of Food Science, 80(3), C584–C593. dynamics in aerobically stored chicken. Journal of Food Science, 81(8), M2006–2014.
Granato, D., Putnik, P., Kovačević, D. B., Santos, J. S., Calado, V., Rocha, R. S., ... Monakhova, Y. B., Ruge, W., Kuballa, T., Ilse, M., Winkelmann, O., Diehl, B., ...
Pomerantsev, A. (2018). Trends in chemometrics: Food authentication, microbiology, Lachenmeier, D. W. (2015). Rapid approach to identify the presence of Arabica and
and effects of processing. Comprehensive Reviews in Food Science and Food Safety, Robusta species in coffee using 1H NMR spectroscopy. Food Chemistry, 182, 178–184.
17(3), 663–677. Nekvapil, F., Brezestean, I., Barchewitz, D., Glamuzina, B., Chiş, V., & Cintă Pinzaru, S.
Guo, J., Yue, T., & Yuan, Y. (2012). Feature selection and recognition from nonspecific (2018). Citrus fruits freshness assessment using Raman spectroscopy. Food Chemistry,
volatile profiles for discrimination of apple juices according to variety and geo- 242, 560–567.
graphical origin. Journal of Food Science, 77(10), C1090–C1096. Novotná, H., Kmiecik, O., Gałązka, M., Krtková, V., Hurajová, A., Schulzová, V., ...
Hallmann, E., & Rembiałkowska, E. (2012). Characterisation of antioxidant compounds in Hajšlová, J. (2012). Metabolomic fingerprinting employing DART-TOFMS for au-
sweet bell pepper (Capsicum annuum L.) under organic and conventional growing thentication of tomatoes and peppers from organic and conventional farming. Food
systems. Journal of the Science of Food and Agriculture, 92(12), 2409–2415. Additives & Contaminants: Part A, 29(9), 1335–1346.
Hu, C., Shi, J., Quan, S., Cui, B., Kleessen, S., Nikoloski, Z., ... Lin, H. (2014). Metabolic Oliveira, J. M., Faria, M., Sá, F., Barros, F., & Araújo, I. M. (2006). C 6-alcohols as varietal
variation between japonica and indica rice cultivars as revealed by non-targeted markers for assessment of wine origin. Analytica Chimica Acta, 563(1), 300–309.
metabolomics. Scientific Reports, 4, 5067. Parlapani, F. F., Haroutounian, S. A., Nychas, G.-J. E., & Boziaris, I. S. (2015).
Huang, S., Zhang, C. P., Li, G. Q., Sun, Y. Y., Wang, K., & Hu, F. L. (2014). Identification of Microbiological spoilage and volatiles production of gutted European sea bass stored
catechol as a new marker for detecting propolis adulteration. Molecules, 19(7), under air and commercial modified atmosphere package at 2 ͦC. Food Microbiology,
10208–10217. 50, 44–53.
Jumhawan, U., Putri, S. P., Yusianto, Marwani, E., Bamba, T., & Fukusaki, E. (2013). Pattamayutanon, P., Angeli, S., Thakeow, P., Abraham, J., Disayathanoowat, T., &
Selection of discriminant markers for authentication of Asian palm civet coffee (Kopi Chantawannakul, P. (2017). Volatile organic compounds of Thai honeys produced
Luwak): A metabolomics approach. Journal of Agricultural and Food Chemistry, 61(33), from several floral sources by different honey bee species. PLoS One, 12(2),
7994–8001. e0172099.
Khalil, M. N., Fekry, M. I., & Farag, M. A. (2017). Metabolome based volatiles profiling in Pedan, V., Weber, C., Do, T., Fischer, N., Reich, E., & Rohn, S. (2017). HPTLC fingerprint
13 date palm fruit varieties from Egypt via SPME GC–MS and chemometrics. Food profile analysis of cocoa proanthocyanidins depending on origin and genotype. Food
Chemistry, 217, 171–181. Chemistry.
Kim, J., Jung, Y., Song, B., Bong, Y. S., Ryu, D. H., Lee, K. S., & Hwang, G. S. (2013). Perestrelo, R., Barros, A. S., Rocha, S. M., & Câmara, J. S. (2014). Establishment of the
Discrimination of cabbage (Brassica rapa ssp. pekinensis) cultivars grown in different varietal profile of Vitis vinifera L. grape varieties from different geographical regions
geographical areas using (1)H NMR-based metabolomics. Food Chemistry, 137(1–4), based on HS-SPME/GC-qMS combined with chemometric tools. Microchemical
68–75. Journal, 116, 107–117.
Kim, J. M., Kim, H. J., & Park, J. M. (2015). Determination of milk fat adulteration with Perestrelo, R., Silva, C., & Câmara, J. S. (2014). A useful approach for the differentiation
vegetable oils and animal fats by gas chromatographic analysis. Journal of Food of wines according to geographical origin based on global volatile patterns. Journal of
Science, 80(9), C1945–1951. Separation Science, 37(15), 1974–1981.
Klockmann, Reiner, E., Bachmann, R., Hackl, T., & Fischer, M. (2016). Food finger- Pisano, P. L., Silva, M. F., & Olivieri, A. C. (2015). Anthocyanins as markers for the
printing: Metabolomic approaches for geographical origin discrimination of classification of Argentinean wines according to botanical and geographical origin.
Hazelnuts (Corylus avellana) by UPLC-QTOF-MS. Journal of Agricultural and Food Chemometric modeling of liquid chromatography-mass spectrometry data. Food
Chemistry, 64(48), 9253–9262. Chemistry, 175, 174–180.
Kuuliala, L., Abatih, E., Ioannidis, A. G., Vanderroost, M., De Meulenaer, B., Ragaert, P., & Pontes, M., Pereira, J., & Câmara, J. S. (2012). Dynamic headspace solid-phase

161
S. Medina et al. Food Chemistry 278 (2019) 144–162

microextraction combined with one-dimensional gas chromatography–mass spec- Tompa, G., Susic, R., Rogelj, I., & Pompe, M. (2013). Cryotrap/SPME/GC/MS method for
trometry as a powerful tool to differentiate banana cultivars based on their volatile profiling of monoterpenes in cheese and their clustering according to geographic
metabolite profile. Food Chemistry, 134(4), 2509–2520. origin. Acta Chimica Slovenica, 60(3), 595–603.
Porep, J. U., Walter, R., Kortekamp, A., & Carle, R. (2014). Ergosterol as an objective Trivedi, D. K., Hollywood, K. A., Rattray, N. J., Ward, H., Trivedi, D. K., Greenwood, J., ...
indicator for grape rot and fungal biomass in grapes. Food Control, 37, 77–84. Goodacre, R. (2016). Meat, the metabolites: An integrated metabolite profiling and
Porto-Figueira, P., Freitas, A., Cruz, C. J., Figueira, J., & Câmara, J. S. (2015). Profiling of lipidomics approach for the detection of the adulteration of beef with pork. Analyst,
passion fruit volatiles: An effective tool to discriminate between species and varieties. 141(7), 2155–2164.
Food Research International, 77, 408–418. Vallverdú-Queralt, A., Medina-Remón, A., Casals-Ribes, I., Amat, M., & Lamuela-
Povolo, M., Pelizzola, V., Ravera, D., & Contarini, G. (2009). Significance of the non- Raventós, R. M. (2011). A Metabolomic Approach Differentiates between
volatile minor compounds of the neutral lipid fraction as markers of the origin of Conventional and Organic Ketchups. Journal of Agricultural and Food Chemistry,
dairy products. Journal of Agricultural and Food Chemistry, 57(16), 7387–7394. 59(21), 11703–11710.
Righetti, L., Rubert, J., Galaverna, G., Hurkova, K., Dall'Asta, C., Hajslova, J., & Stranska- van Ruth, S., Alewijn, M., Rogers, K., Newton-Smith, E., Tena, N., Bollen, M., & Koot, A.
Zachariasova, M. (2018). A novel approach based on untargeted lipidomics reveals (2011). Authentication of organic and conventional eggs by carotenoid profiling.
differences in the lipid pattern among durum and common wheat. Food Chemistry, Food Chemistry, 126(3), 1299–1305.
240, 775–783. Vasta, V., Luciano, G., Dimauro, C., Rohrle, F., Priolo, A., Monahan, F. J., & Moloney, A.
Risticevic, S., Carasek, E., & Pawliszyn, J. (2008). Headspace solid-phase microextraction- P. (2011). The volatile profile of longissimus dorsi muscle of heifers fed pasture,
gas chromatographic-time-of-flight mass spectrometric methodology for geographical pasture silage or cereal concentrate: Implication for dietary discrimination. Meat
origin verification of coffee. Analytica Chimica Acta, 617(1–2), 72–84. Science, 87(3), 282–289.
Röhlig, R. M., & Engel, K.-H. (2010). Influence of the input system (Conventional versus Vetter, W., & Schröder, M. (2010). Concentrations of phytanic acid and pristanic acid are
Organic Farming) on metabolite profiles of maize (Zea mays) Kernels. Journal of higher in organic than in conventional dairy products from the German market. Food
Agricultural and Food Chemistry, 58(5), 3022–3030. Chemistry, 119(2), 746–752.
Rubert, J., Lacina, O., Zachariasova, M., & Hajslova, J. (2016). Saffron authentication Vita, F., Taiti, C., Pompeiano, A., Bazihizina, N., Lucarotti, V., Mancuso, S., & Alpi, A.
based on liquid chromatography high resolution tandem mass spectrometry and (2015). Volatile organic compounds in truffle (Tuber magnatum Pico): Comparison of
multivariate data analysis. Food Chemistry, 204, 201–209. samples from different regions of Italy and from different seasons. Scientific Reports, 5,
Schievano, E., Morelato, E., Facchin, C., & Mammi, S. (2013). Characterization of markers 12629.
of botanical origin and other compounds extracted from unifloral honeys. Journal of Wan, H., Wu, L., Yang, Y., Zhou, G., & Ruan, Y. L. (2018). Evolution of sucrose meta-
Agricultural and Food Chemistry, 61(8), 1747–1755. bolism: The dichotomy of invertases and beyond. Trends in Plant Science, 23(2),
Segato, S., Galaverna, G., Contiero, B., Berzaghi, P., Caligiani, A., Marseglia, A., & Cozzi, 163–177.
G. (2017). Identification of lipid biomarkers to discriminate between the different Wang, S. Y., Chen, C.-T., Sciarappa, W., Wang, C. Y., & Camp, M. J. (2008). Fruit quality,
production systems for asiago PDO cheese. Journal of Agricultural and Food Chemistry, antioxidant capacity, and flavonoid content of organically and conventionally grown
65(45), 9887–9892. blueberries. Journal of Agricultural and Food Chemistry, 56(14), 5788–5794.
Shepherd, L. V. T., Hackett, C. A., Alexander, C. J., Sungurtas, J. A., Pont, S. D. A., Weesepoel, Y., Heenan, S., Boerrigter-Eenling, R., Venderink, T., Blokland, M., & van
Stewart, D., ... Davies, H. V. (2014). Effect of agricultural production systems on the Ruth, S. (2016). Protocatechuic acid levels discriminate between organic and con-
potato metabolome. Metabolomics, 10(2), 212–224. ventional wheat from Denmark. CHIMIA International Journal for Chemistry, 70(5),
Son, H.-S., Hwang, G.-S., Kim, K. M., Ahn, H.-J., Park, W.-M., Van Den Berg, F., ... Lee, C.- 360–363.
H. (2009). Metabolomic studies on geographical grapes and their wines using 1H Wei, F., Furihata, K., Koda, M., Hu, F., Kato, R., Miyakawa, T., & Tanokura, M. (2012).
NMR analysis coupled with multivariate statistics. Journal of Agricultural and Food 13C NMR-based metabolomics for the classification of green coffee beans according
Chemistry, 57(4), 1481–1490. to variety and origin. Journal of Agricultural and Food Chemistry, 60(40),
Souard, F., Delporte, C., Stoffelen, P., Thévenot, E. A., Noret, N., Dauvergne, B., ... 10118–10125.
Stévigny, C. (2018). Metabolomics fingerprint of coffee species determined by un- Wierda, R. L., Fletcher, G., Xu, L., & Dufour, J. P. (2006). Analysis of volatile compounds
targeted-profiling study using LC-HRMS. Food Chemistry, 245, 603–612. as spoilage indicators in fresh king salmon (Oncorhynchus tshawytscha) during storage
Spanik, I., Pazitna, A., Siska, P., & Szolcsanyi, P. (2014). The determination of botanical using SPME-GC-MS. Journal of Agricultural and Food Chemistry, 54(22), 8480–8490.
origin of honeys based on enantiomer distribution of chiral volatile organic com- Willems, J. L., & Low, N. H. (2018). Structural identification of compounds for use in the
pounds. Food Chemistry, 158, 497–503. detection of juice-to-juice debasing between apple and pear juices. Food Chemistry,
Splivallo, R., Bossi, S., Maffei, M., & Bonfante, P. (2007). Discrimination of truffle fruiting 241, 346–352.
body versus mycelial aromas by stir bar sorptive extraction. Phytochemistry, 68(20), Xue, X., Wang, Q., Li, Y., Wu, L., Chen, L., Zhao, J., & Liu, F. (2013). 2-acetylfuran-3-
2584–2598. glucopyranoside as a novel marker for the detection of honey adulterated with rice
Stanimirova, I., Üstün, B., Cajka, T., Riddelova, K., Hajslova, J., Buydens, L. M. C., & syrup. Journal of Agricultural and Food Chemistry, 61(31), 7488–7493.
Walczak, B. (2010). Tracing the geographical origin of honeys based on volatile Yahagi, T., Masada, S., Oshima, N., Suzuki, R., Matsufuji, H., Takahashi, Y., ...
compounds profiles assessment using pattern recognition techniques. Food Chemistry, Hakamatsuka, T. (2016). Determination and Identification of a specific marker
118(1), 171–176. compound for discriminating shrub chaste tree fruit from agnus castus fruit based on
Sun, X., Li, L., Ma, T., Liu, X., Huang, W., & Zhan, J. (2015). Profiles of phenolic acids and LC/MS metabolic analysis. Chemical and Pharmaceutical Bulletin, 64(4), 305–310.
flavan-3-ols for select chinese red wines: A comparison and differentiation according Yang, N., Liu, C., Liu, X., Degn, T. K., Munchow, M., & Fisk, I. (2016). Determination of
to geographic origin and grape variety. Journal of Food Science, 80(10). volatile marker compounds of common coffee roast defects. Food Chemistry, 211,
Tette, P. A., Guidi, L. R., Bastos, E. M., Fernandes, C., & Gloria, M. B. (2017). Synephrine – 206–214.
A potential biomarker for orange honey authenticity. Food Chemistry, 229, 527–533. Zörb, C., Langenkämper, G., Betsche, T., Niehaus, K., & Barsch, A. (2006). Metabolite
Thavarajah, P., & Low, N. H. (2006). Adulteration of apple with pear juice: Emphasis on profiling of wheat grains (Triticum aestivum L.) from organic and conventional agri-
major carbohydrates, proline, and arbutin. Journal of Agricultural and Food Chemistry, culture. Journal of Agricultural and Food Chemistry, 54(21), 8301–8306.
54(13), 4861–4867.

162

Potrebbero piacerti anche