Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 Oct 5.
Published in final edited form as: J Oral Sci. 2012 Mar;54(1):61–70. doi: 10.2334/josnusd.54.61

Discovery of Putative Salivary Biomarkers for Sjögren’s Syndrome Using High Resolution Mass Spectrometry and Bioinformatics

Driss Zoukhri 1),2), Ian Rawe 3), Mabi Singh 1), Ashley Brown 1), Claire L Kublin 1), Kevin Dawson 4),5), William F Haddon 4),6), Earl L White 4),7), Kathleen M Hanley 4),8), Daniel Tusé 4),9), Wasyl Malyj 4), Athena Papas 1)
PMCID: PMC9533682  NIHMSID: NIHMS1838183  PMID: 22466888

Abstract

The purpose of the current study was to determine if saliva contains biomarkers that can be used as diagnostic tools for Sjögren’s syndrome (SjS). Twenty seven SjS patients and 27 age-matched healthy controls were recruited for these studies. Unstimulated glandular saliva was collected from the Wharton’s duct using a suction device. Two μl of salvia were processed for mass spectrometry analyses on a prOTOF 2000 matrix-assisted laser desorption/ionization orthogonal time of flight (MALDI O-TOF) mass spectrometer. Raw data were analyzed using bioinformatic tools to identify biomarkers. MALDI O-TOF MS analyses of saliva samples were highly reproducible and the mass spectra generated were very rich in peptides and peptide fragments in the 750–7,500 Da range. Data analysis using bioinformatic tools resulted in several classification models being built and several biomarkers identified. One model based on 7 putative biomarkers yielded a sensitivity of 97.5%, specificity of 97.8% and an accuracy of 97.6%. One biomarker was present only in SjS samples and was identified as a proteolytic peptide originating from human basic salivary proline-rich protein 3 precursor. We conclude that salivary biomarkers detected by high-resolution mass spectrometry coupled with powerful bioinformatic tools offer the potential to serve as diagnostic/prognostic tools for SjS.

Keywords: MALDI O-TOF, saliva, biomarkers, bioinformatics, Sjögren’s syndrome

Introduction

Secretion of saliva in adequate quantity and quality is of paramount importance for the maintenance of oral health (1–2). Saliva contains a variety of antimicrobial proteins and growth factors that play a crucial role in the homeostasis of the oral cavity (1–3) Saliva serves a number of different functions including lubrication and protection of oral tissues, buffering action and clearance, maintenance of tooth integrity, antibacterial activity, and taste and digestion (1–3). A decrease or lack of salivary gland secretion leads to dry mouth or xerostomia, as observed in Sjögren’s syndrome (SjS).

SjS is a systemic inflammatory disease affecting primarily the lacrimal and salivary glands (4–5). It may exist as a primary disorder (primary SjS) or can be associated with other autoimmune diseases (secondary SjS) such as rheumatoid arthritis, systemic lupus erythematosus, or systemic sclerosis. The etiology of SjS, although still ill-defined, is thought to be multifactorial, involving viral, neural, genetic, and environmental factors (4, 6). This fact adds an extra layer of complexity in diagnosing SjS (4–5). The prevalence of this syndrome varies widely depending on the criteria for classification, but it is estimated that between 1 and 3 million North Americans suffer from this disease (7–8). Clinically, if unmanaged, SjS can have a severe impact on the integrity of the oral cavity (4, 6). SjS patients usually experience difficulty swallowing, severe and progressive tooth decay, and oral infections (particularly fungal). To date there are no treatments or cures for this disease.

SjS is considered to be the most commonly under-diagnosed autoimmune disease. It is not uncommon for there to be a delay of 5 to 8 years after symptom onset before a diagnosis is made (9–10). Furthermore, the diagnostic approach to SjS is complicated because it must include two different goals: firstly, assessment of the ocular and salivary components, and secondly, differentiation between the primary and secondary variants of the syndrome (5). Current diagnostic criteria for primary SjS include subjective (patient reported) and objective (measurable) signs of dry eye and/or dry mouth, presence of autoantibodies against the ribonucleoproteins Ro/SSA and/or La/SSB, and focal lymphocytic sialoadenitis (5). Obviously, these criteria can only detect patients with established disease when glandular destruction has already occurred. Furthermore, most of these tests either lack specificity (unstimulated salivary flow) or are invasive (lip biopsies) (11). Therefore, the development of accurate and non-invasive tests that can lead to earlier diagnosis or predict the onset of disease would be of considerable value to improve treatment and prognosis of SjS patients.

Individual markers have been used successfully in human disease diagnosis, classification, and prognosis. However, it is now recognized that multiplexed profiling (studying multiple genes or proteins to form a marker signature) greatly improves discrimination between patient populations (i.e., diseased vs. healthy) (12–13). Mathematically, a pattern of multiple biomarkers, when compared to a single biomarker, may contain a higher level of discriminatory power (12–13). This is particularly important when studies are conducted on heterogeneous patient populations and deal with complex multifactorial diseases, such as cancer or autoimmune diseases (12–13).

Several genes, proteins, and their metabolic byproducts have been known to change during developing stages of human disease. Recently, proteomic and metabolomic marker profiles of individuals are being recognized as telltale signatures, which can provide early warning detection of disease (14–17). The latest advances in mass spectrometry coupled with the development of algorithms capable of handling the massive data generated by mass spectrometry, offer great hope for accurate and early diagnosis of human diseases.

The aim of the present studies was to determine if biomarkers that can be used for diagnosis of SjS are present in saliva. We used a combination of high-resolution mass spectrometry coupled with patented bioinformatic algorithms to analyze samples collected from well-defined SjS patients and healthy control individuals.

Materials and Methods

Patients

All experiments were conducted in accordance with the Declaration of Helsinki (http://wma.net). Saliva samples were collected from SjS patients and healthy controls. SjS patients were recruited from The Rubenstein Clinic at Tufts University School of Dental Medicine. Informed consent was obtained from all participants in accordance with the TUFTS-NEMC Institutional Review Board guidelines. A total of 27 SjS patients (18 with diagnosed primary SjS and 9 diagnosed with secondary SjS) and 27 healthy controls were recruited for saliva collection. SjS patients’ characteristics are summarized in Table 1.

Table 1.

Summary of Sjögren’s syndrome patients’ characteristics

Primary (n = 18) Secondary (n = 9) †
Age * 63 ± 10 (34–81)# 62 ± 11 (50–81) #
Sex (F/M) 14/0 12/1
Disease duration * § 12 ± 8 (1–30)# 12 ± 6 (3–20) #
Salivary flow rate ‡
 Unstimulated 0.066 ± 0.100 0.017 ± 0.021
 Stimulated 0.564 ± 0.633 0.703 ± 0.659
Open in a new tab
†

Five with rheumatoid arthritis, 1 with systemic lupus erythematosus (SLE), 1 with CREST syndrome (limited scleroderma) and SLE, 1 with fibromyalgia and celiac disease and 1 with liver disease.

*

Mean (years) ± S.D.

§

Since diagnosis.

#

Range (years).

‡

mL/min (mean ± S.D.)

Saliva Collection

The patients were instructed not to eat, smoke, or drink for at least three hours prior to the collection of saliva. Furthermore, all collections were performed at a fixed time of the day to minimize fluctuations related to circadian rhythm in salivary secretion. The patients were seated and the floor of the mouth was isolated with sterile gauge pad. Unstimulated saliva was collected from the submandibular glands (Wharton’s duct) using a suction device. Saliva samples, approximately 0.2 ml, were transferred to a clean 1 ml microfuge tube kept on ice and centrifuged for 20 min at 12,000 rpm in a benchtop-refrigerated centrifuge. The samples were then transferred to clean 1 ml microfuge tubes and kept at −80°C until analysis.

Sample Processing and Mass Spectrometry

A 2 μl saliva sample volume was found to be adequate for analyses. Sample preparation and biomarkers elution was carried out as described by Lopez et al. (16). For reproducibility, each sample was analyzed in triplicate and applied to three different wells on the microtiter plate. Samples were eluted directly from the resin ZipPlate onto disposable MALDIchip™ targets (PerkinElmer, Boston, MA).

Processed saliva samples were analyzed by mass spectrometry on a prOTOF™ 2000 matrix assisted laser desorption/ionization orthogonal time-of-flight (MALDI O-TOF) mass spectrometer (PerkinElmer) as described by Lopez et al. (16). Data acquisition was performed using a mass range of 750–12,000 Daltons.

Processing and Analysis of Spectral Profiles

The raw mass spectral data were analyzed using PROFILE™, a proprietary biomarker discovery platform optimized for the sensitive and accurate identification of disease-specific signatures in high-resolution MALDI-TOF mass spectra. This biomarker discovery platform successfully identified biomarker signatures in Alzheimer’s disease (16), multiple sclerosis (17), and pheochromocytoma (18).

Spectra were baseline-corrected, m/z-aligned, normalized, de-noised, and if present, sample processing artifacts were suppressed. The pre-processed spectra were compared to each other and outlier spectra were identified using a combination of spectral entropy and Euclidean distance calculation-based methods. Spectra that passed the outlier rejection step were used for calculating the average spectrum for each sample. Peak locations were identified in the spectra and peak intensities or a well-defined portion of the area under the curve around the peak were used as input to proprietary feature selection and model building algorithms.

During feature selection and model building, the data set was split into a training set and a test set (two-thirds training and one-third test), and the training set was used to build a classification model capable of classifying samples with optimal sensitivity and specificity. In the feature selection step, algorithms were run to identify features in the qualified spectral data that were clearly different among groups. Typically, the feature selection process identifies numerous features; however, only some may be useful as diagnostic predictors. PROFILE™ can select more than one type of learning algorithm and can build multiple competing classification models. A model consists of a particular combination of features. The algorithms identify the minimum number of features included in the optimal model. After the model was built, its classification performance was evaluated on the test set. This procedure was repeated at least 50 times (50-fold cross-validation) and the statistical result of these analyses yielded sensitivity and specificity figures for the particular model. The contribution of each individual biomarker to the model performance was evaluated using sensitivity analysis that is the re-evaluation of the model performance with the omission of the particular biomarker. Performance of these models was evaluated using external cross-validation with a test set using a different set of samples not included in the original model.

Sample fractionation and Edman sequencing

Saliva samples proteins and peptides were fractionated by high performance liquid chromatography (HPLC) prior to protein sequencing by Edman degradation. A 200 μl aliquot of each sample was placed in an Agilent autosampler vial and fractionated using an Agilent 1100 HPLC system (Palo Alto, CA) equipped with a quaternary pumping system and a photodiode array (PDA) detector. Eluted fractions were evaporated to dryness, then reconstituted and an aliquot of each of these fractions was analyzed by MALDI O-TOF MS, as described above. The mass spectral data were then interrogated for detection of candidate biomarkers. An aliquot of the remainder of the fractions was subjected to Edman sequencing on an Applied BioSystems 494HT instrument.

Results

Reproducibility of sample handling and analysis

For a biomarker profile to be rigorous enough to be diagnostically relevant, the profile has to be extremely reproducible (19–20). Figure 1A shows three sets of mass spectral profiles measured from the same saliva sample that was processed three times. The salivary mass spectra were very rich in peptides and peptide fragments contained mostly in the 750–7,000 m/z range (Fig. 1A). Importantly, the replicate profiles were nearly identical showing the reproducibility of sample processing and data acquisition by the prOTOF MALDI-MS.

Fig. 1.

Fig. 1

Open in a new tab

Reproducibility of sample handling and analysis. A. Shown are mass spectral profiles measured from the same saliva sample that was processed three times. B. Shown are mass spectral profiles measured from the same saliva sample that was processed after 1 week and 9 months storage at −80°C.

Another critical factor that can alter the mass spectral profiles and introduce errors is sample handling/storage (21). Therefore, we processed saliva samples from the same patients, 1 week following collection and storage at −80°C and then after a 9-month storage period at −80°C. Figure 1B shows representative mass spectral profiles from one subject. It is evident from Fig. 1B that both spectral profiles display similar patterns suggesting that long-term storage (−80°C) has no (or minimal) effect on sample integrity.

Taken together, these results show that our protocol for sample preparation/storage and the data acquisition from the MALDI O-TOF MS are reproducible.

Discovery of putative biomarkers for SjS in saliva

Saliva samples from control and SjS patients were then processed for mass spectrometry analyses. The same investigator analyzed the samples in a blind and random manner. Furthermore, to control for reproducibility, each sample was analyzed in triplicate. Figure 2 shows representative mass spectral profiles of saliva samples collected from 3 control subjects and 3 SjS patients.

Fig. 2.

Fig. 2

Open in a new tab

Representative mass spectra of saliva samples collected from 3 control individuals and 3 SjS patients.

A visual inspection of the mass chromatograms of three SjS’ saliva samples and the mass spectral data from three control subjects indicate significant mass peak differences (Fig. 2). In all cases, the control samples and the SjS’ saliva samples contained several peptides and peptide fragments that spanned the 750–10,000 m/z range (Fig. 2).

PROFILE™ identified 4 outlier spectra, 3 of which originated from the same subject (Fig. 3). This subject and the fourth outlier spectrum were removed before further analysis. The remaining spectra were then background-corrected and average spectra were computed.

Fig. 3.

Fig. 3

Open in a new tab

Outlier detection feature of PROFILE™. Four outlier spectra were detected (arrow), 3 of which originated from the same subject.

Pattern recognition and feature selection algorithms of PROFILE™ were applied to detect spectral features distinguishing the two groups of subjects. Model performance was evaluated using 50-fold cross-validation and the model sensitivity/specificity parameters were reported. PROFILE™ successfully produced several classification models. The two best models are summarized in Tables 2 and 3. Both models are composed of 7 putative biomarkers, 3 of which are present in both models (Table 2, underlined m/z values). The 2 models were selected because of their high sensitivity, specificity and accuracy. Table 2 shows the mass to charge ratio (m/z) of the biomarkers selected for the two models generated and Table 3 shows the sensitivity, specificity and accuracy of the biomarker models.

Table 2.

Sjögren’s syndrome models peptide mass generated by PROFILE™

Putative Biomarkers (m/z)
Model 1 902.48, 2,407.30, 2,912.56, 3,655.78, 3,803.38, 4,281.14, 5,942.00
Model 2 902.48, 1,479.76, 2,536.16, 3,655.78, 3,803.38, 4,930.28, 5,843.94
Open in a new tab

Table 3.

Sensitivity, specificity, and accuracy of Sjögren’s syndrome models

Model Sensitivity Specificity Accuracy
1 97.5% 97.8% 97.6%
2 85.0% 91.3% 88.1%
Open in a new tab

One SjS-specific marker (m/z = 3803.38) showed peaks with 0.5 amu periods in the m/z = 3,803–3,808 range (Fig. 4). As expected, another SjS-specific marker was detected in the m/z = 7,606–7,618 range with 1.0 amu periods, which proves that the SjS-specific marker at m/z = 3,803.38 is a doubly charged ion (Fig. 4). This discovery provides not only a novel putative biomarker for SjS but also an unexpected technical advantage. Namely, the doubly charged ion in the MALDI-TOF spectra can be detected at high-sensitivity because it is located between the background peaks where signals from singly charged ions are not detected.

Fig. 4.

Fig. 4

Open in a new tab

Heat map and spectral profile of one putative peptide biomarker. A. Heat map. B. One SjS-specific marker (m/z = 3,803.38) showed peaks with 0.5 amu periods in the m/z = 3,803–3,808 range. C. Another SjS-specific marker was detected in the m/z = 7,606–7,618 range with 1.0 amu periods. Blue traces, SjS samples; red traces, control samples.

These data show that saliva contain putative biomarkers that can serve to distinguish SjS patients from healthy control subjects with high sensitivity and specificity. Noteworthy is that this biomarker was detected in both primary and secondary SjS patients.

Sample fractionation and Edman sequencing

Experiments were conducted to further characterize the molecular identity of the doubly charged ion that was only present in SjS patients’ saliva samples. Saliva samples from two SjS patients were fractionated, separately, by HPLC as described in the Methods section. As shown in Fig. 5, the doubly charged SjS biomarker was detected in HPLC fraction 17. Inset A in Fig. 5 shows expanded mass spectra depicting the location of the doubly charged ion whereas the corresponding 7,606 Da peptide is depicted in inset B. Note that the singly charged ion (equivalent to protonated molecular ion) is now the largest peak in the spectrum.

Fig. 5.

Fig. 5

Open in a new tab

MALDI mass spectra of HPLC fraction 17. Inset A. shows expanded mass spectra depicting the location of the doubly charged ion whereas the corresponding 7,606 Da peptide is depicted in inset B.

The fractions (from both patients) containing the peptide associated with the doubly charged ion were then combined, concentrated and subjected to Edman sequencing. A total of 40 amino acids were then positively identified using Edman degradation sequencing (N-terminus to C-terminus) SPPA KPQG PPPQ GGNQ PQGP PPPP GKPQ GPPP QGGN KPQG. The partial amino acid sequence was searched against the non-redundant GenBank database and was determined to be a good match to the human basic salivary proline-rich protein 3 precursor.

Discussion

Saliva has been extensively used to aid in the diagnosis of several diseases including periodontal diseases, salivary gland disease and dysfunction, viral diseases, sarcoidosis, tuberculosis, lymphoma, gastric ulcers, liver dysfunction, and cancers (22–26). Several studies have reported alterations in SjS salivary electrolytes composition and/or protein concentration suggesting that saliva can be used for diagnosis of SjS (27–29). Recent proteomics studies reported that SjS saliva contain potential biomarkers (30–31). The discovery-based studies reported here confirm the potential usefulness of saliva as a diagnostic/prognostic tool for SjS.

Technologies with high throughput capabilities in genomics and proteomics led to an ability to conduct discovery-based research (32–33). Such technologies include quantitative real-time polymerase chain reaction (RT-qPCR), single-nucleotide polymorphism (SNP) analyses, gene expression arrays, and high-resolution mass spectrometry. Discovery-based research relies on the ability to analyze large quantities of data without a hypothesis, by searching for patterns (genes or proteins) that usefully discriminate between different groups of individuals (32–33).

Proteomics using high-resolution mass spectrometry holds great promise for discovery of biomarkers of human diseases from minimal amounts of biological fluids (34–35). This technology offers numerous advantages: ease of operation, high-throughput capability, requirement of minimal volume of biological fluids, high sensitivity, and high resolution for low molecular weight proteins and peptides. Surface enhanced laser desorption/ionization (SELDI) mass spectrometry is a variant of MALDI that uses commercially prepared chromatographic chip array surfaces to separate proteins in the sample (36–37). Although many different chip arrays (hydrophobic, anionic, cationic, or metal binding) may be used on one sample, the choice of chip array can require some prior knowledge of the changes that may take place between control and test or disease subjects, in order to select the correct chip array to yield proteins of interest (35). Furthermore, compared to MALDI-MS, SELDI-MS have relatively poor resolution.

Using SELDI-MS coupled to two-dimensional difference gel electrophoresis (2D-DIGE), Ryu et al. identified ten putative salivary biomarkers in the SjS group that differed from control individuals (30). Seven of the biomarkers were already known whereas three were novel (30). Similarly, using MALDI-MS, Hu et al. reported that the expression of 16 peptides was different in primary SjS and controls, with 10 of the peptides being overexpressed and 6 suppressed in SjS patients (31).

The issue of molecular characterization of mass spectrometry-based biomarker discovery is still hotly debated in the scientific community. There are two schools of thought: one that advocates for fully identified biomarker molecule (38) and the other favoring discriminating “ion patterns”(12). It is argued that without molecular characterization of the biomarkers identified by mass spectrometry, this methodology will remain empirical and the results will be difficult to validate, reproduce, and standardize (38). As a counterargument, it is possible that using high-resolution mass spectrometry (such as MALDI-TOF), “each ion is assigned a mass with such precision and accuracy that the accurate mass tag becomes an identifier” (12).

In this report, we used mass spectrometry and bioinformatic search algorithms to identify a 7,606 Da biomarker peptide unique to the salivary samples from the SjS subjects. The presence of this unique ion was made even more notable by the observation that the doubly charged ion of the same peptide was also present in the SjS subjects. Using additional analytical tools, we were able to determine a partial peptide sequence for the peptide. When the sequence was searched against the non-redundant GenBank public database, we determined a good match to the human basic salivary proline-rich protein 3 precursor (PRP3). PRPs are divided into acidic, basic, and glycosylated PRPs. Acidic PRPs bind calcium and inhibit hydroxyapatite crystal growth thus contributing to calcium homeostasis in the mouth and the acquired enamel pellicle of erupted teeth (39). PRPs are encoded by six genes with each gene having several alleles giving rise to ~20 secreted proteins showing complex patterns of polymorphism (40). PRPs are synthesized as precursor proteins and undergo proteolysis as well as posttranslational modifications giving rise to basic PRPs and glycosylated PRPs. The proteases responsible for the processing of the precursor proteins are called convertases and several studies have shown that furin (known to be expressed in salivary glands) is one of these convertases (41).

The first ten amino acid of the 7,606 Da biomarker peptide, SPPAKPQGPP, conform to a furin recognition sequence (XPPXXPXXP) which was shown to be present in precursor PRPs and to be cleaved by furin, in vitro and in vivo (41). It is therefore likely that this biomarker peptide is a proteolytic product from one of the salivary PRPs. Furthermore, the fact that this peptide was only present in saliva from SjS patients, suggests that the activity of the convertase(s) responsible for its generation is high in SjS patients’ salivary glands. It is likely that this biomarker is one of those currently detected in one or both models and awaiting identification.

The biomarkers identified in this study were present in samples from primary as well as secondary SjS patients. It is not clear if these biomarkers were present before disease manifestation or are a consequence of the salivary damage associated with the disease. We found no association between the secondary disease, age, smoking history and the presence of these biomarkers. Furthermore, sialogogues (Salagen or Evoxac), antihistamine, and antidepressants were the most commonly used medications in our study population and their use was not correlated with any of the reported biomarker.

In summary, using a discovery-based approach, we report that saliva contains putative biomarkers capable of distinguishing SjS patients from healthy control individuals with high sensitivity and specificity. The next step will be to validate these biomarkers in an independent and larger SjS patient cohort and control individuals. Furthermore, patients with xerostomia but no SjS and patients with another autoimmune disease but no xerostomia should also be included in the validation studies. As the putative biomarker was present in saliva of both primary and secondary SjS patients, another study should address the question whether the same biomarker is present in autoimmune disorders without lacrimal and salivary gland involvement.

Acknowledgments

The authors gratefully acknowledge and appreciate Robin Hodges for critical reading of the manuscript. This study was supported by National Institutes of Health/National Eye Institute grant RO1-EY12383.

References

  • 1.Amerongen AV, Veerman EC (2002) Saliva--the defender of the oral cavity. Oral Dis 8, 12–22. [DOI] [PubMed] [Google Scholar]
  • 2.Dodds MW, Johnson DA, Yeh CK (2005) Health benefits of saliva: a review. J Dent 33, 223–233. [DOI] [PubMed] [Google Scholar]
  • 3.Baum BJ (1993) Principles of saliva secretion. Ann N Y Acad Sci 694, 17–23. [DOI] [PubMed] [Google Scholar]
  • 4.Fox RI, Kang HI (1992) Pathogenesis of Sjögren’s syndrome. Rheum Dis Clin North Am 18, 517–538. [PubMed] [Google Scholar]
  • 5.Vitali C, Bombardieri S, Jonsson R, Moutsopoulos HM, Alexander EL, Carsons SE, Daniels TE, Fox PC, Fox RI, Kassan SS, Pillemer SR, Talal N, Weisman MH, European Study Group on Classification Criteria for Sjögren’s Syndrome (2002) Classification criteria for Sjögren’s syndrome: a revised version of the European criteria proposed by the American-European Consensus Group. Ann Rheum Dis 61, 554–558. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Daniels TE (2000) Evaluation, differential diagnosis, and treatment of xerostomia. J Rheumatol Suppl 61, 6–10. [PubMed] [Google Scholar]
  • 7.Coll J, Rives A, Griñó MC, Setoain J, Vivancos J, Balcells A (1987) Prevalence of Sjogren’s syndrome in autoimmune diseases. Ann Rheum Dis 46, 286–289. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Schaumberg DA, Sullivan DA, Buring JE, Dana MR (2003) Prevalence of dry eye syndrome among US women. Am J Ophthalmol 136, 318–326. [DOI] [PubMed] [Google Scholar]
  • 9.Daniels TE (1996) Sjogren’s syndrome: clinical spectrum and current diagnostic controversies. Adv Dent Res 10, 3–8. [DOI] [PubMed] [Google Scholar]
  • 10.Asmussen K (2001) Monitoring the disease activity. Scand J Rheumatol Suppl 23–26. [DOI] [PubMed] [Google Scholar]
  • 11.Al-Hashimi I, Wright JM, Cooley CA, Nunn ME (2001) Reproducibility of biopsy grade in Sjögren’s syndrome. J Oral Pathol Med 30, 408–412. [DOI] [PubMed] [Google Scholar]
  • 12.Petricoin III EF, Liotta LA (2003) The Vision for a New Diagnostic Paradigm. Clin Chem 49, 1276–1278. [DOI] [PubMed] [Google Scholar]
  • 13.Diamandis EP (2004) How are we going to discover new cancer biomarkers? A proteomic approach for bladder cancer. Clin Chem 50, 793–795. [DOI] [PubMed] [Google Scholar]
  • 14.Koomen JM, Shih LN, Coombes KR, Li D, Xiao L-c, Fidler IJ, Abbruzzese JL, Kobayashi R (2005) Plasma protein profiling for diagnosis of pancreatic cancer reveals the presence of host response proteins. Clin Cancer Res 11, 1110–1118. [PubMed] [Google Scholar]
  • 15.Petricoin III EF, Ardekani AM, Hitt BA, Levine PJ, Fusaro VA, Steinberg SM, Mills GB, Simone C, Fishman DA, Kohn EC, Liotta LA (2002) Use of proteomic patterns in serum to identify ovarian cancer. The Lancet 359, 572–577. [DOI] [PubMed] [Google Scholar]
  • 16.Lopez MF, Mikulskis A, Kuzdzal S, Bennett DA, Kelly J, Golenko E, Dicesare J, Denoyer E, Patton WF, Ediger R, Sapp L, Ziegert T, Lynch C, Kramer S, Whiteley GR, Wall MR, Mannion DP, Della Cioppa G, Rakitan JS, Wolfe GM (2005) High-Resolution Serum Proteomic Profiling of Alzheimer Disease Samples Reveals Disease-Specific, Carrier-Protein-Bound Mass Signatures. Clin Chem [DOI] [PubMed] [Google Scholar]
  • 17.Avasarala JR, Wall MR, Wolfe GM (2005) A distinctive molecular signature of multiple sclerosis derived from MALDI-TOF/MS and serum proteomic pattern analysis: detection of three biomarkers. J Mol Neurosci 25, 119–125. [DOI] [PubMed] [Google Scholar]
  • 18.Brouwers FM, Petricoin EF 3rd, Ksinantova L, Breza J, Rajapakse V, Ross S, Johann D, Mannelli M, Shulkin BL, Kvetnansky R, Eisenhofer G, Walther MM, Hitt BA, Conrads TP, Veenstra TD, Mannion DP, Wall MR, Wolfe GM, Fusaro VA, Liotta LA, Pacak K (2005) Low molecular weight proteomic information distinguishes metastatic from benign pheochromocytoma. Endocr Relat Cancer 12, 263–272. [DOI] [PubMed] [Google Scholar]
  • 19.Diamandis EP (2004) Mass spectrometry as a diagnostic and a cancer biomarker discovery tool: opportunities and potential limitations. Mol Cell Proteomics 3, 367–378. [DOI] [PubMed] [Google Scholar]
  • 20.Baggerly KA, Morris JS, Coombes KR (2004) Reproducibility of SELDI-TOF protein patterns in serum: comparing datasets from different experiments. Bioinformatics 20, 777–785. [DOI] [PubMed] [Google Scholar]
  • 21.Mitchell B, Yasui Y, Li C, Fitzpatrick A, Lampe P (2005) Impact of freeze-thaw cycles and storage time on plasma samples used in mass spectrometry based biomarker discovery projects. Cancer Inform 1, 98–104. [PMC free article] [PubMed] [Google Scholar]
  • 22.Kaufman E, Lamster IB (2002) The diagnostic applications of saliva--a review. Crit Rev Oral Biol Med 13, 197–212. [DOI] [PubMed] [Google Scholar]
  • 23.Balicki R, Grabowska SZ, Citko A (2005) Salivary epidermal growth factor in oral cavity cancer. Oral Oncol 41, 48–55. [DOI] [PubMed] [Google Scholar]
  • 24.Dziemianczyk D, Grabowska SZ, Balicki R (2005) Evaluation of secretory mucin concentration of patients with squamous cell carcinoma oral cavity. Rocz Akad Med Bialymst 50, 334–338. [PubMed] [Google Scholar]
  • 25.Sistig S, Lukac J, Vucicevic-Boras V, Delic D, Kusic Z (2003) Salivary immunoglobulin A and G subclasses in HIV positive patients. Eur J Med Res 8, 543–548. [PubMed] [Google Scholar]
  • 26.Tabak L, Mandel ID, Karlan D, Baurmash H (1978) Alterations in lactoferrin in salivary gland disease. J Dent Res 57, 43–47. [DOI] [PubMed] [Google Scholar]
  • 27.Almstahl A, Wikstrom M (2003) Electrolytes in stimulated whole saliva in individuals with hyposalivation of different origins. Arch Oral Biol 48, 337–344. [DOI] [PubMed] [Google Scholar]
  • 28.Kalk WW, Vissink A, Stegenga B, Bootsma H, Nieuw Amerongen AV, Kallenberg CG (2002) Sialometry and sialochemistry: a non-invasive approach for diagnosing Sjögren’s syndrome. Ann Rheum Dis 61, 137–144. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Mandel ID, Baurmash H (1976) Sialochemistry in Sjögren’s syndrome. Oral Surg Oral Med Oral Pathol 41, 182–187. [DOI] [PubMed] [Google Scholar]
  • 30.Ryu OH, Atkinson JC, Hoehn GT, Illei GG, Hart TC (2006) Identification of parotid salivary biomarkers in Sjögren’s syndrome by surface-enhanced laser desorption/ionization time-of-flight mass spectrometry and two-dimensional difference gel electrophoresis. Rheumatology (Oxford) 45, 1077–1086. [DOI] [PubMed] [Google Scholar]
  • 31.Hu S, Wang J, Meijer J, Ieong S, Xie Y, Yu T, Zhou H, Henry S, Vissink A, Pijpe J, Kallenberg C, Elashoff D, Loo JA, Wong DT (2007) Salivary proteomic and genomic biomarkers for primary Sjögren’s syndrome. Arthritis Rheum 56, 3588–3600. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Pusch W, Flocco MT, Leung SM, Thiele H, Kostrzewa M (2003) Mass spectrometry-based clinical proteomics. Pharmacogenomics 4, 463–476. [DOI] [PubMed] [Google Scholar]
  • 33.Ransohoff DF (2003) Discovery-based research and fishing. Gastroenterology 125, 290. [DOI] [PubMed] [Google Scholar]
  • 34.Aebersold R, Mann M (2003) Mass spectrometry-based proteomics. Nature 422, 198–207. [DOI] [PubMed] [Google Scholar]
  • 35.Aldred S, Grant MM, Griffiths HR (2004) The use of proteomics for the assessment of clinical samples in research. Clin Biochem 37, 943–952. [DOI] [PubMed] [Google Scholar]
  • 36.Engwegen JY, Gast MC, Schellens JH, Beijnen JH (2006) Clinical proteomics: searching for better tumour markers with SELDI-TOF mass spectrometry. Trends Pharmacol Sci 27, 251–259. [DOI] [PubMed] [Google Scholar]
  • 37.Merchant M, Weinberger SR (2000) Recent advancements in surface-enhanced laser desorption/ionization-time of flight-mass spectrometry. Electrophoresis 21, 1164–1177. [DOI] [PubMed] [Google Scholar]
  • 38.Diamandis EP (2003) Point: proteomic patterns in biological fluids: do they represent the future of cancer diagnostics? Clin Chem 49, 1272–1275. [DOI] [PubMed] [Google Scholar]
  • 39.Lendenmann U, Grogan J, Oppenheim FG (2000) Saliva and dental pellicle--a review. Adv Dent Res 14, 22–28. [DOI] [PubMed] [Google Scholar]
  • 40.Minaguchi K, Bennick A (1989) Genetics of human salivary proteins. J Dent Res 68, 2–15. [DOI] [PubMed] [Google Scholar]
  • 41.Cai K, Bennick A (2004) Processing of acidic proline-rich proprotein by human salivary gland convertase. Arch Oral Biol 49, 871–879. [DOI] [PubMed] [Google Scholar]

RESOURCES