Identification of Potential Biomarkers for Ovarian ... - ACS Publications

Article pubs.acs.org/jpr

Identification of Potential Biomarkers for Ovarian Cancer by Urinary Metabolomic Profiling Tao Zhang,†,‡,Δ Xiaoyan Wu,†,Δ Chaofu Ke,† Mingzhu Yin,§ Zhenzi Li,† Lijun Fan,† Wang Zhang,† Haiyu Zhang,† Falin Zhao,∥ Xiaohua Zhou,⊥ Ge Lou,*,§ and Kang Li*,† †

Department of Epidemiology and Biostatistics, Public Health School, Harbin Medical University, Harbin, P. R. China Department of Epidemiology and Biostatistics, Public Health School, Shandong University, Jinan, P. R. China § Department of Gynecology Oncology, The Tumor Hospital, Harbin Medical University, Harbin, P. R. China ∥ Department of Social Medicine, College of Medicine and Health Management, Hangzhou Normal University, Hangzhou, P. R. China ⊥ Department of Biostatistics, School of Public Health and Community Medicine, University of Washington, Seattle, Washington, United States ‡

S Supporting Information *

ABSTRACT: To evaluate the application of urinary metabolomics on discovering potential biomarkers for epithelial ovarian cancer (EOC), urine samples from 40 preoperative EOC patients, 62 benign ovarian tumor (BOT) patients, and 54 healthy controls were collected and analyzed with ultraperformance liquid chromatography quadrupole time-of-flight mass spectrometry (UPLC-QTOF/MS). Good separations were obtained for EOC vs BOT, EOC vs healthy controls analyzed by partial least-squares discriminant analysis, or principal component analysis. Twenty-two ascertained metabolomic biomarkers were found to be disturbed in several metabolic pathways among EOC patients, including nucleotide metabolism (pseudouridine, N4-acetylcytidine), histidine metabolism (L-histidine, imidazol-5-yl-pyruvate), tryptophan metabolism (3-indolelactic acid), and mucin metabolism (3′-sialyllactose and 3-sialyl-N-acetyllactosamine). In addition, the concentrations of some urinary metabolites of 18 postoperative EOC patients among the 40 EOC patients changed significantly compared with those of their preoperative condition, and four of them suggested recovery tendency toward normal level after surgical operation, including N4-acetylcytidine, pseudouridine, urate-3-ribonucleoside, and succinic acid. These metabolites would be highly postulated to be associated with EOC. In conclusion, our study demonstrated that urinary metabolomics analysis by UPLC-QTOF/MS, performed in a minimally noninvasive and convenient manner, possessed great potential in biomarker discovery for EOC. KEYWORDS: metabolomics, epithelial ovarian cancer, urine, biomarker, mass spectrometry

■

INTRODUCTION Epithelial ovarian cancer (EOC), the most common form of malignant ovarian tumor, is the leading cause of death among gynecological cancers.1 Almost 90% of EOC patients were diagnosed at an advanced stage (III/IV), with a poor 5-year survival rate of less than 30%.2 The cancer antigen 125 (CA125) and transvaginal ultrasound are two major techniques used to diagnose EOC currently; however, the identification of EOC was unsatisfied by them in clinical practice. The increased level of CA125 is not present in all EOC women, and it could also occur in women with other cancers (e.g., pancreatic, breast, and lung) as well as benign diseases (e.g., uterine fibroids and benign ovarian tumors (BOT)).3 The diagnostic accuracy of transvaginal ultrasound has been reported to be better than that of CA125, while transvaginal ultrasound is invasive and expensive, and its accuracy mainly depends on the subjective experience of the examiners.4 © XXXX American Chemical Society

Metabolomics, the global quantitative assessment of endogenous small molecule metabolites within a biological system,5 has been successfully utilized in cancer biomarker discovery using specimens of biofluids or tissues,6 such as breast cancer,7 colon cancer,8 oral cancer,9 and prostate cancer.10 Recently, metabolomic analyses for EOC were conducted with mass spectrometry (MS) or nuclear magnetic resonance (NMR), which revealed significantly perturbed metabolic expression in EOC patients compared with healthy women.2,11−16 Among them, two studies concentrated on urinary metabolic signatures of EOC patients. Urinary metabolites are the downstream products of cellular processes, and urinary metabolomic analysis could provide complementary information to metabolomic analysis of serum and tissue. In addition, urinary test is noninvasive, inexpensive, Received: October 11, 2012

A

dx.doi.org/10.1021/pr3009572 | J. Proteome Res. XXXX, XXX, XXX−XXX

Journal of Proteome Research

Article

Table 1. Demographic and Clinical Chemistry Characteristics of Epithelial Ovarian Cancer (EOC) Patients, Benign Ovarian Tumor (BOT) Patients, Healthy Controls, and Postoperative EOC Patients no. of subjects age (median/range) wt (median/range) menopause postpreundocumented CA125 (median/range) FIGO stage I II III IV undocumented histology type serous adenocarcinoma others undocumented histopath differen well differen moder differen poor differen undocumented

preop EOC patients

BOT patients

healthy controls

40 53.9/33.4−68.2 59/35−86

62 41.1/19.6−67.4 61/42.5−93

54 45.0/26.2−57.5 60/45−80

18 51.8/40.0−68.2 60/35−86

21 19

14 35 13 15.3/1.9−249.7

5 45 4 12.1/4.6−336.2

9 9

373.2/9.1−5000

215.9/17.8−2843

7 5 20 2 6

5 1 10 1 1

17 5 18

9 3 6

0 1 19 20

0 1 9 8

Federation of Gynecology and Obstetrics (FIGO) staging system for ovarian cancer:17 stage I, 7 patients; stage II, 5 patients; stage III, 20 patients; stage IV, 2 patients; undocumented, 6 patients. The demographic and clinical chemistry characteristics of enrolled subjects are shown in Table 1. All urine samples were collected in the morning before breakfast, centrifuged at 3000 rpm for 15 min to remove impurities, and then frozen immediately and stored at −80 °C until analysis.

and convenient in clinical practice. All these advantages of urinary metabolomics make it applicable in biomarker discovery. Slupsky et al. found differences of urinary metabolic profiling between 50 female EOC and 72 female volunteers analyzed by NMR.16 Woo et al. reported different metabolite patterns in urine between 9 patients with ovarian cancer and 22 normal controls analyzed by LC-MS.14 However, both studies employed healthy women as their controls and one of them was limited by small sample size. In addition, the results reported in the two studies were diverse, which would provide us evidence that different metabolites related to EOC would be detected from studies with different research designs. Therefore, further research revealing differences in urinary metabolic profiling between EOC patients and other than normal controls is needed, which would provide more information for the complete elucidation of the underlying metabolic mechanisms of EOC. In this study, we performed a urinary metabolomic analysis on ultraperformance liquid chromatography quadrupole timeof-flight mass spectrometry (UPLC-QTOF/MS) to discriminate 40 preoperative EOC patients from 62 women with BOT or 54 healthy controls (HC) and tried to reveal the differential metabolites between these groups. In addition, we investigated the potential of urinary metabolomics in evaluating the metabolic changes between pre- and postoperation in 18 EOC patients.

■

postop EOC patients

Sample Preparation and Pretreatment

Prior to sample preparation and UPLC-QTOF/MS analysis, all the urine samples were randomized along with 18 pooled quality control (QC) urine samples. The pooled QC samples were prepared by mixing equal amounts of urine samples from three EOC patients and three healthy controls. All urine samples were thawed in a 4 °C water bath for 20−30 min and then centrifuged at 14000g for 10 min. An aliquot of 150 μL of supernatant was transferred to a 2 mL centrifuge tube, and then 450 μL of water was added to the tube to make the dilution ratio of urine water be 1:3. The total 600 μL solutions were vortexed for 2 min, placed into the sampling vial, and stored at 4 °C pending UPLC-QTOF/MS analysis. UPLC-QTOF/MS Analysis of Urine Samples

A 2 μL aliquot of the pretreated sample was injected into a HSS T3 2.1 mm × 100 mm × 1.8 μm column (Waters, Milford, USA) held at 35 °C using an Acquity ultraperformance liquid chromatography system (Waters, Milford, USA). The UPLC mobile phase consisted of water with 0.1% formic acid (solution A), and acetonitrile with 0.1% formic acid (solution B). The column was eluted with a linear gradient of 2−20% B for 0−6 min, 20−35% B for 6−7 min, 35−70% B for 7−8 min, 70−98% B for 8−10 min, and kept 98% B for 10−11.5 min. The gradient was then changed to 70% B for 11.5−12 min, 20% B for 12−13 min, and finally to 2% B for 13−15 min. The flow rate was 0.35 mL/min, and all of the samples were maintained at 4 °C during the analysis.

MATERIALS AND METHODS

Urine Collection

Urine samples of 40 EOC patients preoperation and 18 patients postoperation on the seventh day, 62 BOT patients, and 54 healthy controls were collected from Department of Gynecology of Harbin Medical University Tumor Hospital (Harbin, China) between September 2009 and November 2010. All participants signed informed consent forms under local research ethics committee approval. Each EOC patient was diagnosed with different histopathological features and staged according to the International B



Article

Figure 1. Typical UPLC-QTOF/MS chromatograms of urine samples for a healthy woman, a benign ovarian tumor (BOT) patient, and an epithelial ovarian cancer (EOC) patient in the preoperative and postoperative situation.

PLS-DA model by comparing the goodness of fit (R2 and Q2) of the PLS-DA models with the goodness of fit of 100 Y-permutated models.26 Meanwhile, the nonparametric Kruskal−Wallis rank sum test was performed to determine the significance of each metabolite, and the relevant false discovery rates (FDR) based on the p-values were estimated in the context of multiple testing.27 A potential metabolic biomarker was selected when the value of its VIP was more than 1 and FDR was less than 0.05. Statistical analysis was performed on the R platform,28 with the exception that PCA and PLS-DA were carried out on SIMCA-p.

Mass spectrometry was performed using a Waters Micromass Q-TOF (Waters, Manchester, U.K.) equipped with an electrospray ionization source operating in negative-ion mode (ESI−). The source temperature was set at 110 °C with a cone gas flow rate of 50 L/h. Meanwhile, the desolvation gas temperature was 300 °C with a desolvation gas flow rate of 650 L/h. The capillary voltage and the cone voltage were set to 2.8 kV and 35 V, respectively. Centroid data were collected from 50 to 1000 m/z with a scan time of 0.4 s, an interscan delay of 0.1 s, and a lock spray frequency of 10 s.

■

Data Preprocessing and Annotation

The raw UPLC-QTOF/MS ESI data were first transformed to NetCDF files by Databridge (Waters, Manchester, U.K.), and then the files were imported to the xcms package in the R platform for preprocessing, including nonlinear retention time (RT) alignment, matched filtration, peak detection, and peak matching.18 Full width at half-maximum (fwhm) was set to 10, and the retention time window was set to 10 (bw = 10), while the values of other parameters were default.19 The preprocessing results by xcms offered a three-dimensional matrix containing arbitrarily assigned peak indices of RT and mass-to-charge ratio (m/z) pairs, sample names, and ion-intensity information. The R package CAMERA was used for annotation of isotope peaks, adducts, and fragments in the processed peak data by xcms.20,21 Normalization to total peak intensities for each sample was done before statistical analysis.22−24

RESULTS The typical UPLC-QTOF/MS chromatograms are shown in Figure 1. The final data table contained 353 variables (chromatographic peaks). Metabolic Profiles of EOC, BOT, and Healthy Controls

Although the PCA scores plots only showed a tendency of difference in metabolic profiling for EOC vs HC and EOC vs BOT (Figures S1 (A and B) in the Supporting Information), the PLS-DA score plots revealed a clear separation for EOC vs HC (Figure 2A) and EOC vs BOT (Figure 2C). The PLS-DA model for EOC vs HC contained four factors with the performance of R2X = 0.347, R2Ycum = 0.817, and Q2cum = 0.376. And the validation plot presented in Figure 2B strongly indicated the validity of the PLS-DA model, since the Q2 regression line in blue had a negative intercept and all permuted R2-values in green on the left were lower than the original point of the R2-value on the right. The PLS-DA model for EOC vs BOT contained four latent variables with the performance of R2X = 0.332, R2Ycum = 0.776, and Q2cum = 0.301, and the validation plot supported the validity of the PLS-DA model (Figure 2D). For employing metabolomic profiles to discriminate BOT patients from healthy controls, either PCA or PLS-DA showed overlaid results between

Statistical Analysis

Both the unsupervised method (principal component analysis, PCA) and the supervised method (partial least-squares discriminant analysis, PLS-DA) were employed to reveal the global metabolic changes of EOC vs BOT, EOC vs HC, and pre- vs postoperative EOC patients using SIMCA-p 11.5 (Umetrics AB, Umea, Sweden), and the corresponding variable importance in the projection (VIP values) was calculated in the PLS-DA model as well.25 A validation plot was used to assess the validity of the C



Article

Figure 2. (A) PLS-DA three-dimensional scores plot for the top three latent variables discriminates epithelial ovarian cancer (EOC, ●) patients from healthy controls (△), with fitting and predictive performance (4 latent variables, R2X = 0.347, R2Ycum = 0.817, Q2cum = 0.376). (B) Validation plot obtained from 100 permutation tests. (C) PLS-DA three-dimensional scores plot for the top three latent variables discriminates epithelial ovarian cancer (EOC, ●) patients from benign ovarian tumor (BOT, △), with fitting and predictive performance (4 latent variables, R2X = 0.332, R2Ycum = 0.776, Q2cum = 0.301). (D) Validation plot obtained from 100 permutation tests.

Figure 3. (A) PLS-DA scores plot for the first component discriminates preoperative (●) and postoperative (△) EOC patients, with fitting and predictive performance (1 component, R2X = 0.258, R2Ycum = 0.787, Q2cum = 0.612). (B) Validation plot obtained from 100 permutation tests.

patients (for PLS-DA: 1 component, with R2X = 0.258, R2Ycum = 0.787, and Q2cum = 0.612). The validation plot supported the validity of this PLS-DA model (Figure 3B).

these two groups (Figures S1 (D) and S2 in the Supporting Information, respectively). Metabolic Changes between Pre- and Postoperative EOC Patients

Discovery and Identification of Metabolic Biomarkers

Both the PCA scores plot (Figure S1 (C) in the Supporting Information) and the PLS-DA scores plot (Figure 3A) demonstrated excellent separation between pre- and postoperative EOC

A total of 42 urine metabolites were selected as potential biomarkers to classify EOC vs HC (34 metabolites) or EOC vs BOT (23 metabolites), and 13 of them were statistically significant D



Article

Table 2. Urinary Metabolic Biomarkers for EOC vs Healthy Controls, EOC vs BOT, or Pre- vs Postoperative EOC Patientsa EOC vs healthy control marker v80 v26 v56 v160 v159 v78 v117 v272 v73 v63 v345 v197 v67 v342 v45 v57 v74 v59 v110 v44 v49 v348

identity imidazol-5-yl-pyruvate L-histidine

N-acetylneuraminate 9-phosphate N4-acetylcytidine urate-3-ribonucleoside pseudouridine 3-dehydroquinic acid 3-indolelactic acid Nα-acetyl-L-glutamine succinic acid LPA(P-16:0e/0:0) (S)-reticuline taurine prasterone sulfate N-acetylneuraminic acid 3-sialyl-N-acetyllactosamine β-nicotinamide mononucleotide prolylhydroxyproline selenocystathionine 3′-sialyllactose N-acetylgalactosamine 4-sulfate glycodeoxycholate

vipb 1.26 1.58 1.55 1.24 1.57 1.31

p-valuec 0.00727 0.00394 0.00101 0.0116 0.00019 0.00727

1.56 1.31

0.00852 0.00912

1.46 1 1.33 1.55 1.62 1.08 1.94 1.24 1.52 1.23 1.18

0.00217 0.00135 0.00316 0.000196 9.03 × 10−05 0.00466 2.38 × 10−06 0.0121 4.43 × 10−05 0.000981 0.0121

EOC vs BOT

FCd

vip

p-value

pre- vs postoperative FC

1.3 −0.89 1.06 0.37 1.09 9.09

p-value

FC

1.21 1.24 1.08

0.00361 0.00193 0.0713

0.72 1.62 14.63

1.26 1.01 1.14

0.00651 0.0247 0.0124

−5.18 1.33 0.37

1.45 1.16

0.00417 0.001216

1.73 7.51

1.48 1.51

0.005742 0.001617

1.67 2.09

1.66 1.63

0.000102 0.001579

2.44 1.61

1.09

0.0209

−1.33

1.39 1.6

0.005863 0.001617

0.67 2.13

1.53

0.00031

−3.74

−1.16 0.22 2.12 0.87 −16.36 2.52 1.97 0.44 4.17 0.63 2.36 1.46 −2.15

vip

%RSDe 6.5 5.2 5.2 3 7.4 8.1 12.4 12.1 8.2 NA 6.7 24.3 4.4 14.7 5.5 3 7.3 4.4 NA 2.7 3.2 3.9

a

Abbreviations: EOC, epithelial ovarian cancer; BOT, benign ovarian tumor. bVariable importance in the projection (VIP) was obtained from PLS-DA with a threshold of 1.0. cThe p-value was calculated from the nonparametric Kruskal−Wallis rank sum test. dFold change (FC) was calculated from the arithmetic mean values of each group; FC with a positive value indicates that the concentration of a certain metabolite is up-regulated in preoperative EOC compared to healthy controls, BOT, or postoperative EOC. eRelative standard deviation (%RSD) in the levels of marker metabolites in the 18 quality control samples.

with the value of the area of the receiver operating characteristic curve (AUC) obtained by these nine differentiated metabolites of 0.828. Furthermore, the possible matching pathways for these 22 identified biomarkers were searched in online pathway databases of KEGG (http://www.genome.jp/kegg/) and SMPDB (http://www.smpdb.ca/), which mainly involved nucleotide metabolism (pseudouridine, N4-acetylcytidine), histidine metabolism (L-histidine, imidazol-5-yl-pyruvate), tryptophan metabolism (3-indolelactic acid), and mucin metabolism (3′-sialyllactose and 3-sialyl-N-acetyllactosamine).

between pre- and postoperative EOC patients (Table S1 in the Supporting Information). Among them, 22 metabolites were identified through MS/MS (tandem mass spectroscopy) experiments and online databases: 19 metabolites for EOC vs HC and 8 metabolites for EOC vs BOT (Table 2). The average %RSDs of these 22 metabolites were 83.7, 77.2, 74.8, and 93.6 for the preoperative EOC patients, BOT women, health controls, and postoperative EOC patients, respectively, which suggested a consistent intragroup variation among the four groups. The % RSD of these 22 metabolites from urine QC samples varied from 2.7 to 24.3% with a median of 6.5%, which indicated the robustness of our metabolic profiling platform, and this robustness could be suggested by the PCA scores plot comprising pre- and postoperative EOC, BOT, healthy control, and the QCs as well (Figure S4 in the Supporting Information). Detailed information about metabolites identification was shown in Table S2 and Figure S3 in the Supporting Information, and five of them, including L-histidine, N-acetylglutamine, succinic acid, taurine, and 3-sialyl-N-acetyllactosamine were additionally verified by external reference standards. The MS/MS spectra of these five biomarkers in our urine sample matched well with those of the reference standards, respectively (the identification of 3-sialylN-acetyllactosamine was shown in Figure S5 in the Supporting Information). In addition, the concentrations of nine metabolites (imidazol-5-yl-pyruvate, N4-acetylcytidine, pseudouridine, succinic acid, (S)-reticuline, N-acetylneuraminic acid, 3-sialyl-N-acetyllactosamine, β-nicotinamide mononucleotide, and 3′-sialyllactose) were found to be significantly different in EOC patients who were in different FIGO stages (Figure S6 in the Supporting Information), and the classification between EOC patients in early stages (I and II) and those in advanced stages (III and IV) was detected in our study,

■

DISCUSSION Our study suggested that the urinary metabolomics analyzed by UPLC-QTOF/MS could be used to discriminate EOC from BOT/HC, and the 22 ascertained urinary metabolic biomarkers were highly possible to be associated with EOC. In addition, the metabolomic changes were found between pre- and postoperative EOC patients, and some metabolites showed recovery tendency toward the normal level. In line with previous results of metabolomic analysis from serum, urine, or tissue samples,11−16,29 our findings demonstrated good separations between EOC and BOT/HC as well. Meanwhile, we found that the ascertained urinary biomarkers could improve the diagnostic accuracy of CA125, the most commonly used plasma biomarker in the clinic currently,3,4,30 especially for differentiating between EOC and BOT (the values of AUC of combination of eight metabolites and CA125: 0.870; CA125: 0.803; 8 metabolites: 0.728). Our study further confirmed that urinary metabolomics could be used to differentiate EOC patients from healthy controls or BOT patients through urine samples analyzed by UPLC-QTOF/MS. E



Article

Among the 22 ascertained metabolic biomarkers, five of them (3-indolelactic acid, N-acetylglutamine, succinic acid, taurine, prasterone sulfate) have been previously reported,15,16,31−35 in which significantly disturbed concentrations of these five biomarkers in EOC patients compared to normal controls or BOT patients were not completely consistent with our results. However, only one of these studies employed urine metabolomics analyzed by NMR in which succinic acid was reported, and others employed metabolic profiles from serum or tissue samples. In addition, the eight urinary metabolic biomarkers identified to discriminate EOC from BOT in the current study were quite different from the six biomarkers in our recent study of differentiating 80 EOC from 90 BOT women by plasma metabolomics,29 which suggested that urinary metabolomic analysis might provide complementary information to plasma metabolomic analysis for the complete elucidation of the underlying metabolic mechanisms of EOC. Notably, the possible related pathways of the ascertained urinary metabolic biomarkers were further investigated in the current study. Urinary concentrations of some nucleosides, N4-acetylcytidine, pseudouridine, and urate-3-ribonucleoside, were increased in EOC patients compared to healthy controls. These elevated nucleosides should result primarily from the degradation of a large amount of synthesized RNA,36,37 as most of the modified nucleosides coming from the degradation of tRNA were excreted in urine.38 Our results suggested enhanced nucleotide metabolism in EOC patients. Meanwhile, the urinary L-histidine in EOC patients was down-regulated relative to that in healthy controls, which was consistent with the results of urinary metabolomic study on colorectal cancer.33 Imidazol-5-yl-pyruvate, one metabolic product of L-histidine, was up-regulated relative to that in healthy controls, which further indicated the disturbance of the L-histidine metabolic pathway in EOC patients. It has been suggested that the biosynthesis of histidine is inherently linked to the pathways of nucleotide formation through 5-phosphoribosyl 1-pyrophosphate (PRPP).39 In addition, N-acetylglutamine, a liquid-stable source of glutamine,40,41 was found down-regulated in EOC patients, and it is clear that glutamine is engaged in the PRPP-medicated nucleotide metabolism.36 The abnormal L-histidine metabolism and nucleotide metabolism found in our study might indicate the turbulence of the PRPP-medicated histidine-nucleotide superpathway in EOC patients (shown in Figure 4).

patients, since 3-indolelactic acid is a direct degradation product of L-tryptophan.42 Two oligosaccharides, 3′-sialyllactose and 3-sialyl-N-acetyllactosamine, were found to be up-regulated in EOC patients as compared with those in both BOT patients and healthy controls. Oligosaccharides are known to make up a large fraction of the mass of mucins,43 and mucins have been suggested to be associated with the pathogenesis of cancer adenocecinomas.44 It is well-known that elevated serum mucins of CA125 (MUC16) and MUC1 can be found in EOC patients.45 The high urinary concentrations of 3′-sialyllactose and 3-sialyl-N-acetyllactosamine in our study would provide complementary information to serum test of mucins. Urinary concentration of succinic acid in EOC patients was observed to be higher than that in healthy controls. Studies have suggested that accumulated succinic acid can contribute to prolyl hydroxylase (PHD) inhibition and HIF-α hydroxylation, which in turn induces expression of genes that facilitates angiogenesis, metastasis, and metabolism of aggressive tumors.46 β-Nicotinamide mononucleotide, which is involved in the biosynthesis of nicotinamide adenine dinucleotide (NAD+),47 was found elevated in the urine of EOC patients compared to their healthy controls. NAD+, carrying electrons from one reaction to another, plays a critical role in energy production.48 Therefore, the increase of β-nicotinamide mononucleotide, resulting in the elevation of NAD+, would meet larges amount of energy consumption of cancer cells growth. In addition, taurine was significantly up-regulated in EOC patients compared to that in healthy controls, which might indicate the disturbance of taurine metabolism but need to be validated in further studies. Among the differential metabolites of EOC patients before and after surgical operation, four of them (N4-acetylcytidine, pseudouridine, urate-3-ribonucleoside, and succinic acid) showed recovery tendency toward normal level, while others (N-acetylglutamine, 3-indolelactic acid, 3′-sialyllactose, and 3-sialyl-N-acetyllactosamine) showed completely opposite tendency to recovery. Our results demonstrated metabolomic changes occurred between pre- and postoperative EOC patients, but these changes could be attributed to nutritional supplementation, increased oxidative stress, or surgical curative effect.33 The metabolites showing recovery tendency, which are most likely due to the surgical curative effect, would be most highly postulated to be associated with EOC. However, these metabolic changes were based on small number of EOC patients and deserve further investigation in large cohorts in the future. Although age and menopausal status were imbalanced between EOC and controls in this study, the potential 22 biomarkers were still statistically significant after being adjusted for age and menopause in the logisitic regression models (data not shown). In addition, the classification between EOC patients in early stages (I and II) and those in advanced stages (III and IV) was detected in our study; however, the classification of EOC patients who were in different FIGO stages (I, II, III, and IV) was not detected, which might be limited by the relatively small sample size. Further investigations consisting of larger sample sizes are needed to validate our findings. The findings in our study showed the valuable potential of urinary metabolomics to discover biomarkers in discriminating EOC patients from BOT patients or healthy controls. The urinary biomarkers, up- or down-regulated in EOC patients compared to those in BOT patients or healthy controls, would suggest the disturbed metabolic pathways in EOC patients, including metabolisms of nucleotide, histidine, tryptophan, and mucin. Our results also indicated that the urinary metabolomics

Figure 4. Disturbance of the PRPP-medicated histidine-nucleotide superpathway in ovarian cancer patients.

Elevated urinary concentration of 3-indolelactic acid was observed in EOC patients relative to BOT patients, which suggested the disturbed metabolic pathway of tryptophan in EOC F



Article

(7) Claudino, W. M.; Quattrone, A.; Biganzoli, L.; Pestrin, M.; Bertini, I.; Di Leo, A. Metabolomics: available results, current research projects in breast cancer, and future applications. J. Clin. Oncol. 2007, 25 (19), 2840−6. (8) Qiu, Y.; Cai, G.; Su, M.; Chen, T.; Zheng, X.; Xu, Y.; Ni, Y.; Zhao, A.; Xu, L. X.; Cai, S.; Jia, W. Serum metabolite profiling of human colorectal cancer using GC-TOFMS and UPLC-QTOFMS. J. Proteome Res. 2009, 8 (10), 4844−50. (9) Zhou, J.; Xu, B.; Huang, J.; Jia, X.; Xue, J.; Shi, X.; Xiao, L.; Li, W. 1H NMR-based metabonomic and pattern recognition analysis for detection of oral squamous cell carcinoma. Clin. Chim. Acta 2009, 401 (1−2), 8−13. (10) Sreekumar, A.; Poisson, L. M.; Rajendiran, T. M.; Khan, A. P.; Cao, Q.; Yu, J.; Laxman, B.; Mehra, R.; Lonigro, R. J.; Li, Y.; Nyati, M. K.; Ahsan, A.; Kalyana-Sundaram, S.; Han, B.; Cao, X.; Byun, J.; Omenn, G. S.; Ghosh, D.; Pennathur, S.; Alexander, D. C.; Berger, A.; Shuster, J. R.; Wei, J. T.; Varambally, S.; Beecher, C.; Chinnaiyan, A. M. Metabolomic profiles delineate potential role for sarcosine in prostate cancer progression. Nature 2009, 457 (7231), 910−4. (11) Odunsi, K.; Wollman, R. M.; Ambrosone, C. B.; Hutson, A.; McCann, S. E.; Tammela, J.; Geisler, J. P.; Miller, G.; Sellers, T.; Cliby, W.; Qian, F.; Keitz, B.; Intengan, M.; Lele, S.; Alderfer, J. L. Detection of epithelial ovarian cancer using 1H-NMR-based metabonomics. Int. J. Cancer 2005, 113 (5), 782−8. (12) Garcia, E.; Andrews, C.; Hua, J.; Kim, H. L.; Sukumaran, D. K.; Szyperski, T.; Odunsi, K. Diagnosis of early stage ovarian cancer by (1)h NMR metabonomics of serum explored by use of a microflow NMR probe. J. Proteome Res. 2011, 10 (4), 1765−71. (13) Chen, J.; Zhang, X.; Cao, R.; Lu, X.; Zhao, S.; Fekete, A.; Huang, Q.; Schmitt-Kopplin, P.; Wang, Y.; Xu, Z.; Wan, X.; Wu, X.; Zhao, N.; Xu, C.; Xu, G. Serum 27-nor-5beta-cholestane-3,7,12,24,25 pentol glucuronide discovered by metabolomics as potential diagnostic biomarker for epithelium ovarian cancer. J. Proteome Res. 2011, 10 (5), 2625−32. (14) Woo, H. M.; Kim, K. M.; Choi, M. H.; Jung, B. H.; Lee, J.; Kong, G.; Nam, S. J.; Kim, S.; Bai, S. W.; Chung, B. C. Mass spectrometry based metabolomic approaches in urinary biomarker study of women’s cancers. Clin. Chim. Acta 2009, 400 (1−2), 63−9. (15) Guan, W.; Zhou, M.; Hampton, C. Y.; Benigno, B. B.; Walker, L. D.; Gray, A.; McDonald, J. F.; Fernandez, F. M. Ovarian cancer detection from metabolomic liquid chromatography/mass spectrometry data by support vector machines. BMC Bioinf. 2009, 10, 259. (16) Slupsky, C. M.; Steed, H.; Wells, T. H.; Dabbs, K.; Schepansky, A.; Capstick, V.; Faught, W.; Sawyer, M. B. Urine metabolite analysis offers potential early diagnosis of ovarian and breast cancers. Clin. Cancer Res. 2010, 16 (23), 5835−41. (17) Teneriello, M. G.; Park, R. C. Early detection of ovarian cancer. CaCancer J. Clin. 1995, 45 (2), 71−87. (18) Smith, C. A.; Want, E. J.; O’Maille, G.; Abagyan, R.; Siuzdak, G. XCMS: processing mass spectrometry data for metabolite profiling using nonlinear peak alignment, matching, and identification. Anal. Chem. 2006, 78 (3), 779−87. (19) Smith, C. A. LC/MS Preprocessing and Analysis with xcms. Memory 2005, 1−13. (20) Tautenhahn, R.; Böttcher, C.; Neumann, S., Annotation of LC/ ESI-MS Mass Signals. In Bioinformatics Research and Development; Hochreiter, S., Wagner, R., Eds.; Springer: Berlin/Heidelberg, 2007; Vol. 4414, pp 371−380. (21) Kuhl, C.; Tautenhahn, R.; Neumann, S. LC-MS Peak Annotation and Identification with CAMERA. 2010, October, 1−14 (http:// www.bioconductor.org/packages/release/bioc/vignettes/CAMERA/ inst/doc/CAMERA.pdf). (22) Katajamaa, M.; Oresic, M. Data processing for mass spectrometrybased metabolomics. J. Chromatogr., A 2007, 1158 (1−2), 318−28. (23) van den Berg, R. A.; Hoefsloot, H. C.; Westerhuis, J. A.; Smilde, A. K.; van der Werf, M. J. Centering, scaling, and transformations: improving the biological information content of metabolomics data. BMC Genomics 2006, 7, 142.

could provide evidence on ascertaining the metabolites which may relate to EOC, when they show recovery tendency after surgical operation. In conclusion, the urinary metabolomics possessed great potential in biomarker discovery and investigation of the underlying metabolic mechanisms for EOC.

■

ASSOCIATED CONTENT

S Supporting Information *

Strategy of metabolite identification; list of 42 selected biomarkers (Table S1) and detailed identification information of 22 ascertained metabolites (Table S2); PCA score plots for EOC vs healthy controls (Figure S1 (A)), EOC vs BOT (Figure S1 (B)), BOT vs healthy controls (Figure S1 (C)), and BOT vs healthy controls (Figure S1 (D)); PLS-DA score plot and validation plot for BOT vs healthy controls (Figure S2); typical process of identifying potential biomarkers (Figure S3 and S5); PCA scores plot comprising EOC, postoperative EOC, BOT, healthy controls, and quality controls (Figure S4); and metabolite profiles of potential biomarkers among epithelial ovarian cancer (EOC) patients with different FIGO stages (Figure S6). This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*K.L.: address, Department of Epidemiology and Biostatistics, 157 Baojian Road, Harbin 150081, P. R. China; phone, 86451-87502939; fax, 86-451-87502831; e-mail, [email protected]. edu.cn. G.L.: address, Department of Gynecology Oncology, 157 Baojian Road, Harbin 150081, P. R. China; phone, 86451-86298303; fax, 86-451-86298303; e-mail, [email protected]. edu.cn. Author Contributions Δ

These two authors made equal contributions.

Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was funded by the National Natural Science Foundation of China (Project Number 81172767), and sample analysis was supported in part by a Public Health College of Harbin Medical University grant.

■

REFERENCES

(1) Jemal, A.; Siegel, R.; Xu, J.; Ward, E. Cancer statistics, 2010. Ca Cancer J. Clin. 2010, 60 (5), 277−300. (2) Jacobs, I. J.; Menon, U. Progress and challenges in screening for early detection of ovarian cancer. Mol. Cell Proteomics 2004, 3 (4), 355− 66. (3) Sjovall, K.; Nilsson, B.; Einhorn, N. The significance of serum CA 125 elevation in malignant and nonmalignant diseases. Gynecol. Oncol. 2002, 85 (1), 175−8. (4) Van Calster, B.; Timmerman, D.; Bourne, T.; Testa, A. C.; Van Holsbeke, C.; Domali, E.; Jurkovic, D.; Neven, P.; Van Huffel, S.; Valentin, L. Discrimination between benign and malignant adnexal masses by specialist ultrasound examination versus serum CA-125. J. Natl. Cancer Inst. 2007, 99 (22), 1706−14. (5) Nicholson, J. K.; Lindon, J. C. Systems biology: Metabonomics. Nature 2008, 455 (7216), 1054−6. (6) Spratlin, J. L.; Serkova, N. J.; Eckhardt, S. G. Clinical applications of metabolomics in oncology: a review. Clin. Cancer Res. 2009, 15 (2), 431−40. G



Article

(24) Sysi-Aho, M.; Katajamaa, M.; Yetukuri, L.; Oresic, M. Normalization method for metabolomics data using optimal selection of multiple internal standards. BMC Bioinf. 2007, 8, 93. (25) Boulesteix, A. L.; Strimmer, K. Partial least squares: a versatile tool for the analysis of high-dimensional genomic data. Briefings Bioinf. 2007, 8 (1), 32−44. (26) Wiklund, S.; Nilsson, D.; Eriksson, L.; Sjöström, M.; Wold, S.; Faber, K. A randomization test for PLS component selection. J. Chemom. 2007, 21 (10−11), 427−439. (27) Strimmer, K. fdrtool: a versatile R package for estimating local and tail area-based false discovery rates. Bioinformatics 2008, 24 (12), 1461− 2. (28) R Development Core Team. R: A language and environment for statistical computing; R Foundation for Statistical Computing: Vienna, Austria, 2010; ISBN 3-900051-07-0. (29) Zhang, T.; Wu, X.; Yin, M.; Fan, L.; Zhang, H.; Zhao, F.; Zhang, W.; Ke, C.; Zhang, G.; Hou, Y.; Zhou, X.; Lou, G.; Li, K. Discrimination between malignant and benign ovarian tumors by plasma metabolomic profiling using ultra performance liquid chromatography/mass spectrometry. Clin. Chim. Acta 2012, 413 (9−10), 861−868. (30) Medeiros, L. R.; Rosa, D. D.; da Rosa, M. I.; Bozzetti, M. C. Accuracy of CA 125 in the diagnosis of ovarian tumors: a quantitative systematic review. Eur. J. Obstet. Gynecol. Reprod. Biol. 2009, 142 (2), 99−105. (31) Fong, M. Y.; McDunn, J.; Kakar, S. S. Identification of Metabolites in the Normal Ovary and Their Transformation in Primary and Metastatic Ovarian Cancer. PLoS One 2011, 6 (5), e19963. (32) Carrola, J.; Rocha, C. M.; Barros, A. S.; Gil, A. M.; Goodfellow, B. J.; Carreira, I. M.; Bernardo, J.; Gomes, A.; Sousa, V.; Carvalho, L.; Duarte, I. F. Metabolic signatures of lung cancer in biofluids: NMRbased metabonomics of urine. J. Proteome Res. 2011, 10 (1), 221−30. (33) Qiu, Y.; Cai, G.; Su, M.; Chen, T.; Liu, Y.; Xu, Y.; Ni, Y.; Zhao, A.; Cai, S.; Xu, L. X.; Jia, W. Urinary metabonomic study on colorectal cancer. J. Proteome Res. 2010, 9 (3), 1627−34. (34) Chen, T.; Xie, G.; Wang, X.; Fan, J.; Qiu, Y.; Zheng, X.; Qi, X.; Cao, Y.; Su, M.; Xu, L. X.; Yen, Y.; Liu, P.; Jia, W. Serum and urine metabolite profiling reveals potential biomarkers of human hepatocellular carcinoma. Mol. Cell. Proteomics 2011, 10 (7), M110 004945. (35) Ben Sellem, D.; Elbayed, K.; Neuville, A.; Moussallieh, F. M.; Lang-Averous, G.; Piotto, M.; Bellocq, J. P.; Namer, I. J. Metabolomic Characterization of Ovarian Epithelial Carcinomas by HRMAS-NMR Spectroscopy. J. Oncol. 2011, 2011, 174019. (36) Deberardinis, R. J.; Sayed, N.; Ditsworth, D.; Thompson, C. B. Brick by brick: metabolism and tumor cell growth. Curr. Opin. Genet. Dev. 2008, 18 (1), 54−61. (37) Schram, K. H. Urinary nucleosides. Mass Spectrom. Rev. 1998, 17 (3), 131−251. (38) Borek, E. K. S., Atypical transfer RNA’s and their origin in neoplastic cells. Adv. Cancer Res. 1972, 15 (163), 90. (39) Frolkis, A.; Knox, C.; Lim, E.; Jewison, T.; Law, V.; Hau, D. D.; Liu, P.; Gautam, B.; Ly, S.; Guo, A. C.; Xia, J.; Liang, Y.; Shrivastava, S.; Wishart, D. S. SMPDB: The Small Molecule Pathway Database. Nucleic Acids Res. 2010, 38 (Database issue), D480−7. (40) Arnaud, A.; Ramirez, M.; Baxter, J. H.; Angulo, A. J. Absorption of enterally administered N-acetyl-l-glutamine versus glutamine in pigs. Clin. Nutr. 2004, 23 (6), 1303−12. (41) Lopez-Pedrosa, J. M.; Manzano, M.; Baxter, J. H.; Rueda, R. Nacetyl-L-glutamine, a liquid-stable source of glutamine, partially prevents changes in body weight and on intestinal immunity induced by protein energy malnutrition in pigs. Dig. Dis. Sci. 2007, 52 (3), 650−8. (42) Sperner-Unterweger, B.; Neurauter, G.; Klieber, M.; Kurz, K.; Meraner, V.; Zeimet, A.; Fuchs, D. Enhanced tryptophan degradation in patients with ovarian carcinoma correlates with several serum soluble immune activation markers. Immunobiology 2011, 216 (3), 296−301. (43) Gregory, S.; Kelly, N. D. Pantethine: a review of its biochemistry and therapeutic applications. Altern. Med. Rev. 1997, 2 (5), 365−377. (44) Hollingsworth, M. A.; Swanson, B. J. Mucins in cancer: protection and control of the cell surface. Nat. Rev. Cancer 2004, 4 (1), 45−60.

(45) Stimpfl, M.; Schmid, B. C.; Schiebel, I.; Tong, D.; Leodolter, S.; Obermair, A.; Zeillinger, R. Expression of mucins and cytokeratins in ovarian cancer cell lines. Cancer Lett. 1999, 145 (1−2), 133−41. (46) Selak, M. A.; Armour, S. M.; MacKenzie, E. D.; Boulahbel, H.; Watson, D. G.; Mansfield, K. D.; Pan, Y.; Simon, M. C.; Thompson, C. B.; Gottlieb, E. Succinate links TCA cycle dysfunction to oncogenesis by inhibiting HIF-alpha prolyl hydroxylase. Cancer Cell 2005, 7 (1), 77−85. (47) Belenky, P.; Bogan, K. L.; Brenner, C. NAD+ metabolism in health and disease. Trends Biochem. Sci. 2007, 32 (1), 12−9. (48) Hsu, C. P.; Hariharan, N.; Alcendor, R. R.; Oka, S.; Sadoshima, J. Nicotinamide phosphoribosyltransferase regulates cell survival through autophagy in cardiomyocytes. Autophagy 2009, 5 (8), 1229−31.

H


Identification of Potential Biomarkers for Ovarian ... - ACS Publications

Recommend Documents