Content uploaded by Petr G. Lokhov
Author content
All content in this area was uploaded by Petr G. Lokhov
Content may be subject to copyright.
ORIGINAL ARTICLE
Metabolite profiling of blood plasma of patients with prostate
cancer
Petr G. Lokhov •Maxim I. Dashtiev •
Sergey A. Moshkovskii •Alexander I. Archakov
Received: 2 August 2009 / Accepted: 13 October 2009 / Published online: 25 October 2009
ÓSpringer Science+Business Media, LLC 2009
Abstract Prostate cancer is one of the most common
types of cancer in men. It is though extremely important to
search for specific markers including metabolites, which
concentration in blood could be a diagnostic measure. In
this regard, the metabolite profiling of blood plasma was
performed with two groups of people: healthy volunteers
(n=30) and patients with prostate cancer, second stage
(n=40). The profiling protocol included proteins removal
from blood plasma with methanol and direct analysis of
metabolite fractions by mass spectrometry. Identification of
the most abundant metabolites in samples was performed
using an accurate mass tag and an isotope pattern methods.
Cancer-specific metabolites were revealed by statistical
analysis of metabolite intensities in the mass spectra. Six
different metabolites were found to be cancer-specific. Two
metabolites, acylcarnitine and arachidonoyl amine, have
the AUC 0.97 and 0.86, respectively, which are higher than
those from PSA test, 0.59.
Keywords Mass spectrometry Prostate cancer
Blood plasma metabolites
1 Introduction
Prostate cancer is one of the most common types of cancer
in men. Rates of prostate cancer vary widely across the
world and found to be the highest in the US (IARC 2001).
The major problem is that many men who develop prostate
cancer never have symptoms, undergo no therapy, and
eventually die of other causes. A key challenge in cancer
medicine thus is to detect cancer at earlier stage as possi-
ble. The survival rate of patiences with prostate cancer
changes from 33% when it is detected at the advanced
stage to about 100% when is detected at earlier stage (Je-
mal et al. 2004). For early and accurate diagnosis of the
disease and to monitor its progression molecular bio-
markers are often used (Carini 2007). Recent advances in
biomarker discovery provide some high-throughput tech-
nologies, such as: genomics, proteomics and metabolo-
mics. Many studies employ the first two strategies
(Rajcevic et al. 2009; Ahmed 2009; He and Chiu 2003; Seo
and Ginsburg 2005; Emilien et al. 2000) here we will focus
on the metabolomics, namely metabolic profiling.
Metabolomics refers to a global analysis of the low
molecular weight molecules, known as metabolites, in cells,
tissues or fluids. Changes inmetabolite concentrations due to
a disease will add valuable information for biomarker dis-
covery. Mass spectrometry-based metabolic profiling is
nowadays a traditional method in metabolomics allowing
identification of major metabolites in biosamples with high
efficiency and sensitivity (Beecher 2003). In contrast to
classical metabolite studies that are focused only on a few
metabolites, metabolic profiling is aimed to analyze all
metabolites in sample of a given biological system. Metabolic
profiling has already found a wide application in practical
medicine, in particular, in screening newborns for the pres-
ence of congenital failures of metabolism of amino acids,
fatty acids, etc. (Piraud et al. 2003; Schulze et al. 2003; Chace
et al. 2002;2003; Chace and Kalas 2005; German et al. 2003).
Up to date, there is a clear evidence of the correlation of
prostate cancer with a concentration in the blood of
P. G. Lokhov (&)S. A. Moshkovskii A. I. Archakov
Institute of Biomedical Chemistry RAMS, Pogodinskaya Street,
10, Moscow 119121, Russia
e-mail: lokhovpg@rambler.ru
M. I. Dashtiev
Bruker Ltd., Moscow, Russia
123
Metabolomics (2010) 6:156–163
DOI 10.1007/s11306-009-0187-x
metabolites, for example lysophospholipids, androgens,
serotonin, amino acids, including aspartic acid, ornithine
and sarcosine (Osl et al. 2008; Siddiqui et al. 2006; Dizeyi
et al. 2004; Lai et al. 2005; Taylor et al. 2008; Isbarn et al.
2009; Raynaud 2009; Sreekumar et al. 2009). Thus, the
prostate cancer progression is reflected in the low mass
molecules of blood plasma, and further investigations in
this direction such as metabolic profiling can provide new
insights to the cancer diagnostics.
Various protocols of the metabolic analysis of blood
plasma based on mass spectrometry coupled with liquid
(LC–MS) and gas chromatography (GC–MS) are known
(Osl et al. 2008; Gowda et al. 2008; Xue et al. 2008). A
distinct feature of our approach is a direct mass spectro-
metric analysis of metabolite fraction of blood plasma. The
main advantage of this approach is that there is no sepa-
ration prior to mass spectrometry thus reaching maximum
reproducibility for mass spectrometry data. Thus, reliable
changes in concentration of metabolites can be detected
more clearly (Dettmer et al. 2007).
2 Materials and methods
2.1 Samples of blood plasma
Blood for measurement of prostate-specific antigen (PSA)
concentration in men was provided by ‘‘New Medical
Technologies’’ Ltd., Russia. ELISA (test kit ‘‘oncoELISA-
total PSA’’, AlcorBio, Russia) was chosen as a measure of
the PSA concentration. All patients signed an informed
consent to provide their blood samples for research pur-
poses. They were then examined in Clinical Cancer Center
(Voronezh, Russia) by palpation and ultrasound investi-
gation, needle biopsy and then the diagnosis was specified.
For mass spectrometry analysis 40 samples from patients
with prostate cancer stage II, T2NxMO and 30 samples
from healthy patients were selected. All patients were
55–80 years old.
The following protocol for blood sampling was used:
10 ml of blood from cubital vein were collected in glass
tubes, containing sodium citrate. In 15 min a 3.8% citrate
blood was centrifuged for 15 min and 16009gat the room
temperature. After that, 2.5 ml of plasma were aliquoted in
4 eppendorf-like tubes and then frozen and stored at -
80°C. Samples for analysis were frozen-thawed not more
than once.
2.2 Sample preparation of blood plasma for mass
spectrometry
For the deproteinization of blood plasma 100 ll of them
was mixed with 100 ll of water (LiChrosolv, Merck, USA)
and 800 ll of methanol (Fluka, Germany) and then incu-
bated for 10 min at 4°C. Samples with precipitated proteins
were centrifuged for 10 min at 13000 rpm (MiniSpin plus,
Eppendorf, Germany), and then supernatant was trans-
ferred to clean eppendorf tubes (Eppendorf, Germany) and
the solvent was evaporated during 3 h at 45°C in the
SpeedVac (Eppendorf, Germany). The resulting dry resi-
due was dissolved in 100 ll of 95% acetonitrile (Acros
Organics, USA) containing 0.1% formic acid (Fluka,
Germany). Samples were sonicated for better dissolution in
the Bandelin RM 40UH ultrasonic bath (Sonorex Technik,
Germany) twice for 30 s. Then, the samples were centri-
fuged again for 10 min at 13000 rpm (MiniSpin plus) and
the resulting supernatant was used for mass spectrometry
analysis.
2.3 Mass spectrometry and sample processing
Mass spectrometry analysis was carried out on an elec-
trospray hybrid quadrupole time-of-flight mass spectrom-
eter MicrOTOF-Q (Bruker Daltonik GmbH, Germany).
Mass spectrometer was tuned for a mass range of 250–
1500 Da for optimal signal intensity. Both positive and
negative ions were measured. Samples were delivered into
the mass spectrometer by a direct infusion with a syringe
pump (Hamilton Bonaduz, Switzerland). Accumulation time
of one measurement was 1 min at a flow rate of 3 ll/min.
Threshold for peak selection was set to S/N: 10.
Mass spectra were then processed in Data Analysis 3.4
(Bruker Daltonik GmbH, Germany). For the 25 most
intense peaks having clear isotopic distribution of posi-
tively and negatively charged metabolites the difference
between healthy and diseased patients was defined by two-
sided Wilcoxon rank sum test (Matlab, MathWorks,
USA). Two peaks were considered to relate to the same
metabolite if the mass difference does not exceed
0.01 Da. For the metabolite peaks, whose intensity sta-
tistically changing (1 -p[0.95) in the event of illness,
the average values and standard deviations for the groups
corresponding to the sick and healthy patients were cal-
culated. To determine the parameters of laboratory diag-
nosis, such as specificity and sensitivity, as well as the
construction of the ROC (Receiver Operating Character-
istic)-curve and calculating the area under the ROC-curve
(AUC) ‘‘Statistical Package for the Social Sciences
(SPSS)’’ version of 10.0 (SPSS Inc., USA) was used.
These specificity and sensitivity of diagnosis based on the
intensities of the metabolites were compared with the
sensitivity and specificity of diagnosis, based on the
concentration of PSA in the blood of patients. Given the
age of patients with prostate cancer, the standard values of
the PSA concentration were taken from 0 to 5.36 ng/ml
(0–0.16 910
-9
M).
Blood metabolome and prostate cancer 157
123
2.4 Metabolite identification
For mass spectrometric peaks having the highest intensity
and having clear isotopic distribution, the correspondence
to the specific metabolites from the database ‘‘Human
Metabolome Database’’ (http://www.hmdb.ca) (Wishart
et al. 2007) and/or Metlin (Scripps Center for Mass Spec-
trometry, USA; http://metlin.scripps.edu) (Smith et al.
2005) was established. An example of the identification of
metabolite mass, chemical formula and isotopic distribu-
tion is demonstrated in Fig. 1.
3 Results and discussion
On the average 1900 metabolite’s ions were detected in
samples both from healthy and diseased patients. Typical
mass spectra are shown in Fig. 2. The spectra were then
queried against the database and the list of identified
metabolites was retrieved (Table 1). It can be seen that the
area of the most intense peaks both in positive and negative
modes correspond to the fraction of phospholipids. High
number of peaks is observed due to differences in hydro-
carbon chains of fatty acids that present in phospholipids.
The mass range from 450–600 Da corresponds mainly to
lysoforms of phospholipids. Low molecular weight region
(\400 Da) corresponds to metabolites of different
chemical classes that are responsible for various physio-
logical functions.
The metabolites, whose intensity changes significantly
in the event of illness were identified by statistical analysis.
All of the identified differences were in the area of low
molecular metabolites, namely: 302.2442, 304.2602,
377.2680 m/zfor the positively charged metabolites and
307.0452, 367.1486, 369.1631 m/zfor the negatively
charged metabolites. For intensity data of metabolites
mean values and standard deviations were calculated
(Fig. 3).
Results of the identification for these masses are given in
Table 1. It should be noted that for masses 367.1486 and
369.1632 m/zthere are two candidates for each mass. This
is related to the fact that candidates have the same chemical
formula and, consequently, identical isotope distribution
that does not allow to differentiate them with the used in
this study a method of metabolite identification.
The cancer-specific reliable changes in metabolite
intensities allow considering them as potential markers of
disease. In order to determine the possible effectiveness of
such diagnostic data the ROC-curve was plotted using
intensities of metabolite’s mass peaks (Fig. 4). The area
under the ROC-curve is a direct indication of the efficiency
and clinical applicability of the diagnostic method. It is
widely accepted that diagnostic methods are considered to
be clinically applicable when the value for the area under
Fig. 1 Metabolite identification
using accurate mass-tag method
in combination with isotope
pattern. The measured
molecular mass of the
metabolite is queried against a
metabolite database (Human
Metabolome database) (‘1’) that
gives a list of matched
compounds with chemical
formula. Using True Isotopic
Pattern (‘2’) with Smart
Formula 3D (Bruker Daltonik
GmbH, Germany) the
theoretical isotopic distributions
were calculated for chemical
formulas of given candidates
and through the comparing them
with experimental isotopic
pattern the exact molecular
composition and name of
candidate is defined (‘3’,’4’)
158 P. G. Lokhov et al.
123
ROC-curve (AUC) is not less than 0.6, and to have a good
diagnostic measure when the value is greater than 0.8
(Metz 1978).
Our suggested models of the diagnostic system based on
the metabolites with masses of 302.2442 m/zand
304.2602 m/zcan be attributed to the ‘‘good’’ because they
have a values higher than 0.8. Metabolites with masses
377.2680, 367.1486, and 369.1632 m/zare not suitable for
the diagnosing of prostate cancer, because the value for the
AUC is less then 0.6 (Table 2). For the mass 307.0452 m/z
the ROC-curve was not plotted because of no data in the
group of healthy volunteers for this mass (see Fig. 3).
It should be noted that the PSA test for the same group
of patients had AUC 0.59, which is consistent with data
from other sources (0.51 and 0.54 for the test kits of Roche
and Bayer, respectively; Lein et al. 2003), and this fact
illustrate that the PSA test for the diagnosis of prostate
cancer 2nd stage is questionable.
High death rates from prostate cancer are usually due to
a long-term asymptomatic course of the disease that causes
delays in diagnosis. More than 60% of patients treated by a
physician doctor already have metastases in organs. This
risk group mostly consists of men older than 50–55 years
(Hankey et al. 1999). The metabolic profiling of blood
plasma for this target group of patients has been done and
several cancer-specific changes have been revealed.
Identified changes in the concentration of androgens in
the blood of patients with advanced prostate cancer are not
accidental. A number of studies have reported worrisome
associations between low serum testosterone and prostate
cancer. Morgentaler et al. assessed the prostate cancer
prevalence in hypogonadal men in two studies (Morgen-
taler et al. 1996; Morgentaler and Rhoden 2006). The risk
of prostate cancer detection was correlated with the
severity of androgen deficiency. Also, in a subset analysis
of 184 men, a low testosterone-to-PSA ratio was an inde-
pendent predictor of prostate cancer after adjustment for
age and PSA level (Rhoden et al. 2008). Additionally, other
studies have reported that a low serum testosterone level is
associated with pathologic stage and high Gleason score
(Imamoto et al. 2005; Schatzl et al. 2001; Yano et al.
2007). Thus, recorded in this study, the reduction of
androgens in the blood of patients with cancer is a sup-
plement previously revealed prostate cancer connection
with a low concentration of androgens in the blood.
The observed fall in steroid-like substance such as iso-
lithocholic acid in the blood of patients with advanced
prostate cancer was not shown before. The nature of this
phenomenon is difficult to interpret clearly. Perhaps the fall
of isolithocholic acid is a consequence of low levels of
androgens in the blood of a patient, that directly affects the
secretion of biliary lipids (Ohshima et al. 1996) to which
isolithocholic acid relates.
Numerous disorders have been described that lead to
disturbances in energy production and in intermediary
metabolism of the body that are characterized by the pro-
duction and excretion of unusual acylcarnitines. Determi-
nation of the qualitative pattern of acylcarnitines can be of
diagnostic and therapeutic importance. Changes in the
concentration of acylcarnitine including those observed in
cancer patients are described in Vinci et al. (2005), Sachan
and Dodson (1987), and Malaguarnera et al. (2006). This
phenomenon was studied for advanced pathologic stages
when there was a malignant cachexia. In our case, we have
an early stage of cancer and, more likely, the increase in
concentration of acylcarnitine, namely dimethylheptanoyl
Fig. 2 Mass spectrum of negatively charged metabolites of human blood plasma
Blood metabolome and prostate cancer 159
123
carnitine, in blood of patients with prostate cancer is
associated with the above changes in hormonal back-
ground, determined by the low level of androgens. The
correlation of testosterone concentration in blood with the
carnitine and acylcarnitine concentrations was previously
described (Marquis and Fritz 1965; Carter et al. 1980).
Moreover, Carter et al. (1980) found significant negative
correlation between blood plasma testosterone concentra-
tions and blood plasma carnitine in animals, the fact, which
is consistent with our data.
Table 1 Results of metabolite
identification
PC phosphatidylcholine;
PE phosphatidylethanolamine;
RN-butyl or N-propyl alpha-
methyl; MW molecular weight
Measured
adduct MW (m/z)
Matching MW from
database (Da)
Chemical
formula
Metabolite
common name
Wilcoxon test
(1 -p)
Positively charged metabolites
1 288.2906 287.2824 C
17
H
37
NO
2
Sphingosine 0.51
2 302.2442 301.2253 C
16
H
31
NO
4
Dimethylheptanoyl carnitine 1
3 304.2602 303.2562 C
20
H
33
NO Arachidonoyl amine 1
4 360.3225 359.3188 C
24
H
41
NO R-arachidonoyl amine 0.44
5 377.2680 376.2978 C
24
H
40
O
3
Isolithocholic acid 1
6 437.1928 436.2590 C
21
H
41
O
7
P 1-Oleoyl-lysophosphatidic
acid
0.94
7 518.3269 517.3168 C
26
H
48
NO
7
P Linolenoyl lysolecithin 0.22
8 542.3208 541.3168 C
28
H
48
NO
7
P LysoPC 0.73
9 544.3353 543.3325 C
28
H
50
NO
7
P LysoPC 0.58
10 780.5537 779.5464 C
44
H
78
NO
8
P PC 0.26
11 804.5503 803.5465 C
46
H
78
NO
8
P PC 0.57
12 806.5628 805.5621 C
46
H
80
NO
8
P PC 0.16
13 808.5793 807.5778 C
46
H
82
NO
8
P PC 0.38
14 828.5477 827.5465 C
48
H
78
NO
8
P PC 0.11
15 830.5617 829.5621 C
48
H
80
NO
8
P PC 0.13
16 832.5798 831.5778 C
48
H
82
NO
8
P PC 0.35
Negatively charged metabolites
17 303.2238 304.2402 C
20
H
32
O
2
Arachidonic acid 0.10
18 307.0452 308.0410 C
9
H
13
N
2
O
8
P Deoxyuridine monophosphate 0.96
19 367.1486 368.1657 C
19
H
28
O
5
S Testosterone sulfate or
Dehydroepiandrosterone
sulfate
0.99
20 369.1632 370.1814 C
19
H
30
O
5
S Androsterone sulfate or
5a-Dihydrotestosterone
sulfate
0.95
21 397.1911 398.1938 C
16
H
33
NO
8
P Glycerophosphocholine 0.59
22 445.3184 446.3190 C
28
H
43
FO
3
Dihydroxyvitamin D2 0.86
23 480.2977 481.3168 C
23
H
48
NO
7
P LysoPC 0.47
24 530.2886 531.3319 C
27
H
50
NO
7
P LysoPE 0.08
25 532.2867 533.3481 C
27
H
52
NO
7
P LysoPE 0.12
26 557.4414 558.4284 C
35
H
58
O
5
Diglyceride 0.72
27 593.4630 594.5223 C
37
H
70
O
5
Diglyceride 0.94
28 792.5106 793.5985 C
46
H
84
NO
7
P PC 0.54
29 794.5182 795.5778 C
45
H
82
NO
8
P PC 0.05
30 816.5114 817.5621 C
47
H
80
NO
8
P PE 0.66
31 818.5194 819.5778 C
47
H
82
NO
8
P PE 0.81
32 820.5393 821.5929 C
47
H
84
NO
8
P PE 0.52
33 840.4984 841.5621 C
49
H
80
NO
8
P PE 0.62
34 842.5192 843.5778 C
49
H
82
NO
8
P PE 0.92
35 844.5391 845.5934 C
49
H
84
NO
8
P PE 0.84
36 850.4731 851.5465 C
50
H
78
NO
8
P PC 0.92
37 852.4776 853.5621 C
50
H
80
NO
8
P PC 0.91
160 P. G. Lokhov et al.
123
Changes of concentration of various lipids in blood of
patients are well known. Changes in lysophospholipids
level are presumably due to binding and activation by them
the specific G-protein binding receptor followed by growth
and proliferation of cancer cells. This interaction leads the
decrease of concentration of lysophospholipids in blood,
which can be used for diagnostic purposes (Murph et al.
2007; Osl et al. 2008). There are no statistical significant
changes in phospholipids revealed in our research. Proba-
bly, it is connected with lower dynamic range of direct
mass spectrometry in comparison with LC–MS used in
previous works. Also, the used early stages of cancers may
be reason why several detected phospholipids did not reach
statistical significance accepted for biomarkers (Table 1).
Required value ‘0.95’ may be reached at further progres-
sion of cancers.
Statistically reliable change of deoxyuridine mono-
phosphate is most likely associated with a known rise in the
blood of nucleotides in patients with cancer as the primary
degradation products of tRNA (Cho et al. 2009; Hsu et al.
2009; Zheng et al. 2005). Moreover, changes in concen-
trations of some nucleotides in the urine have been pro-
posed for the diagnosis of certain types of cancer (Hsu
et al. 2009; Zheng et al. 2005). Since the deoxyuridine is
detected only in the blood of prostate cancer patients, the
ROC was not plotted and the potential use of this metab-
olite as a marker of prostate cancer was not estimated as
required.
A statistically reliable change of arachidonoyl amine is
not shown previously for patient with prostate cancer.
Given the lack of literature data of arachidonoyl amine
metabolism in men, it is difficult to interpret the role of this
Fig. 3 Graphs of mean values
and standard deviations of the
ion intensities of the blood
metabolites of patients with
prostate cancer (1) and healthy
volunteers (2)
Fig. 4 ROC-curve for the models of diagnostic systems of prostate
cancer based on measurement of intensities of metabolite peaks in the
mass spectrometric profile of blood plasma. To construct the curves of
blood plasma samples, 40 prostate cancer patients and 30 healthy
volunteers were used. ROC-curves for: 1, dimethylheptanoyl carni-
tine; 2, arachidonoyl amine; 3, isolithocholic acid; 4, testosterone
sulfate/dehydroepiandrosterone sulfate; 5, androsterone sulfate/5a-
dihydrotestosterone sulfate
Blood metabolome and prostate cancer 161
123
metabolite in prostate cancer. However, calculated AUC
for the metabolite, which is 0.86, indicates the potential use-
fulness as a marker of early diagnosis of prostate cancer.
4 Concluding remarks
In this work a mass spectrometric metabolic profiling of
blood plasma of patients with prostate cancer was carried
out. The suggested protocol allows observation of dominant
peaks from lysophospholipids, and phospholipids in the
mass spectrometric profile, which are not specific for pros-
tate cancer. However, in the low molecular weight region
two metabolites with masses 302.2442 m/z(301.2253 Da,
acylcarnitine) and 304.2602 m/z(303.2562 Da, arachido-
noyl amine), were identified as potentially suitable markers
for early diagnostic of prostate cancer with the efficiency of
significantly higher than the current PSA test.
Acknowledgments The authors would like to thank ‘‘New Medical
Technologies’’ Ltd. (Russia) and Clinical Cancer Center (Voronezh,
Russia) for the blood samples.
References
Ahmed, F. E. (2009). Sample preparation and fractionation for
proteome analysis and cancer biomarker discovery by mass
spectrometry. Journal of Separation Science, 32, 771–798.
Beecher, C. W. W. (2003). Metabolic profiling: Its role in biomarker
discovery and gene function analysis (pp. 311–335). New York:
Springer.
Carini, C. (2007). Biomarkers: A valuable tool in clinical research
and medical practice. IDrugs, 10, 395–398.
Carter, A. L., Hutson, S. M., Stratman, F. W., & Haning, R. V., Jr.
(1980). Relationship of carnitine and acylcarnitines in ejaculated
sperm to blood plasma testosterone of dairy bulls. Biology of
Reproduction, 23, 820–825.
Chace, D. H., & Kalas, T. A. (2005). A biochemical perspective on
the use of tandem mass spectrometry for newborn screening and
clinical testing. Clinical Biochemistry, 38, 296–309.
Chace, D. H., Kalas, T. A., & Naylor, E. W. (2002). The application of
tandem mass spectrometry to neonatal screening for inherited
disorders of intermediary metabolism. Annual Review of Genom-
ics and Human Genetics, 3, 17–45.
Chace, D. H., Kalas, T. A., & Naylor, E. W. (2003). Use of tandem
mass spectrometry for multianalyte screening of dried blood
specimens from newborns. Clinical Chemistry, 49, 1797–1817.
Cho, S. H., Choi, M. H., Lee, W. Y., & Chung, B. C. (2009). Evaluation
of urinary nucleosides in breast cancer patients before and after
tumor removal. Clinical Biochemistry, 42, 540–543.
Dettmer, K., Aronov, P. A., & Hammock, B. D. (2007). Mass
spectrometry-based metabolomics. Mass Spectrometry Reviews,
26, 51–78.
Dizeyi, N., Bjartell, A., Nilsson, E., et al. (2004). Expression of
serotonin receptors and role of serotonin in human prostate
cancer tissue and cell lines. Prostate, 59, 328–336.
Emilien, G., Ponchon, M., Caldas, C., Isacson, O., & Maloteaux, J. M.
(2000). Impact of genomics on drug discovery and clinical
medicine. QJM, 93, 391–423.
German, J. B., Roberts, M. A., & Watkins, S. M. (2003). Genomics and
metabolomics as markers for the interaction of diet and health:
Lessons from lipids. Journal of Nutrition, 133, 2078–2083.
Gowda, G. A., Zhang, S., Gu, H., Asiago, V., Shanaiah, N., &
Raftery, D. (2008). Metabolomics-based methods for early
disease diagnostics. Expert Review of Molecular Diagnostics,
8, 617–633.
Hankey, B. F., Feuer, E. J., Clegg, L. X., et al. (1999). Cancer
surveillance series: Interpreting trends in prostate cancer. Part I.
Evidence of the effects of screening in recent prostate cancer
incidence, mortality, and survival rates. Journal of the National
Cancer Institute, 91, 1017–1024.
He, Q. Y., & Chiu, J. F. (2003). Proteomics in biomarker discovery
and drug development. Journal of Cellular Biochemistry, 89,
868–886.
Hsu, W. Y., Chen, W. T., Lin, W. D., et al. (2009). Analysis of urinary
nucleosides as potential tumor markers in human colorectal cancer
by high performance liquid chromatography/electrospray ioniza-
tion tandem mass spectrometry. Clinica Chimica Acta, 402, 31–37.
IARC Worldwide Cancer Incidence Statistics—Prostate. (2001).
JNCI cancer spectrum (Vol. 19). Oxford University Press.
Imamoto, T., Suzuki, H., Fukasawa, S., et al. (2005). Pretreatment
serum testosterone level as a predictive factor of pathological
stage in localized prostate cancer patients treated with radical
prostatectomy. European Urology, 47, 308–312.
Isbarn, H., Pinthus, J. H., Marks, L. S., et al. (2009). Testosterone and
prostate cancer: Revisiting old paradigms. European Urology,
56, 48–56.
Jemal, A., Tiwari, R. C., Murray, T., et al. (2004). Cancer Statistics,
2004. CA: A Cancer Journal for Clinicians, 54, 8–29.
Lai, H. S., Lee, J. C., Lee, P. H., Wang, S. T., & Chen, W. J. (2005).
Plasma free amino acid profile in cancer patients. Seminars in
Cancer Biology, 15, 267–276.
Lein, M., Kwiatkowski, M., Semjonow, A., et al. (2003). A
multicenter clinical trial on the use of complexed prostate
Table 2 Parameters of the efficiency of the prostate cancer diagnosis based on the intensities of metabolite peaks in the mass spectrometric
profile
Measured metabolite
mass (m/z)
Metabolite common name Sensitivity (%) Specificity (%) AUC
302.2442 Dimethylheptanoyl carnitine 94.6 96.4 0.97
304.2602 Arachidonoyl amine 86.5 92.9 0.86
377.2680 Isolithocholic acid 21.6 21.4 0.11
367.1486 Testosterone sulfate or dehydroepiandrosterone sulfate 40.5 42.9 0.36
369.1632 Androsterone sulfate or 5a-dihydrotestosterone sulfate 62.2 28.6 0.29
PSA concentration 35.0 83.3 0.59
162 P. G. Lokhov et al.
123
specific antigen in low prostate specific antigen concentrations.
Journal of Urology, 170, 1175–1179.
Malaguarnera, M., Risino, C., Gargante, M. P., et al. (2006). Decrease
of serum carnitine levels in patients with or without gastroin-
testinal cancer cachexia. World Journal of Gastroenterology, 12,
4541–4545.
Marquis, N. R., & Fritz, I. B. (1965). Effects of testosterone on the
distribution of carnitine, acetylcarnitine, and carnitine acetyl-
transferase in tissues of the reproductive system of the male rat.
Journal of Biological Chemistry, 240, 2197–2200.
Metz, C. E. (1978). Basic principles of ROC analysis. Seminars in
Nuclear Medicine, 8, 283–298.
Morgentaler, A., Bruning, C. O., III, & DeWolf, W. C. (1996). Occult
prostate cancer in men with low serum testosterone levels.
Journal of the American Medical Association, 276, 1904–1906.
Morgentaler, A., & Rhoden, E. L. (2006). Prevalence of prostate
cancer among hypogonadal men with prostate-specific antigen
levels of 4.0 ng/mL or less. Urology, 68, 1263–1267.
Murph, M., Tanaka, T., Pang, J., et al. (2007). Liquid chromatography
mass spectrometry for quantifying plasma lysophospholipids:
Potential biomarkers for cancer diagnosis. Methods in Enzymol-
ogy, 433, 1–25.
Ohshima, A., Cohen, B. I., Ayyad, N., & Mosbach, E. H. (1996).
Effect of a synthetic androgen on biliary lipid secretion in the
female hamster. Lipids, 31, 879–886.
Osl, M., Dreiseitl, S., Pfeifer, B., et al. (2008). A new rule-based
algorithm for identifying metabolic markers in prostate cancer
using tandem mass spectrometry. Bioinformatics, 24, 2908–
2914.
Piraud, M., Vianey-Saban, C., Petritis, K., et al. (2003). ESI-MS/MS
analysis of underivatised amino acids: A new tool for the
diagnosis of inherited disorders of amino acid metabolism.
Fragmentation study of 79 molecules of biological interest in
positive and negative ionisation mode. Rapid Communications in
Mass Spectrometry, 17, 1297–1311.
Rajcevic, U., Niclou, S. P., & Jimenez, C. R. (2009). Proteomics
strategies for target identification and biomarker discovery in
cancer. Frontiers in Bioscience, 14, 3292–32303.
Raynaud, J. P. (2009). Testosterone deficiency syndrome: Treatment
and cancer risk. Journal of Steroid Biochemistry and Molecular
Biology, 114, 96–105.
Rhoden, E. L., Riedner, C. E., & Morgentaler, A. (2008). The ratio of
serum testosterone-to-prostate specific antigen predicts prostate
cancer in hypogonadal men. Journal of Urology, 179, 1741–
1744.
Sachan, D. S., & Dodson, W. L. (1987). The serum carnitine status of
cancer patients. Journal of the American College of Nutrition,
6, 145–150.
Schatzl, G., Madersbacher, S., Thurridl, T., et al. (2001). High-grade
prostate cancer is associated with low serum testosterone levels.
Prostate, 47, 52–58.
Schulze, A., Lindner, M., Kohlmuller, D., Olgemoller, K., May-
atepek, E., & Hoffmann, G. F. (2003). Expanded newborn
screening for inborn errors of metabolism by electrospray
ionization-tandem mass spectrometry: Results, outcome, and
implications. Pediatrics, 111, 1399–1406.
Seo, D., & Ginsburg, G. S. (2005). Genomic medicine: Bringing
biomarkers to clinical medicine. Current Opinion in Chemical
Biology, 9, 381–386.
Siddiqui, E. J., Shabbir, M., Mikhailidis, D. P., Thompson, C. S., &
Mumtaz, F. H. (2006). The role of serotonin (5-hydroxytrypta-
mine 1A and 1B) receptors in prostate cancer cell proliferation.
Journal of Urology, 176, 1648–1653.
Smith, C. A., O’Maille, G., Want, E. J., et al. (2005). METLIN: A
metabolite mass spectral database. Therapeutic Drug Monitor-
ing, 27, 747–751.
Sreekumar, A., Poisson, L. M., Rajendiran, T. M., et al. (2009).
Metabolomic profiles delineate potential role for sarcosine in
prostate cancer progression. Nature, 457, 910–914.
Taylor, B. S., Pal, M., Yu, J., et al. (2008). Humoral response profiling
reveals pathways to prostate cancer progression. Molecular &
Cellular Proteomics, 7, 600–611.
Vinci, E., Rampello, E., Zanoli, L., Oreste, G., Pistone, G., &
Malaguarnera, M. (2005). Serum carnitine levels in patients with
tumoral cachexia. European Journal of Internal Medicine, 16,
419–423.
Wishart, D. S., Tzur, D., Knox, C., et al. (2007). HMDB: The Human
Metabolome Database. Nucleic Acids Research (Database
issue), 35, D521–D526.
Xue, R., Lin, Z., Deng, C., et al. (2008). A serum metabolomic
investigation on hepatocellular carcinoma patients by chemical
derivatization followedby gas chromatography/mass spectrometry.
Rapid Communications in Mass Spectrometry, 22, 3061–3068.
Yano, M., Imamoto, T., Suzuki, H., et al. (2007). The clinical
potential of pretreatment serum testosterone level to improve the
efficiency of prostate cancer screening. European Urology, 51,
375–380.
Zheng, Y. F., Yang, J., Zhao, X. J., et al. (2005). Urinary nucleosides
as biological markers for patients with colorectal cancer. World
Journal of Gastroenterology, 11, 3871–3876.
Blood metabolome and prostate cancer 163
123
