CANCER GENOMICS & PROTEOMICS 6: 75-84 (2009)

Surface-enhanced Laser Desorption/Ionisation Time-of-flight Mass Spectrometry to Detect Breast Cancer Markers in Tears and Serum

ANTJE LEBRECHT1, DANIEL BOEHM1, MARCUS SCHMIDT1, HEINZ KOELBL1 and FRANZ H. GRUS2

1Department of Obstetrics and Gynecology and 2Experimental Ophthalmology, Department of Ophthalmology, Johannes Gutenberg University Mainz, D-55101 Mainz, Germany

Abstract. Background: New biomarkers are needed to improve women or women with dense breast tissue (2). The the early detection of breast cancer. This study describes the use discovery of molecular markers that can potentially identify of surface-enhanced laser desorption/ionisation time-of-flight such lesions will provide a real opportunity to treat these mass spectroscopy (SELDI-TOF-MS) in serum and tear fluid. neoplasms in time (3). But the lack of good serum tumor Materials and Methods: Blood and tear fluid of 10 women with markers for breast cancer still creates many problems for its breast cancer and 10 healthy age-matched women were molecular diagnosis in early stages (4). Proteomics screened for potential biomarkers. Blood samples and tear fluid technologies offer the opportunity to discover novel were drawn prior to surgery. SELDI-TOF-MS were used for biomarkers (5). Proteomic analysis of human body fluids has protein profiling with three different active surfaces of the become one of the most promising approaches to the protein chips. The data were analyzed by multivariate statistical discovery of biomarkers for human diseases. Body fluid techniques and artificial neural networks. Results: Complex testing provides several key advantages including low protein and peptide patterns were found on all three surfaces. invasiveness, minimum cost and easy sample collection and We identified the main proteins in tear fluid. Statistically processing (6). significant differences in the protein pattern (p<0.001) were The potential of proteomics has been demonstrated in found between breast cancer patients and healthy controls. The various medical areas (e.g. infectious disease (7), diagnostic pattern differentiated cancer patients from controls Alzheimer’s disease (8), cardiovascular disease (9), and with a specificity and sensitivity of approximately 90% in serum oncology (10-14). In breast cancer, significant differences in and tear fluid. Conclusion: Protein chip technology facilitates protein expression in nipple aspirate fluid of healthy women the discovery of new and better biomarkers in breast cancer. It and those with breast cancer were demonstrated (15). Serum is a promising approach to analyse a large number of patients protein profiles to classify breast cancer vs. normal benign with high sensitivity and specificity. Analysing tear fluid could breast diseases showed 90% sensitivity and 94% specificity show some advantages. (5). Other studies have demonstrated that saliva testing may be useful in breast cancer detection (16, 17). Recent trends in breast cancer incidence, survival and In tears, proteins and peptides play an important role as mortality rates provide evidence that both earlier detection antibacterial medium, for transport of water-insoluble and better treatment of breast cancer have contributed to the molecules and as a protection for the cornea. The tear film is recent marked decline in breast cancer mortality (1). a very complex mixture of proteins, lipids, glycoproteins, Unfortunately small lesions are frequently missed and may neuropeptides and cytokines and is far from just being a not be visible even by mammography, particularly in young simple reflection of blood plasma (31-34). Changes in tear protein expressions could be very important biomarkers to analyse disease states or for diagnosis. Biochemical analysis of tear-film composition might provide a useful approach to Correspondence to: Antje Lebrecht, MD, Johannes Gutenberg the search for disease-specific biomarkers and lead to the University Mainz, Department of Obstetrics and Gynecology, discovery of new diagnostic procedures and treatment Langenbeckstrasse 1, D-55101 Mainz, Germany. Tel: +496131177311, options. Altered protein profiles in tear fluid were found in Fax: +49613117472609, e-mail: [email protected] some ocular and systemic disease (e.g. dry eye, diabetes Key Words: SELDI-TOF-MS, breast cancer, biomarkers, proteomics, mellitus) (18-20). Contrary to tear fluid, removal of abundant tear fluid, serum. proteins such as albumin or immunoglobulin is often desired

1109-6535/2009 $2.00+.40 75 CANCER GENOMICS & PROTEOMICS 6: 75-84 (2009)

Figure 1. SELDI-TOF mass spectra of tear protein profiles. SELDI-TOF-MS pattern of a healthy patient (1-3) and a patient with breast cancer (4- 6). The intensity of proteins (U) is plotted versus the molecular mass in a range up to 6,000 Da analyzed with a low laser energy setting.

prior to proteome analysis of serum. There has been concern from a variety of complex biological materials, e.g. serum, regarding whether less abundant serum proteins are removed tears, saliva, urine and cell lysates. This novel fast along with albumin or other proteins (6). technology requires far less material compared to Surface-enhanced laser desorption/ionization time-of- conventional methods and has higher reproducibility (21). flight mass spectrometry (SELDI-TOF-MS; Ciphergen, Using this technology, we performed a study comparing Fremont, CA, USA) separates proteins on different solid- protein profiles in serum and tear fluid from a group of phase chromatographic surfaces (ProteinChip) and the breast cancer patients and known healthy controls. To the proteins are subsequently ionized and detected by TOF-MS. Authors’ knowledge, no data on protein profiling of tear fluid ProteinChip technology is able to analyze protein profiles in breast cancer patients have been reported to date.

76 Lebrecht et al: SELDI-TOF-MS of Tear and Serum Proteins in Breast Cancer

Figure 2. Detection of the two main tear proteins lysozyme (1) and lipocalin (2) analyzed on the three different array surfaces (H50, Q10, and CM10) at high laser energy.

Figure 3. Heat map of protein expression patterns in tears at each condition. In this graph, the intensity of protein expression is shown in colour values. The x-axis displays the corresponding molecular mass region and the chip surface and the y-axis the different groups (group 1 samples from controls, group 2 samples from cancer patients).

77 CANCER GENOMICS & PROTEOMICS 6: 75-84 (2009)

The aim of the study was to determine whether it was with an approximate volume of 250 to 300 μl can be handled by the possible to protein profile tear and serum specimens using robotic station. The washing steps with the dodecyl eluate were SELDI-TOF-MS and if these protein profiles were possibly performed without using the Biomek. altered in the presence of carcinoma of the breast. ProteinChip analysis. The two eluates from the tear fluid and the three eluates from the serum samples were assayed on three different Patients and Methods chromatographic surfaces: a week cation exchange surface (CM10), a strong anion exchange surface (Q10) and a reversed-phase surface Patient characteristics. This prospective study included 20 patients: (H50). All ProteinChip Arrays were pretreated according to the 10 patients with breast cancer and 10 healthy patients matched for standard protocols of the manufacturer. Binding buffers were 5% age. Informed consent was obtained from all patients participating acetonitrile/0.1% trifluoroacetic acid (H50), 20 mM sodium acetate in the study and the protocols were approved by the institutional buffer (pH 5; CM10), and 50 mMTris (pH 8; Q10). Ethics Committee and conformed to the provisions of the Twenty microliters of the TFA elution and 20 μl of binding Declaration of Helsinki. Blood samples and tear fluid were drawn buffer were applied to each spot using the Biomek. The arrays prior to surgery. Patients’ blood was obtained 24-48 hours before were incubated on the DPC shaking platform for 1 hour and the surgery by peripheral venous puncture and was immediately solution in each well was removed by the Biomek. All wells were centrifuged at 4000 xg for 5 min. The serum was frozen at –80˚C washed with sodium acetate (pH 4) followed by a wash step with until examination. Tear fluid was eluted from Schirmer strips. This distilled water for 5 min to remove buffer salts. After the wells basal secretory test (BST) to collect the tear fluid was also were dry, the Biomek was used to apply 2 μl saturated sinapinic performed 24-48 hours before surgery and the Schirmer strips were acid solution (an energy-absorbing molecule) in 50% acetonitrile stored at –80˚C until use. and 0.5% TFA to each spot. After the spots were air dried, each spot was analyzed in a ProteinChip Reader. Each sample (2 μl per Serum assay. For preparation and purification of the serum samples, spot) was bound to each array surface in duplicate on separate purification kits 100 MB-HIC 8 and 18, and MB-WCX (Bruker arrays and BioProcessors. Daltonics Inc., Billerica, MA, USA) were used. The 100 MB-HIC 8 The dodecyl eluate (2 μl per spot) was applied directly on the and 18 purification kits are based on super-paramagnetic ProteinChips. The washing steps with sodium acetate (pH 4) and microparticles with a highly porous surface functionalized with distilled water were carried out as mentioned above. hydrophobic coatings. The three resulting eluates from the serum samples were also Ten μl binding solution and 5 μl of the serum sample were mixed directly applied to the ProteinChips (2 μl eluate and 2 μl matrix with 5 μl magnetic bead suspension by pipetting up and down five per spot). times. After 1 min, the beads were separated from the supernatant by a magnetic separator for 20 s. Washing with 100 μl wash solution Data acquisition and preprocessing. ProteinChip Arrays were was performed twice. To elute the peptides and proteins, 5 μl 50% analyzed on a PBS-IIc ProteinChip Reader equipped with a acetonitrile were added and mixed thoroughly before transferring ProteinChip Array AutoLoader using the ProteinChip Software the elution to a fresh tube. version 3.2 (Ciphergen Biosystems, Inc.). The AutoLoader is able to The MB-WCX kit is based on super-paramagnetic microparticles analyse up to 24 ProteinChip Arrays (192 spots) at one time. Each with negatively charged functional groups at their surface enabling array was read at two laser intensities: low intensity optimized for cation exchange chromatography. Ten μl binding solution and 10 μl low molecular mass proteins and high intensity for high molecular MB-WCX beads suspension were mixed by pipetting up and down mass proteins. The high-intensity protocol averaged 195 laser shots before 5 μl serum sample were applied and mixed again. Incubation from each spot with a laser intensity of 190, a deflector setting of for 5 min was followed by collecting the supernatant. Washing with 3,000 Da, a detector sensitivity of 9, and a molecular mass detection 100 μl wash solution, collecting the beads from the tube wall and range of 2,000 to 200,000 Da. For low-intensity measurements, the removal of the supernatant was repeated twice. By adding 5 μl laser intensity was set at 180 and the deflector set at 1,500 Da. The elution solution and pipetting up and down ten times, the beads raw data were transferred to the CiphergenExpress Data Manager were dissolved from the tube wall. After incubation for 2 min, the Software version 2.1 (CE; Ciphergen Biosystems) for analysis. supernatants were transferred to a fresh tube and mixed with 5 μl stabilization solution. Data analysis. The CE data manager software was used to normalize the spectra, to automatically detect peaks and to create Tear fluid. The Schirmer strips were eluted overnight in 500 μl the peak cluster lists. The peak intensities were normalized 0.1% dodecylmaltoside and then for two hours in 300 μl 0.1% according to the total ion current. The cluster lists were exported as trifluoroacetid acid (TFA) resulting in two elutions, a TFA elution ASCII files to a statistical analysis program (Statistica, ver. 6.2; and a dodecyl elution. For automatic handling of all binding and StatSoft, Tulsa, OK, USA). The cluster lists contained normalized washing steps with the TFA elution, a robotic laboratory automation peak intensity values for each sample within a group. Based on station (Biomek 2000; Beckman Coulter, Fullerton, CA, USA) was these normalized peak intensities, probabilities based on t-tests and used. Sodium acetate (pH 4) was used as a washing buffer. The multivariate discriminant analysis were calculated. station was extended by an integrated microplate mixer (Micromix The separation between clinical groups for diagnostic purposes 5; Diagnostic Products Company, Los Angeles, CA, USA), which can generally be enhanced by using multimarker panels of protein held the ProteinChip Array BioProcessor (Ciphergen biosystems, biomarkers rather than single biomarkers. The statistical software Inc.). This BioProcessor was equipped with 12 ProteinChip Arrays, package (Statistica) was used to perform a multivariate discriminant each having eight spots. With two BioProcessors, up to 192 wells analysis based on combinations of multiple biomarker peaks. This

78 Lebrecht et al: SELDI-TOF-MS of Tear and Serum Proteins in Breast Cancer discriminant analysis tests the zero hypothesis, namely that mean each surface (CM10, H50, and Q10). For each of these biomarker peaks of the different clinical groups derive from a cluster lists, a multivariate discriminant analysis was multivariate normally distributed population, and also shows which performed which defined a total of 8 peaks with significant of the various groups are statistically different. The discriminant differences between the tear (p<0.0080) and serum analysis selected the eight most important protein biomarkers for best discrimination between the groups. (p<0.0023) protein profiles of control and cancer samples. To assess the diagnostic power of this eight biomarker panel, Figures 4 and 5 show the different protein profiles in serum a neural network was trained using these markers as input and tears of control and cancer samples, respectively. Using neurons. The traditional artificial neural network (ANN) model this biomarker panel for input, an artificial neural network is the most widely used today and is a multiple-layer feed- was trained with a training data set. The performance of the forward network (MLFN) with the back-propagation training trained net was assessed using a test data set. Figures 6 and 7 algorithm (22). In this study, the neural network module show the ROC curves of sera and tears with an AUC of 0.95 provided by the software (Statistica) was used. The network was trained by the eight biomarker peaks and the output neurons and 0.90, and a specificity and sensitivity of approximately defined whether the protein pattern classified the patient as 90% for each in sera and tears, respectively. having breast cancer or not. To assess the performance of the network, the data set was randomly divided into two parts: the Discussion first half of the patients and controls was used as training set, the second as a test set. After completion of training, the success of The findings of this study appear to satisfy its objectives: it the algorithm was tested using the test set, which comprised was possible to protein profile tear and serum specimens samples from patients and controls to which the neural network had not previously been exposed. using mass spectrometry and these protein profiles were The software generates a receiver operating characteristic (ROC) somewhat altered in the presence of carcinoma of the breast. curve by plotting sensitivity against 1 – specificity. A test that The results of the pilot study suggest that protein profiling perfectly discriminates between two groups would yield a curve that using SELDI technology and other mass spectrometry-based coincided with the left and top sides of the plot. A global assessment approaches could be applied to tear fluid and/or serum of the performance of the test (sometimes called diagnostic biomarker discovery with regard to breast cancer. Based on accuracy) is given by the area under the ROC curve (AUC). A the evidence provided in the Results section, the perfect marker with complete group separation would have an ROC AUC value of 1. investigation suggests that tear fluid did not require special sample preparation as serum samples did. Results The technology primarily used to date, namely two- dimensional (2-D) gel electrophoresis, allows analysis and Serum and tear samples were analyzed on three different quantification of proteins and the establishment of a protein ProteinChip Array chemistries as described. Figure 1 map. However, this technology is very time consuming and is exemplarily shows the SELDI-TOF mass spectra of tear limited by problems in reproducibility, which is particularly protein profiles at low laser energy. The figure also illustrates important when it is used in clinical routine with many the complexity of tear protein profiles in the low molecular samples. Furthermore, the sensitivity of 2-D electrophoresis mass range and the variability from patient to patient. limits the analysis to proteins larger than 8 to 10 kDa (20). In In tears, SELDI-TOF-MS succeeded in identifying the main contrast, the SELDI-TOF-MS technology allows rapid sample tear proteins, lysozyme and lipocalin, for the three different analysis because very small sample volumes can be directly chip surfaces as shown in Figure 2. Furthermore, Figure 3 applied to the ProteinChip Array surfaces and the process can shows a heat map of tear eluate, where the protein expressions easily be automated for high-throughput analysis (23). are shown as pseudocolors for the molecular mass region To identify tears and serum biomarkers in this study, between 3,000 and 70,000 Da (x-axis) and relative peak comparisons were made between the breast cancer individuals intensity (y-axis) for the different groups (CRTL=1 and and the healthy controls. We found statistically significant CA=2) for the various conditions. The heat map provides a biomarker clusters which discriminated between breast cancer more comprehensive overview of the complex protein patterns patients and healthy controls in both tear and serum samples. found on the different array surfaces. At the present time, we do not know the identity of these ion In conclusion, the results shown in Figures 1 to 3 provide signals. After validating the biomarker profiles in a larger group evidence that SELDI-TOF-MS ProteinChip Array analysis of patients, we will identify the proteins in a second step. is very useful for the analyses of tear and serum protein For breast cancer, early diagnosis is much more significant profiles. than any treatment (24). The main diagnostic measures such The other objective of this study was to search for proteins as self-examination, clinical examination and mammography or peptides that are differentially expressed in healthy women do little for patients whose tumors are small and current and those with breast cancer. Cluster lists were generated for serum tumor markers such as CA 15-3 and CEA do not have each condition (low- and high-energy laser settings) and for enough sensitivity or specificity for these patients (4).

79 CANCER GENOMICS & PROTEOMICS 6: 75-84 (2009)

Figure 4. Protein profiles of serum samples from controls (group 1) and cancer patients (group 2).

Figure 5. Protein profiles of tear samples from controls (group 1) and cancer patients (group 2).

80 Lebrecht et al: SELDI-TOF-MS of Tear and Serum Proteins in Breast Cancer

Figure 6. ROC curves of serum samples.

Figure 7. ROC curves of tear samples.

81 CANCER GENOMICS & PROTEOMICS 6: 75-84 (2009)

As in other tumor entities, in breast cancer some authors 12 Petricoin EF, Ardekani AM, Hitt BA, Levine PJ, Fusaro VA, tried to find protein profiles to discriminate between breast Steinberg SM, Mills GB, Simone C, Fishman DA, Kohn EC and cancer patients and healthy controls (5, 6). Serum samples, Liotta LA: Use of proteomic patterns in serum to identify ovarian cancer. Lancet 359: 572-577, 2002. nipple aspirate and saliva have been used as diagnostic 13 Chen YD, Zheng S, Yu JK and Hu X: Artificial neural networks fluid. Besides serum samples, we chose tear fluid because it analysis of surface-enhanced laser desorption/ionization mass is easily and noninvasively obtained and the lack of highly spectra of serum protein pattern distinguishes colorectal cancer abundant proteins obviates the need for purification of the from healthy population. Clin Cancer Res 10: 8380-8385, samples. Comparable to the published data (25-30), we 2004. found significantly different protein profiles in breast 14 Yu JK, Cheng YD and Zheng S: An integrated approach to the cancer patients and in healthy controls, both in tear fluid detection of colorectal cancer utilizing proteomics and and serum. The specificity and the sensitivity were both bioinformatics. World J Gastroenterol 10: 3127-3131, 2004. 15 Pawlik TM, Fritsche H, Coombes KR, Xiao L, Krishnamurthy over 90% . S, Hunt KK, Pusztai L, Cheng J, Clarke CH, Arun B, Hung M The results of this study suggest that SELDI-TOF-MS and Kuerer HM: Significant differences in nipple aspirate fluid may be a useful tool in the development of the detection of protein expression between healthy women and those with breast tear and serum biomarkers in breast cancer. cancer demonstrated by time-of-flight mass spectrometry. Breast Cancer Res Treat 89: 149-157, 2005. References 16 Streckfus CF, Bigler LR and Zwick M: The use of surface- enhanced laser desorption/ionization time-of-flight mass 1 Chu KC, Tarone RE, Kessler LG, Ries LAG, Hynkey BF, Miller spectrometry to detect putative breast cancer markers in saliva: a BA and Edwards BK: Recent trends in U.S. Breast Cancer feasibility study. J Oral Pathol Med 35: 292-300, 2006. Incidence, survival and Mortality Rates. J Natl Cancer Inst 88: 17 Streckfus CF and Bigler LR: Saliva as a diagnostic fluid. Oral 1571-1579, 1996. Disease 8: 69-76, 2002. 2 Antman K and Shea S: Screening mammography under age 50. 18 Herber S, Grus FH, Sabuncuo P and Augustin AJ: Changes in the JAMA 281: 1470-1472, 1999. tear protein patterns of diabetic patients using two-dimensional 3 Jinong L, Zhao J, Yu X, Lange J, Kuerer H, Krishnamurthy S, elecrophoresis. Adv Exp Med Biol 506: 623-626, 2002. Schilling E, Khan SA, Sukumar S and Chan DW: Identification 19 Stolwijk TR, Kuizenga A van Haeringen NJ, Kijlstra A, of biomarkers for breast cancer in nipple aspiration and ductal Oosterhuis JA and van Best JA: Analysis of tear fluid proteins lavage fluid. Clin Cancer Res 11: 8312-8320, 2005. in insulin-dependent diabetes mellitus. Acta Ophthalmol 72: 4 Hu Y, Zhang S, Yu J, Liu J and Zheng S: SELDI-TOF-MS: the 357-362, 1994. proteomics and bioinformatics approaches in the diagnosis of 20 Grus FH, Podust VN, Bruns K, Lackner K, Fu S, Dalmasso EA, breast cancer. Breast 14: 250-255, 2005. Wirthlin A and Pfeiffer N: SELDI-TOF-MS ProteinChip array 5 Clarke CH, Buckley JA and Fung ET: SELDI-TOF-MS proteomics profiling of tears from patients with dry eye. Invest Ophthalmol of breast cancer. Clin Chem Lab Med 43: 1314-1320, 2005. Vis Sci 46: 863-876, 2005. 6 Hu S, Loo JA and Wong DT: Human body fluid proteome 21 Grus FH, Joachim SC and Pfeiffer N: Analysis of complex analysis. Proteomics 6: 6326-6353, 2006. autoantibody repertoires by surface-enhanced laser desorption/ 7 Bernhard OK, Diefenbach RJ and Cunningham AL: New ionization-time of flight mass spectrometry. Proteomics 3: 957- insights into viral structure and virus-cell interactions through 961, 2003. proteomics. Expert Rev Proteomics 2: 577-588, 2005. 22 Rumelhart D, Hinton G and Williams J: Learning representations 8 Ho L, Sharma N, Blackman L, Festa E, Reddy G and Pasinetti by back-propagating procedures. Nature 323: 533-536, 1986. GM: From proteomics to biomarker discovery in Alzheimer’s 23 Wulfkuhle JD, Paweletz CP, Steeg PS, Petricoin EF and Liotta L: disease. Brain Res Brain Res Rev 48: 360-369, 2005. Proteomic approaches to the diagnosis, treatment, and 9 Anderson L: Candidate-based proteomics in the search for monitoring of cancer. Adv Exp Med Biol 532: 59-68, 2003. biomarkers of cardiovascular disease. J Physiol 563: 23-60, 2005. 24 Smith RA, Cokkinides V and Eyre HJ: American Cancer 10 Petricoin EF 3rd, Ornstein DK, Paweletz CP, Ardekani A, Society. American Cancer Society guidelines for the early Hackett PS, Hitt BA, Velassco A, Trucco C, Wiegand L, Wood detection of cancer. CA Cancer J Clin 54: 41-52, 2004. K, Simone CB, Levine PJ, Linehan WM, Emmert-Buck MR, 25 de Noo ME, Deelder A, van der Werff M, Ozalp A, Mertens B Steinberg SM, Kohn EC and Liotta LA: Serum proteomic and Tollenaar R: MALDI-TOF serum protein profiling for the patterns for detection of prostate cancer. J Natl Cancer Inst 94: detection of breast cancer. Onkologie 29: 501-506, 2006. 1576-1578, 2002. 26 Belluco C, Petricoin EF, Mammano E, Facchiano F, Ross- 11 Semmes OJ, Feng Z, Adam BL, Banez LL, Bigbee WL, Campos Rucker S, Nitti D, Di Maggio CD, Liu C, Lise M, Liotta LA and D, Cazares LH, Chan DW, Grizzle WE, Izbicka E, Kagan J, Whiteley G: Serum proteomic analysis identifies a highly Malik G, McLerran D, Moul JW, Partin A, Prasanna P, sensitive and specific discriminatory pattern in stage 1 breast Rosenzweig J, Sokoll LJ, Srivastava S, Srivastava S, Thompson cancer. Ann Surg Oncol 14: 2470-2476, 2007. I, Welsh MJ, White N, Winget M, Yasui Y, Zhang Z and Zhu L: 27 He J, Gornbein J, Shen D, Lu M, Rovai LE, Shau H, Katz J, Evaluation of serum protein profiling by surface-enhanced laser Whitelegge JP, Faull KF and Chang HR: Detection of breast desorption/ionization time-of-flight mass spectrometry for the cancer biomarkers in nipple aspirate fluid by SELDI-TOF and detection of prostate cancer: I. Assessment of platform their identification by combined liquid chromatography-tandem reproducibility. Clin Chem 51: 102-112, 2005. mass spectrometry. Int J Oncol 30: 145-154, 2007.

82 Lebrecht et al: SELDI-TOF-MS of Tear and Serum Proteins in Breast Cancer

28 Mendrinos S, Nolen JDL, Styblo T, Carlson G, Pohl J, Lewis M 31 Janssen PT and van Bijsterveld OP: Origin and biosynthesis of and RitchieJ: Cytologic findings and protein expression profiles human tear fluid proteins. Invest Ophthalmol Vis SCI 24: 623- associated with ductal carcinoma of the breast in ductal lavage 630, 1983. specimens using surface-enhanced laser desorption and 32 Sariri Rand Ghafoori H: Tear proteins in health, disease, and ionisation-time of flight mass spectrometry. Cancer 105: 178- contact lens wear. Biochemistry 73: 381-392, 2008. 183, 2005. 33 Van Haeringen NJ: Clinical biochemistry of tears. Surv 29 Alexander H, Stegner AL, Wagner-Mann C, Du Bois GC, Ophthalmol 26: 84-96, 1981. Alexander S and Sauter ER: Proteomic analysis to identify breast 34 Tiffany JM: The normal tear film. Dev Ophthalmol 41: 1-20, cancer biomarkers in nipple aspirate fluid. Clin Cancer Res 10: 2008. 7500-7510, 2004. 30 Kuerer HM, Goldknopf IL, Fritsche H, Krishnamurthy S, Sheta EA and Hunt KK: Identification of distinct protein expression patterns in bilateral matched pair breast ductal fluid specimens Received July 22, 2008 from women with unilateral invasive breast carcinoma. Cancer Revised December 24, 2008 95: 2276-2282, 2002. Accepted December 29, 2008

83