Identification and Confirmation of Increased Fibrinopeptide A Serum Protein Levels in Gastric Cancer Sera by Magnet Bead Assisted MALDI-TOF Mass Spectrometry Matthias P. A. Ebert,*,† Dagmar Niemeyer,| So1 ren O. Deininger,| Thomas Wex,‡ Claudia Knippig,‡ Juliane Hoffmann,‡ Jo1 rg Sauer,| Wolfgang Albrecht,| Peter Malfertheiner,‡ and Christoph Ro1 cken§ Department of Medicine II, Klinikum rechts der Isar, Technische Universita¨t Mu ¨ nchen, Mu ¨ nchen, Department of Gastroenterology, Hepatology and Infectious Diseases, Otto-von-Guericke University Magdeburg, 39120 Magdeburg, Institute of Pathology, Charite, Berlin, Germany, and Bruker Daltonik GmbH, Bremen, Germany Received January 12, 2006
Gastric cancer is the second most common malignancy and prognosis remains dismal. The reasons for the poor prognosis are the lack of sensitive serum markers for early detection and screening of high-risk individuals as well as the limited treatment options in advanced cancer stages. Using MALDITOF mass spectrometry after prefractionation of sera with magnet hydrophobic C8 coated beads sera from 14 patients with gastric cancer and 14 healthy controls mass spectra were generated. A peptide fragment was found to be highly elevated in cancer sera and was identified as fibrinopeptide A. To confirm proteome analysis of gastric cancer sera, we then screened a larger series of patients with gastric cancer (n ) 99), high-risk individuals (n ) 13) and normal controls (n ) 111) for fibrinopeptide A serum levels. Interestingly, the mean logarithmic concentrations of serum fibrinopeptide A levels were significantly higher in cancer patients (mean 3.636 ( 0.3738; p < 0.0001) and high-risk individuals (mean 3.569 ( 0.4722; p < 0.05) compared to normal controls (mean 3.303 ( 0.4012). In contrast, we observed no association of fibrinopeptide A levels with tumor stage, tumor location, presence of regional or distant metastasis, and Lauren type of gastric cancer. In conclusion, MALDI-TOF mass spectrometry of prefractionated gastric cancer sera allows the identification of potential biomarkers that may lead to the development of serum based tests for screening of high-risk individuals. Keywords: gastric cancer • proteome • diagnosis • screening • MALDI • serum
Introduction Despite its decreasing incidence in certain regions of the world, gastric cancer remains overall the second most frequent cancer and prognosis is poor.1 Most patients are diagnosed with advanced disease in which treatment options are limited contributing to the overall 5 year survival rate of less than 25%. Identifying early cancers which can be treated with curative resection or screening of individuals with an increased risk of developing gastric cancer are the most promising approaches for improving the management of this malignancy.1,2 From epidemiological analysis, it has been demonstrated that firstdegree relatives carry an almost 3-5-fold increased risk of developing gastric cancer.3 Furthermore, a risk gastritis phenotype has been described in which individuals with gastritis * To whom correspondence should be addressed. Department of Medicine II, Klinikum rechts der Isar, Technische Universita¨t Mu ¨nchen, Ismaningerstr. 22, D-81675 Mu ¨ nchen, Germany. Tel: +49-89-4140-4872, Fax: +49-894140-4871. E-mail:
[email protected]. † Technische Universita¨t Mu ¨ nchen. ‡ Otto-von-Guericke University Magdeburg. § Institute of Pathology. | Bruker Daltonik GmbH.
2152
Journal of Proteome Research 2006, 5, 2152-2158
Published on Web 07/26/2006
predominantly in the corpus of the stomach exhibit a strongly increased risk of developing gastric cancer.4 Thus, these patient groups would benefit from a screening program aimed at prevention of gastric cancer, however, to date no sensitive serum marker for gastric cancer has been identified. With the development of new proteomic based tools to analyze tissue and blood samples from cancer patients, recently, we and other groups have reported the identification of biomarkers or biomarker patterns which allow the identification and detection of cancer patients.5-8 Among others, the surface-enhanced laser desorption ionization/time-of-flight mass spectrometry (SELDI/TOF-MS) has been used to identify differential peptide and protein expression in various biological fluids, such as serum, plasma, urine and pancreatic juice.5-8 The advantages of this method are its high through-put capability and the small sample size necessary for the analysis. Apart from this technique, recently, a further approach has been developed in which sera can also be analyzed by MALDI-TOFMS after fractionation of the serum using magnetic beads.9 Using this modification, recently, other groups have reported the identification of biomarkers and patterns in various condi10.1021/pr060011c CCC: $33.50
2006 American Chemical Society
research articles
Increased Fibrinopeptide A Levels in Gastric Cancer
tions, such as brain tumors, asthma and leukemia.9 We used this technique to screen for potential biomarkers of gastric cancer. After identification of increased fibrinopeptide A serum levels by proteome analysis, we confirmed the role of this marker in a large set of sera collections from gastric cancer patients, high risk individuals, and normal controls.
Material and Methods Blood Samples. Sera for protein profiling were collected from 14 patients with histologially confirmed gastric cancer (11 male, 3 female, median age 63 years, range 51-78) and 14 healthy individuals (4 male, 10 female, median age 62.5, range 40-70 years) undergoing upper GI endoscopy for dyspepsia in which gastric cancer was excluded. Sera for assessment of fibrinopeptide A serum levels were obtained from further 99 patients with gastric cancer (67 male, 32 female), with a median age of 65.5 years (range 41-84). Sera were also collected from a high-risk group (3 male, 10 female) comprising of 4 patients with corpusdominant gastritis and 9 first-degree gastric cancer relatives with a median age of 57.5 years (range 25-80 years). Sera were also collected from healthy controls (71 males, 40 females) undergoing upper GI endoscopy for dyspepsia. The median age was 71 years (range 24-86). In all healthy control individuals gastric cancer was ruled out by endoscopy and histological analysis. The normal controls were free of malignant disease and denied a personal or family history of cancer. In the gastric cancer patients diagnosis was confirmed by histological diagnosis (CR) and staging of the tumors was performed according to routine management of cancer patients. This study was performed according to the guidelines of the Ethics Committee of the Medical Faculty of the University of Magdeburg and all patients data were anonymized to protect patients’ identity. Serum Pretreatment with C8 Magnetic Beads. Serum samples (5 µL) from the 14 patients with gastric cancer and the 14 healthy individuals were prefractionated using magnetic bead-based C8-hydrophobic chromatography resins (Bruker Daltonik, Bremen, Germany) according to the manufacturer’s recommendations.10 The magnet beads were supplied in a standard kit including a standard protocol and washing buffers. A 5 µL serum sample was mixed with 10 µL binding solution, then 5 µL MB-C8 was added and the solution was carefully mixed. Next, the tube was placed in a magnetic bead separator to allow separation of the unbound solution and the supernatant was removed. The beads were then washed three times with 100 µL wash buffer and afterwards proteins and peptides were eluted from the magnetic beads with 5 µL of 50% acetonitrile. The eluted sample was diluted 1:10 in matrix solution R-cyano-4-hydroxycinnamic acid (0.6 g/L in 2:1 ethanol:acetone). Then 0.5 µL of the mixture was spotted onto a SCOUT 600 µm prestructured sample support (AnchorChip target, Bruker Daltonik) and remained for 5 min at room temperature to dry.10 Mass Spectrometry. After processing the serum samples as outlined above the sera were analyzed using a MALDI-TOF mass spectrometer (Ultraflex, Bruker Daltonik, Bremen, Germany) equipped with a pulsed ion extraction ion source. Ionization was performed by irradiation with a nitrogen laser which operates at 50 Hz. Spectrum detection was done in the linear mode and external calibration using a defined set of peptide/protein standards was performed in the range of 1000 to 12 000 Da. Spectra were acquired with 100 shots at a fixed laser power. Automated data acquisition was performed with the FlexControl software in automatic mode. Data were ana-
lyzed using the ClinProTools (Bruker Daltonik) software. The mass spectra obtained from cancer individuals and healthy controls were analyzed using the genetic algorithm as previously reported.10,11 The MALDI-MS/MS Spectrum was recorded on an Autoflex II TOF/TOF instrument (equipped similarly as the Ultraflex) in LIFT mode. 200 Laser shots for the parent 1600 laser shots for the fragments were summed up. Database Search. The MALDI-MS/MS spectrum was subjected to a database search via the Mascot (Matrixscience, UK) database search engine. The search parameter were: no enzyme specificity, 25 ppm mass tolerance for the parent mass and 0.2 Da for the fragment masses. No fixed or variable modifications were selected. The NCBInr database was used for the search.12 Determination of Serum Fibrinopeptide A Protein Levels. Serum samples from patients with gastric cancer and high risk individuals were stored in aliquots at -28 °C until usage. Ageand gender-matched control serum samples were obtained from patients undergoing upper GI endoscopy for dyspepsia. Serum samples for fibrinopeptide A measurement in gastric cancer patients, high risk individuals, and controls were randomly selected and investigators were blinded with respect to the group assignment. Fibrinopeptide A serum levels were determined using the commercial fibrinopeptide A ELISA-kit from HiSS Diagnostics GmbH (Freiburg, Germany) in accordance to the manufacturer’s protocol with slight modifications. Instead of 500 µL sample, 50 µL were used for the pretreatment with bentonite suspension. The resulting supernatants were finally diluted to 1:500 using the diluent solution of the kit. About a quarter of samples exhibited fibrinopeptide A levels exceeding the maximal value of the standard curve. From those samples analysis was repeated using dilution of 1:1000 or 1:2000. Mean values from duplicate measurements were calculated, entered into the database and statistically analyzed by appropriate tests as outlined below. Statistical Analysis. Statistical analyses were performed with the GraphPad InStat software. Serum fibrinopeptide A levels were logarithmically transformed for conversion to a normal distribution. Clinical and biochemical data of patients are presented as mean (( s.d.). Welch’s t test or Fisher’s exact test was used to compare the data of the different groups. Twoway analysis of variance (ANOVA) was applied to analyze the correlation between clinicopathological factors and the logarithmic concentrations of serum fibrinopeptide A. Differences of p < 0.05 were considered to be statistically significant.13
Results Identification of Fibrinopeptide A in Gastric Cancer Sera by Proteome Analysis. Peptide profiling of gastric cancer sera and sera of healthy controls was performed by magnetic bead assisted prefractionation of sera and subsequent MALDI-TOF mass spectrometry. Comparison of the spectra of the two different groups showed that one peak with a m/z value of 1466 Da was significantly upregulated in the tumor sera (Figure 1). The exact mass of that peak was determined by a MALDI-TOF measurement in reflectron mode. The spectrum was recorded directly from one of the preparations used for the profiling experiment. The m/z value of the peptide was measured as 1465.64 Da (Figure 1). Subsequently, a MALDI-MS/MS spectrum from that signal was recorded (Figure 2). This spectrum was submitted to a database search. The Mascot search engine reported the peptide to be a partial sequence of fibrinopeptide A with the sequence ADSGEGDFLAEGGGVR (Figure 3). The Journal of Proteome Research • Vol. 5, No. 9, 2006 2153
research articles
Ebert et al.
Figure 1. (A) Mass spectra generated by MALDI-TOF-MS after prefractionation of sera by C8 magnetic beads. Increased levels of a mass with approximately 1466 Da in cancer vs normal sera. (B) Large view of the mass identified in cancer sera. (C) 3D view of mass spectra, demonstrating increased levels of the mass in cancer vs normal sera.
Figure 2. MS/MS view of 1466 m/z. The associated amino acid information in given. Database searching identified the peptide as fragment of fibrinogen A.
MASCOT score was reported to be 94, the significance threshold was reported to be 48. The expect value, the probability of a false positive identification, was reported to be 1.2e-6, indicating a clear identification. Serum Fibrinopeptide A Levels are Increased in Gastric Cancer and High-Risk Patients. To confirm the role of fibrinopeptide A serum levels for the identification of gastric cancers, 2154
Journal of Proteome Research • Vol. 5, No. 9, 2006
we determined serum fibrinopeptide A levels using an ELISA in 99 gastric cancer patients, 14 patients with an increased risk of developing gastric cancer and in 111 healthy noncancer controls. Serum fibrinopeptide A levels were logarithmically transformed for conversion to a normal distribution. The mean fibrinopeptide A serum level in the normal noncancer controls was 3.303 ((0.4012). In the gastric cancer patients serum levels
Increased Fibrinopeptide A Levels in Gastric Cancer
research articles
Figure 3. Protein sequence of fibrinogen A. Underlined sequence indicates sequence of fibrinopeptide A. Sequence in red is the peptide fragment that was identified in the serum of gastric cancer patients (NCBI Accession No. 650770B).
Figure 4. Quantification of fibrinopeptide A serum levels in cancer patients, high risk individuals and normal controls. Increased levels were found in cancer patients and high risk individuals compared to normal controls.
(mean 3.636 ( 0.3738) were significantly higher (p < 0.0001) and, interestingly, the mean serum level in the high-risk patients (mean 3.569 ( 0.4722) was also significantly increased compared to healthy controls (p ) 0.0378). Serum levels in cancer patients and patients at risk were not significantly different (Figure 4). Apart from this significant difference between cancer and risk patients vs normal controls, the analysis of various subgroups of cancer patients did not reveal any further difference (Figures 5, 6). Thus, the serum fibrinopeptide A levels in cancer patients were independent of tumor stage, the presence of lymph node or distant metastasis, location of the primary tumor and Lauren type of differentiation (Figures 5, 6).
Discussion Despite recent efforts in the development of new diagnostic and therapeutic strategies in gastric cancer, its overall prognosis is poor and 5 year survival rates are between 10 and 28%.1,2 Until to date no valid tumor markers have been identified facilitating the early detection and screening of gastric cancer. Current tumor markers, such as CEA, CA19-9, or Ca 72-4, allow diagnosis of gastric cancer in less than 60% of individuals, and in stage I cancers the sensitivity of these markers is even
below 20%.14-16 To improve early diagnosis and screening of high risk individuals and, thus, improve the overall poor prognosis of gastric cancer patients, new serum markers are required. While molecular markers based on the detection of genetic, epigenetic and/or molecular alterations in the pathogenesis of gastric cancer have not been useful for the detection of gastric cancers, the recent developments in the global analysis of the proteome in tissue and serum of cancer patients have raised hopes that through these techniques new biomarkers may be identified.17 In this regard, the method of surface enhanced laser desorption and ionization/time-of-flight mass spectrometry (SELDI-TOF-MS) has been extensively used to identify biomarkers and biomarker patterns for the clinical management of cancer patients.5-8 This method is advantageous over the MALDI analysis since it allows the rapid analysis of low volume serum samples. In this method Proteinchip Arrays are used which exhibit different chromatographic properties, such as anion exchange, cation exchange, hydrophilic, or hydrophobic surfaces. After binding to these surfaces serum subproteome analysis is performed by time-of-flight mass spectrometry. Although this approach has been used by many different groups in order to find biomarkers for various cancers, there has also been a lot of controversy regarding its reproducibility, sensitivity and specificity.5-8,18 A rather new approach in this field is the prefractionation of the serum proteome based on magnet beads instead of Proteinchip arrays and the subsequent analysis by MALDITOF-MS. This method has also been successfully used for the identification of biomarkers and biomarker patterns in malignant and nonmalignant diseases, such as leukemia, brain tumors and asthma.9-11 In this study, we used this approach to screen for potential biomarkers of gastric cancer and identified increased levels of a fragment of fibrinopeptide A in sera of gastric cancer patients. We were able to confirm the increased levels of fibrinopeptide A with an ELISA in a large study population including gastric cancer patients, patients at risk and healthy controls. However, no significant association was found between fibrinopeptide A serum levels and tumor stage, grade of differentiation or Lauren type of cancer. Fibrinogen is produced by hepatocytes and is present in the circulation at a concentration of around 9 µM and a half-life of approximately 100 h. During tissue injury fibrinogen is cleaved to fibrin by thrombin with the release of fibrinopeptides A and B.19 Other proteases may also degrade fibrin, including Journal of Proteome Research • Vol. 5, No. 9, 2006 2155
research articles
Ebert et al.
Figure 5. Analysis of fibrinopeptide A levels in cancer with regard to location and Lauren type of gastric cancer. No significant difference between cardia vs noncardia or diffuse vs intestinal type gastric cancers was found.
Figure 6. Analysis of fibrinopeptide A levels in cancer with regard to tumor stage. No significant difference in fibrinopeptide A levels with regard to local tumor growth (T-category; A), nodal spread (N-category; B), distant metastasis (M-category; C), or overall UICC tumor stage (D) was found.
neutrophil elastase, tryptase, matrix metalloproteinases and cathepsin D and G. The enhanced levels of fibrinopeptide A in cancer sera may therefore reflect the increased expression and activation of thrombin and proteases, which have been implicated in tumor biology by previous groups. Furthermore, fibrinopeptide A may also exert direct tumor-promoting effects 2156
Journal of Proteome Research • Vol. 5, No. 9, 2006
through its function as a mitogen for endothelial cells, smooth muscle cells and fibroblasts.20 In a previous study, we had used SELDI-TOF-MS to screen for biomarker patterns and biomarkers for gastric cancer and identified a fragment of prothrombin in the sera of these patients.5,21 In the present study, we used a novel technique,
research articles
Increased Fibrinopeptide A Levels in Gastric Cancer
based on the prefractionation of the cancer serum proteome by magnetic bead assisted MALDI TOF to screen for upregulated markers. Both studies revealed different results, since in the SELDI study we used SAX2 chips that exhibit anionexchanging characteristics,5 whereas in the present study we used magnetic hydrophobic C8 coated beads that do not necessarily bind the same proteins as the SAX2 chips. In the present study, however, we found a single marker that was increased in all cancer sera, and further analysis revealed that this peptide was fibrinopeptide A. This marker was then confirmed using a commercial ELISA and subsequent studies confirmed that this peptide is present in other cancers as well.22-24 Our study is a proof-of-concept study using a novel approach to identify a potential biomarker apart from SELDI analysis. To confirm the potential role of this marker, we chose to focus on one single marker. Subsequent studies will address other potential markers of this analysis for this malignancy. Both studies, however, support the hypothesis that changes in the coagulation cascade are present in patients with gastric cancer.21,25 While we had previously also confirmed the role of thrombin light chain A for the identification of stage I cancers, we also analyzed fibrinopeptide A serum levels in sera of individuals with an increased risk of developing gastric cancers. Thus, fibrinopeptide A levels were determined in patients with a gastritis phenotype bearing an increased risk of developing gastric cancer and in a group of first-degree relatives.3,4 Interestingly, both groups of patients also had significantly increased fibrinopeptide A levels compared to healthy controls. Inasmuch as these individuals are also characterized by a strong inflammation in the gastric mucosa, compared to normal and healthy individuals, the increased fibrinopeptide A levels might also reflect these inflammatory changes which also contribute to the increased risk of developing cancer.26 There are several limitations to our study. Increased fibrinopeptide A serum levels are not specific for gastric cancer. In fact, a number of recent studies have demonstrated that increased fibrinopeptide A levels can be found in hepatocellular, ovarian and urothelial cancers.22-24 Therefore, we can only speculate that increased fibrinopeptide A levels may indicate potential malignant disease. Furthermore, there is a significant overlap of the standard deviations of the respective biomarker in cancer and normal controls, which indicates that this marker may not be very useful for diagnostic purposes. A further important limitation of our study is the fact that differentiation of chronic inflammatory diseases from malignant disease was not analyzed in this study. However, the focus of the present study was the identification of potential cancer markers by a novel magnetic bead assisted MALDI analysis in a proof-of-concept study. The issue of differentiation of chronic inflammatory diseases from neoplastic diseases is of great importance and will need large patient populations to identify potential biomarkers. Furthermore, the source of the biomarker may limit its relevance to clinical application. In a study by Kaplan et al., plasma fibrinopeptide A levels were determined in blood samples that were placed in plastic tubes at 37 °C with and without hirudin for up to 150 min.27 The aim of the study was to assess the contribution of thrombin formation to platelet release during incubation without agitation. The authors observed a strong increase of plasma fibrinopeptide A levels after 60 min. Thus, analyzing cancer sera for potential biomarkers may be of limited relevance. However, this experimental study on normal hemostasis reflects carefully selected in vitro conditions. In contrast, in cancer patients a persistent
activation of coagulation is present which leads to increased fibrinopeptide A levels in a number of malignancies, such as ovarian, gastric, and hepatocellular cancer.22-24 Moreover, a recent publication indicates that this increased activation status is associated with an increased risk of developing cancer of the digestive tract,28 which confirms our observation that individuals with an increased risk of developing gastric cancer also exhibit significantly increased fibrinopeptide A serum levels. In our study, increased fibrinopeptide A levels were identified in cancer sera using MALDI-TOF-MS, and we believe that this increased marker should be confirmed using the same samples. Therefore, we chose to assess fibrinopeptide A levels in cancer sera, and sera from healthy controls and high risk individuals with a commercially available ELISA. In conclusion, in this study we identified increased fibrinopeptide A levels in gastric cancer patients and high-risk individuals using MALDI-TOF-MS and ELISA, confirming previous reports that alterations in the coagulation system are present during the early stages of gastric carcinogenesis.
Acknowledgment. M. Ebert is supported by the Heisenberg-Program of the DFG (Eb 187/5-1, 5-2). Further support was granted by the Start-Up Program (NBL-3) of the BMBF, Germany. References (1) Hohenberger, P.; Gretschel, S. Lancet 2003, 362, 305-315. (2) Ebert, M. P.; Malfertheiner, P. Aliment. Pharmacol. Ther. 2002, 16, 1059-1066. (3) Yatsuya, H.; Toyoshima, H.; Tamakoshi, A.; Kikuchi, S.; Tamakoshi, K.; Kondo, T.; Mizoue, T.; Tokui, N.; Hoshiyama, Y.; Sakata, K.; Hayakawa, N.; Yoshimura, T. Br. J. Cancer 2004, 91, 929-934. (4) Meining, A. G.; Bayerdorffer, E.; Stolte, M. Eur. J. Gastroenterol. Hepatol. 1999, 11, 717-720. (5) Ebert, M. P.; Meuer, J.; Wiemer, J. C.; Schulz, H. U.; Reymond, M. A.; Traugott, U.; Malfertheiner, P.; Rocken, C. J. Proteome Res. 2004, 3, 1261-1266. (6) Adam, B. L.; Qu, Y.; Davis, J. W.; Ward, M. D.; Clements, M. A.; Cazares, L. H.; Semmes, O. J.; Schellhammer, P. F.; Yasui, Y.; Feng, Z.; Wright, G. L. Cancer Res. 2002, 62, 3609-3614. (7) Vlahou, A.; Schellhammer, P. F.; Mandrinos, S.; Patel, K.; Kondylis, F. I.; Gong, L.; Nasim, S.; Wright, G. L. Am. J. Pathol. 2001, 158, 1491-1502. (8) Li, J.; Zhang, Z.; Rosenzweig, J.; Wang, Y. Y.; Chan, D. W. Clin. Chem. 2002, 48, 1296-1304. (9) Pusch, W.; Flocco, M. T.; Leung, S. M.; Thiele, H.; Kostrzewa, M. Pharmacogenomics 2003, 4, 463-476. (10) Zhang, X.; Leung, S. M.; Morris, C. R.; Shigenaga, M. K. J. Biomol. Tech. 2004, 15, 167-175. (11) Villanueva, J.; Philip, J.; Entenberg, D.; Chaparro, C. A.; Tanwar, M. K.; Holland, E. C.; Tempst, P. Anal. Chem. 2004, 76, 15601570. (12) Kuster, B.; Mortensen, P.; Andersen, J. S.; Mann, M. Proteomics 2001, 1, 641-650. (13) Siegel, S. Nonparametric statistics for behavioral sciences. McGrawHill: New York; 1956. (14) Marrelli, D.; Rovello, F.; de Stefano, A.; Farnetani, M.; Marosi, L.; Messano, A.; Pinto, E. Oncology 1999, 57, 55-62. (15) Nakajiima, K.; Ochiai, T.; Suzuki, T.; Shimada, H.; Hayashi, H.; Yasumoto, A.; Takeda, A.; Hishikawa, E.; Isono, K. Tumor Biol. 1998, 19, 464-469. (16) Istigami, S.; Natsugoe, S.; Hokita, S.; Che, X.; Tokuda, K.; Nakajo, A.; Iwashige, H.; Tokushige, M.; Watanabe, T.; Takao, S.; Aikou, T. J. Clin. Gastroenterol. 2001, 32, 41-44. (17) Chen, J.; Rocken, C.; Malfertheiner, P.; Ebert, M. P. Dig. Dis. 2004, 22, 380-385. (18) Diamandis, E. Mol. Cell. Proteomics 2004, 3, 367-378. (19) Herrick, S.; Blanc-Brude, O.; Gray, A.; Laurent, G. Int. J. Biochem. Cell Biol. 1999, 31, 741-746. (20) Sporn, L. A.; Bunce, A. A.; Francis, C. W. Blood 1995, 86, 18021810. (21) Ebert, M. P.; Lamer, S.; Meuer, J.; Malfertheiner, P.; Reymond, M.; Buschmann, T.; Rocken, C.; Seibert, V. J. Proteome Res. 2005, 4, 586-590.
Journal of Proteome Research • Vol. 5, No. 9, 2006 2157
research articles (22) Theodorescu, D.; Wittke, S.; Ross, M. M.; Walden, M.; Conaway, M.; Just, I.; Mischak, H.; Frierson, H. F. Lancet Oncol. 2006, 7, 230-40. (23) Bergen, H. R.; Vasmatzis, G.; Cliby, W. A.; Johnson, K. L.; Oberg, A. L.; Muddiman, D. C. Dis. Markers 2003, 19, 239-49. (24) Orvisky, E.; Drake, S. K.; Martin, B. M.; Abdel-Hamid, M.; Ressom, H. W.; Varghese, R. S.; An, Y.; Saha, D.; Hortin, G. L.; Loffredo, C. A.; Goldman, R. Proteomics 2006, 6, 2895-902. (25) De Cicco, M. Crit. Rev. Oncol. Hematol. 2004, 50, 187-196.
2158
Journal of Proteome Research • Vol. 5, No. 9, 2006
Ebert et al. (26) El-Omar, E. M.; Oien, K.; Murray, L. S.; El-Nujumi, A.; Wirz, A.; Gillen, D.; Williams, C.; Fullarton, G.; McColl, K. E. Gastroenterology 2000, 118, 22-30. (27) Kaplan, K. L.; Drillings, M.; Lesznik, G. J. Clin. Invest. 1981, 67, 1561-1568. (28) Miller, G. J.; Bauer, K. A.; Howarth, D. J.; Cooper, J. A.; Humphires, S. E.; Rosenberg, R. D. J. Thromb. Haemost. 2004, 2, 2107-2114.
PR060011C
