By Kyselova, Zuzana Mechref, Yehia; Kang, Pilsoo; Goetz, John A; Dobrolecki, Lacey E; Sledge, George W; Schnaper, Lauren; Hickey, Robert J; Malkas, Linda H; Novotny, Milos V
BACKGROUND: Glycosylated proteins play important roles in cell- to-cell interactions, immunosurveillance, and a variety of receptor- mediated and specific protein functions through a highly complex repertoire of glycan structures. Aberrant glycosylation has been implicated in cancer for many years. METHODS: We performed specific MALDI mass spectrometry (MS)-based glycomic profile analyses of permethylated glycans in sera from breast cancer patients (12, stage I; 11, stage II; 9, stage III; and 50, stage IV) along with sera from 27 disease-free women. The serum glycoproteins were enzymatically deglycosylated, and the released glycans were purified and quantitatively permethylated before their MALDI-MS analyses. We applied various statistical analysis tools, including ANOVA and principal component analysis, to evaluate the MS profiles.
RESULTS: Two statistical procedures implicated several sialylated and fucosylated N-glycan structures as highly probable biomarkers. Quantitative changes according to a cancer stage resulted when we categorized the glycans according to molecular size, number of oligomer branches, and abundance of sugar residues. Increases in sialylation and fucosylation of glycan structures appeared to be indicative of cancer progression. Different statistical evaluations confirmed independently that changes in the relative intensities of 8 N-glycans are characteristic of breast cancer (P
CONCLUSIONS: MS-based N-glycomic profiling of serum-derived constituents appears promising as a highly sensitive and informative approach for staging the progression of cancer.
(c) 2008 American Association for Clinical Chemistry
Glycosylated proteins play important roles in cell-to-cell interactions, immunosurveillance, and a variety of receptor- mediated and specific protein functions through a highly complex repertoire of glycan structures (1-5). Correspondingly, metabolic dysfunctions and disease states may be reflected in the appearance of abnormal glycans or altered quantitative proportions within the glycome. Such observations provide the rationale for recent developments in the field of functional glycomics (6-10). Differential glycomic measurements between healthy and disease states may even have significant clinical diagnostic potential, as exemplified by the recent utilization of capillary electrophoresis in diagnosing liver cirrhosis (11).
Certain links between cancer diseases and altered protein glycosylation have been noted for several years (12-16) for both N- linked and O-linked glycoconjugates, whereas the most advanced knowledge of abnormal glycosylation in this set of diseases has primarily come from the investigations of tumor tissues and cell cultures. Detecting cancer-related changes in patients’ blood has been a distinct goal of more recent studies but has been less commonly pursued because of methodological difficulties with high- sensitivity measurement of glycans. Coincidentally, some recently identified cancer biomarkers in human serum are glycoproteins (17- 20), usually the mucin-type, large molecules.
In this study, we report that the profiles of N-glycans released from glycoproteins of human serum are highly indicative of the different conditions of breast cancer. Using small (10 [mu]L) serum aliquots, we performed sensitive and quantitative mass spectrometry (MS)5 measurements of the constituent profiles of N-glycans originating from circulating proteins. These profiles could be further evaluated statistically through pattern recognition techniques [principal component analysis (PCA)] in terms of breast cancer disease stage. The N-glycan data cluster remarkably well for different stages, which further differ from a data set recorded from individuals apparently free of the disease. The clusters readily distribute into groups that correlate with the stage of breast disease as determined by pathological assessment. This type of clinically useful information could be obtained from a small volume of blood serum, without biopsy or tumor removal. To implicate certain oligosaccharide structures as biomarkers, we carried out additional statistical evaluations (data mining) through nonparametric ROC and ANOVA analyses for approximately 50 individual N-glycans. Finally, for a comparison, we also examined the glycomic profiles of breast cancer cell lines, both invasive and noninvasive, to determine whether they resemble the glycan patterns derived from patient specimens.
Materials and Methods
MATERIALS
Trypsin (EC: 3.4.21.4) and PNGase F (EC: 3.5.1.52) were obtained from Sigma. We purchased the matrix, 2,5-dihydroxybenzoic acid (DHB), trifluoroethanol, sodium hydroxide, CHAPS, Tris-HCl, sodium pyrophosphate, EDTA, and EGTA from Aldrich; chloroform, iodomethane, and sodium chloride from EM Science; dithiothreitol (DTT) and iodoacetamide (IAA) from Bio-Rad Laboratories; ammonium bicarbonate from Mallinckrodt Chemical Company, and acetonitrile from Fisher Scientific.
BLOOD SERUM SAMPLES AND CLINICAL DIAGNOSIS
Blood serum collections from disease-free female volunteers and women diagnosed with different breast cancer stages were performed by a clinical team according to an institutional review board- approved clinical trial. Venous blood samples were taken during the morning fasting state, with minimal stasis in evacuated tubes. After at least 30 min, but within 2 h, the tubes were centrifuged at 20 [degrees]C for 12 min at 120[degrees]g. Sera were stored frozen in plastic vials at -80 [degrees]C before use in consecutive measurements.
We generated glycomic profiles for samples derived from women in one of two categories: (1) disease-free woman at low risk for developing breast cancer and (2) postmenopausal women with confirmed disease stratified according to their having noninvasive or invasive breast cancer. We divided breast cancer patients into 4 subgroups (I- IV) based on severity of disease (21). We collected samples from 27 healthy individuals and from 12 patients in stage 1,11 in stage II, 9 in stage III, and 50 in stage IV. A detailed characterization of the healthy volunteers and patients with their clinical diagnoses can be found in Tables 1-2 in the Data Supplement that accompanies the online version of this article at http://www.clinchem.org/ content/ vol54/issue7.
CANCER CELL LINES
We lysed cancer cell lines (normal mammary epithelial, MCFlOA; invasive, MDA-MB-231, MDA-MB-435; noninvasive, 578T, ADR-RES, BT549, and T47D) in a CHAPS-based buffer (150 mmol/L NaCl, 0.5% CHAPS, 50 mmol/L Tris (pH 7.5), 10 mmol/L sodium pyrophosphate, 1 mmol/L EDTA, 1 mmol/L EGTA). We then subjected lysates to ultracentrirugation at 16400g, separating the samples into cytosolic and membrane- associated proteins, and 200-[mu]g protein aliquots (determined by Bradford assay) underwent the same procedures described for blood serum.
TRYPSIN DIGESTION
Human serum samples were reduced and alkylated before the addition of trypsin as described (22). Briefly, 10 [mu]L of human serum was dried and then resuspended in 25 [mu]L of 25 mmol/L ammonium bicarbonate, 25 [mu]L trifluoroethanol, and 2.5 [mu]L of 200 mmol/L DTT. We incubated the samples for 45 min at 60 [degrees]C, added 10 [mu]L of 200 mmol/L LAA, and left the samples at room temperature for 1 h in the dark. We added 2.5 [mu]L DIT to remove excess IAA and again left the samples for l h in the dark. Finally, we added 300 [mu]L water to dilute samples and 100 [mu]L ammonium bicarbonate stock solution to adjust pH. Proteolytic digestion was performed using 1 [mu]g/[mu]L [or 1:50 (wt/wt) ratio] proteomics-grade trypsin dissolved in 1 mmol/L HCl and incubated at 37 [degrees]C overnight (at least 18 h). Afterward, enzyme activity was quenched by incubation at 95 [degrees]C for 10 min and allowed to cool at room temperature.
RELEASE OF M-GLYCANS FROM GLYCOPROTEINS AND PURIFICATION
The N-glycans were enzymatically released from human serum samples as previously described (23). Briefly, we added 5 mU PNGase F to the reaction mixture, which was subsequently incubated overnight (18-22 h) at 37 [degrees]C. We removed deglycosylated peptides from the reaction mixture using C18 Sep-Pak(R) cartridges (Waters) that were preconditioned with ethanol and deionized water. We further purified the aqueous eluent containing released N- glycans using activated charcoal microcolumns (Harvard Apparatus). The columns were conditioned with acetonitrile and equilibrated with 0.1% trifluoroacetic acid (TFA). After trapping of the diluted sample, the microcolumns were washed with 0.1% TPA, and the samples were eluted in a minimum volume of 50% acetonitrile containing 0.1% TFA.
CAPILLARY PERMETHYLATION
The N-glycans derived from human blood serum samples, as described above, were finally dried under vacuum and subsequently permethylated using a capillary approach, which involves the use of fused silica capillary reactor (500 [mu]m i.d.; Polymicro Technologies) packed with sodium hydroxide beads (22). We used a Hamilton 100-[mu]L syringe and a syringe pump from KD Scientific to introduce the sample into the reactor. Dried samples were resuspended in 50 [mu]L DMSO to which 0.3 [mu]L water and 22 [mu]L methyl iodide were added before infusion through reactors. This permethylation procedure minimizes oxidative degradation and “peeling” reactions and prevents the need for excessive clean-up (22). Finally, permethylated Nglycans were extracted with chloroform and washed repeatedly with water. MALDI-TOF/TOF MS INSTRUM ENTATION
We resuspended dried permethylated samples in 2 [mu]L methanol:water solution (50:50) containing 1 mmol/L sodium acetate and spotted 0.5-[mu]L sample aliquots directly on the MALDI plate, mixed with an equal volume of 2,5-DHB matrix [prepared by suspending 10 mg of DHB in 1 mL watermethanoi (50:50)], and dried under vacuum.
Mass spectra were acquired in the positive-ion mode on the Applied Biosystems 4700 Proteomic Analyzer. This instrument is equipped with Nd:YAG (355 nm) laser. Argon was used as a collision gas in the tandem MS measurements, and the collision cell pressure was set to 6.5 x 10^sup -6^ torr. The acquired spectra were the average of 1000 laser shots.
DATA EVALUATION
We further processed MS data using DataExplorer 4.0 (Applied Biosystems) and an software tool developed in-house (PeakCalc 2.0). We performed PCA using MarkerView (ABI), allowing the visualization of multivariate information. We used supervised PCA methods with a prior knowledge of the sample groups, such as healthy vs diseased. MS data were weighted using the natural logarithm of the peak intensities. The peak intensities were also scaled using the Pareto option, in which the average value is subtracted from each value and the difference divided by the square root of the standard deviation. This option is suited for MS data, since it prevents intense peaks from completely dominating PCA. Variation within groups was expressed as SE (24).
We also used ROC curve analysis AccuROC 2.5 (Accumetric Corporation) to assess the sensitivity and selectivity of the potential diagnostic variables, and data were statistically analyzed using a single-factor ANOVA test. The difference between the 2 groups of data was considered statistically significant when P values were
Results
SERUM GLYCOMIC PROFILES
In the differential measurements of glycomic data for different patient groups, it was essential to achieve a high degree of methodological precision, so that the individual variations (patient- to-patient), or “profile individuality,” were not overshadowed by the measurement errors. Fortunately, our recently reported glycan mapping procedure (22) yields quantitative reproducibiliry with
We were able to identify and quantify approximately 50 different N-glycan structures, which cover all typical structural types, including high-mannose, hybrid, and complex-type entities. As shown previously (26), the high-energy collision process used in this type of tandem mass spectrometry (MS/MS) yields reliable structural identification of each recorded glycan. MS/MS was employed here to identify the majority of the structures (data not shown). In comparing visually the glycomic profiles of the healthy individuals with those suffering from breast cancer, we were able to note that, within the molecular mass range of 1500-5000 mlz, there was an overall decrease in abundance of smaller N-glycans (m/z: 1500-2700) compensated by an increase in the abundance of larger structures. Representative profiles are illustrated in Fig. 1, A and B, with a healthy profile compared to stage I and IV, respectively. These relative ion intensity changes in a profile (small N-glycans changing into larger structures due to different addition of sugar residues) were enhanced from stage I to stage IV. This general observation is consistent with a published report (13) by Rye and Walker as well as more recent studies on glycosylation in tumor tissues (6, 8, 9). Obviously, the trend of adding fucosyl and sialyl residues to certain glycoprotein structures in tumor tissues, observed in these investigations, was also observed here.
PCA OF N-GLYCAN PROFILES
In assessing more precisely the patterns of N-glycans that could be characteristic for specific cancer conditions, we turned to PCA, which is a chemometric tool designed to reduce the dimensionality in a data set by defining a reduced set of projection axes that allow a maximum amount of the variance information originally in the data set to be retained (27). The acquired profile data, measured for all subjects (healthy and breast cancer patients) were subjected to PCA as described above (Fig. 2) followed by cluster analysis. The numbers of subjects in the early stages of breast cancer (stage 1,12; stage 11,11; stage III, 9) were lower than for stage IV (50) and individuals free of the disease (27). The final clustering of our data indicates that there are significant differences in the glycomic profiles of healthy individuals in comparison to those of individuals in the early stages of breast cancer. This observation supports the notion of the diagnostic potential for recognizing different stages of breast cancer and perhaps even detecting early onset through glycomic analysis of serum specimens. Data clustering was achieved despite the heterogeneity of the sample population studied here. This aspect is believed to be an advantage, since such heterogeneity reflects a true representation of a breast cancer population. However, the presence of such heterogeneity might also prompt challenging data interpretation.
GLYCAN POTENTIAL BIOMARKER IDENTIFICATION THROUGH DATA MINING
We performed a further statistical evaluation of putative biomarkers through nonparametric ROC procedure (28-30). Fig. 3 shows some examples of ROC analysis of N-glycan markers, comparing healthy individuals and stage IV breast cancer patients. Thus, the N-glycan with m/z 3864 had an area under the curve (AUC) value of 0.97, indicating its high diagnostic accuracy (Fig. 3A), whereas another structure (m/z 2111) had an AUC value of only 0.49, similar to random variation (Fig. 3C). An AUC value of 0.88 was calculated for the N-glycan with m/z 1835, making it only a moderately accurate test (Fig. 3B).
In Table 1, we list the AUCs for the N-glycans that were evaluated by ROC analysis and their P values from single-factor ANOVA. We have taken into consideration only the N-glycans for which comparison of nondiseased and diseased experimental groups yielded P values
A further correlation of the favorable AUC values with P values from ANOVA implicates 8 N-glycans whose relative intensity changes profoundly during breast cancer development (Fig. 4A). These particular N-glycans thus represent strong biomarker candidates, as the occurrence of their specific structures has been confirmed by 2 independent statistical approaches, both accepted widely in the biomedical literature.
Whereas data on glycan modifications in cancer cell lines and tumors, including fucosylation and sialylation, are available (31- 35), it seemed prudent to supplement our investigations on serum glycan levels by the parallel analyses of a human mammary epithelium cell (control) and cancer cell lines (invasive and noninvasive) using the same analytical procedure. Examining the 8 N-glycan structures implicated in the blood serum as pertinent to cancer, we found similar trends for 6 of these structures in the extracts prepared from invasive (MDA-MB-231, MDA-MB-435) and noninvasive (578T, ADR-RES, BT549, T47D) breast cancer cells (Fig. 4B). The invasive cell lines, capable of forming tumor metastases distant from their original site of transformation, were found to have a more dramatic change in the glycomic pattern compared with the noninvasive cells. Moreover, N-glycans associated with m/z 3951 and 4226 are not observed in normal cell lines, suggesting their potential role during the tumorigenesis process.
STRUCTURAL CORRELATIONS
Grouping N-glycan profile constituents into categories according to molecular size, number of antennas, and sugar residue abundance (Fig. 5) can be indicative of certain general trends of disease progression. The overall trend in increased N-glycan size due to the addition of sialic acid residues and enhanced fucosylation is clearly evident. A noticeable fact is that all 8 N-glycans selected by AUC values and ANOVA are sialylated to a different degree (mono- , di-, tri-, and tetrasialylated). Moreover, 5 of these structures are fucosylated (2 of them difucosylated), supporting the general notion of fucosylation involvement during progression of cancer in a different organ (33).
Fig. 1. MALDI mirror mass spectra of N-glycans derived from a 10- [mu]L aliquot of the serum of a healthy individual (upper trace) and a stage I breast cancer patient (lower trace) (A) and a healthy individual (upper trace) and a stage IV breast cancer patient (lower trace) (B).
[black square] N-acetylglucosamine; *, mannose; [white circle], galactose; [white triangle up], fucose; *, N-acetylneuraminic acid.
Fig. 2. PCA score plot of MALDI/MS N-glycan profiles derived from the sera of healthy individuals and breast cancer : patients at different stages of the disease. PC1, principal component 1; PC2, principal component 2.
Discussion
Procedural simplicity, combined with sensitivity, information content, and accuracy, has been among the most valued attributes of clinical diagnostic and prognostic methods. Most procedures for breast cancer detection rely on pathology-based criteria such as tumor size and grade, lymph node status, etc. (36) and histological evaluation of hormone receptors and plasminogen activators (37) after biopsy or surgery, which are used to predict patients’ chances for survival. Detection of biomarkers in circulation has been limited to single-protein indicators, such as CA 15-3, MUC1, carcinoembryonic antigen (CEA), BR 27.29, tissue polypeptide antigen (TPA), tissue polypeptide-specific antigen (TPS), and the shed form of human epidermal growth factor receptor 2 (HER-2), which all provide (with the possible exception of CA 15-3 (38) and MUC1 (8)) neither the sensitivity nor the specificity needed for early-stage disease (39). Consequently, mammography and histopathology still remain the primary modalities for detecting breast cancer at an early stage of the disease.
Fig. 3. Representative ROC analysis AUC plots for different N- glycan structures reflecting high-accuracy (A), moderate-accuracy (B), and low-accuracy (C) tests.
AUCs are expressed as mean (SE). Symbpis as in Fig. 1.
Additional validation for clinically relevant glycomic mapping from serum samples comes from the correspondence of implicated N- glycan structures with those found in cancer cell lines in association with membrane glycoproteins. As shown in Fig. 4 and Fig. 5, and verified through statistical criteria, a more intense glycosylation is associated with cancerous cells, in agreement with recent observations made elsewhere (6, 8, 9) on glycosylation due to cancer progression. In our measurements, the invasive cell lines capable of forming tumors away from their original site of transformation represented a more extreme change in glycomic pattern compared with the noninvasive cells.
Table 1. Evaluation of diagnostic potential of specific N- glycans according to AUC.
Fig. 4. Relative intensity changes of 8 N-glycans with high- accuracy ROC analysis AUC values (0.9
Symbols as in Fig. 1.
Six N-glycan structures from the cancer cell lines are shared with those implicated as significant in our serum measurements (8 major N-glycans). The corresponding differences in glycomic maps revealed a substantial increase of rucosylation (both within the core component and the branched segments) with malignant transformation. Interestingly, increased rucosylation has also been associated with pancreatic cancer (33), colorectal cancer (40), human leukocyte cancer (34), and renal carcinomas (35 J. At this point in time, measurements of various sialylated structures in serum present a less dear picture. Since various sialylated oligosaccharides were implicated in different malignant cells as both O- and N-linked structures (7-9, 31-35, 40), our future studies will concentrate on these structural types.
Fig. 5. Bar graphs of N-glycan relative intensity changes derived from the blood sera of healthy Individuals vs breast cancer patients according to their structural type.
The lower trace is a zoomed segment of the upper trace.
In conclusion, the MS-based glycomic profiling of serum-derived constituents described here provides a Highly sensitive and informative approach to differential evaluation of cancer conditions. Such an approach can be based on the sound knowledge of aberrant glycobiology of cancerous cells that shed their surface glycoproteins into the surrounding biofluids. Although it is not yet clear which serum glycoproteins are primarily responsible for the diagnostically distinct N-glycans, statistical analyses of selected N-glycans (or their patterns) provide potential for developing diagnostic and prognostic procedures. Whereas MS instrumentation is relatively expensive, the method provides nearly absolute structural information at a moderate level of measurement throughput. PCA, ROC, and ANOVA statistical analyses of the MS data independently confirmed 8 N-glycans (P
Grant/Funding Support: This work was supported in part by grants GM24349 from the National Institutes of Health (NIH) and RR018942 from the National Center for Research Resources, NIH, as a contribution from the National Center for Glycomics and Glycoproteomics at Indiana University to M. Novotny. This work was also supported in part by grant ROl CA83199 to L. Malkas from the National Cancer Institute, NIH, by a Biomedical Research Grant from the Indiana University Medical School to R. Hickey, and a Department of Defense Center of Excellence Grant W81XWH-04-1-0468 to G. Sledge. The initial stages of this investigation were also aided by a grant from the 21st Century Fund of the State of Indiana. This work was also supported in part by a fellowship from Merck Research Laboratories for P. Kang.
Financial Disclosures: None declared.
Acknowledgments: The authors thank Milan Madera for the development of PeakCalc 2.0.
5 Nonstandard abbreviations: MS, mass spectrometry; PCA, principal component analysis; DHB, 2,5-dihydroxybenioic add; DTT, dithiothreitol; IAA, iodoacetamide; TFA, trifluoroacetic acid; AUC, area under the curve.
References
1. Clark GF, Dell A, Morris HR, Patankar M, Oehninger S, Seppala M. Viewing AIDS from a grycobiological perspective: potential linkages to the human fetoembryonic defence system hypothesis. Mol Hum Reprod 1997;3:5-13.
2. Zachara NE, Hart GW. The emerging significance of O-GlcNAc in cellular regulation. Chem Rev 2002;102:431-8.
3. Calarese DA. Antibody domain exchange is an immunologkal solution to carbohydrate cluster recognition. Science (Wash D) 2003;300: 2065-71.
4. Pochec E, Litynska A, Amoresano A, Casbarra A. Glycosylation profile of integrin alpha 3 beta 1 changes with melanoma progression. Biochim Biophys Acta 2003;1643:113-23.
5. Rudd PM, Elliott T, Cresswell P, Wilson IA, Dwek RA. Glycosylation and the immune system. Science (Wash D) 2001;291:2370- 6.
6. Mechref Y, Novotny MV. Structural investigation of grycoconjugates at high sensitivity. Chem Rev 2002;102:320-70.
7. Dwek MV, Lacey HA, Leathern AJ. Breast cancer progression is associated with a reduction in the diversity of sialylated and neutral oligosaccharides. Clin Chim Acta 1998;271:191-202.
8. Baldus SE, Wienand JR, Werner JP, Landsberg S, Drebber U, Hanisch FG, Dienes HP. Expression of MUC1, MUC2 and oligosaccharide epitopes in breast cancer prognostic significance of a sialylated MUC1 epitope. Int J Ontol 2005;27: 1289-97.
9. Handerson T, Camp R, Harigopal M, Rimm D, Pawelek J. Beta 1,6- branched oligosaccharides are increased in lymph node metastases and predict poor outcome in breast carcinoma. Clin Cancer Res 2005;11:2969-73.
10. Uematsu R, Furukawa J, Nakagawa H, Shinohara Y, Deguchi K, Monde K, Nishimura S. High throughput quantitative glycomics and glycoform-focused proteomics of murine dermis and epidermis. Mol Cell Proteomics 2005;4: 1977-89.
11. Callewaert N, Van Vlierberghe H, Van Hecke A, Laray W, Delanghe J, Contreras R. Noninvasive diagnosis of liver cirrhosis using DNA sequencer-based total serum protein glycomics. Nature (Lond) Med 2004;10:429 -34.
12. Hakim AA. A mechanism by which human breast carcinoma cells escape the host immune system. Neoplasms 1988;35;691-705.
13. Rye PD, Walker RA. Analysis of glycoproteins released from benign and malignant human breast: changes in size and fucosylation with malignancy. Eur J Cancer Clin Oncol 1989;25: 65-72.
14. Hakomori S. Tumor malignancy defined by aberrant grycosylation and sphingo(glyco)-lipid metabolism, Cancer Res 1996;56:5309-18.
15. Kobata A. A retrospective and prospective view of grycopathology. Glycoconj J 1998;4:323-31.
16. Alper J. Glycobiology: turning sweet on cancer. Science (Wash DC) 2003;301:159-60.
17. Yang Z, Harris LE, Palmer-Toy DE, Hancock WS. Multilectin affinity chromatography for characterization of multiple glycoprotein biomarker candidates In serum from breast cancer patients. Clin Chem 2006;52:1897-905.
18. Wong NK, Renouf DV, Lehmann S, Hounsell EF. Characterization of the oligosaccharides associated with the human ovarian tumor marker CA125. J Biol Chem 2003;278:28619-34.
19. Shariat SF, Canto El, Kattan MW, Slawin KM. Beyond prostate- specific antigen: new serologic biomarkers for improved diagnosis and management of prostate cancer. Rev Urol 2004;6:58-72.
20. An HJ, Miyamoto S, Lancaster KS, Kirmiz C, Li B, Lam KS, Leiserowitz GS, et al. Profiling of grycans in serum for the discovery of potential biomarkers for ovarian cancer. J Proteome Res 2006; 5:1626-35.
21. Fleming ID, Phillips JL, Menck HR, Murphy GP, Winchester DP. The National Cancer Data Base report on recent hospital cancer program progress toward complete American Joint Committee on Cancer/ TNM staging. Cancer 1997;80: 2305-10.
22. Kang P, Mechref Y, Klouckova I, Novotny MV. Solid-phase permethylation of grycoconjugates for mass-spectrometric analysis. Rapid Commun Mass Spectrom 2005;19:3421-8.
23. Mechref Y, Novotny MV. Mass spectrometric mapping and sequencing of N-linked oligosaccharides derived from submicrogram amounts of glycoproteins. Anal Chem 1998;70:455-63.
24. Higgins JP, Thompson SG, Quantifying heterogeneity in a meta- analysis. Stat Med 2002;21: 1539-58.
25. Wada Y, Azadi P, Costello C, Dell EA, Dwek RA, Geyer H, et al. Comparison of the methods for profiling glycoprotein glycans; HUPO Human Disease Glycomics/Proteome Initiative multi- institutional study. Glycobiology 2007;17:411-22. 26. Mechref Y, Kang P, Novotny MV. Differentiating structural isomere of sialylated glycans by matrixassisted laser desorption/ionization time-of- flight/ time-of-flight tandem mass spectrometry. Rapid Commun Mass Spectrom 2006;20:1381-9.
27. Musumaira G, Barresi V, Condorelli DF, Scire S. A biojnformatic approach to the identification of candidate genes for the development of new cancer diagnostics. Biol Chem 2003;384:321- 7.
28. Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982;143:29- 36.
29. Pepe MS, Cai T, Longton G. Combining predictors for classification using the area under the receiver operating characteristic curve. Biometrics 2006; 62:221-9.
30. Swets M. Measuring the accuracy of diagnostic systems. Science (Wash D) 1988;40:1285-93.
31. Shaw L, Yousefi S, Dennis JW, Schauer R. CMP-N- acetylneuraminic acid hydroxylase activity determines the wheat germ agglutinin-binding phenotype in two mutants of the lymphoma cell line MDAY-D2. Glycoconj J 1991;8:434-41.
32. Takano R, Muchrnore EA, Dennis JW. Sialylation and malignant potential in tumor cell glycosylarton mutants. Glycobiology 1994;4:665-74.
33. Mas E, Pasqualirti E, Caillol N, El Battari A, Crotte C, Lombarde D, Sadoulet MO. Fucosyftransferase activities in human pancreatic tissue: comparative study between cancer tissues and established tumoral cell lines. Glycobiology 1998;8:605-13.
34. Liu F, Zhang Y, Zhang X-Y, Chen H-L. Transfection of the nm23- H1 gene into human hepatocardnoma cell line inhibits the expression of sialyl Lewis X, al,3 fucosyhransferase VII, and metastatic potential. J Cancer Res Clin Oncol 2002; 128:189-96.
35. Saito S, Yamashita S, Endoh M, Yamato T, Hoshi S, Ohyama C, Watanabe R, et al. Clinical significance of ST3Gal IV expression in human renal cell carcinoma. Oncol Rep 2002;9:1251-5.
36. Elston CW, Ellis IQ, Finder SE. Pathological prognostic factors in breast cancer [Review]. Crit Rev Oncol Hematol 1999;31:209-23.
37. Isaacs C, Stearns V1 Haves DF. New prognostic factors for breast cancer recurrence. Semin Oncol 2001;28;53-67.
38. Duffy MJ. Serum tumor markers in breast cancer: are they of clinical value? Clin Chem 2006;52: 345-51.
39. Cheung KL, Graves CR, Robertson JF. Tumour marker measurements in the diagnosis and monitoring of breast cancer. Cancer Treat Rev 2000; 26:91-102.
40. Izawa M, Kumamoto K, Mitsuoka C, Kanamori C, Kanamori A, Ohmori K, Ishida H, et al. Expression of sialyf 6-sulfo Lewis X is inversely correlated with conventional sialyl Lewis X expression in human colorectal cancer. Cancer Res 2000;60: 1410-6.
Zuzana Kyselova,1 Yehia Mechref,1* Pilsoo Kang,1 John A. Goetz,1 Lacey E. Dobrolecki,2 George W. Siedge,2,4 Lauren Schnaper,3 Robert J. Hickey,2,4 Linda H. Malkas,2,4 and Milos V. Novotny1,4*
1 National Center for Glycomics and Glycoproteomics, Department of Chemistry, Indiana University, Bloomincrton, IN; 2 Department of Medicine, Indiana University School of Medicine, Indianapolis, IN; 3 Breast Cancer Center, Greater Baltimore Medical Center, Baltimore, MD; 4 Indiana University Cancer Center, Indianapolis, IN.
* Address correspondence to these authors at: Department of Chemistry, Indiana University, 800 E. Kirkwood Ave., Bloomington, IN 47405- Fax 812-855-8300; e-mail [email protected], [email protected].
Received February 7, 2007; accepted April 8, 2008.
Previously published online at DOI: 10.1373/clinchem.2007.087148
Copyright American Association for Clinical Chemistry Jul 2008
(c) 2008 Clinical Chemistry. Provided by ProQuest Information and Learning. All rights Reserved.
Comments