This web only provides the extract of this article. If you want to read the figures and tables, please reference the PDF full text on Blackwell Synergy. Thank you.
- Original Article -
Application of surface-enhanced laser desorption/ionization time-of-flight-based serum proteomic array technique for the early diagnosis of prostate cancer
Yu-Zhuo Pan1, Xue-Yuan Xiao2, Dan Zhao1, Ling Zhang1, Guo-Yi Ji1, Yang Li1, Bao-Xue Yang1, Da-Cheng He2, Xue-Jian Zhao1
1Research Center of Prostate Diseases, Department of Reproductive Pathophysiology, School of Basic Medicine, Jilin
University, Changchun 130021, China
2Research Institute of Cytobiology of the Academy of Life Science, Research Institute of Proteomics, Normal University,
Beijing, 100875, China
Abstract
Aim: To identify the serum biomarkers of prostate cancer (PCa) by protein chip and bioinformatics.
Methods: Serum samples from 83 PCa patients and 95 healthy men were taken from a mass screening in Changchun, China.
Protein profiling was carried out using surface-enhanced laser desorption/ionization time-of-flight mass spectrometry
(SELDI-TOF MS). The data of spectra were analyzed using two bioinformatics tools.
Results: Eighteen serum differential proteins were identified in the PCa group compared with the control group
(P < 0.01). There were four proteins at the higher serum level and 14 proteins at the lower serum level in the PCa group. A decision tree
classification algorithm that used an eight-protein mass pattern was developed to correctly classify the samples. A sensitivity
of 92.0 % and a specificity of 96.7 % for the study group were obtained by comparing the PCa and control groups.
Conclusion: We identified new serum biomarkers of PCa. SELDI-TOF MS coupled with a decision tree
classification algorithm will provide a highly accurate and innovative approach for the early diagnosis of PCa.
(Asian J Androl 2006 Jan; 8: 45-51)
Keywords: prostate cancer; early diagnosis; protein chip; biomarker; serum
Correspondence to: Dr Xue-Jian Zhao, Prostate Diseases Prevention and Treatment Research Center, Jilin University, Changchun 130021,
China.
Tel/Fax: +86-431-563-2348
E-mail: pro-2@jlu.edu.cn
Dr Da-Cheng He, Research Institute of Cytobiology of Academy of Life Science, Research Institute of Proteomics, Normal University,
Beijing 100875, China
Tel: +86-10-5880-9729, Fax: +86-10-5880-5042
E-mail: dhe@bnu.edu.cn
These authors contributed equally to this work.
Corresponding authors who contributed equally to this work according to the cooperating protocol.
Received 2005-01-19 Accepted 2005-10-14
DOI: 10.1111/j.1745-7262.2006.00103.x
1 Introduction
Prostate cancer (PCa) is one of the main forms of
cancer affecting old men worldwide. In the USA, in 2003,
PCa had the highest cancer incidence with 220 900 new
cases and the second-highest cancer-related mortality rate,
with 28 900 deaths [1]. The retrospective research of
Gu et al. [2] showed that PCa incidence in Beijing, in the
1990s was five times higher than that in the 1950s. We
retrospectively investigated all PCa cases from eight
hospitals in Changchun, China from 1986 to 2001, and found
the number of cases had increased 4.6-fold in the
1999-2001 period vs. the 1986-1989 period [3]. We also
organized a mass screening in Changchun for PCa in men
aged 50 years or older with routine serum
prostate-specific antigen (PSA) tests. The detection rate of PSA was
greater than 1.7 %, and 18.8 % of the cases involved
osseous metastasis [4, 5].
Currently, the serum PSA test for mass screening for
PCa appears to be controversial. Recently, Stamey
et al. [6] announced that the serum PSA test as the standard
detection test for PCa was out of date in the USA,
based on 20 years of experience; in our experience, the biomarker
still plays an important role in the mass screening of PCa
in China. However, with the rapid emergence of proteomics, it is appealing to many investigators to
explore new and more effective specific proteins to
accurately detect PCa. The use of surface-enhanced laser
desorption/ionization time-of-flight mass spectrometry
(SELDI-TOF MS), for example, requires less labor, and
has a high throughput and excellent reproducibility. Initial
researches [7-9] on the identification of
biomarkers for cancers using this technique have been very promising,
which has encouraged us to further verify the
availability of MS. This technique has been successfully applied
for exploring new protein markers for early detection of
PCa in the USA. However, the proteomic approach has
not yet been used to identify the protein markers of PCa
in a cohort of Chinese men.
In this study, we analyzed 178 serum samples using
SELDI-TOF MS to explore the marker proteins for the
detection of PCa for the first time in China.
2 Materials and methods
2.1 Serum samples
One hundred and seventy-eight serum samples were
randomly taken from a serum bank in the Research Cen
ter of Prostate Diseases at Jilin University (Changchun,
China) and were kept at -70 ºC. Of the 178 samples, 95
were taken from healthy men and 83 from PCa patients,
respectively, confirmed by PSA test and histology
examination through prostate needle biopsy. The healthy
control (n = 95, age range 52-81 years, mean 63 years)
and PCa (n = 83) groups were age-matched. The PCa
group was subsequently divided into the organ-confined
PCa group (n = 45; T1/T2; age range 50-89 years, mean
67 years) and the non-organ-confined PCa group
(n = 37; T3/T4; age range 48-91 years, mean 70 years). The
PSA ranges were 0.01-1.00 ng/mL (mean 0.59 ng/mL)
for the control, 0.00-22.76 ng/mL (mean
9.62 ng/mL) for the organ-confined PCa group,
and 0.00-108.00 ng/mL (mean 77.21 ng/mL) for the
non-organ-confined PCa group. The study was conducted in accordance with
the ethics of the Declaration of Helsinki and with the
approval of the Jilin University Bethune Medical Services
Ethics Committee.
2.2 SELDI-TOF protein profiling
Immobilized metal affinity capture array (IMAC)-Cu
metal binding chips (Ciphergen Biosystems, Fremont, CA,
USA) were used and put into a bioprocessor (Ciphergen
Biosystems, Fremont, CA, USA), a device that holds eight
chips and allows for the application of large volume of serum to each chip array. The chips were coated with 50 µL
of 100 mmol/L CuSO4 on each array and agitated for 5 min
at room temperature. The chips were rinsed three times
with deionized H2O, then 150 µL of 100 mmol/L sodium
acetate buffer (pH 4.0) was added to each array and shaken
for 5 min to remove the unbound copper. Serum samples
for SELDI-TOF analysis were prepared by vortexing 5 µL
of serum with 10 µL of 8 mol/L urea and 1 %
3-[(3-cholamidopropyl) dimethylammonio]-1-propanesulfonic
acid in phosphate-buffered saline (PBS) in a tube at 4 ºC
for 30 min. Ten microliters of the serum/urea mixture
was applied to each well with 90 µL PBS, and the
bioprocessor was sealed and shaken on a platform shaker
for 30 min. The serum/urea mixture was discarded, and
the chips were washed twice, 5 min each wash cycle, in
150 µL of 100 mmol/L sodium acetate buffer as described
above. The chips were removed from the bioprocessor,
washed twice with deionized H2O, air-dried, and stored
in the dark at room temperature until subjected to
SELDI-TOF analysis.
Before SELDI-TOF MS analysis, 0.5 µL of saturated
solution of sinapinic acid (Fluka, Milwaukee, WI, USA)
in 50 % (v/v) acetonitrile, and 0.5 % trifluoroacetic acid
was applied onto each spot twice and air-dried between
applications. The chips were then placed in the PBS-II SELDI-TOF MS (Ciphergen Biosystems, Fremont, CA,
USA) operated in the positive ion mode. Time-of-flight
spectra were generated by averaging 90 laser shots
collected on each spot with a laser intensity setting of 195,
detector sensitivity setting of 9, and a lag time focusing
of 900 ns. The spectra were calibrated using the
All-in-1 protein molecular mass standard (Ciphergen Biosystems, Fremont, CA, USA). To compensate for slight
spot-to-spot variations, if any, the spectra were also normalized
using the total ion current method in the mass to charge
(m/z) range of 1 500-30 000 (IMAC-Cu surface) with
subtracted baseline. The reproducibility of the
SELDI-TOF system was determined using the pooled normal
serum quality control sample.
Peaks were identified after mass calibration,
background subtraction and normalization using the
clustering and alignment function of ProteinChip Biomarker
Wizard software 3.1 (Ciphergen Biosystems, Fremont,
CA, USA). The settings used were as follows:
signal/noise ratio in the first pass: 3;
minimum peak threshold: 5 %; cluster mass window: 0.3 %; and
signal/noise in the second pass: 1.5. The peaks detected within
± 0.15 %
m/z units of each other across the
spectra were considered one cluster, and a particular cluster
was represented by its average
m/z value. The peak information,
including m/z and intensity values, was exported into ProteinChip
Biomarker Pattern 4.0 (Ciphergen Biosystems, Fremont,
CA, USA) for statistical analysis. The reproducibility of
the SELDI-TOF system was determined, through which
each chip included a control sample (aliquoted from a
single collection from each individual), permitting
estimation of assay variability and reproducibility.
2.3 Bioinformatics and biostatistics
Construction of the classification tree was accomplished
with the 308 peaks per spectrum by the ProteinChip
Biomarker Wizard software (Ciphergen Biosystems, Fremont, CA, USA). Classification
trees split up a dataset into two bins or nodes, using one rule once in the form
of a question. The splitting decision was
defined by the presence or absence and the intensity levels of one
peak. The splitting process continued till terminal nodes or leaves were produced or
further splitting had no gain. The
classification of terminal nodes was
determined by the group
("class") of samples (i.e. PCa or control)
representing the majority of samples in the corresponding node. The
data from the Biomarker Wizard software were analyzed
with the Biomarker Pattern software based on the Gini
clustering classification tree (sensitivity: 0.05; method:
no independent testing-exploratory tree).
Comparisons between groups were performed by analysis of variance with the Biomarker Wizard software.
Statistical significance was defined as
P < 0.05.
Variable importance scores reflect the contribution
of each variable makes in classifying or predicting the
target variable, with the contribution stemming from both
the variable¡¯s role as a primary splitter and its role as a
surrogate to any of the primary splitters. The
variable used to split the root node is ranked as the most important.
This variable received a zero score, indicating that it did
not play any role in the analysis as either primary
splitters or surrogates.
3 Results
3.1 Differences in protein mass spectrum
The Seldi-TOF protein profiles of the PCa and control groups are shown in Figure 1. The SELDI-TOF
technology is particularly effective in resolving low
molecular weight (< 10 kDa). Figure 1A is a representative
protein spectrum showing the protein masses
between 2 000 kDa and 20 000 kDa of a single serum specimen.
The SELDI-TOF Biomarker Wizard program detected 308
peaks per spectrum for each corresponding sample.
Eighteen serum differential proteins were clearly identified in the
PCa group compared with the control group
(P < 0.01). However, the use of each single peak from the 18
proteins could not completely differentiate the PCa group
from the control group. The 18 proteins from the PCa
group included 14 proteins in low abundance and 4
proteins in high abundance (Table 1).
3.2 Construction and analysis of taxonomic tree
As Figure 2 shows, the classification tree that was
reflected through eight masses (M1 669, M4 300, M5 923,
M6 250, M6 652, M7 782, M15 868 and M145 875) calculated by the Biomarker Pattern program generated nine
terminal nodes (Figure 2). Of the eight masses, six (M4
300, M5 923, M6 250, M6 652, M7 782 and M15 868) were identified as being consistent with those in the
corresponding 18 differential proteins for PCa
(Table 1). Figure
3 shows the spectrum and gel maps representing the six
masses. As a result of their absence among the 18
differential proteins, the remaining two masses (M1 669 and
M145 875) are not shown. The split criteria from Node 1
to Node 8 (7.66 [M5 923], 2.535 [M15 868], 24.46 [M7
782], 9.852 [M6 250], 0.1 [M145 875], 1.66 [M4 300],
4.303 [M6 652] and 0.854 [M1 669]) directed the nine
terminal nodes. The terminal nodes 1, 3, 4, 6 and
8 represented PCa (Figure 2). The remaining terminal nodes were
normal (Figure 2). The diagnosis sensitivity and specifity
for PCa were 92.0 % and 96.7 %, respectively. The
variable importance scores of the eight masses and the
corresponding diagnosis sensibility and speciality of each
mass are summarized in Table 2. Node 1 (M5 923) had
the highest importance scores.
4 Discussion
PSA is a tissue-specific antigen of the prostate. The
serum PSA test has been used as the standard approach
for the detection of PCa for many years. Such a simple
approach has been successful for mass screening of PCa
[9]. A PSA level of 4.0-10.0 ng/mL is a significant
indication for PCa in the early period of the disease, when
clinical treatments are optimal [10]. However, many cases
of non-cancer diseases are covered in this interval, which
influences the differentiation between non-cancer diseases
and PCa. In order to overcome such drawbacks,
researchers have attempted to make use of new parameters, such
as PSA density and free PSA/total PSA, to replace the
original PSA test [8]. However, these attempts are just
adjuvant for the optimization of the PSA test and cannot
resolve the essentially low specificity of PSA. Most
studies have shown that the specificity of the PSA test is
25 % [8]. Stamey et al. [6] pointed out that it is
necessary to explore more specific proteins and more efficient
methods with the gradual disappearance of the
predominance of the PSA approach.
Proteomics techniques, especially protein array, have
been widely applied in the field of searching biomarkers
for early cancer detection [7, 11-15]. Adam et
al. [7] reported promising results with sensitivity of 83 % and
specificity of 97 % using the SELDI-TOF pattern for
serum biomarkers of PCa detection. Petricoin et
al. [14] and Qu et al. [16] reported a 95 % sensitivity at
78-83 % specificity and a 97-100 % sensitivity at 97-100 %
specificity, respectively. These data indicate that the
proteomic technique for PCa detection is superior to the
conventional PSA technique in terms of sensitivity.
Adam et al. [7] used nine masses at m/z
ratios of 4 475, 5 074, 5 382, 7 024, 7 820, 8 141, 9 149, 9 507 and
9 656, whereas Petricoin et al. [14] selected seven masses
at m/z ratios of 2 092, 2 367, 2 582, 3 080, 4 819, 5 439
and 18 220. Qu et al. [16] identified 12 major masses at
m/z ratios of 9 656, 9 720, 6 542, 6 797, 6 949, 7 024,
8 067, 8 356, 3 963, 4 080, 7 885 and 6 991 in
differentiating non-cancer from cancer, and nine masses at
m/z ratios of 7 820, 4 580, 7 844, 4 071, 7 054, 5 298, 3 486,
6 099 and 8 943 in differentiating healthy individuals from
patients with benign prostatic hyperplasia (BPH). Of note
is that, in addition to masses at m/z ratios of 7 820 identified
by Petricoin et al. [14] and Qu et
al. [16], no cross-talk between those identified by Petricoin
et al. [14], Adam et al. [7] or Qu
et al. [16] was found. Furthermore, although Adam
et al. [7] and Qu et al. [16] used the same
chip for serum extraction and the same instrument for
peak identification, their distinguishing peaks are clearly
different. In this study, we found that only three masses
at m/z ratios of 5 074, 7 850 and 9 507, as shown in Table
1, are matched with that reported by Adam et
al. [7]. We also found that the peak mass at an
m/z ratio of 7850 is much closer to the peak mass of 7 820 identified by
Adam et al. [7] and Qu et al.[16].
The results from the study of Adam et al. [7]
demonstrated that proteomic array is an optimal modality for
the diagnosis of PCa. In our study, the IMAC-Cu chip
was adopted to collect serum proteins from 178 samples.
We identified 18 differential expression proteins for the
PCa group, including 14 lower-expression proteins and
4 higher-expression proteins. Through further
classification using Biomarker Pattern software, we developed
a taxonomic tree with eight protein biomarkers and
confirmed that a combination of these eight proteins could
accurately screen PCa patients, with a sensitivity of 93 %
and specificity of 96 %. This method is superior to
using each biomarker alone, such as the 5 923 kDa protein
with a sensitivity of only 56.2 %. In terms of specificity,
our data collected from a cohort of Chinese PCa patients
are comparable to results reported by others [7, 12, 14].
In the present study, our results demonstrated that
the use of cluster analysis for 308 peaks per spectrum
could construct the optimal classification tree model. We
found eight biomarkers, six of which were located at
superior layers of the taxonomic tree with statistical
significance in discriminating between the PCa and normal
groups. Deletion of the other two proteins results in
decreased sensitivity and specificity of the decision tree
to 80 % and 81 %, respectively, indicating the
importance of the combined use of multiple marker proteins.
Our data showed that the correct resolution of a single
protein with molecular mass 5 923.29 kDa was just
56.23 %. This might explain the low specificity of the
PSA test. Our data demonstrated that the SELDI-TOF
pattern facilitated the exploration of various proteins
simultaneously. The cluster analysis of these proteins
could identify special proteins which may be not useful
alone, but may be beneficial when incorporated with other
proteins for the diagnosis of malignancies. The model of
a combination of multiple biomarkers is superior to that of
a single biomarker. Other researches [7, 11, 12, 14] have
also shown the advantage of multiple biomarkers in
different cancers.
In conclusion, this study represents the first
demonstration that the SELDI-TOF-based serum proteomic
array technique is effective for the diagnosis of PCa in
Chinese men. Such an approach is useful for the
research of other cancers requiring biomarker analysis.
Acknowledgment
This study was supported by the Special Technical
Cooperation Items between the Chinese and Japanese
Governments (59th item).
References
1 Jemal A, Murray T, Samuels A, Ghafoor A, Ward E, Thun MJ.
Cancer statistics, 2003. CA Cancer J Clin 2003; 53: 5-26.
2 Gu FL. Changing constituents of genitourinary cancer in
recent 50 years in Beijing. Chin Med J (Engl) 2003; 116:
1391-3. Erratum in: Chin Med J (Engl). 2003; 116:1643.
3 Li XM, Zhang L, Li J, Li Y, Wang HL, Ji GY,
et al. Measurement of serum zinc improves prostate cancer detection
efficiency in patients with PSA levels between 4ng/mL and 10
ng/mL. Asian J Androl. 2005; 7:323-8.
4 Zhang HF, Wang HL, Xu N, Li SW, Ji GY, Li XM,
et al. Mass screening of 12,027 elderly men for prostate carcinoma by
measuring serum prostate specific antigen. Chin Med J (Engl)
2004; 117: 67-70.
5 Kuwahara M, Tochigi T, Kawamura S, Ogata Y, Xu N, Wang
H, et al. Mass screening for prostate cancer: a comparative
study in Natori, Japan and Changchun, China. Urology 2003;
61: 137-41.
6 Stamey TA, Caldwell M, McNeal JE, Nolley R, Hemenez M,
Downs J. The prostate specific antigen era in the United
States is over for prostate cancer: what happened in the last 20
years? J Urol 2004; 172: 1297-301.
7 Adam BL, Qu Y, Davis JW, Ward MD, Clements MA, Cazares
LH, et al. Serum protein fingerprinting coupled with a
pattern-matching algorithm distinguishes prostate cancer from benign
prostate hyperplasia and healthy men. Cancer Res 2002; 62:
3609-14.
8 Zhang L, Ji G, Li X, Wang W, Gao H, Pan Y,
et al. Early diagnosis of prostate cancer using free/total prostate-specific
antigen ratio with population-based screening data. Zhonghua
Nan Ke Xue 2004; 10: 582-5.
9 Stamey TA, Yang N, Hay AR, McNeal JE, Freiha FS, Redwine
E. Prostate-specific antigen as a serum marker for
adenocarcinoma of the prostate. N Engl J Med 1987; 317: 909-16.
10 Stamey TA, Kabalin JN, McNeal JE, Johnstone IM, Freiha F,
Redwine EA, et al. Prostate specific antigen in the diagnosis
and treatment of adenocarcinoma of the prostate. II. Radical
prostatectomy treated patients. J Urol 1989; 141: 1076-83.
11 Xiao X, Liu D, Tang Y, Guo F, Xia L, Liu J, He D.
Development of proteomic patterns for detecting lung cancer. Dis
Markers 2003; 19: 33-9.
12 Xiao XY, Zhao X, Liu J, Guo F, Liu D,He DC. Discovery of
laryngeal carcinoma by serum proteomic pattern analysis. Sci
China C Life Sci, 2004, 47: 219-23.
13 Petricoin EF 3rd, Ornstein DK, Paweletz CP, Ardekani A,
Hackett PS, Hitt BA, et al. Serum proteomic patterns for
detection of prostate cancer. J Natl Cancer Inst 2002, 94:
1576-8.
14 Petricoin EF, Ornstein DK, Liotta LA. Clinical proteomics:
applications for prostate cancer biomarker discovery and
detection. Urol Oncol 2004; 22: 322-8.
15 Petricoin EF, Ardekani AM, Hitt BA, Levine PJ, Fusaro VA,
Steinberg SM, et al. Use of proteomic patterns in serum to
identify ovarian cancer. Lancet 2002; 359: 572-7.
16 Qu Y, Adam BL, Yasui Y, Ward MD, Cazares LH, Schellhammer
PF, et al. Boosted decision tree analysis of surface-enhanced
laser desorption/ionization mass spectral serum profiles
discriminates prostate cancer from noncancer patients. Clin Chem
2002; 48:1835-43.
|