Analyst: Volume 144 Number 14 21 July 2019 Pages 4125-4448
Analyst: Volume 144 Number 14 21 July 2019 Pages 4125-4448
Analyst
rsc.li/analyst
ISSN 0003-2654
PAPER
José S. Câmara et al.
Volatomic pattern of breast cancer and cancer-free tissues
as a powerful strategy to identify potential biomarkers
Analyst
PAPER
Breast cancer (BC), ranked as the fifth amongst all cancers, remains at the top of women’s cancers
worldwide followed by colorectal, lung, cervix, and stomach cancers. The main handicap of most of the
screening/diagnostic methods is based on their low sensitivity and specificity and the invasive behavior
of most sampling procedures. The aim of this study was to establish the volatomic pattern of BC and
cancer-free (CF) tissues (n = 30) from the same patients, as a powerful tool to identify a set of volatile
organic metabolite (VOM) potential BC biomarkers which might be used together or complement with
the traditional BC diagnostics strategies, through the integration of chromatographic data, obtained by
solid-phase microextraction followed by gas chromatography-mass spectrometry (SPME/GC-qMS),
with chemometric tools. A total of four metabolites: limonene, decanoic acid, acetic acid and furfural
presented the highest contribution towards discrimination of BC and CF tissues (VIP > 1, p < 0.05). The
discrimination efficiency and accuracy of BC tissue metabolites was ascertained by ROC curve analysis
that allowed the identification of some metabolites with high sensitivity and specificity. The results
Received 7th February 2019, obtained with this approach suggest the possibility of identifying endogenous metabolites as a platform
Accepted 11th May 2019
to find potential BC biomarkers and pave the way to investigate the related metabolomic pathways in
DOI: 10.1039/c9an00263d order to improve BC diagnostic tools. Moreover, deeper investigations could unravel novel mechanistic
rsc.li/analyst insights into the disease pathophysiology.
This journal is © The Royal Society of Chemistry 2019 Analyst, 2019, 144, 4153–4161 | 4153
Paper Analyst
coverage in BC tissues can be maximized by combining Sigma Aldrich (St Louis, MO, USA), respectively. The C8–C20
different technologies for metabolic profiling, namely gas alkane solution (concentration of 40 mg L−1 in n-hexane) was
chromatography mass-spectrometry (GC-qMS), and the results purchased from Fluka (Buchs, Switzerland). The digital stir-
can be used to classify BC helping to identify new prognostic ring plate (Cimarec™) was supplied by Thermo Scientific
and predictive markers and to discover new targets for future (Waltham, MA, USA) while the SPME holder for manual
therapeutic interventions.9 Among them, the study of volatile sampling and 75 µm Carboxen/polydimethylsiloxane fiber
organic metabolites (VOMs) present in biological samples, (CAR/PDMS) were purchased from Supelco (Bellefonte, PA, USA).
namely in saliva, urine, breath and tissues can be useful for
cancer diagnosis, in particular for BC. The most common pro- Subjects and tissue collection
cedure used in the extraction of volatile compounds in biologi-
cal samples is solid-phase microextraction (SPME), normally in To investigate the BC tissue metabolomic profile, 30 samples
the headspace mode (HS-SPME) being used in several biologi- from patients with breast cancer (BC, n = 30, age range 44–85,
cal matrices.10–13 GC-qMS was used to screen salivary volatiles average 67), and 30 samples from cancer-free tissues (CF, n =
for putative BC as an exploratory study involving geographi- 30, age range 44–85, average 67) without malignant infiltration
cally distant populations,14 also to establish the metabolomic were resected from each patient. The resected samples were
signature of human BC cell lines11 and to discriminate divided into the active carcinoma and cancer-free tissues
different types of cancer based on urinary volatomic biosigna- outside the tumor margin and were immediately frozen in
tures.13 In the first study, up to 120 VOMs from distinct chemi- liquid nitrogen, in a total set of 60 samples. The tissues were
cal families, with significant variations among the groups, stored at −80 °C until extraction. These samples were obtained
were identified,14 whereas Silva et al.11 and Porto-Figueira at the Pathological Anatomy Unit of Hospital Dr Nélio
et al.13 identified 60 and 130 VOMs in BC cell lines and urine, Mendonça (Funchal, Portugal) according to Table 1. All experi-
respectively. On the other hand, Budczies et al.15,16 used a gas ments were performed in accordance with the standard
chromatography time of flight mass spectrometry (GC-TOFMS) Guidelines from Declaration of Helsinki and approved by the
framework to evaluate the glutamate enrichment as a new institutional ethics committee of Hospital Dr Nélio Mendonça
diagnostic opportunity in BC and to accomplish comparative and University of Madeira. All the participants of this study
metabolomics of estrogen receptor positive (ER+) and estrogen were informed about the investigation and a signed informed
receptor negative (ER−) samples. Budczies et al.16 identified 19 consent was obtained from all human participants prior to
VOMs and the GC-TOFMS based analysis of metabolites sample collection. Using the TNM (tumor, node, and meta-
present in BC tissues revealed significant differences in central stasis) staging approach, the examined BC cases included five
metabolism in the more aggressive ER− compared to the ER+ of stage IA, ten of stage IIA, one of stage IIIA B, seven of stage
type. The detected changes included the metabolism of gluta- IIB, five of stage IIIB and two of stage IIIC.
mine with a decrease in the concentration of glutamine and
an increase in the concentration of glutamate and 2-hydroxy- Extraction of metabolites from breast tissues
glutaric acid.16 In turn, Dougan et al.8 used GC-MS to evaluate The HS-SPME extraction conditions were based on a developed
the detectability, reliability, and distribution of metabolites method previously established in our laboratory.10,12 Briefly,
measured in pre-diagnostic plasma samples in a pilot study of tissue samples were thawed and then portions of 100 mg were
women listed in the Northern California site of the BC Family weighed into 20 mL vials together with 17% NaCl (w/v),
Registry. In this study, 661 VOMs were detected, 338 (51%) of 1000 µL of ultrapure water and 100 µL of the internal standard
them were found in all samples, and 490 (74%) in more than (IS, 4-methyl-2-pentanol, 1.6 mg L−1). The pH was adjusted to
80% of samples. The aim of this study was to establish the 2 with small amounts of 5 M HCl. Then, the vial was capped
volatomic pattern of BC tissue and CF tissue samples collected with a Teflon (PTFE) septum using a screw cap and the SPME
after surgery, from the same patient (to minimize the inter- fiber was introduced and exposed into the headspace over
ference of epigenetic and external factors), in order to find a 75 min at 50 °C at 800 rpm (0.5 mm × 0.1 mm bar). After this
set of volatile metabolites to be used as potential BC bio- period, the fiber was removed from the vial and inserted into
markers, using HS-SPME/GC-qMS combined with multivariate the GC injection port and the VOMs extracted were desorbed
statistical tools. This high-throughput strategy might have the
potential to be applied in a clinical environment as a diagnos-
Table 1 List of collected tissue samples from breast cancer patients
tic approach or as a complementary way with the current diag-
nostic methods to improve the diagnostic decision. Samples BC tissue Cancer-free tissue
Number 30 30
Age (range, median) (44–85, 65) (44–85, 65)
Experimental Histological grade IA (5) Not applicable
(number of samples) IIA (10)
Materials and reagents IIIA (1)
IIB (7)
Sodium chloride (NaCl), hydrochloric acid (HCl) and 4-methyl- IIIB (5)
2-pentanol were supplied by Panreac (Barcelona, Spain) and IIIC (2)
4154 | Analyst, 2019, 144, 4153–4161 This journal is © The Royal Society of Chemistry 2019
Analyst Paper
for 10 min at 250 °C. Each sample was analyzed in duplicate Finally, the selected metabolites were used for the metabolic
and blanks were performed before each analysis. pathway analysis to identify the most relevant metabolic path-
ways involved in BC.
Gas chromatography quadrupole mass-spectrometry (GC-qMS)
conditions
After the extraction procedure, the SPME fiber with the ana- Results and discussion
lytes was inserted into the injection port of an Agilent
Technologies 6890N Network gas chromatograph system (Palo Tissue metabolomic pattern based on GC-qMS
Alto, CA, USA) where the VOMs were desorbed at 250 °C for A total of twenty-nine VOMs were identified in BC tissue and
10 min. The gas chromatograph was equipped with a 60 m × CF tissue samples which were classified in several chemical
0.25 mm I.D. × 0.25 µm film thickness, BP-20 (SGE, families, namely phenols, benzene derivatives, carbonyl com-
Dortmund, Germany) fused silica capillary column and inter- pounds, acids, alcohols and furanic compounds as shown in ESI
faced with an Agilent 5975 quadrupole inert mass selective Fig. S1.† Data were processed using software (NIST, 2005; Mass
detector. The following oven temperature profile was set: (a) Spectral Search Program V.2.0d) which provides quality matching
5 min at 45 °C; (b) temperature was increased to 150 °C, at using advanced spectral matching algorithms background sub-
a rate of 2 °C min−1 (hold for 10 min); (c) 150 °C for 10 min; traction and KI comparison. The KI were determined through
(d) temperature was increased to 220 °C, at a rate of the injection of a C8–C20 alkane solution in order to confirm the
7 °C min−1; and (e) 220 °C for 10 min for a total GC run identity of the VOMs by comparison with the literature (e.g. pher-
time of 87.5 min. The column flow was constant at obase). It can be observed in Table 2 that the KI of the identified
1.3 mL min−1 using helium (He, N60, Air Liquide, Portugal) as VOMs, when compared to the ones found in the literature for a
a carrier gas. The injection port was operated in the splitless similar column (BP-20), are closer to one another, ensuring that
mode and held at 250 °C. For the 5975 MS system, the operat- a good identification was achieved.
ing temperatures of the transfer line, quadrupole and ioniza- The highest contribution for the volatile profile was from
tion source were 270, 150 and 230 °C, respectively, while elec- phenols in the BC group, whereas for the CF group it was from
tron impact mass spectra were recorded at 70 eV ionization benzene derivatives and acids. The main VOMs identified for
voltage and the ionization current was 10 µA. Data acquisition these families were phenol, toluene and acetic acid, respect-
was performed in the scan mode (30–200 m/z). The electron ively. These metabolites were already identified in several
multiplier was set to the autotune procedure. Metabolite reports in the literature in biological matrices such as
identification was accomplished by manual interpretation urine,10,12 cancer cell lines,11,18 saliva,14 and exhaled
through single ion monitoring (SIM) of spectra and matching breath.18,19 Several studies reported the potential of VOMs
against the Agilent MS ChemStation Software, equipped with a from different biological matrices on differentiation between
NIST05 mass spectral library with a similarity threshold higher cancer patients and healthy subjects. In addition, different
than 80% and comparison with commercially available stan- endogenous metabolites were identified as potential cancer
dard samples when available. A series of C8–C20 n-alkanes biomarkers which might be used in cancer diagnosis. These
were analyzed using the same extraction procedure to establish metabolites can be analysed through different analytical plat-
the Kovats indices (KI), and to confirm the identity of the forms including GC-MS, LC-MSMS and NMR. Table 2 shows
VOMs by comparison with the literature. The analyses were the identification of VOMs, as well as the minimum and
performed in duplicate and the results are expressed by means maximum relative GC peak areas for each metabolite and
± standard deviation. group. It can be observed that most metabolites were identi-
fied in all samples with a frequency of occurrence (FO) higher
Statistical analysis than 90%, where the maximum relative area was obtained for
Statistical analysis was performed using the web server p-tert-butyl-phenol in the CF group and phenol for the BC
Metaboanalyst 4.0.17 The multivariate statistical analysis, group. For the CF group, the major VOMs identified include
namely the principal component analysis (PCA), the partial acetic acid, phenol and p-tert-butyl-phenol.
least squares-discriminant analysis (PLS-DA) and the orthog- Some of these have been already reported in the literature,
onal projections to latent structures discriminant analysis namely acetic acid was found as a discriminant when associ-
(OPLS-DA) were applied on the tissue metabolomic profile ated with controls (CTL) in a study conducted by Ahmed
dataset to provide insights into the groups under study. The et al.20 that investigated the possibility of faecal VOMs as
metabolites with VIP scores higher than 1.0 selected by the potential diagnostic biomarkers for inflammatory bowel
PLS-DA analysis were used for the pathway analysis. disease. Mochalski et al.21 investigated the emission of volatile
Furthermore, hierarchical cluster analysis by K-means of the compounds from gastric cancer tissues and non-cancerous
two groups in study was carried out and Pearson’s correlation tissues with the aim of identifying characteristic chemical pat-
was used to build the heat map with the aim of identifying BC terns associated with gastric cancer. Furthermore, Silva
clustering patterns. The receiver operating characteristic (ROC) et al.10,12 also reported phenols as the major chemical family
curves were attained to verify which metabolites had the identified in urine from the oncologic group. Furthermore,
highest sensitivity/specificity for a potential BC diagnosis. Raman et al.22 studied faecal VOMs in obese humans and
This journal is © The Royal Society of Chemistry 2019 Analyst, 2019, 144, 4153–4161 | 4155
Paper Analyst
Table 2 Identified metabolites of tissue samples from BC patients and cancer-free (CF) through GC-qMS, minimum (Min) and maximum (Max), vari-
ation of relative peak areas (relative to internal standard, RSD < 10%) regarding the BC group and their frequency of occurrence (in %). Identification
mode and the Kovats index for each identified VOM and the literature values for a similar GC column
CF BC Frequency
of occurrence
Peak no. IECa RTb (min) IDc KIcal d KIlit e VOM f Min Max Min Max Variation (FO)/%
A1 41, 69 9.77 MS 990 983 Methyl isobutyrate 0.03 1.81 0.02 1.66 ↓ 71
A2 91 11.24 Std, MS 1029 1042 Toluene 0.03 7.62 0.01 17.17 ↑ 100
A3 44, 56 13.12 Std, MS 1073 1075 Hexanal 0.03 6.82 0.04 2.09 ↓ 99
A7 41, 70 18.48 Std, MS 1172 1168 Heptanal 0.04 0.17 0.05 0.19 ↑ 13
A8 68, 93 18.73 St, MS 1176 1198 Limonene 0.01 4.46 0.01 1.03 ↓ 92
A10 43, 33 19.84 St, MS 1187 — 2-Methyl-1-propanol 0.01 1.44 0.01 0.51 ↓ 79
A11 81 20.69 MS 1197 1229 2-Pentyl-furan 0.01 0.07 0.01 0.13 ↑ 77
A12 105, 120 21.39 MS 1209 — Trimethylbenzene 0.01 5.41 0.01 1.70 ↓ 64
A13 119, 134 23.18 St, MS 1237 1234 o-Cymene 0.01 0.84 0.01 0.33 ↓ 24
A14 42, 55 24.36 MS 1254 1255 1-Pentanol 0.01 0.03 0.01 0.23 ↑ 50
A15 43, 56 24.90 St, MS 1262 1280 Octanal 0.01 0.24 0.01 9.19 ↑ 34
A16 56 30.20 St, MS 1338 1360 1-Hexanol 0.01 0.05 0.01 1.17 ↑ 27
A17 41, 57 31.89 St, MS 1362 1385 Nonanal 0.02 0.48 0.17 2.36 ↑ 52
A18 117, 132 34.45 MS 1396 — p-Cymenene 0.01 0.76 0.01 0.15 ↓ 100
A19 43, 60 36.42 St, MS 1428 1450 Acetic acid 0.62 42.75 0.10 12.28 ↓ 100
A20 96 36.61 St, MS 1431 1455 Furfural 0.01 5.87 0.04 18.03 ↑ 100
A21 57 38.53 St, MS 1461 1487 2-Ethyl-1-hexanol 0.01 0.50 0.01 3.20 ↑ 2
A22 43, 57 38.77 St, MS 1465 1484 Decanal n.d.g n.d. 0.14 0.15 — 72
A23 77, 106 39.81 St, MS 1480 1495 Benzaldehyde 0.01 0.36 0.01 0.33 ↓ 54
A24 41, 56 42.63 MS 1525 1553 1-Octanol 0.01 0.17 0.01 1.57 ↑ 74
A25 77, 105 47.36 MS 1602 1607 Acetophenone 0.01 0.19 0.01 0.42 ↑ 79
A26 60, 73 58.70 St, MS 1802 1829 Hexanoic acid 0.01 1.25 0.02 5.77 ↑ 100
A27 94 69.54 St, MS 1968 1965 Phenol 0.51 23.32 0.45 44.08 ↑ 39
A28 60, 73 72.22 St, MS 2004 2013 Octanoic acid 0.01 0.25 0.01 6.51 ↑ 76
A29 107 73.00 MS 2009 2017 4-Methyl-phenol 0.01 1.08 0.01 3.72 ↑ 45
A30 60, 73 75.65 St, MS 2027 — Nonanoic acid 0.01 0.18 0.01 0.64 ↑ 29
A31 155, 170 75.71 MS 2027 — 1,6,7-Trimethyl-naphthalene 0.02 0.41 0.01 0.37 ↓ 39
A32 60, 73 78.15 St, MS 2043 — Decanoic acid 0.01 0.11 0.02 0.52 ↑ 100
A33 135, 150 78.49 St, MS 2045 — p-tert-Butyl-phenol 0.02 53.76 0.03 20.71 ↓ 100
a
IEC – ion extraction chromatogram. b Retention time (min). c ID – metabolite identification using a standard compound (St) or mass spectra of
the NIST library search (MS). d Kovats index relative to n-alkanes (C8–C20) on a BP-20 capillary column. e Kovats index relative to that reported in
the literature for an equivalent capillary column. f VOM – volatile organic metabolite. g n.d. – not detected.
identified also acetic acid and phenol as major metabolites. On where the samples are classified into different groups (ESI
the other hand, P. Porto-Figueira et al.13 also identified p-xylene, Fig. S2A†). Then, the top four metabolites (limonene, decanoic
o-xylene, acetic acid, phenol and p-tert-butyl-phenol in urine acid, acetic acid and furfural) with the highest contribution for
from cancer patients, with some of them presenting higher group discrimination were selected with variable importance
values in cancer patients. Regarding the relative area values in projection (VIP > 1) (ESI Fig. S2B†). Furthermore, Fig. 1
obtained for the BC group, most VOMs were down-regulated shows the comparison of three illustrative VOMs with a higher
with the exception of acetic acid, toluene, octanal and nonanal. VIP score from BC tissue and CF tissue. It can be observed that
for a majority of metabolites their level is higher in the CF
Multivariate statistical analysis of tissue metabolomic profile group rather than in the BC group. Moreover, orthogonal
The statistical analysis was performed using the Metaboanalyst partial least squares-discriminant (OPLS-DA) analysis was
4.0 17 web server as described in the Experimental section. applied on the tissue metabolomic profile dataset to maximize
Only the VOMs with FO higher than 90% were considered for the separation of BC and CF groups. Significant group separ-
the statistical analysis, in a total of 8 VOMs (toluene, hexanal, ation was observed in the OPLS-DA score plot between BC
limonene, p-cymenene, acetic acid, furfural, hexanoic acid and patients and CF group indicating intrinsic metabolic altera-
decanoic acid). Initially, data were transformed by log trans- tions in each group (Fig. 2A). To attest the robustness of the
formation and mean centering approaches, before being sub- model, a random permutation test with 1000 permutations
jected to multivariate statistical analysis. Partial least squares- was performed with OPLS-DA (Fig. 2B). The permutation test
discriminant analysis (PLS-DA) was used as a supervised clus- yielded R2 (represents goodness of fit) as 0.717 and Q2 (rep-
tering method to verify the existence of an altered metabolite resents predictive ability) as 0.691 indicating that the model is
pattern. Additionally, this type of statistical analysis takes into not over fitted and has a relative good predictive ability to dis-
account the variance/covariance between samples of groups tinguish between study groups.
4156 | Analyst, 2019, 144, 4153–4161 This journal is © The Royal Society of Chemistry 2019
Analyst Paper
Fig. 2 Loading score plots of orthogonal projection to latent structure discriminant analysis (OPLS-DA) and (B) model validation by the permutation
test based on 1000 permutations of VOMs obtained by GC-qMS of tissue samples from the 2 groups under study.
This journal is © The Royal Society of Chemistry 2019 Analyst, 2019, 144, 4153–4161 | 4157
Paper Analyst
Fig. 3 Receiver operating characteristic (ROC) curves for the predictive model. (A) A combination metabolites model calculated from the logistic
regression analysis using the 4 metabolites selected by the VIP (>1.0) values, (B) ROC curve for the top 4 metabolites (limonene, decanoic acid,
acetic acid and furfural) with the highest ability to discriminate BC patients against the CF.
Fig. 4 (A) Plot of the predicted class probabilities for all samples using the OPLS-DA biomarker model based on AUC and (B) box plot of the predic-
tive accuracy (with an average of 0.908) of the biomarker model based on 100 cross validations.
their respective groups with the exception of O10, O5 and O18 most important pathways with the highest impact on BC. As
with a probability score ranging from 0.982 to 0.595. shown, sulfur and pyruvate metabolisms were the most impor-
Also, the heat map was constructed with selected VOMs by tant metabolic pathways explaining the BC carcinogenesis. In
VIP > 1, using Pearson’s correlation, providing intuitive visual- Fig. 6B we can observe the most important metabolomics
ization of the data set and the correlations between samples routes producing the most significant volatile metabolites in
and VOMs (Fig. 5). BC (indicated by KEGG accession number).
Finally, the metabolic pathway analysis was performed It can be observed that the pathway with the highest impact
using the metabolites with VIP values higher than 1 in order was the pyruvate and sulfur metabolism due to acetic acid.
to explore which pathways changed in BC. Fig. 6A shows the This metabolite according to the human metabolome database
4158 | Analyst, 2019, 144, 4153–4161 This journal is © The Royal Society of Chemistry 2019
Analyst Paper
Fig. 5 Heat map visualization and hierarchical clustering analysis using the four metabolites with significance ( p < 0.05) by Pearson’s distance ana-
lysis (A8 – limonene; A19 – acetic acid; A32 – decnoic acid: A20 – furfural).
Fig. 6 (A) The metabolome view map of significant altered metabolic pathways observed in tissue samples from BC and CF groups and (B) the pyru-
vate metabolism. The map was generated using the reference map by KEGG (Homo sapiens); C00033 represents acetic acid.
is normally found in most tissues (liver, kidney, among others) CF tissues using GC-qMS combined with multivariate statisti-
and in several biofluids, namely saliva14 and urine.10 cal tools (PLS-DA and OPLS-DA). Twenty-nine metabolites were
Moreover, limonene was also included as a significant metab- identified and multivariate statistical analysis revealed some
olite belonging to the class of monoterpenoids involved in metabolites significantly altered in BC patients. Of the metab-
monooxygenase activity through cytochrome P450 and the olites identified, limonene, decanoic acid, acetic acid and fur-
mevalonate pathway.32,33 fural showed the highest sensitivity and specificity to discrimi-
Based on the results obtained, a successful discrimination nate BC patients. Also, the analysis of the plots leads to a
of tissue samples was achieved, according to the group metabolomic pattern comprising an array of some biochemical
showing that the volatomic tissue profile can be a useful pathways altered in BC patients. The metabolic pathway ana-
approach to identify potential BC biomarkers. lysis indicated that the discriminatory metabolites could orig-
inate from several dysregulated pathways in BC such as those
involved in pyruvate and sulphur metabolism, and limonene
Conclusions degradation.
In addition, the analysis of cancer and non-cancer tissues
This study has enabled the untargeted assessment of the meta- from the same subject can also aid in balancing the effect of
bolomic tissue profile from BC patients when compared with external interferents, such as diet or environmental exposure
This journal is © The Royal Society of Chemistry 2019 Analyst, 2019, 144, 4153–4161 | 4159
Paper Analyst
and help in identifying potential biomarkers. These results ovarian and breast cancers, Clin. Cancer Res., 2010, 16,
suggest a possibility of identifying endogenous metabolites as 5835–5841.
a platform to discover potential BC biomarkers and pave the 4 L. Lavra, A. Catini, A. Ulivieri, R. Capuano, L. Baghernajad
way to investigate the related metabolomic pathways to Salehi, S. Sciacchitano, A. Bartolazzi, S. Nardis, R. Paolesse,
improve the diagnostic tools of BC. E. Martinelli, et al., Investigation of VOCs associated with
different characteristics of breast cancer cells, Sci. Rep.,
2015, 5, 13246.
Author contributions 5 A. McCartney, A. Vignoli, L. Biganzoli, R. Love, L. Tenori,
C. Luchinat and A. Di Leo, Metabolomics, in breast cancer:
Catarina Silva: Conception of studies and experimental design, A decade in review, Cancer Treat. Rev., 2018, 67, 88–96.
BC and CF tissue extractions, sample analysis by HS-SPME/GC- 6 M. M. Koek, R. H. Jellema, J. van der Greef, A. C. Tas and
qMS, data analysis and manuscript preparation. Rosa T. Hankemeier, Quantitative metabolomics based on gas
Perestrelo: Design of experiments and data analysis. Pedro chromatography mass spectrometry: status and perspec-
Silva: Design of experiments, statistical analysis and interpret- tives, Metabolomics, 2011, 7, 307–328.
ation. Filipa Capelinha: Histological classification of BC 7 Y. Huang, Y. Li, Z. Luo and Y. Duan, Investigation of bio-
tissues. Helena Tomás: Co-supervisor of the work, conception markers for discriminating breast cancer cell lines from
of studies, experimental design, and manuscript preparation. normal mammary cell lines based on VOCs analysis and
José S. Câmara: Main supervisor of the work, conception of metabolomics, RSC Adv., 2016, 6, 41816–41824.
studies, experimental design, and manuscript preparation. 8 M. M. Dougan, Y. Li, L. W. Chu, R. W. Haile,
A. S. Whittemore, S. S. Han, S. C. Moore, J. N. Sampson,
I. L. Andrulis, E. M. John, et al., Metabolomic profiles in
Conflicts of interest breast cancer:a pilot case-control study in the breast cancer
family registry, BMC Cancer, 2018, 18, 532.
There are no conflicts to declare.
9 C. Denkert, E. Bucher, M. Hilvo, R. Salek, M. Orešič,
J. Griffin, S. Brockmöller, F. Klauschen, S. Loibl,
D. K. Barupal, et al., Metabolomics of human breast
Acknowledgements cancer: new approaches for tumor typing and biomarker
This work was supported by the FCT-Fundação para a Ciência discovery, Genome Med., 2012, 4, 37.
e a Tecnologia ( project PEstOE/QUI/UI0674/2019, CQM, 10 C. L. Silva, M. Passos and J. S. Câmara, Solid phase micro-
Portuguese Government funds, and INNOINDIGO/0001/2015), extraction, mass spectrometry and metabolomic
the Madeira 14-20 Program ( project PROEQUIPRAM – Reforço approaches for detection of potential urinary cancer bio-
do Investimento em Equipamentos e Infraestruturas markers–a powerful strategy for breast cancer diagnosis,
Científicas na RAM – M1420-01-0145-FEDER-000008) and by Talanta, 2012, 89, 360–368.
the ARDITI-Agência Regional para o Desenvolvimento da 11 C. L. Silva, R. Perestrelo, P. Silva, H. Tomás and
Investigação Tecnologia e Inovação through the project J. S. Câmara, Volatile metabolomic signature of human
M1420-01-0145-FEDER-000005 – Centro de Química da breast cancer cell lines, Sci. Rep., 2017, 7, 43969.
Madeira – CQM+ (Madeira 14-20). The authors also acknowl- 12 C. L. Silva, M. Passos and J. S. Câmara, Investigation of
edge the FCT for the Ph.D. grant SFRH/BD/97039/2013 and urinary volatile organic metabolites as potential cancer bio-
PostDoctoral fellowship SFRH/BPD/97387/2013 given to markers by solid-phase microextraction in combination
Catarina L. Silva and Rosa Perestrelo. Pedro Silva acknowl- with gas chromatography-mass spectrometry, Br. J. Cancer,
edges ARDITI through the Ph.D grant under the M1420 2011, 105, 1894–1904.
Project – 09-5369-FSE-000001. 13 P. Porto-Figueira, J. A. M. Pereira and J. S. Câmara,
Exploring the potential of needle trap microextraction
combined with chromatographic and statistical data to
References discriminate different types of cancer based on urinary
volatomic biosignature, Anal. Chim. Acta, 2018, 1023, 53–
1 F. Bray, J. Ferlay, I. Soerjomataram, R. L. Siegel, L. A. Torre 63.
and A. Jemal, Global cancer statistics 2018: GLOBOCAN 14 C. Cavaco, J. A. M. Pereira, K. Taunk, R. Taware, S. Rapole,
estimates of incidence and mortality worldwide for 36 H. Nagarajaram and J. S. Câmara, Screening of salivary
cancers in 185 countries, Ca–Cancer J. Clin., 2018, 68, 394– volatiles for putative breast cancer discrimination: an
424. exploratory study involving geographically distant popu-
2 M. J. Duffy, Serum Tumor Markers in Breast Cancer: Are lations, Anal. Bioanal. Chem., 2018, 410, 1–10.
They of Clinical Value?, Clin. Chem., 2006, 52, 345–351. 15 J. Budczies, B. M. Pfitzner, B. Györffy, K.-J. Winzer,
3 C. M. Slupsky, H. Steed, T. H. Wells, K. Dabbs, C. Radke, M. Dietel, O. Fiehn and C. Denkert, Glutamate
A. Schepansky, V. Capstick, W. Faught and M. B. Sawyer, enrichment as new diagnostic opportunity in breast
Urine metabolite analysis offers potential early diagnosis of cancer, Int. J. Cancer, 2015, 136, 1619–1628.
4160 | Analyst, 2019, 144, 4153–4161 This journal is © The Royal Society of Chemistry 2019
Analyst Paper
This journal is © The Royal Society of Chemistry 2019 Analyst, 2019, 144, 4153–4161 | 4161