Multiple Sclerosis Resource Centre

Welcome to the Multiple Sclerosis Resource Centre. This website is intended for international healthcare professionals with an interest in Multiple Sclerosis. By clicking the link below you are declaring and confirming that you are a healthcare professional

You are here

Serum proteomics in multiple sclerosis disease progression

Journal of Proteomics, Volume 118, 6 April 2015, Pages 2-11


Multiple sclerosis (MS) is associated with chronic degeneration of the central nervous system and may cause permanent neurological problems and considerable disability. While its causes remain unclear, its extensive phenotypic variability makes its prognosis and treatment difficult. The identification of serum proteomic biomarkers of MS progression could further our understanding of the molecular mechanisms related to MS disease processes. In the current study, we used isobaric tagging for relative and absolute protein quantification (iTRAQ) methodology and advanced multivariate statistical analysis to quantify and identify potential serum biomarker proteins of MS progression. We identified a panel of 11 proteins and combined them into a classifier that best classified samples into the two disease groups. The estimated area under the receiver operating curve of this classifier was 0.88 (p-value = 0.017), with 86% sensitivity and specificity. The identified proteins encompassed processes related to inflammation, opsonization, and complement activation. Results from this study are in particular valuable to design a targeted Multiple Reaction Monitoring mass spectrometry based (MRM-MS) assay to conduct an external validation in an independent and larger cohort of patients. Validated biomarkers may result in the development of a minimally-invasive tool to monitor MS progression and complement current clinical practices.

Biological significance

A hallmark of multiple sclerosis is the unpredictable disease course (progression). There are currently no clinically useful biomarkers of MS disease progression; most work has focused on the analysis of CSF, which requires an invasive procedure. Here, we explore the potential of proteomics to identify panels ofserumbiomarkers of disease progression in MS. By comparing the protein signatures of two challenging to obtain, but well-defined, MS phenotypic groups at the extremes of progression (benign and aggressive cases of MS), we identified proteins that encompass processes related to inflammation, opsonization, and complement activation. Findings require validation, but are an important step on the pathway to clinically useful biomarker discovery.

This article is part of a Special Issue entitled: Protein dynamics in health and disease. Guest Editors: Pierre Thibault and Anne-Claude Gingras.

Graphical abstract





  • Proteomic biomarkers may improve our understanding of multiple sclerosis progression.
  • We analyzed serum protein abundances of two well-phenotyped groups of MS patients.
  • Proteins were quantified using iTRAQ-MALDI-TOF/TOF.
  • We identified 11 potential markers and build a classifier with an AUC of 0.88.
  • The identified proteins encompass interesting processes and pathways.

Abbreviations: AUC - area under the receiver operating curve, CSF - cerebrospinal fluid, EN - Elastic Net, FDR - false discovery rate, iTRAQ - isobaric tagging for relative and absolute protein quantification, LIMMA - linear models for microarray analysis, LOOCV - leave-one-out cross-validation, MALDI - matrix-assisted laser desorption ionization, MRM-MS - Multiple Reaction Monitoring mass spectrometry, MS - multiple sclerosis, TOF - time-of-flight.

Keywords: Serum proteomics, iTRAQ mass spectrometry, Multiple sclerosis, MS progression, Biomarkers, Classifier.

1. Introduction

Multiple sclerosis (MS) results in chronic degeneration of the central nervous system, causing considerable disability, and has no known cause or cure [1] . The extensive phenotypic variability and unpredictability of the course of disease in any given individual are considered hallmarks of MS, making therapeutic decisions, the prognosis, and even ‘life-planning’ difficult. At the molecular level, despite extensive work, it still remains unclear as to the exact mechanisms implicated with the progression of MS[2] and [3]. The application of large-scale quantitative proteomics technologies in MS research has the potential to further our understanding of this disease[3] and [4]and identify potentially useful biomarkers of disease progression. Ultimately, validated serum proteomic biomarkers of MS progression may complement and improve current patient care, for instance by helping identify patients in most need of aggressive interventions.

Proteomic analyses of biological samples of MS patients and experimental models have started to emerge in the last decade, although most have focused on the study of a limited number of pre-selected biomarkers [3] . With the ongoing growth of extremely sensitive high throughput proteomics technologies, an alternative approach of emerging interest is the application of mass spectrometric untargeted technologies to conduct a broad, discovery driven, and unbiased identification of disease biomarkers[3], [4], [5], and [6]. To date, most proteomics studies have focused on the analysis of cerebrospinal fluid (CSF)[5] and [6]. While the CSF might be considered more tissue specific, sampling requires an uncomfortable, invasive lumbar puncture; with risks ranging from headache (common) to nerve damage and paraplegia (rare) [7] . Consequently, regular CSF sampling is unacceptable for most patients and their physicians. Although the serum represents an attractive alternative to CSF, both for relative convenience and the potential for biomarker discovery, it reflects the collective expression of all tissue and cell types[8] and [9], lacking tissue specificity. Thus, to be successful, a serum study needs to be based on groups of well-phenotyped MS patients. To date relatively few studies in MS have been conducted using serum[4] and [10], especially in the context of disability progression[4], [5], [6], and [11].

We set out to apply mass spectrometric untargeted technology to identify potential serum biomarkers (or ‘signatures’) of disease progression in MS by comparing two well-phenotyped clinical extremes of disease progression — ‘benign’ and ‘aggressive’ MS in this proof-of-principle study.

2. Materials and methods

2.1. Study population and design

This was a retrospective analysis of prospectively collected data and biological samples. Patients were selected according to the following criteria: i) diagnosed with definite relapsing-onset MS by an MS neurologist (Poser [12] or McDonald [13] criteria); ii) fulfilled criteria for either benign or aggressive MS, defined as: minimal disability (Expanded Disability Status Scale Score, EDSS ≤ 3) despite 20 + years of disease duration (benign MS) or severe disability (EDSS ≥ 6) within 10 years of MS onset (aggressive MS); iii) availability of a banked serum sample, drawn prior to exposure to any ‘disease modifying’ drugs for MS.

Demographic and clinical characteristics were derived from the British Columbia Multiple Sclerosis (BCMS) database[14], [15], [16], [17], and [18], a large population-based longitudinal MS database established in 1980, linking the four original MS clinics in BC (Vancouver, Victoria, Kelowna and Prince George to end of 2004). Biological samples were provided by some patients; typically those visiting the Vancouver site, these were stored within the Neuroimmunology Biobank at the University of British Columbia. By cross-referencing these two sources, we identified patients fulfilling study criteria.

In accordance with the institutional ethical approval, once the biological samples and clinical information were linked, all patient identifiers were removed, creating a fully anonymized cohort which could not be re-identified. Only the following patient characteristics were retained, with all time-related factors recorded in full years (no dates were retained): sex; age at symptoms onset (years); age at sample draw; disease duration at sample draw; sample storage time; clinical course (group allocation ‘A’ or ‘B’); sample drawn before clinical course was confirmed (yes vs no). The clinical course (benign vs aggressive) determined the group allocation and was concealed from those involved with the sample analyses (proteomics) or statistical analyses, who were informed groups A and B represented two different ‘types’ of MS patients (no specifics were given other than the basic demographics needed to balance the groups during the proteomic analysis, see below).

2.2. iTRAQ study design

Protein relative quantitations were obtained using iTRAQ-MALDI-TOF/TOF methodology, which allows simultaneous processing of eight samples per experimental run. Given the number of samples in this study, two independent iTRAQ runs were used to process all samples. To construct protein ratios that are comparable across both experimental runs, a reference sample was processed together with 7 patient samples in each iTRAQ run. The reference sample consisted of a pool of serum from the 14 individuals in this study and thus, ensures the identification of most proteins in the analyzed samples. Samples were run in 2 separate iTRAQ batches, with each batch having a balanced spread of patients, based on sex, sample storage-time, anonymized grouping (‘A’ vs ‘B’), age and disease duration (time from symptom onset) at sample collection. Patient and reference samples within each run were randomly labeled with the eight iTRAQ reagents.

2.3. iTRAQ data acquisition

Blood samples were processed as previously described[19] and [20]. Briefly, peripheral blood samples were drawn into vacuum tubes without any anticoagulation agents and allowed to clot for a minimum of 30 min at room temperature prior to processing. Serum was separated and stored at − 80 °C until selected for analysis. Serum samples were depleted of the 14 most abundant plasma proteins (albumin, fibrinogen, transferin, IgG, IgA, IgM,haptoglobin, α2-macroglobulin, α1-acid glycoprotein, α1-antitrypsin, apolipoprotein-AI, apolipoprotein-AII, complement C3 and apolipoprotein B) by immuno-affinity chromatography (Genway Biotech; San Diego, CA); digested with trypsin and labeled with iTRAQ reagents according to the manufacturer's protocol (Applied Biosystems; Foster City, CA). Labeled peptides were pooled, acidified to pH 2.5–3.0 with 6 M phosphoric acid (ACP Chemicals Inc; Montreal, QC, Canada), and separated by 2D liquid chromatography.

iTRAQ labeled peptides were fractionated by strong cation exchange chromatography (SCX) using a 100 mm × 4.6 mm ID polysulphoethyl A column packed with 5 μm beads (300 Å poresize). A 120 min linear gradient was used to separate the peptides in the first dimension. The 20 to 30 fractions containing the highest concentration of peptides (based on their UV absorbance at 215 nm) were selected and their volumes were reduced to 150 μL in preparation for the subsequent nanoscale reversed-phase chromatography separation. Peptides were desalted on the LC system by loading each fraction onto a C18 PepMap guard column (300 μm ID × 5 mm, 5 μm particle size, 100 Å pore size, LC Packings, Amsterdam) and washing for 15 min at 50 μL/min with a mobile phase A′, consisting of water:acetonitrile:TFA (98:2:0.1 (v/v)). The trapping column was then switched into the flow stream which operated at 200 nL/min, and the peptide mixture was loaded onto a Magic C18 nano LC analytical column (15 cm, 5 μm particle size, 100 Å pore size; Michrom Bioresources Inc., Auburn CA, USA). Peptides were eluted with a 3-step linear gradient: 0–45 min with 5% to 15% B′ (acetonitrile:water:TFA 98:2:0.1, v/v); 45–100 min with 15% to 40% B′, and 100–105 min with 40% to 75% B′. The eluent was spotted directly onto AB Sciex's 384-spot targets for 30 s using a Probot microfraction collector (LC Packings, Amsterdam, Netherlands). The matrix solution (3 mg/mL α-cyano-4-hydroxycinnamic acid (Sigma-Aldrich, St Louis, MO USA) in 50% ACN, 0.1% TFA), was then added at 0.75 μL per spot. The samples were analyzed in by a 4800 MALDI TOF/TOF mass spectrometer (Applied Biosystems), acquisition time ranging from 35 to 40 h.

ProteinPilot™ software v4.0 with the integrated Paragon™ Search Algorithm and Pro Group™ Algorithm  [21] (Applied Biosystems) was used to process the resulting data and to search against the UniProt database [22] to identify peptides and proteins. The precursor mass tolerance was set to 150 ppm and the iTRAQ fragment tolerance was set to 0.2 Da. Identification parameters were set for trypsin cleavages and cysteine alkylation by methyl methanethiosulfonate. Modifications, substitutions, and number of missed cleavages allowed are not limited to a fixed value by Paragon algorithm  [21] . The detected protein threshold was set toUnused ProtScore > 0.70 (equivalent to 80.0% confidence interval). ProteinPilot™ was used to calculate protein ratios by the weighted geometric means of the unique peptides contributing to the identification of the protein, after performing a bias correction and using background correction factor that reduces the ratio compression.

2.4. Statistical analysis

The set of proteins detected in both iTRAQ runs were considered for further analysis. Although the association between sex and disease progression (benign vs aggressive) in our study was not significant (Fisher's exact test p-value is 0.19), our data was not sex-balanced with only 3 male samples in the aggressive MS group. We therefore have used a moderatedt-test to prefilter any potential sex-associated proteins in this modest-sized cohort (robust LIMMA, with a conservative p-value < 0.1) [23] . Further, proteins with a total standard deviation below 0.15 are mainly reflecting the typical technical variation that characterizes iTRAQ experiments previously reported in the literature[6] and [24]. Since these proteins lack relevant biological variation, they were not considered in subsequent statistical analyses. To be more conservative and robust to outlying observations, instead of computing the standard deviation, we computed the median absolute deviation (MAD), i.e., median of the absolute differences of each protein ratio to the overall median.

All proteins passing these prefiltering steps were considered as candidate markers to build a classifier using a regularized logistic regression model, called Elastic Net [25] . Unlike the classicalt-tests, Elastic Net evaluates thejointcontribution of all proteins and selects a panel of proteins associated with multiple sclerosis progression. Moreover, Elastic Net builds a classifier by assigning a weight to each of the selected proteins, which can be used to classify samples into benign or aggressive multiple sclerosis. A principal component analysis (PCA) is used to illustrate thejointcontribution of the selected proteins in differentiating the two disease groups.

Classification performance measures, including sensitivity, specificity and area under the receiver operating curve (AUC) were estimated by a leave-one-out cross-validation (LOOCV). Each sample was left out (test set) once for classification and the remaining samples (training set) were used to select a panel and build a classifier using Elastic Net. The process was repeated until all samples were left out. Then, the test error was estimated from the average all samples' classification errors. Although the performance of the classifier was not evaluated in an independent cohort, the test sample of the LOOCV was not considered to select proteins and estimate their contribution to the classifier score, ensuring reliable performance estimates [26] . Since the prefiltering steps were not directly related to the proteomic distinction of MS phenotypes and considering the small number of male samples in the study, we opted to leave these steps out of the cross-validation.

The biological processes enriched in the biomarker proteins were evaluated using MetaCore™ (Thomson Reuters), a data-mining and pathway analysis software. The significance of each identified process was evaluated by its corresponding enrichment p-value. The resulting 10 most significant gene ontology cellular processes, which were identified based on the proteins that differentiated the extremes of disease progression – i.e., ‘benign’ vs ‘aggressive’ MS – were then investigated.

All of the statistical analyses were implemented using R version 3.1.0. The AUC was computed using the ROCR package [27] . The ggplot2 package was used to illustrate the relative protein ratios of the selected panel in Fig. 2 [28] . The study was approved by the Clinical Research Ethics Board at the University of British Columbia (#H11-02239).

3. Results

A total of 14 serum samples (n = 7 from each MS phenotype) fulfilled the selection criteria described in the previous section and were used in this proof-of-principle study. Characteristics of the two groups are shown in Table 1 .

Table 1 Characteristics of the benign and aggressive multiple sclerosis patients.

  Aggressive (n = 7) Benign (n = 7)
Number of females 4 7
Age at symptom onset, years 38 29
Age at sample draw, years (SD) 42 (4.62) 45 (7.64)
Disease duration at sample draw, years (SD) 4 (1.80) 16 (10.20)
Sample storage time, years (SD) 17 (6.47) 21 (3.78)
Number of samples drawn before the clinical course (benign or aggressive) was confirmed 0 5

A total of 108 proteins were identified and quantified in both iTRAQ runs. A total of 10 proteins that showed a negligible overall biological variation among patient samples were prefiltered out. In addition, 15 proteins with significant differential protein concentrations between male and female samples were also prefiltered out (robust LIMMA, p-val < 0.1). Fig. 1 summarizes the quality of the identification of these sets of proteins and indicates the number of peptides that supported the proteins' identifications in each iTRAQ run. Results show that 75% of the 83 proteins remaining after prefiltering were identified based on 5 or more distinct peptides in one iTRAQ experiment, and 61% in the other. In addition, only 8% of the 83 proteins were identified based on less than 2 peptides in one experiment, and 11% in the other one.


Fig. 1 Proportion of proteins identified using different peptide counts. Proportion of proteins identified based on different peptide counts within the analyzed set of proteins (Analyzed, in blue) and the Elastic Net selected protein panel (Panel, in red) in both iTRAQ experiments (top and bottom, respectively).

Through the Elastic Net model, 11 out of 83 candidate proteins were selected to form a panel, allowing an integrated classifier score to be built ( Table 2 ). With the exception of Platelet glycoprotein V (GP5, UniProt: P40197 , identified with a unique peptide in one experiment), all other proteins in the Elastic Net panel were identified based on of 2 or more unique peptide sequences in both iTRAQ experiments, demonstrating confidence in the identification of the selected proteins ( Fig. 1 and Table 3 ). Further details on protein identification, quantitation and variation are given in Table 3 . Some of the proteins were identified and quantified with a low coverage (e.g., GP5), illustrating the need of further technical validation with different instrumentation.

Table 2 Proteomic panel. Accession numbers, protein names and gene symbols from UniProtKB/Swiss-Prot. Mean and standard deviation, in parenthesis, of raw relative proteins ratios within each group are given in the last two columns.

Accession (UniProt ID) Protein name Gene symbol Mean (SD) aggressive Mean (SD) benign
P07996 Thrombospondin-1 THBS1 1.17 (0.19) 0.89 (0.14)
P02775 Platelet basic protein PPBP 1.09 (0.19) 0.85 (0.14)
P17936 Insulin-like growth factor-binding protein 3 IGFBP3 1.12 (0.14) 0.93 (0.21)
P06276 Cholinesterase BCHE 1.10 (0.19) 0.90 (0.15)
P02750 Leucine-rich alpha-2-glycoprotein LRG1 0.84 (0.11) 1.06 (0.25)
P08185 Corticosteroid-binding globulin CBG 0.78 (0.15) 1.20 (0.67)
Q14624 Iter-alpha-trypsin inhibitor heavy chain H4 ITIH4 0.92 (0.17) 1.22 (0.45)
P06702 Protein S100-A9 S100A9 1.34 (1.60) 0.62 (0.80)
P40197 Platelet glycoprotein V GP5 1.35 (0.25) 1.04 (0.28)
Q15485 Ficolin-2 FCN2 0.95 (0.30) 0.69 (0.17)
P18428 Lipopolysaccharide-binding protein LBP 1.11 (0.51) 1.26 (0.28)

Table 3 Quality control parameters of the selected panel. Unused, Coverage and Error Factor are three quality parameters calculated by ProteinPilot within each iTRAQ run. Averages of these measures across runs are shown. ‘Unused’ is a measure of detection confidence with values of 0.70 equivalent to an 80.0% confidence interval. ‘Coverage’ represents the percentage of the protein sequence covered by the list of unique distinct peptides. ‘Error Factor’ is a measure of confidence in protein quantitation related with the between-peptide variation. ‘Peptide Count’ shows the number of unique peptides, excluding miscleavages, used for protein identification and quantitation within each iTRAQ run. ‘Length’ and ‘PI/molecular mass’ contain the number of amino acids in each sequence and the isoelectric point/molecular mass (kDa) for each protein, respectively.

Accession Unused Coverage Error factor Peptide count iTRAQ1/iTRAQ2 Length PI/molecular mass
P07996 18.95 19.44 0.96 17/9 1170 4.71/129,383
P02775 7.10 47.65 0.62 7/7 128 9.04/13,894
P06276 3.69 10.55 1.00 2/4 602 7.12/68,418
P17936 4.82 22.34 0.88 4/3 291 9.03/31,674
P02750 15.37 37.32 0.62 13/9 347 6.45/38,178
P08185 11.13 26.66 0.65 9/4 405 5.64/45,141
Q14624 65.04 51.02 0.64 44/31 930 6.51/103,357
P06702 5.90 46.05 0.81 2/4 114 5.71/13,242
P40197 0.48 8.84 1.55 2/1 560 9.73/60,959
Q15485 1.21 12.46 0.82 2/3 313 6.31/34,001
P18428 6.56 20.38 0.97 7/2 481 6.23/53,384

Fig. 2 shows the relative protein rations of all samples for the selected panel of proteins, with 7 of the 11 proteins more abundant, on average, in the aggressive compared to the benign MS groups (platelet glycoprotein V (GP5, UniProt: P40197 ), ficolin-2 (FCN2, UniProt: Q15485 ), thrombospondin-1 (THBS1, UniProt: P07996 ), protein S100-A9 (S100A9, UniProt: P06702 ), platelet basic protein (PPBP, UniProt: P02775 ), insulin-like growth factor-binding protein 3 (IGFBP3, UniProt: P17936 ), cholinesterase (BCHE, UniProt: P06276 )), and 4 less abundant, on average, in the aggressive MS group (lipopolysaccharide-binding protein (LBP, UniProt: P18428 ), leucine-rich alpha-2-glycoprotein (LRG1, UniProt: P02750 ), corticosteroid-binding globulin (CBG, UniProt: P08185 ), inter-alpha-trypsin inhibitor heavy chain H4 (ITIH4, UniProt: Q14624 )) in patients with ‘aggressive’ MS, compared to those with ‘benign’ MS.


Fig. 2 Scatter plots of panel proteins. Relative protein concentrations (in log2-scale) of the proteins in the Elastic Net panel. The horizontal lines represent the median protein concentration within each group. Group names ‘A’ and ‘B’ correspond to aggressive and benign MS, respectively.

Since the Elastic Net model considered the joint contribution of the selected proteins, the 11 identified markerstogetherdifferentiated the two MS groups ( Fig. 3 ), even if this is not so evident from each single protein marker (with some univariate fold-changes being relatively small). In addition, Elastic Net integrated the 11 panel proteins into a classifier that achieved a satisfactory classification performance, estimated by LOOCV ( Fig. 4 ). According to this estimated performance, the Elastic Net classifier can achieve an estimated AUC of 0.88 (p-value of Mann–Whitney U-statistic equal to 0.017, [29] ), with 86% sensitivity and specificity.


Fig. 3 Principal component analysis. The joint contribution of the 11 proteins in the panel to differentiate the aggressive (red points) from benign (blue points) MS is illustrated by the first two principal components.


Fig. 4 Receiver operating curve (ROC). Leave-one-out cross-validation is used to evaluate the proposed methodology. Within each cross validation fold, a training set is used to select and build a classifier score to test each sample left out. The ROC summarizes the results of the classification on the test samples. The red dot represents a threshold with 86% sensitivity and 86% sensitivity.

Based on the results produced by MetaCore, there were 100 significantly enriched gene ontology cellular processes associated with the 11 proteins in the identified panel (FDR < 0.05). Among the top 10 of these processes, opsonization (associated with phagocytosis) was significantly enriched ( Fig. 5 ). This was due to the contribution of ficolin-2 (FCN2, UniProt: Q15485 ) and lipopolysaccharide-binding protein (LBP, UniProt: P18428 ) to the panel (the former being more abundant, and the latter less abundant, on average, in aggressive MS, as compared to benign). In addition, mostly due to ficolin-2, complement activation was enriched (the second and fourth most significantly enriched processes). Since thrombospondin-1 (THBS1, UniProt: P07996 ) and protein S100-A9 [also known as S100 calcium binding protein A9, S100A9, UniProt: P06702 ] (on average, more abundant in aggressive MS) and lipopolysaccharide-binding protein (on average, less abundant in aggressive MS) were in the panel, chronic inflammatory response, tumor necrosis factor biosynthesis, and macrophage activation were significantly enriched as well. Insulin-like growth factor-binding protein 3 (IGFBP3, UniProt: P17936 ) (on average, more abundant in aggressive MS) was also related to many of the top 10 processes, such as smooth muscle cell migration, insulin-like growth factor receptor signaling, and the glucose metabolic process.


Fig. 5 Top 10 most enriched cellular processes. The cellular processes enriched by the biomarker panel proteins were investigated using the MetaCoreTMsoftware. The 10 most enriched cellular processes (y-axis) and the associated false discovery rate (FDR) (− log10(FDR) in the x-axis) are shown here.

4. Discussion

We explored and compared the protein signatures of two well-defined phenotypic groups of MS patients at the extremes of disease progression. Using an advanced mass spectrometry-based untargeted (unbiased) multiplexing methodology (iTRAQ-MALDI-TOF/TOF) and solid statistical techniques, we identified and combined a panel of 11 candidate protein markers into a classifier score to classify patient samples as ‘benign’ or ‘aggressive’ MS progression phenotypes. The identified proteomic biomarker panel could be of interest in furthering our understanding of the disease processes in MS. Moreover, results from this study can be used to design a targeted MRM-MS assay to conduct an external validation in a larger and independent cohort of patients [30] .

Within the panel of candidate proteins, seven were significantly enriched in the serum of those with aggressive multiple sclerosis and four were enriched in those with benign multiple sclerosis. There are relatively few proteomics studies exploring the associations between protein levels and disease progression in MS. As a result, we were unable to find a study that used the ‘extremes of disease’ approach with which to compare findings. Nonetheless, some of our observations concur with the broader literature. For example, we observed that leucine-rich alpha-2-glycoprotein (LRG1, UniProt: P02750 ) is, on average, more abundant in individuals with benign vs aggressive MS. Others have reported that this protein was upregulated in the serum of 20 MS patients vs 20 controls [31] (without MS), however no clinical or phenotypic details were provided.

Opsonization (i.e., the process by which a pathogen is ‘labeled’ for destruction by a phagocyte) and complement activation pathways were significantly enriched in the MetaCore analysis, both of which have been associated with demyelinating diseases such as MS [32] . For instance, myelin phagocytosis is considered a pathological hallmark of MS [33] and has been reported as more ‘efficient’ in the MS brains vs healthy donor brains [33] . Ficolin-2 (FCN2, UniProt: Q15485 ) protein, more abundant, on average, in our aggressive MS patients (vs benign) contributed to the identification of the opsonization pathway. This protein has also been associated with innate immunity [34] which has increasingly been recognized as important in MS [35] . Another interesting observation was the identification of thrombospondin-1 (THBS1, UniProt: P07996 ) and platelet glycoprotein V (GP5, UniProt: P40197 ) — both of which are involved in blood coagulation. Upregulation of the coagulation cascade has been reported in MS-related studies, with most focused on neuropathology or the animal model of MS (Experimental Allergic Encephalomyelitis, EAE)[36] and [37]. For example, fibrin, the final product of the coagulation cascade is abundantly deposited in MS plaques[38], [39], [40], and [41]and plays a critical role in the induction of neuroinflammation and axonal damage[36], [42], and [43]. A growing number of studies describe the protective effects of fibrin inhibition in EAE[36], [37], and [44]. Together, these data suggest that the anticoagulants may also have a protective effect in MS by inhibiting the pathogenic effects of fibrin, possibly mediated through activation of central nervous system innate immunity and induction[36] and [44]. The wider field of biomarker discovery in MS in relation to disease progression is broad and growing[45], [46], and [47]. For example, the role of autoantibodies has been explored with the presence of specific IgM autoantibodies present in sera of MS patients being associated with changes in monthly MRI scans measured over 6 months in a cohort of 40 untreated relapsing-remitting MS patients [47] . Other specific biomarkers understudy in relation to disease course or drug response have included: cell surface markers (e.g. myelin oligodendrocyte glycoprotein (MOG)-reactive CD45RO + T lymphocytes), alpha B-crystallin, neurofilament peptides, tumor necrosis factor and other cytokines measurable in peripheral blood or serum [46] . Overall, the general lack of studies exploring proteomics in the context of disease progression highlights the unique contribution of our study.

This study examines serum samples from a small but very unique cohort of patients — being at the ‘extremes’ of the disease course and free from disease-modifying drugs, which particularly for those with aggressive MS, is rare in today's therapeutic era. Obtaining samples from patients while still drug naive is important in order to differentiate effects directly related to actual disease progression, avoiding any potential bias or confounding due to drug exposure. This is also relevant given that those with a more aggressive disease course are more likely to be exposed to drug treatment (compared to those with a benign disease course [17] ). This type of indication bias is challenging and can confound the relationship between any potential biomarker and the actual underlying disease progression vs. effects of treatment [17] . However, this comes with limitations, including a small sample size and lack of access to an independent validation cohort. We purposely selected a cohort of patients with relapsing-onset MS only, excluding those with primary-progressive MS, to minimize any variation related to disease course. Others have shown that the proteomic signature is affected by disease course — in particular, large differences between primary progressive (PP) MS and relapsing-onset MS have been reported [48] , with PPMS surprisingly showing greater similarity to Amyotrophic lateral sclerosis (ALS) than relapsing-onset MS.

By selecting the ‘extremes’ of disease progression, our two patients groups differed (unsurprisingly) by age, sex and disease duration. For instance, by definition of disease progression, the disease durations at the time that the samples were collected differ between groups. For instance, the ‘benign’ group had a long follow-up – at least 20 years – in order to meet the definition, and the ‘aggressive group’ usually exhibit ‘severe disability’ early in their disease course. In addition, all those with benign MS in our study were women, reflecting the epidemiology of both MS and benign MS in general[49] and [50]. In particular, the potential confounding of sex was tested in our statistical analysis and potential sex-associate proteins were prefiltered. However, other possible unavailable confounders, such as presence of co-morbid conditions, may also exist [51] . That being said, insights into why these two groups of patients display such phenotypically different rates of progression are of major importance. It remains possible that those in the ‘benign’ group could progress overtime further in their disease or that the ‘aggressive’ individuals could reach even higher levels of disability. However, it is not the aim of this study to examine the temporal patterns of a proteomic signature and clinical variables for these two groups of patients.

5. Conclusion

Our goal was to conduct an unbiased study to identify a panel of serum proteins associated with MS disease progression. Using advanced and rigorous technological and statistical methods, we identified and built a score based on a panel of 11 proteins to classify samples from MS patients into two extreme disease groups. Longitudinal studies are required to understand if these proteins are a consequence of disease progression or whether they are potential disease predictors. The outcomes of this study can improve our understanding of the underlying mechanisms of MS progression. In addition, results are in particular valuable to design a targeted MRM-MS proteomics assay to conduct an external validation. If further validated, a proteomic panel of MS progression could result in a much needed, minimally-invasive clinical tool to complement current therapeutic guidelines.

Conflict of interest

None of the authors have any direct conflicts of interest in the research reported in this article.

H.T. has received: research support from the National Multiple Sclerosis Society, Canadian Institutes of Health Research, and UK MS Trust; speaker honoraria and/or travel expenses to attend conferences from the Consortium of MS Centres (2013), the MS Society of Canada, endMS Summer School (2012, 2014), the National MS Society (2012, 2014), Bayer Pharmaceutical (speaker, 2010, honoraria declined), Teva Pharmaceuticals (speaker 2011), ECTRIMS (2011, 2012, 2013), UK MS Trust (2011), the Chesapeake Health Education Program, US Veterans Affairs (2012, honorarium declined), Novartis Canada (2012), Biogen Idec (2014, honorarium declined), American Academy of Neurologists (annual meeting speaker, 2013, 2014, honorarium declined). Unless otherwise stated, all speaker honoraria are either donated to an MS charity or to an unrestricted grant for use by her research group.


We thank the University of Victoria Genome BC Proteomics Centre, which is supported by Genome Canada and Genome BC through Science and Technology Innovation Centre funding, for conducting the iTRAQ experiments. C.H.B. holds the Don and Eleanor Rix BC Leadership Chair in Biomedical and Environmental Proteomics.

We gratefully acknowledge the BC MS Clinic neurologists who contributed to the study via the BCMS database through patient examination and data collection (current members listed here):

UBC MS Clinic: A. Traboulsee, MD, FRCPC (UBC Hospital MS Clinic Director and Head of the UBC MS Programs); A-L. Sayao, MD, FRCPC; V. Devonshire, MD, FRCPC; S. Hashimoto, MD, FRCPC (UBC and Victoria MS Clinics); J. Hooge, MD, FRCPC (UBC and Prince George MS Clinic); L. Kastrukoff, MD, FRCPC (UBC and Prince George MS Clinic); J. Oger, MD, FRCPC

Kelowna MS Clinic: D. Adams, MD, FRCPC; D. Craig, MD, FRCPC; S. Meckling, MD, FRCPC

Prince George MS Clinic: L. Daly, MD, FRCPC

Victoria MS Clinic: O. Hrebicek, MD, FRCPC; D. Parton, MD, FRCPC; K Atwell-Pope, MD, FRCPC. We also thank Anna-Marie Bueno for her help in coordinating the data collection.

The views expressed in this paper do not necessarily reflect the views of each individual acknowledged.

H.T. is funded by the Multiple Sclerosis Society of Canada (Don Paty Career Development Award); is a Michael Smith Foundation for Health Research Scholar and a Canada Research Chair for Neuroepidemiology and Multiple Sclerosis. G.C.F is a Canada Research Chair with Canada Foundation for Innovation in Statistical Genomics.


  • [1] A. Compston, A. Coles. Multiple sclerosis. Lancet. 2008;372:1502-1517 Crossref
  • [2] M.A. Friese, B. Schattling, L. Fugger. Mechanisms of neurodegeneration and axonal dysfunction in multiple sclerosis. Nat Rev Neurol. 2014;10:225-238 Crossref
  • [3] A.S. Farias, F. Pradella, A. Schmitt, L.M. Santos, D. Martins-de-Souza. Ten years of proteomics in multiple sclerosis. Proteomics. 2014;14:467-480 Crossref
  • [4] C.E. Teunissen, M.J. Koel-Simmelink, T.V. Pham, J.C. Knol, M. Khalil, A. Trentini, et al. Identification of biomarkers for diagnosis and progression of MS by MALDI-TOF mass spectrometry. Mult Scler. 2011;17:838-850 Crossref
  • [5] A.C. Kroksveen, J.A. Opsahl, T.T. Aye, R.J. Ulvik, F.S. Berven. Proteomics of human cerebrospinal fluid: discovery and verification of biomarker candidates in neurodegenerative diseases using quantitative proteomics. J Proteomics. 2011;74:371-388 Crossref
  • [6] V. Singh, R.Q. Hintzen, T.M. Luider, M.P. Stoop. Proteomics technologies for biomarker discovery in multiple sclerosis. J Neuroimmunol. 2012;248:40-47 Crossref
  • [7] R.W. Evans. Complications of lumbar puncture. Neurol Clin. 1998;16:83-105 Crossref
  • [8] M.H. Elliott, D.S. Smith, C.E. Parker, C. Borchers. Current trends in quantitative proteomics. J Mass Spectrom. 2009;44:1637-1660
  • [9] L.M. Chahine, M.B. Stern, A. Chen-Plotkin. Blood-based biomarkers for Parkinson's disease. Parkinsonism Relat Disord. 2014;20(Suppl. 1):S99-S103 Crossref
  • [10] T. Sakurai, A. Kimura, M. Yamada, A. Koumura, Y. Hayashi, Y. Tanaka, et al. Identification of antibodies as biological markers in serum from multiple sclerosis patients by immunoproteomic approach. J Neuroimmunol. 2011;233:175-180 Crossref
  • [11] P. Lourenco, A. Shirani, J. Saeedi, J. Oger, W.E. Schreiber, H. Tremlett. Oligoclonal bands and cerebrospinal fluid markers in multiple sclerosis: associations with disease course and progression. Mult Scler. 2013;19:577-584 Crossref
  • [12] C.M. Poser, D.W. Paty, L. Scheinberg, W.I. McDonald, F.A. Davis, G.C. Ebers, et al. New diagnostic criteria for multiple sclerosis: guidelines for research proposals. Ann Neurol. 1983;13:227-231 Crossref
  • [13] W.I. McDonald, A. Compston, G. Edan, D. Goodkin, H.P. Hartung, F.D. Lublin, et al. Recommended diagnostic criteria for multiple sclerosis: guidelines from the International Panel on the diagnosis of multiple sclerosis. Ann Neurol. 2001;50:121-127 Crossref
  • [14] H. Tremlett, V. Devonshire. Is late onset multiple sclerosis associated with a worse outcome?. Neurology. 2006;67:954-959 Crossref
  • [15] H. Tremlett, D. Paty, V. Devonshire. Disability progression in multiple sclerosis is slower than previously reported. Neurology. 2006;66:172-177 Crossref
  • [16] H. Tremlett, Y. Zhao, P. Rieckmann, M. Hutchinson. New perspectives in the natural history of multiple sclerosis. Neurology. 2010;74:2004-2015 Crossref
  • [17] A. Shirani, Y. Zhao, M.E. Karim, C. Evans, E. Kingwell, M.L. van der Kop, et al. Association between use of interferon beta and progression of disability in patients with relapsing – remitting multiple sclerosis. JAMA. 2012;308:247-256
  • [18] A. Shirani, Y. Zhao, E. Kingwell, P. Rieckmann, H. Tremlett. Temporal trends of disability progression in multiple sclerosis: findings from British Columbia, Canada (1975–2009). Mult Scler. 2012;18:442-450 Crossref
  • [19] G. Cohen Freue, M. Sasaki, A. Meredith, O. Gunther, A. Bergman, et al. Proteomic signatures in plasma during early acute renal allograft rejection. Mol Cell Proteomics. 2010;9:1954-1967
  • [20] L.B. Ohlund, D.B. Hardie, M.H. Elliott, D.S. Smith, J.D. Reid, et al. Standard operating procedures and protocols for the preparation and analysis of plasma samples using the iTRAQ methodology. A. Aivanov, A. Lazarev (Eds.) Sample preparation in biological mass spectrometry (Springer, New York, 2010)
  • [21] I.V. Shilov, S.L. Seymour, A.A. Patel, A. Loboda, W.H. Tang, S.P. Keating, et al. The Paragon Algorithm, a next generation search engine that uses sequence temperature values and feature probabilities to identify peptides from tandem mass spectra. Mol Cell Proteomics. 2007;6:1638-1655 Crossref
  • [22] C. UniProt. Activities at the Universal Protein Resource (UniProt). Nucleic Acids Res. 2014;42:D191-D198
  • [23] G.K. Smyth. Linear models and empirical bayes methods for assessing differential expression in microarray experiments. Stat Appl Genet Mol Biol. 2004;3 [Article3]
  • [24] X. Song, J. Bandow, J. Sherman, J.D. Baker, P.W. Brown, M.T. McDowell, et al. iTRAQ experimental design for plasma biomarker discovery. J Proteome Res. 2008;7:2952-2958 Crossref
  • [25] H. Zou, T. Hastie. Regularization and variable selection via the elastic net. J R Stat Soc Ser B (Stat Methodol). 2005;67:301-320 Crossref
  • [26] J. Gareth, D. Witten, T. Hastie, R. Tibshirani. An introduction to statistical learning. (Springer, New York, 2013)
  • [27] T. Sing, O. Sander, N. Beerenwinkel, T. Lengauer. ROCR: visualizing classifier performance in R. Bioinformatics. 2005;21:3940-3941 Crossref
  • [28] H. Wickham. ggplot2: elegant graphics for data analysis. (Springer, New York, 2009)
  • [29] S.J. Mason, N.E. Graham. Areas beneath the relative operating characteristics (ROC) and relative operating levels (ROL) curves: statistical significance and interpretation. Q J Roy Meteorol Soc. 2002;128:2145-2166 Crossref
  • [30] G.V. Cohen Freue, A. Meredith, D. Smith, A. Bergman, M. Sasaki, K.K. Lam, et al. Computational biomarker pipeline from discovery to clinical implementation: plasma proteomic biomarkers for cardiac transplantation. PLoS Comput Biol. 2013;9:e1002963 Crossref
  • [31] K.C. O'Connor, S.M. Roy, C.H. Becker, D.A. Hafler, A.B. Kantor. Comprehensive phenotyping in multiple sclerosis: discovery based proteomics and the current understanding of putative biomarkers. Dis Markers. 2006;22:213-225 Crossref
  • [32] D.A. Hendrickx, N. Koning, K.G. Schuurman, M.E. van Strien, C.G. van Eden, J. Hamann, et al. Selective upregulation of scavenger receptors in and around demyelinating areas in multiple sclerosis. J Neuropathol Exp Neurol. 2013;72:106-118 Crossref
  • [33] D.A. Hendrickx, K.G. Schuurman, M. van Draanen, J. Hamann, I. Huitinga. Enhanced uptake of multiple sclerosis-derived myelin by THP-1 macrophages and primary human microglia. J Neuroinflammation. 2014;11:64 Crossref
  • [34] D.C. Kilpatrick, J.D. Chalmers. Human L-ficolin (ficolin-2) and its clinical significance. J Biomed Biotechnol. 2012;2012:138797
  • [35] R. Gandhi, A. Laroni, H.L. Weiner. Role of the innate immune system in the pathogenesis of multiple sclerosis. J Neuroimmunol. 2010;221:7-14 Crossref
  • [36] D. Davalos, K. Akassoglou. Fibrinogen as a key regulator of inflammation in disease. Semin Immunopathol. 2012;34:43-62
  • [37] J. Han, R.M. Danell, J.R. Patel, D.R. Gumerov, C.O. Scarlett, J.P. Speir, et al. Towards high-throughput metabolomics using ultrahigh-field Fourier transform ion cyclotron resonance mass spectrometry. Metabolomics. 2008;4:128-140 Crossref
  • [38] L. Claudio, C.S. Raine, C.F. Brosnan. Evidence of persistent blood–brain barrier abnormalities in chronic-progressive multiple sclerosis. Acta Neuropathol. 1995;90:228-238
  • [39] D. Gay, M. Esiri. Blood–brain barrier damage in acute multiple sclerosis plaques. An immunocytological study. Brain. 1991;114(Pt 1B):557-572
  • [40] C. Marik, P.A. Felts, J. Bauer, H. Lassmann, K.J. Smith. Lesion genesis in a subset of patients with multiple sclerosis: a role for innate immunity?. Brain. 2007;130:2800-2815 Crossref
  • [41] C.M. Vos, J.J. Geurts, L. Montagne, E.S. van Haastert, L. Bo, P. van der Valk, et al. Blood–brain barrier alterations in both focal and diffuse abnormalities on postmortem MRI in multiple sclerosis. Neurobiol Dis. 2005;20:953-960 Crossref
  • [42] R.A. Adams, J. Bauer, M.J. Flick, S.L. Sikorski, T. Nuriel, H. Lassmann, et al. The fibrin-derived gamma377–395 peptide inhibits microglia activation and suppresses relapsing paralysis in central nervous system autoimmune disease. J Exp Med. 2007;204:571-582 Crossref
  • [43] K. Akassoglou, R.A. Adams, J. Bauer, P. Mercado, V. Tseveleki, H. Lassmann, et al. Fibrin depletion decreases inflammation and delays the onset of demyelination in a tumor necrosis factor transgenic mouse model for multiple sclerosis. Proc Natl Acad Sci U S A. 2004;101:6698-6703 Crossref
  • [44] R.A. Adams, C. Schachtrup, D. Davalos, I. Tsigelny, K. Akassoglou. Fibrinogen signal transduction as a mediator and therapeutic target in inflammation: lessons from multiple sclerosis. Curr Med Chem. 2007;14:2925-2936 Crossref
  • [45] O. Fernandez, R. Martin, A. Rovira, S. Llufriu, A. Vidal-Jordana, V.E. Fernandez-Sanchez, et al. Biomarkers in multiple sclerosis: an update for 2014. Rev Neurol. 2014;58:553-570
  • [46] I. Raphael, J. Webb, O. Stuve, W. Haskins, T. Forsthuber. Body fluid biomarkers in multiple sclerosis: how far we have come and how they could affect the clinic now and in the future. Expert Rev Clin Immunol. 2015;11:69-91
  • [47] F. Lolli, B. Mulinacci, A. Carotenuto, B. Bonetti, G. Sabatino, B. Mazzanti, et al. An N-glucosylated peptide detecting disease-specific autoantibodies, biomarkers of multiple sclerosis. Proc Natl Acad Sci U S A. 2005;102:10273-10278 Crossref
  • [48] J.L. Bennett, G.P. Owens. Cerebrospinal fluid proteomics: a new window for understanding human demyelinating disorders?. Ann Neurol. 2012;71:587-588 Crossref
  • [49] G.S. Ramsaransing, J. De Keyser. Benign course in multiple sclerosis: a review. Acta Neurol Scand. 2006;113:359-369 Crossref
  • [50] S. Menon, A. Shirani, Y. Zhao, J. Oger, A.L. Traboulsee, M.S. Freedman. UBC Hospital MS Clinic neurologists and Tremlett H. Characterising aggressive multiple sclerosis. J Neurol Neurosurg Psychiatry. 2013;84:1192-1198 Crossref
  • [51] L. Yan, S. Tonack, R. Smith, S. Dodd, R.E. Jenkins, N. Kitteringham, et al. Confounding effect of obstructive jaundice in the interpretation of proteomic plasma profiling data for pancreatic cancer. J Proteome Res. 2009;8:142-148 Crossref


a Faculty of Medicine, Department of Medicine, Division of Neurology, University of British Columbia, Vancouver, BC V5Z 1M9, Canada

b PROOF Centre of Excellence, Vancouver, BC V6Z 1Y6, Canada

c Department of Medicine, Division of Respiratory Medicine, Canada

d University of Victoria Genome BC Proteomics Centre, Victoria, BC V8Z 7X8, Canada

e Department of Statistics, University of British Columbia, Vancouver, BC V6T 1Z4, Canada

lowast Corresponding author at: Department of Statistics, University of British Columbia, 3152 Earth and Science Building, 2207 Main Mall, Vancouver, BC V6T 1Z4, Canada. Tel.: + 604 822 3710.

This article is part of a Special Issue entitled: Protein dynamics in health and disease. Guest Editors: Pierre Thibault and Anne-Claude Gingras.

Search this site

Stay up-to-date with our monthly e-alert

If you want to regularly receive information on what is happening in MS research sign up to our e-alert.

Subscribe »

About the Editors

  • Prof Timothy Vartanian

    Timothy Vartanian, Professor at the Brain and Mind Research Institute and the Department of Neurology, Weill Cornell Medical College, Cornell...
  • Dr Claire S. Riley

    Claire S. Riley, MD is an assistant attending neurologist and assistant professor of neurology in the Neurological Institute, Columbia University,...
  • Dr Rebecca Farber

    Rebecca Farber, MD is an attending neurologist and assistant professor of neurology at the Neurological Institute, Columbia University, in New...

This online Resource Centre has been made possible by a donation from EMD Serono, Inc., a business of Merck KGaA, Darmstadt, Germany.

Note that EMD Serono, Inc., has no editorial control or influence over the content of this Resource Centre. The Resource Centre and all content therein are subject to an independent editorial review.

The Grant for Multiple Sclerosis Innovation
supports promising translational research projects by academic researchers to improve understanding of multiple sclerosis (MS) for the ultimate benefit of patients.  For full information and application details, please click here

Journal Editor's choice

Recommended by Prof. Brenda Banwell

Causes of death among persons with multiple sclerosis

Gary R. Cutter, Jeffrey Zimmerman, Amber R. Salter, et al.

Multiple Sclerosis and Related Disorders, September 2015, Vol 4 Issue 5