ANTICANCER RESEARCH 27: 1247-1256 (2007)

Review Proteomic Approaches for Serum Biomarker Discovery in Cancer

PRIYANKA MAURYA, PAULA MELEADY, PAUL DOWLING and MARTIN CLYNES

National Institute for Cellular Biotechnology, Dublin City University, Glasnevin, Dublin 9, Ireland

Abstract. Monitoring the protein expression pattern in tumour (CEA), was identified in 1965 in patient serum for the cells by proteomics technologies offers opportunities to discover detection of colorectal cancer (2). Other biomarkers potentially new biomarkers for the early detection and discovered in the 1970s and 1980s include prostate-specific diagnosis of cancer. Different proteomic tools such as 2D- antigen (PSA) for prostate cancer, CA-19 for colorectal and PAGE, 2D-DIGE, SELDI-ToF-MS technology, protein arrays, pancreatic cancer, CA-15-3 for breast cancer and CA-125 ICAT, iTRAQ and MudPIT have been used for differential for ovarian cancer. However, not all biomarkers are analysis of various biological samples, including cell lysates, effective in all clinical situations. For example, PSA is well cell secretome (conditioned medium), serum, plasma, tumour established in clinical practice, but approximately one third tissue and nipple aspirate fluid, to better understand the of patients with an elevated PSA level often undergo molecular basis of cancer pathogenesis and the validation and unnecessary medical procedures because they do not have characterisation of disease-associated proteins. In recent years, a malignant form of prostate cancer (3). Many types of there has been a large increase in cancer-related publications cancer, such as lung carcinoma and melanoma do not have dealing with new biomarker discovery for cancer, therefore, in any significant biomarkers available to screen at the early this review we have focused on the contribution of proteomics stage of disease. Identification of new tumour biomarkers technologies in serum and conditioned medium-based with predictive value is necessary to allow early detection oncology research particularly for lung, breast, melanoma and and treatment of cancer. pancreatic cancer. Proteomics Technologies Although many effective therapies are present for early detection and diagnosis, cancer remains a major cause of With the recent developments in electrophoresis, imaging, death and disease. Cancer is a complex disease that reflects protein labelling, protein array-based approaches and mass the genetic, as well as protein changes within a cell. Gene spectrometric technologies, along with developments in expression data gives us limited relevant information since genomic and protein bioinformatics, proteomics may proteins are the main functional units performing all provide powerful information for improved biomarker biological process in the cell or organism and may have discovery. Several proteomics technologies including two- post-transcriptional event(s) and post-translational dimensional polyacrylamide gel electrophoresis (2D- modification(s) that contribute to the biological activity of PAGE), surface enhanced laser desorption/ionisation time proteins. Protein expression patterns are also changed of flight (SELDI-ToF), protein arrays, isotope coded affinity specifically and significantly in response to every disease (1). tags (ICAT), iTRAQ and multidimensional protein The first protein cancer marker, carcinoembryonic antigen identification technology (MudPIT) are the approaches being implemented in cancer research. 2D-PAGE and SELDI-ToF are the main technologies used in serum cancer research, however other technologies such as protein arrays, Correspondence to: Paula Meleady, National Institute for Cellular ICAT, iTRAQ and MudPIT also offer great potential for Biotechnology, Dublin City University, Glasnevin, Dublin 9, future biomarker discovery in cancer. Ireland. Tel: +353 1 7005700, Fax: +353 1 7005484, e-mail: [email protected] Two-dimensional electrophoresis. 2D-PAGE is the most Key Words: Serum, cancer, proteomics, SELDI, 2D PAGE, 2D widely used proteomics technique to study the proteome as DIGE, review. well as cancer biomarkers (4-8). 2D-PAGE remains

0250-7005/2007 $2.00+.40 1247 ANTICANCER RESEARCH 27: 1247-1256 (2007) challenging mainly because of its low sensitivity and to their surface chemistries. After a short incubation period, reproducibility. Modified 2D electrophoresis by fluorescent unbound proteins and unspecific substances are washed away tagging to proteins, differential gel electrophoresis (DIGE), with an appropriate buffer and water. The ToF reader offers increased throughput, ease of use, reproducibility, records the time-of-flight and calculates the accurate and accurate quantitation of protein expression differences molecular weight of proteins/peptides in the form of a (9). This system enables the separation of two or three spectral map containing mass to charge ratios (m/z) and fluorescently labelled protein samples (Cy2, Cy3 and Cy5) intensities corresponding to each bound protein/peptide. on the same gel. Differential analysis software identifies the Biomarker Wizard software analyses the spectral map and differentially expressed protein targets that can be trypsin- detects differentially expressed protein/peptides with digested and readily identified using mass spectrometry by statistical significance. An increase in published research generating peptide mass fingerprints (PMF) using matrix using SELDI-ToF in the past few years itself has assisted laser desorption ionization time-of-flight mass demonstrated its potential for the early detection of cancer, spectrometry (MALDI-ToF MS), a technique that is both especially for low molecular weight cancer-associated relatively easy to use and reasonably sensitive for identifying proteins. For example, applications of SELDI-ToF have been proteins. Additionally other mass spectrometry techniques demonstrated for the early detection of prostrate (14, 15), such as electrospray ionization (ESI-MS/MS) are capable of breast (16, 17), bladder (18) ovarian (19, 20), pancreatic (21), providing amino acid sequence information on peptide and lung (22) cancer biomarkers. However, there is some fragments of the parent protein (10). controversy over this technology such as its reproducibility, Although 2D-PAGE-based techniques have a reasonable the bioinformatics used, the possibility of over-fitting, the level of throughput, there are a number of difficulties potential bias in the samples, as well as how this could inherent to the technique such as separation of low abundant possibly fit into a routine diagnostic lab (23-25). proteins as it is difficult to enrich for these proteins. Membrane proteins are also difficult to separate due to poor Protein array technology. Protein arrays are being used for solubility. Efforts have been made to overcome these drug discovery, biomarker identification and molecular limitations. For example, low abundant proteins can be profiling of cellular material (26-29). Protein arrays are identified using higher protein concentrations, and applying generated by spotting antibodies (26) or other affinity fractionation methods (7). Moreover, membrane proteins reagents, such as aptamers, purified proteins (30), peptides can be identified to some extent by using commercially (28) or fractionated proteins (27), onto some sort of matrix available mild detergents such as oligooxyethylene, either on flat solid phases or in capillary systems to generate sulfobetaine, dodecyl maltoside or decaethylene glycol mono protein arrays. The sample is applied to the array to allow hexadecyl, as use of strong detergents like SDS interfere with specific binding of the sample to the array and then the the isoelectric focusing of proteins (11). Additional problems arrays are washed to remove the unbound fraction. The with 2D-electrophoresis include insufficient resolution to process can also be reversed whereby the protein samples separate multiple species originating from a single protein of interest are spotted onto the matrix and then probed with with post-translational modifications, such as those with different affinity reagents (31). The wider application of carbohydrates, difficulties in detecting proteins with protein arrays in biomedical research is still limited, partly molecular masses <120 kDa and those with pI values <4 or because of the cost of producing antibodies and the limited >9, low visualisation of less-abundant proteins, and co- availability of antibodies with high specificity and high migration of proteins to the same spots (12). Conventional affinity for the target. Additionally, the difficulties 2D-electrophoresis shows only protein expression and cannot associated with preserving proteins in their biologically detect protein-protein interactions or protein function active conformation before analysis with protein arrays without using particular methods such as affinity further limits the application of this technology as a routine electrophoresis (13). proteomic strategy.

SELDI-ToF MS. This technique allows proteins/peptides to Isobaric Tag Labelling Technology be profiled from different biological samples on a variety of ICAT. Isotope-coded affinity tags (ICAT) use stable isotope chemically (e.g. anionic, cationic, hydrophobic, hydrophilic, labelling to perform quantitative analysis of paired protein metal affinity capture) or biochemically (e.g. immobilized samples. They contain a protein-reactive group, an ethylene antibody, receptor, DNA, enzyme) defined chromatographic glycol linker and a biotin tag (32). Two different isotope surfaces. A small amount of sample of interest is loaded onto tags are generated by using linkers that contain either eight ProteinChipì arrays that selectively bind different subsets of hydrogen atoms (d0, light reagent) or eight deuterium proteins in crude samples by adsorption, partition, atoms (d8, heavy reagent) which bind covalently to cysteine electrostatic interaction or affinity chromatography according moieties of amino acid within protein(s). Both samples are

1248 Maurya et al: Serum Proteomics and Cancer (Review) mixed, digested with trypsin, fractionated by avidin affinity hydrophobicity. Eluted peptides are identified by tandem chromatography and then these differentially tagged mass spectrometry and the technique is extremely sensitive peptides are scanned in a mass spectrometer. Spectral peak and reproducible. One of the major weaknesses of MudPIT analysis in single mass spectrometric (MS) mode of the is in identifying quantitative differences in protein isotopically resolved peptides from the two different sources expression across protein mixtures (37). In a comparative enables quantitation of the relative amounts of the peptide study with 2D-electrophoresis, MudPIT has been reported and hence the protein levels. Differentially expressed to demonstrate superior detection efficiency (36). proteins are then identified by tandem MS (MS/MS) sequencing. One weakness of ICAT is that only cysteine- Implementation of Proteomics Technologies in containing peptides can be labelled. Approximately 10% of Serum Cancer Research proteins do not have cysteine, therefore they will not be detected by ICAT. Lung cancer. Lung cancer is clinically divided into two major histological types, non-small cell (NSCLC) and small cell iTRAQ. A similar isobaric labelling technology to ICAT, lung cancer (SCLC). NSCLC accounts for 75-85% of lung called iTRAQ, that labels amine residues in peptides has cancer patients and consists of several subtypes, been recently developed (33). iTRAQ contains a set of four predominantly squamous cell carcinomas, adenocarcinomas isobaric reagents and therefore can analyze up to four and large cell carcinomas, which are treated in the same protein samples at one time. After trypsin digestion, manner. Small cell lung cancer accounts for 15-25% of lung samples are labelled with four independent iTRAQ cancer patients, often has neuroendocrine components and reagents. The reporter groups of the iTRAQ reagents will is primarily treated with chemotherapy and/or radiotherapy. split from the peptide and generate small fragments for Many lung cancers are composed of histologically mixed each sample with mass/charges (m/z) of 114, 115, 116, and tumour types consisting of non-small cell and small cell 117. The intensity of each of these peaks represents the components. Dyspnoea, cough and thoracic pain are early quantity of small reporter group fragments and thus signs of lung cancer, however they are not specific to all represents the quantity of a peptide sample. Peaks in the cases, while haemoptysis often indicates advanced disease. spectrum graph are used to identify peptide sequences and There is no satisfactory marker for the early detection of therefore protein sequences. By comparing the amounts of lung cancer. However, many widely used biomarkers are peptides labelled with each iTRAQ reagent, quantitative available for differential diagnosis and sub typing, i.e. CEA, differences can be readily measured. A comparative analysis CYFRA 21-1, SCC, NSE and ProGRP (38-41), but offer of iTRAQ and ICAT suggests that the information low specificity and/or sensitivity. generated by the two methods is complementary. Both Various putative biomarkers from serum and conditioned methods have some advantages and disadvantages over the medium have been identified using proteomics as a other. Unlike iTRAQ, ICAT is preferred for low abundant searching tool. Dihydrodiol dehydrogenase (DDH) was proteins including signalling molecules, however shown to be secreted by the adenocarcinoma cell line, A549, overlapping peaks in the MS spectrum can compromise the by performing 2D electrophoresis on conditioned medium quality of results. On the other hand, apart from non- followed by mass spectrometry (42). Following this specific nature of labelling, iTRAQ requires lengthy sample observation, the levels of DDH mRNA and protein processing separately that increases the chances of expression were found to be significantly higher in 15 experimental variation (34). NSCLC cancer tissues and protein levels were also significantly increased in the serum of the NSCLC patients MudPIT. MudPIT is a non-gel approach that uses compared to non-malignant lung tumour and healthy multidimensional high-pressure liquid chromatography controls. Since mRNA expression of DDH was also (LC/LC) separation, tandem mass spectrometry and reported to be elevated in NSCLC tumour specimens, this database searching (35). MudPIT permits a rapid and raises the possibility of using DDH as a tissue marker and a simultaneous separation and identification of proteins and novel serological marker for NSCLC detection (42, 43). peptides in a complex mixture without the need for pre- or Protein gene product 9.5 (PGP 9.5) is a neurospecific post-separation labelling, which is not possible in 2D- polypeptide and was proposed as a marker for non-small electrophoresis, ICAT or iTRAQ (36). The complex protein cell lung cancer, based on its expression in tumour tissue mixture is digested with a specific protease, then peptide (44, 45). PGP 9.5 and autoantibodies against PGP 9.5 were fragments are allowed to separate in parallel with two- observed in sera from lung cancer patients through an dimensional liquid chromatography using a strong cation antibody-based reactivity assay against lung adenocarcinoma exchange (SCX) column that separate peptides based on proteins, resolved by 2D-PAGE (46). However PGP 9.5 is charge, and then by a reverse phase (RP) column based on not limited to lung cancer, as it has also been reported in

1249 ANTICANCER RESEARCH 27: 1247-1256 (2007) pancreatic cancer (47). PGP 9.5 was also shown to be malignant. This result yielded 100% sensitivity. The positive present at the cell surface, as well as secreted in conditioned predictive value for this sample set was 98%. In another medium by A549 (46). study 2D-DIGE analysis of serum samples obtained from 39 In another study, approximately 250 and 100 different patients with breast cancer and 35 controls revealed that proteins were detected in HSA- and IgG-depleted serum proapolipoprotein A-I, transferrin, and hemoglobin were samples from lung cancer patients using 2D-LCMS/MS and up-regulated and three proteins, apolipoprotein A-I, 1D-LCMS/MS respectively (48). Similarly, the serum apolipoprotein C-III, and haptoglobin a2 were down- samples from lung cancer patients were analyzed with 2D- regulated in cancer patients (42). However, in cross PAGE after depletion of highly abundant plasma proteins validation with routine clinical immunochemical reactions, by immunoaffinity chromatography and fractionation by the serum levels of apolipoprotein A-I and haptoglobin anion-exchange chromatography; 58 gene products, could not be detected, indicating the superiority and including the classic plasma proteins and the tissue-leakage sensitivity of the 2D-DIGE technique. proteins catalase, clusterin, ficolin, gelsolin, lumican, A pattern of three serum biomarkers (two up-regulated tetranectin, triosephosphate isomerase and vitronectin, were at 8.1 kDa and 8.9 kDa and one down-regulated at 4.3 kDa) identified from this study (49). were identified from the SELDI profiles of 169 serum SELDI proteomic patterns of lung cancer patient serum samples from 103 breast cancer patients, 41 healthy women, have also been assessed to distinguish lung cancer patients and from 25 benign breast cancer patients that distinguished from healthy individuals. Three protein peaks from the patients from controls (16). The sensitivity and specificity profiles of 30 lung cancer patients and 51 age- and sex- after cross-validation within the sample group matched healthy individuals achieved 93.3% sensitivity in (bootstrapping), using a random subset of the data to build detection of lung cancer patients (50). Similarly, five protein the model and testing it with the remaining data, were peaks at 11493, 6429, 8245, 5335 and 2538 Da were chosen found to be 93% and 91%, respectively. However, a new set automatically as a biomarker pattern from the profile of a of samples in a later study confirmed the up-regulation of training set composed of 208 serum samples, including 158 the 8.1 kDa and 8.9 kDa biomarker species; subsequently lung cancer patients and 50 healthy individuals, and this the 8.9 kDa species was identified to be a complement pattern provided sensitivity of 91.4% in the detection of component of C3a (desArg) and the 8.1 kDa species as a C- non-small cell lung cancers (NSCLC) (51). terminal-truncated form of C3a (desArg) (68). In a later study also using SELDI-TOF MS, the combination of an Breast cancer. The classical pathological methods that are independent cancer biomarker, Ca 15.3, with the serum used to predict survival, development of metastatic disease biomarkers, 4.286 kDa and 4.302 kDa possibly or to guide selection of primary therapy (52, 53) in patients corresponding to the 4.3 kDa peak identified by Li et al. with breast cancer rely on anatomical staging of cancer and (16), and 8.919 kDa and 8.961 kDa possibly corresponding include the Nottingham Prognostic Index (54), Adjuvant to the 8.9 kDa peak also identified by Li et al. (16), Online (AO) (55) and the St. Gallen criteria (56). The significantly improved breast cancer diagnosis (69). In current St. Gallen derived algorithm for selection of another study, four peaks, CA1 (17.3 kDa), CA2 (26.2 kDa), adjuvant systemic therapy for early breast cancer patients CA3 (5.7 kDa), and CA4 (8.9 kDa), were chosen as relies on tumour size and grade, nodal status, menopausal potential biomarkers from the SELDI profile of 49 breast status, peritumoural vessel invasion, endocrine status and cancer patients, 51 patients with benign breast diseases and HER2 (epidermal growth factor receptor 2) status. HER2 33 healthy women to build a prediction model using is the most prominent and commonly used biomarker for artificial neural networks and discriminant analysis (70). breast cancer detection (57). However MMP-2 (matrix One hundred percent sensitivity and specificity were metalloproteinase-2 immuno-reactive protein), absence of observed in a training set while a blind test set, showed oestrogen and progesterone receptors and high expression 76.47% sensitivity and 90.00% specificity. of Ki-67 (Mib-1) antigen, osteopontin (OPN), urikinase type Two genes, BRCA-1 and BRCA-2, have been observed to plasmonogen activator and its inhibitors (PAI-1 and 2) and be strong tumour suppressors in men and women (71, 72). cathepsins (B and L) have also been indicated as prognostic Fifteen serum samples from women with BRCA-1 mutations biomarkers for breast cancer (58-66). who developed breast cancer (BRCA-1 Ca) and 15 from In a proteomics study of breast cancer serum, two those who did not (Carrier), 16 from normal volunteers proteins, HSP27 (up-regulated) and 14-3-3 sigma (down- (NL), and 16 from women with sporadic breast cancer regulated) were identified using 2D-PAGE coupled with (SBC) were profiled with SELDI-TOF enabling the MALDI-TOF-MS (67). Comparison of the expression differentiation between BRCA-1 Ca vs. Carrier with 87% patterns of these two proteins correctly classified 97% of the sensitivity and 87% specificity and BRCA-1 Ca vs. SBC controls as not cancer and 100% of cancer samples as patients with 94% sensitivity and 100% specificity (73).

1250 Maurya et al: Serum Proteomics and Cancer (Review)

Similar results with ~86% discrimination between women 2D-PAGE was used for protein separation and with BRCA-1 breast cancers from the 15 non-cancer approximately 130 sialylated glycoproteins were identified BRCA-1 carriers were later observed (74). using ÌLCMS/MS. The sialylated plasma protease C1 For optimization and individualization of therapeutic inhibitor was identified to be down-regulated in cancer decisions, a multiprotein complex (including haptoglobin, serum. Changes in glycosylation sites in cancer serum were C3a complement fraction, transferrin, apolipoprotein C1 also observed by glycopeptide mapping using ÌLC-ESI- and apolipoprotein A1)-based model developed on SELDI TOF-MS where the N83 glycosylation of R1-antitrypsin was profiles of denatured and fractionated early postoperative found to be down-regulated. serum samples from 81 high-risk early breast cancer patients Patients with cancer have been found to frequently was shown to correctly predict the outcome in 83% of develop autoantibodies and the identification of panels of patients for metastatic relapse (75). tumour autoantigens may have a role in the early diagnosis of cancer and immunotherapy. DEAD-box protein 48 Pancreatic cancer. Poor prognosis in pancreatic cancer is (DDX48), which is highly similar to eukaryotic initiation attributed to the fact that most patients do not develop factor 4A, was observed in 63.64% of patients with newly overt symptoms until the disease has disseminated or caused diagnosed pancreatic cancer and in only 1.9% of normal local organ dysfunction (76). Currently CA 19-9 is the controls using an antibody-based reactivity assay (82). accepted serum marker for pancreatic cancer, but it was Following on from this observation, a large sample set was approved only for monitoring treatment response by the analysed to evaluate DDX48 as a diagnostic antigen and U.S. Food and Drug Administration (77, 78). 33.33% of pancreatic cancer patients, 10% of colorectal Approximately 80-90% of people with pancreatic cancer cancer patients, 6.67% of gastric cancer patients and 6.67% have elevated levels of this marker in their blood (79). of hepatocellular cancer patients were positive for anti- Additionally, current methods of diagnosis, including CA DDX48 autoantibody reactivity using a purified DDX48 19-9, are ineffective for identifying small, surgically antigen-based ELISA assay, while none of the 20 chronic resectable cancers. pancreatitis patients, 30 lung cancer patients, and 60 normal To identify potential serum markers for pancreatic individuals were positive for this assay. Therefore, the cancer, serum samples from 3 pancreatic cancer patients detection of autoantibodies to DDX48 in serum may and 3 normal and healthy individuals were analysed using improve clinical diagnosis of pancreatic cancer (82). 2D-DIGE coupled with MALDI/TOF/TOF-MS and 24 Sera from 49 pancreatic cancer patients and 54 unaffected unique up-regulated proteins and 17 unique down- individuals were profiled using SELDI-ToF-MS (83). Based regulated proteins were identified in cancer serum (8). on the SELDI profiles, a classification model was generated Increased levels of apolipoprotein E, R-1-antichymotrypsin with classification and regression tree and logistic regression and inter-R-trypsin inhibitor in serum proteome analysis of methods and this resulted in differentiation of diseased 20 patients with pancreatic cancer and 14 controls were also patients with 100% sensitivity. The specificity was 93.5% found to be associated with pancreatic cancer by western using a decision tree algorithm and 100% using a logistic blot analysis; the use of these proteins resulted in a regression model. In another study, the two most sensitivity of 82.6% and a specificity of 100% in pancreatic discriminating protein peaks (m/z 3146 and 12861) from the cancer diagnosis. In another study, samples from 32 normal SELDI profiles of serum samples from 60 patients with and 30 pancreatic cancer patients were analysed using 2D- resectable pancreatic adenocarcinoma, 60 age- and sex- PAGE and 9 spots out of 154 commonly overexpressed matched patients with non-malignant pancreatic diseases and proteins discriminated 100% of pancreatic cancer samples 60 age- and sex-matched healthy controls, differentiated and 94% of normal samples (80). Fibrinogen-Á, a protein patients with pancreatic cancer from healthy controls with a associated with the hypercoagulable state of pancreatic sensitivity of 78% and a specificity of 97%. This was found to cancer, was identified as one of these nine spots and was be significantly better than the current standard serum found to discriminate all cancer cases from normal sera, marker, CA19-9, on the same sample set (with a 65% successfully indicating the potential of fibrinogen-Á in sensitivity and 85% specificity) (84). The combination of the pancreatic cancer prediction. two peaks and CA19-9 resulted in slightly improved Differential glycoprotein expression has also been specificity. In a further study on pancreatic cancer, a set of observed in human cancer serum (81). Sialylated four mass peaks at 8766, 17272, 28080 and 14779 m/z, whose glycoproteins from highly abundant protein-depleted serum mean intensities differed significantly in the SELDI profiles samples of normal and pancreatic cancer individuals were of 245 plasma samples including diseased and normal extracted using three different lectins [wheat germ individuals, were selected as a biomarker profile using a agglutinin (WGA), elderberry lectin (SNA), Maackia support vector machine learning algorithm which accurately amurensis lectin, (MAL)] after HPLC-based fractionation. discriminated cancer patients from healthy controls in a

1251 ANTICANCER RESEARCH 27: 1247-1256 (2007) training cohort (sensitivity of 97.2% and specificity of 94.4%) 4 Lilley KS, Rassaq A and Dupree P: Two-dimensional gel and in a validation cohort (sensitivity of 90.9% and a electrophoresis: recent advances in sample preparation, specificity of 91.1%) (85). The combination of CA19-9 with detection and quantitation. Curr Opin Chem Biol 6: 46-50, 2002. 5 Gharbi S, Gaffney P, Yang A, Zvelebil MJ, Cramer R, these differentially expressed peaks resulted in 100% Waterfield MD and Timms JF: Evaluation of two-dimensional detection of pancreatic cancer cases (29/29 samples), differential gel electrophoresis for proteomic expression including early-stage (stages I and II) tumours. analysis of a model breast cancer cell system. Mol Cell Proteomics 1: 91-98, 2002. Melanoma. Currently no protein serum marker is available 6 Somiari RI, Sullivan A, Russell S, Somiari S, Hu H, Jordan R, for surveillance of melanoma progression in early-stage George A, Katenhusen R, Buchowiecka A, Arciero C, Brzeski melanoma (86, 87). A study of serum specimens of 49 early- H, Hooke J and Shriver C: High-throughput proteomic analysis stage (AJCC stage I and II) patients, including 25 patients of human infiltrating ductal carcinoma of the breast. Proteomics 3(10): 1863-1873, 2003. with melanoma recurrence and 24 without evidence of 7 Qin S, Ferdinand AS, Richie JP, O'Leary MP, Mok SC and Liu disease, following resection were profiled using SELDI-ToF- BC: Chromatofocusing fractionation and two-dimensional MS (86). The differential pattern of peaks among patients difference gel electrophoresis for low abundance serum with recurrence and without recurrence were used for proteins. Proteomics 5(12): 3183-3192, 2005. predicting the chances of cancer recurrence and resulted in 8 Yu KH, Rustgi AK and Blair IA: Characterization of proteins a sensitivity of 72% and a specificity of 75% that was in human pancreatic cancer serum using differential gel significant (86). Similarly, 205 serum samples from 101 early- electrophoresis and tandem mass spectrometry. J Proteome Res stage (American Joint Committee on Cancer (AJCC) stage 4(5): 1742-1751, 2005. 9 Unlu M, Morgan ME and Minden JS: Difference gel I) and 104 advanced stage (AJCC stage IV) melanoma electrophoresis: a single gel method for detecting changes in patients were analysed using SELDI-ToF and MALDI-ToF protein extracts. Electrophoresis 18(11): 2071-2077, 1997. MS. SELDI profiles were used to train artificial neural 10 Mann M, Hendrickson RC and Pandey A: Analysis of proteins networks (ANN). Based on these ANN algorithms, 88% of and proteomes by mass spectrometry. Annu Rev Biochem 70: stage assignment was correctly predicted; 80% of stage III 437-473, 2001. samples could be correctly as progressors or non-progressors 11 Luche S, Santoni V and Rabilloud T: Evaluation of nonionic and 82% of stage III progressors were correctly identified. and zwitterionic detergents as membrane protein solubilizers in two-dimensional electrophoresis. Proteomics 3(3): 249-253, The prediction accuracy was much improved compared to 2003. the cases predicted by the conventional marker S-100‚ (88) 12 Gygi SP, Corthals GL, Zhang Y, Rochon Y and Aebersold R: where only 21% of the stage III progressors were detected. Evaluation of two-dimensional gel electrophoresis-based proteome analysis technology. Proc Natl Acad Sci USA 97(17): Future Directions 9390-9395, 2000. 13 Kameshita I, Ishida A and Fujisawa H: Analysis of protein- Significant progress has been made in proteomics protein interaction by two-dimensional affinity electrophoresis. technology development in the last decade and this has Anal Biochem 262(1): 90-92, 1998. enabled researchers to move forward to a better 14 Petricoin EF 3rd, Ornstein DK, Paweletz CP, Ardekani A, understanding of the disease. Several potential cancer type- Hackett PS, Hitt BA, Velassco A, Trucco C, Wiegand L, Wood specific biomarkers for the early detection of the disease or K, Simone CB, Levine PJ, Linehan WM, Emmert-Buck MR, for therapy decision-making have been proposed using 2D- Steinberg SM, Kohn EC and Liotta LA: Serum proteomic patterns for detection of prostate cancer. J Natl Cancer Inst electrophoresis or SELDI approaches on patient serum. 94(20): 1576-1578, 2002. Newer technologies, such as protein arrays, ICAT, iTRAQ 15 Cazares LH, Adam BL, Ward MD, Nasim S, Schellhammer PF, and MudPIT, will increase this potential for biomarker Semmes OJ and Wright GL Jr: Normal, benign, preneoplastic, discovery in cancer. and malignant prostate cells have distinct protein expression profiles resolved by surface enhanced laser desorption/ionization References mass spectrometry. Clin Cancer Res 8(8): 2541-2552, 2002. 16 Li J, Zhang Z, Rosenzweig J, Wang YY and Chan DW: 1 Banks RE, Dunn MJ, Hochstrasser DF, Sanchez JC, Blackstock Proteomics and bioinformatics approaches for identification of W, Pappin DJ and Selby PJ: Proteomics: new perspectives, new serum biomarkers to detect breast cancer. Clin Chem 48(8): biomedical opportunities. Lancet 356(9243): 1749-1756, 2000. 1296-1304, 2002. 2 Gold P and Freedman SO: Demonstration of tumour-specific 17 Ricolleau G, Charbonnel C, Lode L, Loussouarn D, Joalland antigens in human colonic carcinoma by immunological MP, Bogumil R, Jourdain S, Minivielle S, Campone M, tolerance and adsorption techniques. J Exp Med 121: 439-462, Deporte-Fety R, Campion L and Jezequel P: Surface-enhanced 1965. laser desorption/ionization time of flight mass spectrometry 3 Harris R and Lohr KN: Screening for prostate cancer: an protein profiling identifies ubiquitin and ferritin light chain as update of the evidence for the US Preventative Service Task prognostic biomarkers in node-negative breast cancer tumours. Force. Ann Intern Med 137: 917-929, 2002. Proteomics 6(6): 1963-1975, 2006.

1252 Maurya et al: Serum Proteomics and Cancer (Review)

18 Zhang YF, Wu DL, Guan M, Liu WW, Wu Z, Chen YM, 33 Ross PL, Huang YN, Marchese JN, Williamson B, Parker K, Zhang WZ and Lu Y: Tree analysis of mass spectral urine Hattan S, Khainovski N, Pillai S, Dey S, Daniels S, Purkayastha profiles discriminates transitional cell carcinoma of the bladder S, Juhasz P, Martin S, Bartlet-Jones M, He F, Jacobson A and from noncancer patient. Clin Biochem 37(9): 772-779, 2004. Pappin DJ: Multiplexed protein quantitation in Saccharomyces 19 Petricoin EF 3rd, Ardekani AM, Hitt BA, Levine PJ, Fusaro cerevisiae using amine-reactive isobaric tagging reagents. Mol VA, Steinberg SM, Mills GB, Simone C, Fishman DA, Kohn Cell Proteomics 3(12): 1154-1169, 2004. EC and Liotta LA: Use of proteomic patterns in serum to 34 DeSouza L, Diehl G, Rodrigues MJ, Guo J, Romaschin AD, identify ovarian cancer. Lancet 359(9306): 572-577, 2002. Colgan TJ and Siu KW: Search for cancer markers from 20 Rai AJ, Zhang Z, Rosenzweig J, Shih Ie M, Pham T, Fung ET, endometrial tissues using differentially labeled tags iTRAQ and Sokoll LJ and Chan DW: Proteomic approaches to tumour cICAT with multidimensional liquid chromatography and marker discovery. Arch Pathol Lab Med 126: 1518-1526, 2002. tandem mass spectrometry. J Proteome Res 4(2): 377-386, 2005. 21 Rosty C, Christa L, Kuzdzal S, Baldwin WM, Zahurak ML, 35 McDonald WH and Yates JR 3rd: Shotgun proteomics and Carnot F, Chan DW, Canto M, Lillemoe KD, Cameron JL, Yeo biomarker discovery. Dis Markers 18(2): 99-105, 2002. CJ, Hruban RH and Goggins M: Identification of 36 Koller A, Washburn MP, Lange BM, Andon NL, Deciu C, hepatocarcinoma-intestine-pancreas/pancreatitis-associated Haynes PA, Hays L, Schieltz D, Ulaszek R, Wei J, Wolters D protein I as a biomarker for pancreatic ductal adenocarcinoma and Yates JR 3rd: Proteomic survey of metabolic pathways in by protein biochip technology. Cancer Res 62(6): 1868-1875, rice. Proc Natl Acad Sci USA 99(18): 11969-11974, 2002. 2002. 37 Washburn MP, Ulaszek RR and Yates JR 3rd: Reproducibility 22 Zhukov TA, Johanson RA, Cantor AB, Clark RA and Tockman of quantitative proteomic analyses of complex biological MS: Discovery of distinct protein profiles specific for lung mixtures by multidimensional protein identification technology. tumours and pre-malignant lung lesions by SELDI mass Anal Chem 75(19): 5054-5061, 2003. spectrometry. Lung Cancer 40(3): 267-279, 2003. 38 Ochnio J, Roginska E, Kwiek S and Rowinska-Zakrzewska E: 23 Diamandis EP: Mass spectrometry as a diagnostic and a cancer Value of the determination of carcinoembryonic antigen (CEA) biomarker discovery tool: opportunities and potential in the serum for the diagnosis of lung cancer; evaluation of its limitations. Mol Cell Proteomics 3(4): 367-378, 2004. extent and prognosis in conservatively treated patients. 24 Baggerly KA, Morris JS and Coombes KR: Reproducibility of Pneumonol Pol 52(7): 313-319, 1984. SELDI-TOF protein patterns in serum: comparing datasets 39 Pujol JL, Molinier O, Ebert W, Daures JP, Barlesi F, Buccheri G, from different experiments. Bioinformatics 20(5): 777-785, Paesmans M, Quoix E, Moro-Sibilot D, Szturmowicz M, Brechot 2004. JM, Muley T and Grenier J: CYFRA 21-1 is a prognostic 25 Liotta LA, Lowenthal M, Mehta A, Conrads TP, Veenstra TD, determinant in non-small-cell lung cancer: results of a meta- Fishman DA and Petricoin EF 3rd: Importance of analysis in 2063 patients. Br J Cancer 90(11): 2097-2105, 2004. communication between producers and consumers of publicly 40 Barlesi F, Gimenez C, Torre JP, Doddoli C, Mancini J, Greillier available experimental data. J Natl Cancer Inst 97(4): 310-314, L, Roux F and Kleisbauer JP: Prognostic value of combination 2005. of Cyfra 21-1, CEA and NSE in patients with advanced non- 26 Mendoza LG, McQuary P, Mongan A, Gangadharan R, small cell lung cancer. Respir Med 98(4): 357-362, 2004. Brignac S and Eggers M: High-throughput microarray-based 41 Molina R, Auge JM, Filella X, Vinolas N, Alicarte J, Domingo enzyme-linked immunosorbent assay (ELISA). Biotechniques JM and Ballesta AM: Pro-gastrin-releasing peptide (proGRP) 27(4): 778-788, 1999. in patients with benign and malignant diseases: comparison with 27 Madoz-Gurpide J, Wang H, Misek DE, Brichory F and Hanash CEA, SCC, CYFRA 21-1 and NSE in patients with lung cancer. SM: Protein based microarrays: a tool for probing the proteome Anticancer Res 25(3A): 1773-1778, 2005. of cancer cells and tissues. Proteomics 1(10): 1279-1287, 2001. 42 Huang LJ, Chen SX, Huang Y, Luo WJ, Jiang HH, Hu QH, 28 Houseman BT, Huh JH, Kron SJ and Mrksich M: Peptide chips Zhang PF and Yi H: Proteomics-based identification of secreted for the quantitative evaluation of protein kinase activity. Nat protein dihydrodiol dehydrogenase as a novel serum markers of Biotechnol 20(3): 270-274, 2002. non-small cell lung cancer. Lung Cancer 54(1): 87-94, 2006. 29 Woodbury RL, Varnum SM and Zangar RC: Elevated HGF 43 Hsu NY, Ho HC, Chow KC, Lin TY, Shih CS, Wang LS and levels in sera from breast cancer patients detected using a Tsai CM: Overexpression of dihydrodiol dehydrogenase as a protein microarray ELISA. J Proteome Res 1(3): 233-237, prognostic marker of non-small cell lung cancer. Cancer Res 2002. 61(6): 2727-31, 2001. 30 Ge H: UPA, a universal protein array system for quantitative 44 Addis BJ, Hamid Q, Ibrahim NB, Fahey M, Bloom SR and detection of protein-protein, protein-DNA, protein-RNA and Polak JM: Immunohistochemical markers of small cell protein-ligand interactions. Nucleic Acids Res 28(2): e3, 2000. carcinoma and related neuroendocrine tumours of the lung. J 31 Paweletz CP, Charboneau L, Bichsel VE, Simone NL, Chen T, Pathol 153: 137-150, 1987. Gillespie JW, Emmert-Buck MR, Roth MJ, Petricoin EF 3rd 45 Hibi K, Liu Q, Beaudry GA, Madden SL, Westra WH, Wehage and Liotta LA: Reverse phase protein microarrays which SL, Yang SC, Heitmiller RF, Bertelsen AH, Sidransky D and capture disease progression show activation of pro-survival Jen J: Serial analysis of gene expression in non-small cell lung pathways at the cancer invasion front. Oncogene 20(16): 1981- cancer. Cancer Res 58: 5690-5694, 1998. 1989, 2001. 46 Brichory F, Beer D, Le Naour F, Giordano T and Hanash S: 32 Gygi SP, Rist B, Gerber SA, Turecek F, Gelb MH and Aebersold Proteomics-based identification of protein gene product 9.5 as R: Quantitative analysis of complex protein mixtures using a tumour antigen that induces a humoral immune response in isotope-coded affinity tags. Nat Biotechnol 17(10): 994-999, 1999. lung cancer. Cancer Res 61(21): 7908-7912, 2001.

1253 ANTICANCER RESEARCH 27: 1247-1256 (2007)

47 Tezel E, Hibi K, Nagasaka T and Nakao A: PGP9.5 as a prognostic 65 Borstnar S, Vrhovec I, Svetic B and Cufer T: Prognostic value factor in pancreatic cancer. Clin Cancer Res 6: 4764-4767, 2000. of the urokinase-type plasminogen activator, and its inhibitors 48 Fujii K, Nakano T, Kanazawa M, Akimoto S, Hirano T, Kato and receptor in breast cancer patients. Clin Breast Cancer 3(2): H and Nishimura T: Clinical-scale high-throughput human 138-146, 2002. plasma proteome analysis: lung adenocarcinoma. Proteomics 66 Foekens JA, Kos J, Peters HA, Krasovec M, Look MP, 5(4): 1150-1159, 2005. Cimerman N, Meijer-van Gelder ME, Henzen-Logmans SC, 49 Okano T, Kondo T, Kakisaka T, Fujii K, Yamada M, Kato H, van Putten WL and Klijn JG: Prognostic significance of Nishimura T, Gemma A, Kudoh S and Hirohashi S: Plasma cathepsins B and L in primary human breast cancer. J Clin proteomics of lung cancer by a linkage of multi-dimensional Oncol 16(3): 1013-1021, 1998. liquid chromatography and two-dimensional difference gel 67 Rui Z, Jian-Guo J, Yuan-Peng T, Hai P and Bing-Gen R: Use electrophoresis. Proteomics 6(13): 3938-3948, 2006. of serological proteomic methods to find biomarkers associated 50 Xiao X, Liu D, Tang Y, Guo F, Xia L, Liu J and He D: with breast cancer. Proteomics 3(4): 433-439, 2003. Development of proteomic patterns for detecting lung cancer. 68 Li J, Orlandi R, White CN, Rosenzweig J, Zhao J, Seregni E, Dis Markers 19(1): 33-39, 2003-2004. Morelli D, Yu Y, Meng XY, Zhang Z, Davidson NE, Fung ET 51 Yang SY, Xiao XY, Zhang WG, Zhang LJ, Zhang W, Zhou B, and Chan DW: Independent validation of candidate breast Chen G and He DC: Application of serum SELDI proteomic cancer serum biomarkers identified by mass spectrometry. Clin patterns in diagnosis of lung cancer. BMC Cancer 5: 83, 2005. Chem 51(12): 2229-2235, 2005. 52 Jemal A, Siegel R, Ward E, Murray T, Xu J, Smigal C and Thun 69 Mathelin C, Cromer A, Wendling C, Tomasetto C and Rio MC: MJ: Cancer statistics, 2006. CA Cancer J Clin 56: 106-130, 2006. Serum biomarkers for detection of breast cancers: A 53 Abramovitz M and Leyland-Jones B: A systems approach to prospective study. Breast Cancer Res Treat 96(1): 83-90, 2006. clinical oncology: focus on breast cancer. Proteome Sci 4: 5, 2006. 70 Hu Y, Zhang S, Yu J, Liu J and Zheng S: SELDI-TOF-MS: the 54 Galea MH, Blamey RW, Elston CE and Ellis IO: The proteomics and bioinformatics approaches in the diagnosis of Nottingham Prognostic Index in primary breast cancer. Breast breast cancer. Breast 14(4): 250-255, 2005. Cancer Res Treat 22: 207-219, 1992. 71 Miki Y, Swensen J, Shattuck-Eidens D, Futreal PA, Harshman 55 Ravdin PM, Siminoff LA, Davis GJ, Mercer MB, Hewlett J, K, Tavtigian S, Liu Q, Cochran C, Bennett LM, Ding W et al: A Gerson N and Parker HL: Computer program to assist in strong candidate for the breast and ovarian cancer susceptibility making decisions about adjuvant therapy for women with early gene BRCA1. Science 266(5182): 66-71, 1994. breast cancer. J Clin Oncol 19: 980-991, 2001. 72 Wooster R, Bignell G, Lancaster J, Swift S, Seal S, Mangion J, 56 Goldhirsch A, Glick JH, Gelber RD, Coates AS, Thurlimann B Collins N, Gregory S, Gumbs C and Micklem G: Identification and Senn HJ: Panel members: Meeting highlights: international of the breast cancer susceptibility gene BRCA2. Nature expert consensus on the primary therapy of early breast cancer. 378(6559): 789-792, 1995. Ann Oncol 16: 1569-1583, 2005. 73 Becker S, Cazares LH, Watson P, Lynch H, Semmes OJ, Drake 57 Tetu B and Brisson J: Prognostic significance of HER-2/neu RR and Laronga C: Surfaced-enhanced laser desorption/ oncoprotein expression in node-positive breast cancer. The ionization time-of-flight (SELDI-TOF) differentiation of serum influence of the pattern of immunostaining and adjuvant protein profiles of BRCA-1 and sporadic breast cancer. Ann therapy. Cancer 73(9): 2359-2365, 1994. Surg Oncol 11(10): 907-914, 2004. 58 Talvensaari-Mattila A, Paakko P, Hoyhtya M, Blanco-Sequeiros 74 Laronga C, Becker S, Watson P, Gregory B, Cazares L, Lynch G and Turpeenniemi-Hujanen T: Matrix metalloproteinase-2 H, Perry RR, Wright GL Jr, Drake RR and Semmes OJ: immunoreactive protein: a marker of aggressiveness in breast SELDI-TOF serum profiling for prognostic and diagnostic carcinoma. Cancer 83(6): 1153-1162, 1998. classification of breast cancers. Dis Markers 19(4-5): 229-238, 59 McGuire WL, Tandon AK, Allred DC, Chamness GC and Clark 2003-2004. GM: How to use prognostic factors in axillary node-negative 75 Goncalves A, Esterni B, Bertucci F, Sauvan R, Chabannon C, breast cancer patients. J Natl Cancer Inst 82(12): 1006-1015, 1990. Cubizolles M, Bardou VJ, Houvenaegel G, Jacquemier J, 60 Sahin AA, Ro J, Ro JY, Blick MB, el-Naggar AK, Ordonez Granjeaud S, Meng XY, Fung ET, Birnbaum D, Maraninchi D, NG, Fritsche HA, Smith TL, Hortobagyi GN and Ayala AG: Viens P and Borg JP: Postoperative serum proteomic profiles Ki-67 immunostaining in node-negative stage I/II breast may predict metastatic relapse in high-risk primary breast carcinoma. Significant correlation with prognosis. Cancer 68(3): cancer patients receiving adjuvant chemotherapy. Oncogene 549-557, 1991. 25(7): 981-989, 2006. 61 Wintzer HO, Zipfel I, Schulte-Monting J, Hellerich U and von 76 Sakorafas GH, Tsiotou AG and Tsiotos GG: Molecular biology Kleist S: Ki-67 immunostaining in human breast tumours and of pancreatic cancer; oncogenes, tumour suppressor genes, its relationship to prognosis. Cancer 67(2): 421-428, 1991. growth factors, and their receptors from a clinical perspective. 62 Fedarko NS, Jain A, Karadag A, Van Eman MR and Fisher Cancer Treat Rev 26: 29-52, 2000. LW: Elevated serum bone sialoprotein and osteopontin in 77 Pleskow D, Berher HJ, Gyves J, Allen E, McLean A and colon, breast, prostate, and lung cancer. Clin Cancer Res 7(12): Podolsky DK: Evaluation of a serologic marker, CA19-9, in the 4060-4066, 2001. diagnosis of pancreatic cancer. Ann Intern Med 110: 704-709, 63 Rittling SR and Chambers AF: Role of osteopontin in tumour 1989. progression. Br J Cancer 90(10): 1877-1881, 2004. 78 Nazli O, Bozdag AD, Tansug T, Kir R and Kaymak E: The 64 Choong PF and Nadesapillai AP: Urokinase plasminogen diagnostic importance of CEA and CA 19-9 for the early activator system: a multifunctional role in tumour progression diagnosis of pancreatic carcinoma. Hepatogastroenterology 47: and metastasis. Clin Orthop Relat Res 415: S46-58, 2003. 1750-1752, 2000.

1254 Maurya et al: Serum Proteomics and Cancer (Review)

79 Steinberg W: The clinical utility of the CA 19-9 tumour- 85 Honda K, Hayashida Y, Umaki T, Okusaka T, Kosuge T, associated antigen. Am J Gastroenterol 85: 350-355, 1990. Kikuchi S, Endo M, Tsuchida A, Aoki T, Itoi T, Moriyasu F, 80 Bloomston M, Zhou JX, Rosemurgy AS, Frankel W, Muro- Hirohashi S and Yamada Y: Possible detection of pancreatic Cacho CA and Yeatman TJ: Fibrinogen gamma overexpression cancer by plasma protein profiling. Cancer Res 65(22): 10613- in pancreatic cancer identified by large-scale proteomic analysis 10622, 2005. of serum samples. Cancer Res 66(5): 2592-2599, 2006. 86 Wilson LL, Tran L, Morton DL and Hoon DS: Detection of 81 Zhao J, Simeone DM, Heidt D, Anderson MA and Lubman differentially expressed proteins in early-stage melanoma DM: Comparative serum glycoproteomics using lectin selected patients using SELDI-TOF mass spectrometry. Ann NY Acad sialic acid glycoproteins with mass spectrometric analysis: Sci 1022: 317-322, 2004. application to pancreatic cancer serum. J Proteome Res 5(7): 87 Caputo E, Lombardi ML, Luongo V, Moharram R, Tornatore 1792-1802, 2006. P, Pirozzi G, Guardiola J and Martin BM: Peptide profiling in 82 Xia Q, Kong XT, Zhang GA, Hou XJ, Qiang H and Zhong epithelial tumour plasma by the emerging proteomic RQ: Proteomics-based identification of DEAD-box protein 48 techniques. J Chromatogr B Analyt Technol Biomed Life Sci as a novel autoantigen, a prospective serum marker for 819(1): 59-66, 2005. pancreatic cancer. Biochem Biophys Res Commun 330(2): 526- 88 Mian S, Ugurel S, Parkinson E, Schlenzka I, Dryden I, 532, 2005. Lancashire L, Ball G, Creaser C, Rees R and Schadendorf D: 83 Bhattacharyya S, Siegel ER, Petersen GM, Chari ST, Suva LJ Serum proteomic fingerprinting discriminates between clinical and Haun RS: Diagnosis of pancreatic cancer using serum stages and predicts disease progression in melanoma patients. proteomic profiling. Neoplasia 6(5): 674-686, 2004. J Clin Oncol 23(22): 5088-5093, 2005. 84 Koopmann J, Zhang Z, White N, Rosenzweig J, Fedarko N, Jagannath S, Canto MI, Yeo CJ, Chan DW and Goggins M: Serum diagnosis of pancreatic adenocarcinoma using surface- enhanced laser desorption and ionization mass spectrometry. Received December 19, 2006 Clin Cancer Res 10(3): 860-868, 2004. Accepted February 14, 2007

1255