Expression of soluble CD163 (sCD163), a macrophage/microglia biomarker, is increased in inflammatory conditions, and sCD163 levels in the cerebrospinal fluid (CSF) have recently been shown to be elevated in patients with multiple sclerosis (MS): the sCD163 CSF/serum ratio was elevated in patients with relapsing-remitting MS (RRMS), primary progressive MS (PPMS), and clinically isolated syndrome (CIS) compared with symptomatic controls.
To investigate the contributions of the sCD163 CSF/serum ratio to a biomarker panel focusing on inflammation and axonal degeneration in newly diagnosed MS; thus optimising a diagnostic biomarker panel for MS.
After a full MS diagnostic work-up, including collection of paired samples of CSF and serum, 125 patients were included in this study. Patients were divided into groups based on their diagnosis, and patients with normal clinical and paraclinical findings were defined as symptomatic controls. Serum and CSF levels, ratios, and indices of sCD163, CXCL13, osteopontin, neopterin, and CSF levels of neurofilament light polypeptide were determined by enzyme-linked immunosorbent assays (ELISAs). For sCD163 the results constitute a post-hoc analysis of already published data.
All tested biomarkers, notably the sCD163 ratio, the CXCL13 ratio, the NEO ratio, the CSF level of NfL, the IgG index, and the serum level of OPN, were significantly correlated to RRMS, PPMS, and/or CIS. The individual biomarkers in single tests had a lower performance than the IgG index, however, their combined receiver operating characteristic (ROC) curve demonstrated excellent diagnostic discriminatory power.
The biomarker panel showed distinct profiles for each patient group and could be a valuable tool for clinical differentiation of MS subgroups. The combined ROC analysis showed that sCD163 contributes positively as a diagnostic marker to a panel of established MS biomarkers. Patients with PPMS were demonstrated to have significantly elevated levels of both inflammatory and degenerative markers.
Citation: Stilund M, Gjelstrup MC, Petersen T, Møller HJ, Rasmussen PV, Christensen T (2015) Biomarkers of Inflammation and Axonal Degeneration/Damage in Patients with Newly Diagnosed Multiple Sclerosis: Contributions of the Soluble CD163 CSF/Serum Ratio to a Biomarker Panel. PLoS ONE 10(4): e0119681. https://doi.org/10.1371/journal.pone.0119681
Academic Editor: Markus Reindl, Medical University of Innsbruck, AUSTRIA
Received: July 11, 2014; Accepted: January 15, 2015; Published: April 10, 2015
Copyright: © 2015 Stilund et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited
Data Availability: All relevant data are within the paper and its Supporting Information files.
Funding: This study was funded by The Danish Multiple Sclerosis Society, grant number(s): 8990 and 14567, funder URL: http://scleroseforeningen.dk/, The Riisfort Foundation, grant no. 15399, URL: http://www.riisfort.dk/riisfort-fonden/, the Aase and Einar Danielsen’s Foundation; grant no 904587; URL: http://www.danielsensfond.dk/, The Jascha Foundation, grant no. 3099 URL: http://www.jaschafonden.dk/, and the Dagmar Marshalls Fond, grant no. N/A URL: N/A (it is a small Danish Foundation; applications via solicitor Oluf Engell; Bruun & Hjejle; Amagertorv 24 ; 1160 KøbenhavnK). HJM and Aarhus University have received royalties from IQ-products, NL. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: This study was funded by The Danish Multiple Sclerosis Society, grant number(s): 8990 and 14567, funder URL: http://scleroseforeningen.dk/, The Riisfort Foundation, grant no. 15399, URL: http://www.riisfort.dk/riisfort-fonden/, the Aase and Einar Danielsen’s Foundation; grant no 904587, URL: http://www.danielsensfond.dk/, The Jascha Foundation, grant no. 3099 URL: http://www.jaschafonden.dk/, and the Dagmar Marshalls Fond, grant no. N/A URL: N/A (it is a small Danish Foundation; applications via solicitor Oluf Engell; Bruun & Hjejle; Amagertorv 24; 1160 KøbenhavnK). HJM and Aarhus University have received royalties from IQ-products, NL. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. This does not alter the authors' adherence to PLOS ONE policies on sharing data and materials.
Multiple sclerosis (MS) is the most common demyelinating disease of the central nervous system (CNS) in young adults. The cause of the disease is unknown, yet it is generally assumed to be complex interactions between environmental factors and genetic susceptibility [1–3]. At the time of diagnosis, the disease course of patients with MS is characterized by either being remitting-relapsing (RRMS) or progressive: primary progressive MS (PPMS)—or in some cases with previous attacks, secondary progressive MS (SPMS). Patients with clinically isolated syndrome (CIS) often progresses to MS (approximately 85%) . Currently, MS diagnostic workup consists of an anamnestic support for spreading of neurological symptoms in time and space, MRI findings, and CSF/serum biomarkers such as an increased IgG index. A pathognomonic biomarker for MS has never been identified, yet substantial effort has been put into testing candidate molecules in various body fluids from patients. Several different molecules have been suggested as potential diagnostic biomarkers, but they still need to be validated in patient cohorts [5–9]. As suggested in previous reviews, the research in the field of MS biomarkers should be directed towards earlier and more accurate diagnosis to initiate disease modifying therapy [10–12].
A characteristic feature of MS lesions (plaques) is the abundant presence of macrophages in all types (I-IV) of MS lesions . Macrophages are components of the innate immune system and as tissue specific cells in the CNS, their functions are multifaceted, ranging from pro-inflammatory and antigen presenting (M1 pathway) to anti-inflammatory and growth promoting (M2 pathway) [14, 15]. Both pro- and anti- inflammatory macrophages as well as intermediate stages have been shown to be present in active MS lesions [16, 17].
We have recently demonstrated a significant association between levels of the sCD163 CSF/serum ratio (sCD163 ratio) and MS/CIS . Here, in the same patient cohort, we present further analyses on serum and CSF to evaluate the diagnostic contributions of sCD163 and the sCD163 ratio to a panel of established biomarkers focusing on inflammation and axonal damage. The biomarkers chosen for the present study were sCD163, CXCL13, neopterin (NEO), neurofilament light polypeptide (NfL), and osteopontin (OPN). This selection of biomarkers was based on previously published reports [9, 17–54] using the following criteria: 1) known MS diagnostic properties, 2) association with macrophage activity, and 3) detectability in CSF and/or serum by means of a previously validated quantifying method. Biomarkers and their characteristics are listed in table 1 and shown in Fig 1.
Fig 1 shows the biomarkers sCD163, CXCL13, NEO, NfL and OPN and their relation to monocyte activation, macrophage migration, and interactions with macrophages/microglia in active (acute/chronic) MS lesions. Biomarker characteristics are listed below: Abbreviations: sCD163 (soluble CD163), mCD163 (membrane bound CD163), NEO (neopterin), NfL (neurofilament light polypeptide), OPN (osteopontin), Th-cells (T helper cells), MS (multiple sclerosis), CSF (cerebrospinal fluid), EC (epithelial cell), Ag (antigen), SC (Schwann cell), OND (other neurological disease), ROS (reactive oxygen species).
The objectives of the present study were to investigate the contributions of the sCD163 ratio to a biomarker panel focusing on inflammation and axonal degeneration in patients with newly diagnosed MS. The study was performed on a large cohort of patients diagnosed with remitting relapsing MS (RRMS), primary progressive MS (PPMS), or clinically isolated syndrome (CIS). All cases were compared with a group of symptomatic controls (SC) . The applicability of these biomarkers in MS diagnostics, individually and collectively, was evaluated.
Materials and Methods
The study was conducted in accordance with the Ethical Declaration of Helsinki and all patients gave written, informed consent. The study and the material for informed consent were approved by The Central Denmark Region Committee on Biomedical Research Ethics (journal number: 20090210).
Patients admitted for further diagnostic investigations to the MS clinic, Department of Neurology, Aarhus University Hospital were consecutively included in the study during a period from September 2009 to December 2011. A full diagnostic work-up included medical history, clinical examination, MRI of the entire neural axis, CSF analyses (cells, protein, immunoglobulin G (IgG) index), and evoked potentials as recommended . CSF and MRI examination were evaluated according to the revised MacDonald criteria  and an EDSS (extended disability status scale) score was assessed according to Kurtzke . Mean total number of MRI white matter lesions (presented in table 2) were registered by fluid-attenuated inversion recovery (FLAIR) sequences on MRI.
In total 183 patients agreed to participate in this study and in accordance with consensus guidelines  serum and CSF samples were collected and frozen at -70°C. After aliquotation, all analyses were performed immediately after thawing of the aliquots and no samples were subjected to further freeze-thaw cycles. Demographics and paraclinical findings are summarised in table 2.
This patient cohort was also the cohort in the recent study of sCD163 levels and macrophage activity in MS . In the present study five patients from the original cohort were excluded: four patients with SPMS (as these were too few to be representative for this MS subgroup) and one patient with RRMS (due to lack of sample material). None of the patients in the cohort received methylprednisolone or other immune-modulating therapeutics for at least a month before the time of sampling.
Levels of sCD163 were analysed by ELISA as previously described . In brief, rabbit anti-CD163 (2 mg/L, SK Moestrup Aarhus University) was coated into microtitre wells and serum samples (diluted 1:101) or CSF samples (diluted 1:4) were added and incubated for 1 h at RT. Monoclonal anti-CD163 (clone number: GHI/61.3 μg/mL) was added followed by incubation for 1 h at RT with horseradish peroxidase-labelled goat anti-mouse antibodies (0.125 μg/mL; Dako, Glostrup, Denmark).
Levels of CXCL13, NEO, and OPN were analysed in both serum and CSF and levels of NfL were only analysed in CSF by enzyme-linked immunosorbent assay (ELISA) following the instructions given by the manufacturers. The available kits for NfL at the time of analysis were restricted to be used only with CSF. Samples were run in duplicates and a coefficient of variation (CV) was calculated, accepting only values ≤15%. Intra assay variations were calculated from six individual measurements of a known standard on each plate and values ≤15% were accepted (ranges: CXCL13 (2.02–8.57), NEO (3.54–13.67), NF-Light (0.58–5.94%), and OPN (2.00–9.80%)). Samples with values exceeding the highest point of the standard curve were diluted and reanalysed. The diagnoses were established before the results of the sCD163 analyses were received. For the other four biomarkers each plate contained 36 randomly selected samples, and each sample were labelled with a study ID and assayed in a manner blinded to the clinical status of the patients.
CXCL13 levels (R&D Systems, Minneapolis, DCX130) were measured in a competitive ELISA assay by adding 100 μL assay diluent and 50 μL of sample to each well, followed by a two hour incubation period. After 4 × wash, 200 μL of conjugate (mAb-HPR) were added, followed by two hours incubation. After 4 × wash, 200 μL substrate solution (TMB) were added and the plate was incubated for 30 min protected from light. Finally, 50 μL of stop solution (H2SO4) were added and the absorbance was measured on an ELISA plate reader (Thermo Scientific, Multiscan FC) at 450 and 540 nm. Concentrations were calculated by linear regression in Excel.
NEO levels (IBL Int. GMBH, Hamburg, RE59321) were measured by adding 100 μL enzyme conjugate (NEO), 20 μL of sample and 50 μL of NEO antiserum (pAb-HPR) to each well, followed by a 90 min incubation period. After 4 × wash, 150 μL of substrate solution (TMB) were added and the plate was incubated for 10 min protected from light. Finally, 150 μL of stop solution (H2SO4) were added and the absorbance was measured on an ELISA plate reader (Thermo Scientific, Multiscan FC) at 450 and 620 nm. Concentrations were calculated by 4-parametric statistical regression in GraphPad Prism.
NfL (NF-light) (UmanDiagnostics, Umea, UD51001) levels were measured by adding 50 μL sample dilution buffer and 50 μL of sample to each well, followed by one hour incubation. After 3 × wash, 100 μL of tracer (mAb) were added, followed by 45 min incubation. After 3 × wash, 100 μL conjugate (HPR) were added and the plate was incubated for 30 min. After 3 × wash, 100 μL substrate solution (TMB) were added and the plate was incubated for 15 min protected from light. Finally, 50 μL of stop solution (H2SO4) were added and the absorbance was measured on an ELISA plate reader (Thermo Scientific, Multiscan FC) at 450 and 540 nm. Concentrations were calculated by 4-parametric statistical regression in GraphPad Prism.
OPN levels (R&D Systems, Minneapolis, DOST00) were measured by adding 100 μL assay diluent and 50 μL of sample (diluted 1:25) to each well, followed by a two hour incubation period.
After 4 × wash, 200 μL of conjugate (pAb-HPR) were added, followed by two hours incubation. After 4 × wash, 200 μL substrate solution (TMB) were added and the plate was incubated for 30 min protected from light. Finally, 50 μL of stop solution (H2SO4) were added and the absorbance was measured on an ELISA plate reader (Thermo Scientific, Multiscan FC) at 450 and 540 nm. Concentrations were calculated by 4-parametric statistical regression in GraphPad Prism.
The manufacturer comments that OPN levels in serum are validated but may be reduced as a consequence of proteolytic cleavage during clotting.
Collection of data and statistical analysis
Data were stored and handled according to the Danish law on personal data. During collection of demographic and biochemical data we used the Electronic Patient Journal (EPJ). Descriptions of MRI data was conducted by a neuro-radiologist and confirmed by a senior neurologist who viewed all scans in the IMPAX system at the Department of Neurology, AUH.
For data collection we used Microsoft Excel and all statistical analysis was performed using STATA12. Please see Figure A in S1 Dataset for full data. Spearman correlation was used to correlate variables that were on an ordinal scale and Pearson pairwise correlation analyses were used to correlate continuous variables. This was done to identify, firstly, biomarkers that were correlated to age and or gender so that these could be adjusted for in our regression analysis. Secondly, correlations were performed to identify whether biomarkers were correlated to CIS/MS groups and thirdly, whether there was any biomarker intercorrelation. Please see tables B—AA in S1 Dataset for STATA coding and output. Since biomarkers in CSF and serum were positively skewed, we performed correlation and regression analyses on LOG-transformed data, except the intercorrelation analysis that was performed without the LOG-transformation. Only biomarkers that correlated significantly to CIS, RRMS or PPMS were included in the regression analysis, table AB in S1 Dataset. The geometric means (GM) of the SC group were used as baseline and the difference is presented in fold with 95% confidence intervals (95%CI) and p-values .
For calculations of ratios (i.e. sCD163 ratio) the CSF concentration was divided by the serum concentration, and indices were calculated by dividing the biomarker CSF/serum ratio by the albumin ratio.
The receiver operating characteristic (ROC) analyses were performed for the variables that were most significantly different from the SC group in the regression analysis. The parameter results for patients with RRMS, PPMS, and CIS are combined as true positives and plotted against SC as true negatives. A logistic regression analysis was performed for the combined diagnostic properties for each biomarker and the IgG index, and the entire biomarker panel including the IgG index . For a view on the data generated please see tables AC and AD in S1 Dataset for STATA do-files and output.
Soluble CD163 was detectable in CSF and serum from all individuals, as reported in . The sCD163 levels, ratio and index values are presented in Fig 2a–2d. Median levels and ranges are also presented in table 3. Median serum levels of sCD163 for the patients with RRMS, PPMS, and CIS were somewhat lower than for the SC group. However, the sCD163 ratio was elevated, particularly for the patients with PPMS. For the SC group the median level of sCD163 in serum was 1.70 mg/L (range 0.97–3.73 mg/L). This corresponds well with the sCD163 serum reference interval for healthy subjects (0.7–3.9 mg/L) .
Fig 2 shows box plots of all biomarkers including CSF and serum concentrations as well as ratios and index values. The box plots show levels of all the parameters for each group, RRMS, PPMS, CIS and SC. The horizontal line in each box is the median, whiskers fences the upper and lower quartiles. The data on sCD163 have previously been presented in . Fig 2a–2d shows the values for sCD163, Fig 2e–2h shows the values for CXCL13, Fig 2i–2l shows the values for NEO, Fig 2m shows the values for NfL CSF, and Fig 2n–2qshows the values for OPN. For all biomarkers, there were some overlaps in levels between patient groups, and symptomatic controls. Abbreviations: RRMS (relapsing-remitting MS), PPMS (primary-progressive MS), CIS (clinically isolated syndrome), SC (symptomatic controls), CSF (cerebrospinal fluid), sCD163 (soluble CD163), NEO (neopterin), NfL (neurofilament light polypeptide), OPN (osteopontin).
CXCL13 was detectable in serum from all individuals, but CXCL13 in CSF was undetectable in samples from one patient each from the RRMS and CIS patient groups, respectively, and in samples from five patients from the SC group. For calculating statistics the lower detection limit of the assay (0.0016 μg/L) was used where concentrations of proteins were below the lower detection limit. The CXCL13 levels, ratio and index values are presented in Fig 2e–2h. Median levels and ranges are also presented in table 3. The CXCL13 levels in the CSF were highly elevated for the patients with RRMS (particularly), PPMS and CIS, compared with the SC group. For the SC group the median level of CXCL13 in serum was 57.48 μg/L (range 29.5–201.4 μg/L).
NEO was detectable in CSF and serum from all individuals. The NEO levels, ratio and index values are presented in Fig 2i–2l. Median levels and ranges are also presented in table 3. In a pattern similar to that of sCD163, median serum levels of NEO for the patients with RRMS, PPMS, and CIS were somewhat lower than for the SC group. However, the NEO ratio was elevated, particularly for the patients with PPMS. For the SC group the median level of NEO in serum was 1.56 μg/L (range 0.634–2.696 μg/L).
Neurofilament light polypeptide (NfL)
NfL was detectable in CSF from all individuals. The NfL CSF levels are shown in Fig 2m. Median levels and ranges are also presented in table 3. The levels of NfL CSF were elevated for the patients with RRMS, PPMS and notably also for the patients with CIS, as compared with the SC group. For the SC group the median level of NfL CSF was 0.59 mg/L (range 0.233–1.376 μg/L).
OPN was detectable in CSF and serum from all individuals. The OPN levels, ratio and index values are presented in Fig 2n–2q. Median levels and ranges are also presented in table 3. Interestingly, OPN levels in serum for the patients with RRMS, PPMS, and for the patients with CIS, were elevated compared with the SC group. For the SC group the median level of OPN in serum was 18.05 μg/L (range 5.54–35.15 μg/L).
Spearman and Pearson correlation analyses for the biomarkers are presented without and with the Bonferroni correction in table 4. Biomarker variables that were correlated (r>20; see table 4) to gender or age were adjusted accordingly in the linear regression analysis.
Patients in the PPMS group were significantly older than patients in the RRMS, CIS and SC groups; and levels of sCD163 in both serum and CSF were found to correlate with age, however this was not the case for the sCD163 ratio.
The levels of sCD163 in serum and CSF were not significantly correlated to gender or any of the clinical measures: time since last attack, EDSS, mean no. of relapses, mean disease duration, enhancing lesions or the total number of MRI white matter lesions (TNL). There were moderate to strong correlations between most biomarker variables and CSF protein, CSF cells and IgG index.
Pearson pairwise correlation analysis of biomarkers is presented without and with the Bonferroni correction in table 5. sCD163 CSF and the sCD163 ratio were correlated to most of the other biomarkers, moderately correlated to OPN CSF, NEO CSF and the NEO ratio (p<0.001). Interestingly, NEO CSF and the NEO ratio were moderately correlated to CXCL13 CSF and the CXCL13 ratio. NfL CSF was weakly correlated to sCD163, CXCL13 and NEO levels. NfL CSF and index was also moderately correlated to EDSS.
Linear regression with relevant age and/or gender correction
The geometric mean (GM) values of the sCD163 ratio was significantly increased for the patients with RRMS: 1.36 fold (95%CI: 1.17–1.59; p<.001), CIS: 1.33 fold (95%CI: 1.07, 1.65; p = 0.010), and especially PPMS: 1.57 fold (95%CI: 1.23–1.97; p<0.001).
Of the other variables, most notably the CXCL13 ratio was elevated 13 fold (95%CI: 4.19–39.88, p<0.001) in patients with RRMS, but also significantly elevated in patients with PPMS: 5.52 fold (95%CI: 1.29–23.53; p = 0.021) and CIS: 6.86 fold (95%CI: p = 0.003). The NEO ratio was significantly elevated in all groups, especially RRMS: 1.44 (95%CI: 1.22–1.70; p<0.001). The NfL CSF was notably increased for patients with PPMS: 2.49 fold (95%CI: 1.96–3.18; p<0.001).
It is noteworthy, that OPN was a variable in which serum levels were elevated compared with the SC group, significantly for the CIS group: 1.62 fold (95%CI: 1.32–1.98; p<0.001).
The biomarkers with the optimal differentiating properties were established from the regression analysis: the sCD163 ratio, the CXCL13 ratio, the NEO ratio, the NfL CSF, and the OPN serum. These parameters were further analyzed in a biomarker profile (Fig 3), a ROC analysis (Fig 4), and using logistic regression analyses (Fig 5).
The biomarker profile was constructed as an overview representing the major characteristics of each group in the cohort. The profile was determined by categorizing the GM levels of the sCD163 ratio, the CXCL13 ratio, the NEO ratio, the NfL CSF, the OPN serum, and the IgG index for each patient group in the following manner: Very high (significantly increased GM, p <0.001); High (significantly increased GM, p <0.01); Moderate (significantly increased GM, p <0.05); Low (zero is included in the CI). Abbreviations: NEO (neopterin), NfL (neurofilament light polypeptide), OPN (osteopontin), SC (symptomatic controls), RRMS (relapsing-remitting MS), PPMS (primary-progressive MS), CIS (clinically isolated syndrome), CI (confidence interval), GM (geometric means).
Fig 4 shows the ROC curves generated for the sCD163 ratio (4a), the CXCL13 ratio (4b), the NEO ratio (4c), the NfL CSF (4d), the OPN serum (4e), and the IgG index (4f). The data on sCD163 have previously been presented in . AUC, with 95% CI, is given for each parameter. The parameter results for patients with RRMS, PPMS, and CIS are combined as true positives and plotted against SC as true negatives. The diagonal dividing the ROC space represents the random event. Abbreviations: ROC (receiver operating characteristic), AUC (area under the curve), RRMS (relapsing-remitting MS), PPMS (primary-progressive MS), CIS (clinically isolated syndrome), SC (symptomatic controls), CSF (cerebrospinal fluid), CI (confidence interval).
Fig 5 shows the combined ROC curves generated for the IgG index and the sCD163 ratio (5a), the CXCL13 ratio (5b), the NEO ratio (5c), the NfL CSF (5d), the OPN serum (5e), and All biomarkers (5f). AUC, with 95% CI, is given for each parameter. The parameter results for patients with RRMS, PPMS, and CIS are combined as true positives and plotted against SC as true negatives. The diagonal dividing the ROC space represents the random event. Abbreviations: ROC (receiver operating characteristic), AUC (area under the curve), RRMS (relapsing-remitting MS), PPMS (primary-progressive MS), CIS (clinically isolated syndrome), SC (symptomatic controls), CI (confidence interval).
The biomarker profile was determined by categorizing the GM for each patient group in the following manner: Very high (significantly increased GM, p <0.001); High (significantly increased GM, p = [0.01;0.001]; Moderate (significantly increased GM, p =] 0.05; 0.01]; and Low (zero is included in the CI). Fig 3 demonstrates a distinct profile for each patient group, distinguishing all patient groups from the SC group and also, to some extent, differentiating between patients with RRMS, PPMS, and CIS.
ROC analysis and logistic regression on biomarker performance
To examine the sensitivity and specificity of each biomarker a ROC analysis was performed for each biomarker (Fig 4). The sCD163 ratio (Fig 4a) had an area under the curve (AUC) of 0.73 (95%CI: 0.64–0.83) and this was higher than for OPN serum (Fig 4d; AUC 0.69 (95%CI: 0.59–0.78)), equal to the NEO ratio (Fig 4c; AUC 0.74 (95%CI: 0.65–0.83)) and lower than the CXCL13 ratio (Fig 4b; 0.80 (95%CI: 0.73–0.88)), and the NfL CSF (Fig 4e; AUC 0.86 (95%CI: 0.80–0.93)). All individual biomarkers had a lower performance than the IgG index (Fig 4f; AUC 0.89 (95%CI: 0.84–0.95)).
The ROC analyses were followed by a logistic regression model (see methods and table AD in S1 Dataset) in which each biomarker was combined with the IgG index, and finally in a combined analysis with all biomarkers (Fig 5). The combined analyses of sCD163 and the IgG index (Fig 5a) gave an AUC of 0.91 (95%CI: 0.86–0.96) which was higher than the CXCL13 levels and the IgG index combined (Fig 5b; AUC 0.89 (95%CI: 0.84–0.95)), equal to the NEO ratio (Fig 5c; AUC: 0.91 (95%CI: 0.85–0.96), and lower than the NFL index (Fig 5d; AUC: 0.95 (95%CI: 0.93–0.99)) and the OPN serum levels (Fig 5e; AUC: 0.95 (95%CI: 0.91–0.98). The combined biomarker panel (Fig 5f) had an AUC of 0.97 (95%CI: 0.93–0.99)—higher than that of all the individual biomarker AUCs and also higher than all of the individual biomarkers combined with the IgG index alone.
The biomarkers selected for this study have all documented discriminating potency providing differentiation between diseased and healthy subjects. As pointed out by Xia et al  this sort of study primarily focuses on the development of a biomarker panel. As a marker of inflammation, sCD163 levels are elevated in a range of diseases where inflammation contributes to the morbidity and disability [62, 63]. As a marker for monocyte / macrophage activity in disease sCD163 has proven to be a reliable biomarker . In our recently published study , we demonstrated that sCD163 is a fair marker for MS disease, and so, the contribution of this study to MS diagnostics is not the individual biomarker levels as such, but rather the increased discriminatory powers in the combination of biomarkers.
Contributions of sCD163 to a panel of biomarkers for inflammation and axonal degeneration/damage
For sCD163 the results constitute a post-hoc analysis of already published data . In the present study we show that sCD163 contributes to the differentiation between SC and patients diagnosed with CIS or MS as shown in Fig 3. The ROC analyses of the individual biomarkers shows that with respect to specificity and sensitivity, the sCD163 ratio scores on a par with the established biomarkers regarding diagnostic properties (see Fig 4a–4f). According to the criteria outlined by Xia et al  none of the biomarkers analysed could be categorized as excellent (AUC = 0.9–1.0); the IgG index, the NfL index, and the CXCL13 ratio could be categorized as good (AUC = 0.8–0.9), the sCD163 ratio and the NEO ratio could be categorized as fair (AUC = 0.7–0.8), whereas, finally, the OPN serum level could be categorized as poor (AUC = 0.6–0.7) . However, as demonstrated in Fig 5a–5e, the panel of combined biomarkers performs the diagnostic differentiation to a very high level, elevating the diagnostic properties to excellent for most biomarkers (only the CXCL13 ratio is exempted), when combined with the IgG index. The combination of all tested biomarkers (Fig 5f) resulted in an AUC of 0.97, illustrating that in a multifactorial disease, such as MS, the combination of multiple biomarkers into a singularised multivariable model provides a high level of diagnostic discrimination and reliability. Interestingly, in the present study OPN serum shows a poor discriminatory capacity when viewed as a single marker, yet combined with IgG index it becomes excellent. This is probably due to the fact that it OPN is especially elevated in serum of patients with CIS (Fig 2n) as is not the case for the IgG index.
Overall, we find it noteworthy that (except for OPN) the biomarker ratios and indexes (i.e. levels in CSF in relation to levels in serum) were optimal in differentiating between patient groups rather than levels in either CSF or serum. The well-established MS diagnostic marker, the IgG index, also measures a relation between levels in the CNS and in the systemic circulation. This underscores that measuring biomarker levels concomitantly in both compartments is advantageous.
In the present study patients with PPMS had significantly elevated sCD163 CSF median levels compared with SC. This is also the case for the NEO CSF levels. Interestingly, the levels of sCD163 and NEO in serum and CSF are decreased / increased in similar ways (Fig 2a, 2b, 2i and 2j). It has recently been reported, that the release of NEO triggers the release of TNFα, which is associated with the release of sCD163 . The levels of NEO and sCD163 in both serum and CSF were found to be comparable, supporting a concerted inflammatory response. Especially in PPMS, where sCD163 is known to be up regulated during the pro-inflammatory response , and the release of NEO is known to contribute to the orchestration of oxidative stress .
For the other biomarkers analysed, the results were as follows: the median levels of the CXCL13 ratio were significantly increased for all patient groups compared with SC. This increase attained the highest level of significance for patients with RRMS, followed by CIS and PPMS. This is in agreement with the general hypothesis that the pathological mechanism in these two groups is mainly driven by inflammation of which CXCL13 is a marker .
There were no significant differences between GM levels of NfL in patients with PPMS, CIS and RRMS, yet they were all significantly elevated compared with SC. This is interesting, as NfL has been proposed as a marker for relapse in MS . However, that study had some overlap in NfL levels between groups as is also the case in the present study. The levels of NfL were also significantly correlated to no. of attacks (r = 0.32) but there was no correlation to time since last attack (table 4).
For OPN, we found that patients with CIS had significantly elevated GMs of OPN in serum as compared with SC, as was also the case in the age adjusted linear regression model (table 6). Furthermore, the GM levels in CIS were also significantly elevated when compared with the GM levels in the RRMS and PPMS groups. It is noteworthy, that in a recent study by Kivisakk et al  the CIS patients had the lowest levels of OPN compared with patients with MS. However, that study is based on a patient cohort undergoing immune-modulating treatment, and thus the two cohorts are not comparable. The present study substantiates that PPMS also has an inflammatory component, as illustrated by the significantly elevated levels of the CXCL13 ratio, the sCD163 ratio and the NEO ratio as compared with the SC groups. The role of macrophages/microglia in progressive MS disease has recently been reviewed  and it is suggested that the damage unfolded during progression of the MS disease is due both to chronic inflammation and to neurodegeneration . Axonal transection is a common feature in MS pathology  and axonal damage/loss plays a key role in MS disability development. Thus, macrophages/microglia are intimately related to axonal damage in MS lesions and as MHC class II-positive cells  they are abundantly present in acute and chronically active MS lesions [70, 72]. It has been proposed that macrophages/microglia are involved in T cell activation  but also that they actively hinder lymphocyte proliferation illustrating the complex orchestration of pathology where macrophages/microglia play a pivotal role.
We are aware, that the results obtained with the present panel of biomarkers needs to be validated in a larger, independent cohort before application into a clinical setting; particularly as the sample numbers in the present patient groups are rather small for PPMS and SPMS. Extrapolation of results of the analyses to a wider range of patients including patients with i.e. non-demyelinating neurological or infectious disease of the CNS should also be explored in further studies.
In conclusion we propose that the sCD163 ratio contributes as a biomarker in differentiating newly diagnosed patients with MS/CIS from SC. The biomarker panel showed different profiles for each patient group and could be a valuable tool for clinical differentiation of MS subgroups. All tested biomarkers in the present study were significantly correlated to RRMS, PPMS, and/or CIS and their combined ROC curve analysis demonstrated excellent diagnostic discriminatory power. Finally, the study supports that PPMS has an inflammatory component as the levels of all biomarkers (except OPN) were significantly elevated in PPMS.
We thank laboratory technicians Kirsten Bank Petersen, and Eva Lykke Petersen for their excellent practical support.
We thank Jens Kirkeby for his kind assistance in the drawing of figures.
We thank everyone at the MS clinic at the Department of Neurology, Aarhus University Hospital.
Conceived and designed the experiments: MS TC TP HJM. Performed the experiments: MS TC HJM MCG. Analyzed the data: MS TC TP MCG. Contributed reagents/materials/analysis tools: MS TC HJM TP PVR. Wrote the paper: MS TC TP PVR MCG HJM.
- 1. Compston A, Coles A. Multiple sclerosis. Lancet. 2002;359(9313):1221–31. pmid:11955556
- 2. Compston A, Coles A. Multiple sclerosis. Lancet. 2008;372(9648):1502–17. pmid:18970977
- 3. Kipp M, van der Valk P, Amor S. Pathology of multiple sclerosis. CNS Neurol Disord Drug Targets. 2012;11(5):506–17. pmid:22583433
- 4. Miller D, Barkhof F, Montalban X, Thompson A, Filippi M. Clinically isolated syndromes suggestive of multiple sclerosis, part I: natural history, pathogenesis, diagnosis, and prognosis. Lancet neurology. 2005;4(5):281–8. pmid:15847841
- 5. Teunissen CE, Koel-Simmelink MJ, Pham TV, Knol JC, Khalil M, Trentini A, et al. Identification of biomarkers for diagnosis and progression of MS by MALDI-TOF mass spectrometry. Multiple sclerosis. 2011;17(7):838–50. pmid:21505015
- 6. Tumani H, Hartung HP, Hemmer B, Teunissen C, Deisenhammer F, Giovannoni G, et al. Cerebrospinal fluid biomarkers in multiple sclerosis. Neurobiology of disease. 2009;35(2):117–27. pmid:19426803
- 7. Ziemann U, Wahl M, Hattingen E, Tumani H. Development of biomarkers for multiple sclerosis as a neurodegenerative disorder. Progress in neurobiology. 2011;95(4):670–85. pmid:21524682
- 8. Shaw CE, Dunbar PR, Macaulay HA, Neale TJ. Measurement of immune markers in the serum and cerebrospinal fluid of multiple sclerosis patients during clinical remission. Journal of neurology. 1995;242(2):53–8. pmid:7707089
- 9. Stangel M, Fredrikson S, Meinl E, Petzold A, Stuve O, Tumani H. The utility of cerebrospinal fluid analysis in patients with multiple sclerosis. Nature reviews Neurology. 2013;9(5):267–76. pmid:23528543
- 10. Katsavos S, Anagnostouli M. Biomarkers in Multiple Sclerosis: An Up-to-Date Overview. Multiple sclerosis international. 2013;2013:340508. pmid:23401777
- 11. Harris VK, Sadiq SA. Disease biomarkers in multiple sclerosis: potential for use in therapeutic decision making. Molecular diagnosis & therapy. 2009;13(4):225–44.
- 12. Avsar T, Korkmaz D, Tutuncu M, Demirci NO, Saip S, Kamasak M, et al. Protein biomarkers for multiple sclerosis: semi-quantitative analysis of cerebrospinal fluid candidate protein biomarkers in different forms of multiple sclerosis. Mult Scler. 2012;18(8):1081–91. pmid:22252467
- 13. Lucchinetti C, Brück W, Parisi J, Scheithauer B, Rodriguez M, Lassmann H. Heterogeneity of multiple sclerosis lesions: implications for the pathogenesis of demyelination. Ann Neurol. 2000;47(6):707–17. pmid:10852536
- 14. Lieury A, Chanal M, Androdias G, Reynolds R, Cavagna S, Giraudon P, et al. Tissue remodeling in periplaque regions of multiple sclerosis spinal cord lesions. Glia. 2014;62(10):1645–58. pmid:24910450
- 15. Miron VE, Boyd A, Zhao J-W, Yuen TJ, Ruckh JM, Shadrach JL, et al. M2 microglia and macrophages drive oligodendrocyte differentiation during CNS remyelination. Nat Neurosci. 2013;16(9):1211–8. pmid:23872599
- 16. Vogel DYS, Vereyken EJF, Glim JE, Heijnen PDAM, Moeton M, van der Valk P, et al. Macrophages in inflammatory multiple sclerosis lesions have an intermediate activation status. J Neuroinflammation. 2013;10:35. pmid:23452918
- 17. Zhang Z, Zhang Z-Y, Schittenhelm J, Wu Y, Meyermann R, Schluesener HJ. Parenchymal accumulation of CD163+ macrophages/microglia in multiple sclerosis brains. J Neuroimmunol. 2011;237(1–2):73–9. pmid:21774995
- 18. Stilund M, Reuschlein AK, Christensen T, Moller HJ, Rasmussen PV, Petersen T. Soluble CD163 as a Marker of Macrophage Activity in Newly Diagnosed Patients with Multiple Sclerosis. PloS one. 2014;9(6):e98588. pmid:24886843
- 19. Galea J, Cruickshank G, Teeling JL, Boche D, Garland P, Perry VH, et al. The intrathecal CD163-haptoglobin-hemoglobin scavenging system in subarachnoid hemorrhage. Journal of neurochemistry. 2012;121(5):785–92. pmid:22380637
- 20. Davis BH, Zarev PV. Human monocyte CD163 expression inversely correlates with soluble CD163 plasma levels. Cytometry Part B, Clinical cytometry. 2005;63(1):16–22. pmid:15624200
- 21. Moller HJ. Soluble CD163. Scandinavian journal of clinical and laboratory investigation. 2012;72(1):1–13. pmid:22060747
- 22. Högger P, Sorg C. Soluble CD163 inhibits phorbol ester-induced lymphocyte proliferation. Biochemical and biophysical research communications. 2001;288(4):841–3. pmid:11688984
- 23. Legler DF, Loetscher M, Roos RS, Clark-Lewis I, Baggiolini M, Moser B. B cell-attracting chemokine 1, a human CXC chemokine expressed in lymphoid tissues, selectively attracts B lymphocytes via BLR1/CXCR5. The Journal of experimental medicine. 1998;187(4):655–60. pmid:9463416
- 24. Brettschneider J, Czerwoniak A, Senel M, Fang L, Kassubek J, Pinkhardt E, et al. The chemokine CXCL13 is a prognostic marker in clinically isolated syndrome (CIS). PloS one. 2010;5(8):e11986. pmid:20700489
- 25. Krumbholz M, Theil D, Cepok S, Hemmer B, Kivisakk P, Ransohoff RM, et al. Chemokines in multiple sclerosis: CXCL12 and CXCL13 up-regulation is differentially linked to CNS immune cell recruitment. Brain: a journal of neurology. 2006;129(Pt 1):200–11.
- 26. Festa ED, Hankiewicz K, Kim S, Skurnick J, Wolansky LJ, Cook SD, et al. Serum levels of CXCL13 are elevated in active multiple sclerosis. Multiple sclerosis. 2009;15(11):1271–9. pmid:19805441
- 27. Alvarez E, Piccio L, Mikesell RJ, Klawiter EC, Parks BJ, Naismith RT, et al. CXCL13 is a biomarker of inflammation in multiple sclerosis, neuromyelitis optica, and other neurological conditions. Multiple sclerosis. 2013;19(9):1204–8. pmid:23322500
- 28. Kowarik MC, Cepok S, Sellner J, Grummel V, Weber MS, Korn T, et al. CXCL13 is the major determinant for B cell recruitment to the CSF during neuroinflammation. J Neuroinflammation. 2012;9:93. pmid:22591862
- 29. Carlsen HS, Baekkevold ES, Morton HC, Haraldsen G, Brandtzaeg P. Monocyte-like and mature macrophages produce CXCL13 (B cell-attracting chemokine 1) in inflammatory lesions with lymphoid neogenesis. Blood. 2004;104(10):3021–7. pmid:15284119
- 30. Romme Christensen J, Bornsen L, Khademi M, Olsson T, Jensen PE, Sorensen PS, et al. CSF inflammation and axonal damage are increased and correlate in progressive multiple sclerosis. Multiple sclerosis. 2013;19(7):877–84. pmid:23178691
- 31. Sellebjerg F, Bornsen L, Khademi M, Krakauer M, Olsson T, Frederiksen JL, et al. Increased cerebrospinal fluid concentrations of the chemokine CXCL13 in active MS. Neurology. 2009;73(23):2003–10. pmid:19996075
- 32. Edwards KR, Goyal J, Plavina T, Czerkowicz J, Goelz S, Ranger A, et al. Feasibility of the use of combinatorial chemokine arrays to study blood and CSF in multiple sclerosis. PloS one. 2013;8(11):e81007. pmid:24278364
- 33. Ragheb S, Li Y, Simon K, VanHaerents S, Galimberti D, De Riz M, et al. Multiple sclerosis: BAFF and CXCL13 in cerebrospinal fluid. Multiple sclerosis. 2011;17(7):819–29. pmid:21372118
- 34. Khademi M, Kockum I, Andersson ML, Iacobaeus E, Brundin L, Sellebjerg F, et al. Cerebrospinal fluid CXCL13 in multiple sclerosis: a suggestive prognostic marker for the disease course. Multiple sclerosis. 2011;17(3):335–43. pmid:21135023
- 35. Murr C, Widner B, Wirleitner B, Fuchs D. Neopterin as a marker for immune system activation. Current drug metabolism. 2002;3(2):175–87. pmid:12003349
- 36. Werner-Felmayer G, Werner ER, Fuchs D, Hausen A, Reibnegger G, Wachter H. Neopterin formation and tryptophan degradation by a human myelomonocytic cell line (THP-1) upon cytokine treatment. Cancer research. 1990;50(10):2863–7. pmid:2110500
- 37. Hoffmann G, Wirleitner B, Fuchs D. Potential role of immune system activation-associated production of neopterin derivatives in humans. Inflammation research: official journal of the European Histamine Research Society [et al]. 2003;52(8):313–21. pmid:14504669
- 38. Giovannoni G, Lai M, Kidd D, Thorpe JW, Miller DH, Thompson AJ, et al. Daily urinary neopterin excretion as an immunological marker of disease activity in multiple sclerosis. Brain. 1997;120:1–13. pmid:9055793
- 39. Ott M, Demisch L, Engelhardt W, Fischer PA. Interleukin-2, soluble interleukin-2-receptor, neopterin, L-tryptophan and beta 2-microglobulin levels in CSF and serum of patients with relapsing-remitting or chronic-progressive multiple sclerosis. Journal of neurology. 1993;241(2):108–14. pmid:8138825
- 40. Dobson R, Topping J, Davis A, Thompson E, Giovannoni G. Cerebrospinal fluid and urinary biomarkers in multiple sclerosis. Acta neurologica Scandinavica. 2013;128(5):321–7. pmid:23465040
- 41. Huber C, Batchelor JR, Fuchs D, Hausen A, Lang A, Niederwieser D, et al. Immune response-associated production of neopterin. Release from macrophages primarily under control of interferon-gamma. The Journal of experimental medicine. 1984;160(1):310–6. pmid:6429267
- 42. Lee MK, Xu Z, Wong PC, Cleveland DW. Neurofilaments are obligate heteropolymers in vivo. The Journal of cell biology. 1993;122(6):1337–50. pmid:8376466
- 43. Teunissen CE, Khalil M. Neurofilaments as biomarkers in multiple sclerosis. Multiple sclerosis. 2012;18(5):552–6. pmid:22492131
- 44. Huizinga R, van der Star BJ, Kipp M, Jong R, Gerritsen W, Clarner T, et al. Phagocytosis of neuronal debris by microglia is associated with neuronal damage in multiple sclerosis. Glia. 2012;60(3):422–31. pmid:22161990
- 45. Semra YK, Seidi OA, Sharief MK. Heightened intrathecal release of axonal cytoskeletal proteins in multiple sclerosis is associated with progressive disease and clinical disability. J Neuroimmunol. 2002;122(1–2):132–9. pmid:11777557
- 46. Lycke JN, Karlsson JE, Andersen O, Rosengren LE. Neurofilament protein in cerebrospinal fluid: a potential marker of activity in multiple sclerosis. Journal of neurology, neurosurgery, and psychiatry. 1998;64(3):402–4. pmid:9527161
- 47. Norgren N, Sundstrom P, Svenningsson A, Rosengren L, Stigbrand T, Gunnarsson M. Neurofilament and glial fibrillary acidic protein in multiple sclerosis. Neurology. 2004;63(9):1586–90. pmid:15534240
- 48. Bartos A, Fialova L, Soukupova J, Kukal J, Malbohan I, Pitha J. Antibodies against light neurofilaments in multiple sclerosis patients. Acta Neurol Scand. 2007;116(2):100–7. pmid:17661795
- 49. Ehling R, Lutterotti A, Wanschitz J, Khalil M, Gneiss C, Deisenhammer F, et al. Increased frequencies of serum antibodies to neurofilament light in patients with primary chronic progressive multiple sclerosis. Multiple sclerosis. 2004;10(6):601–6. pmid:15584481
- 50. Rittling SR. Osteopontin in macrophage function. Expert reviews in molecular medicine. 2011;13:e15. Epub 2011/05/07. pmid:21545755
- 51. Sodek J, Ganss B, McKee MD. Osteopontin. Critical reviews in oral biology and medicine: an official publication of the American Association of Oral Biologists. 2000;11(3):279–303.
- 52. Schack L, Stapulionis R, Christensen B, Kofod-Olsen E, Skov Sorensen UB, Vorup-Jensen T, et al. Osteopontin enhances phagocytosis through a novel osteopontin receptor, the alphaXbeta2 integrin. Journal of immunology. 2009;182(11):6943–50. pmid:19454691
- 53. Sinclair C, Mirakhur M, Kirk J, Farrell M, McQuaid S. Up-regulation of osteopontin and alphaBeta-crystallin in the normal-appearing white matter of multiple sclerosis: an immunohistochemical study utilizing tissue microarrays. Neuropathology and applied neurobiology. 2005;31(3):292–303. pmid:15885066
- 54. Kivisakk P, Healy BC, Francois K, Gandhi R, Gholipour T, Egorova S, et al. Evaluation of circulating osteopontin levels in an unselected cohort of patients with multiple sclerosis: relevance for biomarker development. Multiple sclerosis. 2014;20(4):438–44. pmid:24005026
- 55. Teunissen C, Menge T, Altintas A, Alvarez-Cermeno JC, Bertolotto A, Berven FS, et al. Consensus definitions and application guidelines for control groups in cerebrospinal fluid biomarker studies in multiple sclerosis. Multiple sclerosis. 2013;19(13):1802–9. pmid:23695446
- 56. Polman CH, Reingold SC, Banwell B, Clanet M, Cohen JA, Filippi M, et al. Diagnostic criteria for multiple sclerosis: 2010 revisions to the McDonald criteria. Ann Neurol. 2011;69(2):292–302. pmid:21387374
- 57. Kurtzke JF. Rating neurologic impairment in multiple sclerosis: an expanded disability status scale (EDSS). Neurology. 1983;33(11):1444–52. pmid:6685237
- 58. Teunissen CE, Petzold A, Bennett JL, Berven FS, Brundin L, Comabella M, et al. A consensus protocol for the standardization of cerebrospinal fluid collection and biobanking. Neurology. 2009;73(22):1914–22. pmid:19949037
- 59. Moller HJ, Hald K, Moestrup SK. Characterization of an enzyme-linked immunosorbent assay for soluble CD163. Scandinavian journal of clinical and laboratory investigation. 2002;62(4):293–9. pmid:12476928
- 60. R N. Stata tip 1: The eform() option of regress Stata Technical Journal: StataCorp LP; 2003 [cited 2003]. Available: http://www.stata-journal.com/sjpdf.html?articlenum=st0054.
- 61. Xia J, Broadhurst DI, Wilson M, Wishart DS. Translational biomarker discovery in clinical metabolomics: an introductory tutorial. Metabolomics. 2013;9(2):280–99. pmid:23543913
- 62. Hiraoka A, Horiike N, Akbar SM, Michitaka K, Matsuyama T, Onji M. Soluble CD163 in patients with liver diseases: very high levels of soluble CD163 in patients with fulminant hepatic failure. Journal of gastroenterology. 2005;40(1):52–6. pmid:15692789
- 63. Moller HJ, de Fost M, Aerts H, Hollak C, Moestrup SK. Plasma level of the macrophage-derived soluble CD163 is increased and positively correlates with severity in Gaucher's disease. European journal of haematology. 2004;72(2):135–9. pmid:14962251
- 64. Moller HJ, Aerts H, Gronbaek H, Peterslund NA, Hyltoft Petersen P, Hornung N, et al. Soluble CD163: a marker molecule for monocyte/macrophage activity in disease. Scandinavian journal of clinical and laboratory investigation Supplementum. 2002;237:29–33. pmid:12570164
- 65. Etzerodt A, Rasmussen MR, Svendsen P, Chalaris A, Schwarz J, Galea I, et al. Structural basis for inflammation-driven shedding of CD163 ectodomain and tumor necrosis factor-alpha in macrophages. The Journal of biological chemistry. 2014;289(2):778–88. pmid:24275664
- 66. Bielekova B, Komori M, Xu Q, Reich DS, Wu T. Cerebrospinal fluid IL-12p40, CXCL13 and IL-8 as a combinatorial biomarker of active intrathecal inflammation. PloS one. 2012;7(11):e48370. pmid:23226202
- 67. Malmestrom C, Haghighi S, Rosengren L, Andersen O, Lycke J. Neurofilament light protein and glial fibrillary acidic protein as biological markers in MS. Neurology. 2003;61(12):1720–5. pmid:14694036
- 68. Correale J. The role of microglial activation in disease progression. Multiple sclerosis. 2014;20(10):1288–95. pmid:24812046
- 69. Centonze D, Muzio L, Rossi S, Furlan R, Bernardi G, Martino G. The link between inflammation, synaptic transmission and neurodegeneration in multiple sclerosis. Cell death and differentiation. 2010;17(7):1083–91. pmid:19927157
- 70. Trapp BD, Peterson J, Ransohoff RM, Rudick R, Mork S, Bo L. Axonal transection in the lesions of multiple sclerosis. The New England journal of medicine. 1998;338(5):278–85. pmid:9445407
- 71. Fabriek BO, Van Haastert ES, Galea I, Polfliet MMJ, Döpp ED, Van Den Heuvel MM, et al. CD163-positive perivascular macrophages in the human CNS express molecules for antigen recognition and presentation. Glia. 2005;51(4):297–305. pmid:15846794
- 72. Trapp BD, Nave K-A. Multiple sclerosis: an immune or neurodegenerative disorder? Annu Rev Neurosci. 2008;31:247–69. pmid:18558855
- 73. Bo L, Mork S, Kong PA, Nyland H, Pardo CA, Trapp BD. Detection of MHC class II-antigens on macrophages and microglia, but not on astrocytes and endothelia in active multiple sclerosis lesions. Journal of neuroimmunology. 1994;51(2):135–46. pmid:8182113