Abstract
Objective: To determine if CNS-derived proteins present in the CSF of multiple sclerosis (MS) patients reflect different pathologic processes of MS and if these proteins could be useful as biologic markers of disease activity.
Methods: Concentrations of the neurofilament light protein (NFL), glial fibrillary acidic protein (GFAP), S100B, and the neuron-specific enolase protein (NSE) were determined in the CSF of 66 MS patients and 50 healthy control subjects with immunoassays.
Results: The mean levels of the NFL were increased during all stages of MS compared with controls (p < 0.001), peaking almost 10 times higher during acute relapses. The highest levels of GFAP were found during the secondary progressive course (p < 0.001) with a strong correlation with neurologic deficits (Expanded Disability Status Scale score, r = 0.73, p < 0.001). No increase of S100B or NSE protein was found in the CSF of MS patients compared with control subjects.
Conclusions: Increased level of NFL is a general feature of MS, indicating continuous axonal damage during the entire course of the disease with the most profound damage during acute relapses. GFAP may serve as a biomarker for disease progression, probably reflecting the increasing rate of astrogliosis.
There is evidence that glial responses as well as axonal damage are important features in the pathogenesis of multiple sclerosis (MS). Whereas permanent neurologic deficits in MS are mainly a result of axonal loss, different MR techniques1,2⇓ as well as immunohistochemical studies3,4⇓ also reveal axonal damage in early stages of MS. The role of the astrocytes in the pathogenesis of MS has attracted less interest. Astrocytes are activated in acute lesions and participate in gliotic scar formation of chronic lesions. CSF examinations might add important information, which, in conjunction with MR data, could elucidate these issues further. Previous investigations of biologic CSF markers in MS have mainly focused upon the inflammatory response in MS. The purpose of our study was to investigate if soluble CNS-derived proteins are useful as biologic markers of disease activity and hence reveal information about the ongoing pathologic processes in MS.
The CNS-derived proteins that we selected and quantified in CSF of MS patients were the neurofilament light protein (NFL), the neuron-specific enolase protein (NSE), the glial fibrillary acidic protein (GFAP), and S100B. The neurofilament protein is composed of a triplet protein that is the most abundant cytoskeletal component in large myelinated axons and is only to a minor extent expressed in neural cell bodies. Increased CSF levels of the light neurofilament subunit were reported in neurodegenerative diseases,5–7⇓⇓ acute CNS infections,8 and autoimmune CNS diseases including MS.9–11⇓⇓ Whereas increased CSF NFL levels are considered to be a marker of axonal damage, NSE is an energy metabolic protein of neural cell bodies. S100B is a calcium binding protein abundantly expressed in astroglial cells, which might serve as a marker for astrocyte activity, and GFAP is the intermediate filament of fibrillary astrocytes and is considered to be the morphologic basis of astrogliosis and the main protein constituent of the chronic MS lesion.
Materials and methods.
Patients and sampling of CSF.
The study was approved by the ethical committee of the University of Gothenburg, Sweden, and informed consent was obtained from all patients and healthy control subjects. The patients were consecutively recruited at the Department of Neurology, Sahlgrenska University Hospital, Gothenburg, and all were clinically examined and had a lumbar spinal tap. The neurologic deficits were scored with the Expanded Disability Status Scale (EDSS),12 and the clinical syndrome of the relapse was classified as mono- or polyfocal and according to topography into optic neuritis, myelopathy, brainstem syndrome, or hemispheric syndrome. The CSF samples were obtained by a lumbar spinal tap in close relation to the neurologic examination. The first 12 mL of CSF was carefully mixed; after centrifugation, fractions were snap-frozen to avoid influence from proteolysis and stored in 0.5-mL aliquots at −80 °C until analyzed. All samples analyzed were thawed once only.
Sixty-six patients (mean age 39.6 years) with MS13 were included in the study and 50 healthy blood donors served as controls. Demographic data of patients and control subjects are presented in table 1. Forty-one patients had relapsing–remitting disease course, of which 18 were in remission, that is, they had no clinical relapses within the last 3 months (RRMS-rem), and 23 had acute relapse (RRMS-rel). Twenty-five patients had secondary progressive MS (SPMS). One RRMS-rel patient and one RRMS-rem patient had ongoing interferon-β treatment at the time of their inclusion. Thirteen patients of the RRMS-rel group, six women and seven men (28 to 45 years old, mean 32.7 years), with MS duration of 0.5 to 25 years (mean 8.6 years) and EDSS score of 1.5 to 6.5 (median 3), were prospectively followed. In this group, EDSS scores and CSF samples were also obtained at approximately 3 weeks and at 3 months after the first examination. Six prospectively followed patients received treatment of their relapses with IV methylprednisolone, and three of them started on interferon-β treatment during their follow-up.
Table 1 Characteristics of patients and healthy control subjects
NFL assay.
Concentrations of NFL in CSF were analyzed according to a previously described sandwich ELISA.5 In brief, capturing antibody (hen anti-NFL IgG) was absorbed to microtest plates, after which CSF samples or reference NFL were incubated for 2 hours at room temperature. Rabbit polyclonal anti-NFL IgG was used as the secondary antibody, and the incubation time was 1 hour. Bound secondary antibody was detected using peroxidase-conjugated donkey anti-rabbit IgG. The standard curve ranged from 125 to 16,000 ng/L. The sensitivity of the assay was 125 ng/L. Samples with levels below the detection limit were assigned this value.
GFAP assay.
GFAP was measured with a previously described ELISA procedure.14 In brief, capturing antibody (hen anti-GFAP IgG) was absorbed to microplates. CSF samples or reference GFAP was added and incubated for 2 hours at room temperature. Rabbit anti-GFAP IgG was then added and incubated for 1 hour at room temperature. Captured secondary antibody was detected using peroxidase-conjugated donkey anti-rabbit IgG. The sensitivity of the GFAP assay was 16 ng/L.
S100B and NSE assay.
S100B protein and NSE concentrations were measured by commercially available luminescence immunoassays (Sangtec Medical, Bromma, Sweden), and the sensitivity was 0.02 and 1.0 μg/L, respectively. Values were normal if the level of S100B was <5 μg/L and the corresponding level for NSE was <10.5 μg/L.
A high degree of specificity has previously been reported for the GFAP and NFL assays, with regard to the other prevalent intermediate filament proteins in CNS.5,14,15⇓⇓ The results of NSE and S100B are based on commercially available assays. In our hands, the reproducibility of all of these assays is high, with a coefficient of variation below 0.10.
Statistics.
The Mann–Whitney U test, corrected for ties, was used to compare patient groups, and Wilcoxon’s signed rank test, corrected for ties, was used to compare changes over time. Spearman’s rank correlation coefficient, corrected for ties, was used to discern relationships between CSF markers and clinical or demographic variables. The level of 124 ng/L was applied in patients with CSF NFL levels below the detection limit of the assay to calculate the discriminatory level. This level was applied to determine the specificity and sensitivity of the NFL assay to identify relapses. Partial correlation coefficients were determined to control the reciprocal dependence between different variables. Analyses were made with SPSS 10.1 software (Chicago, IL).
Results.
NFL.
The mean CSF NFL levels were increased in all three MS groups compared with controls (p < 0.001; figure, A). The highest CSF NFL levels were recorded in RRMS patients during acute relapses (mean 1,727 ng/L, SD 1,711 ng/L, range <125 to 6,217 ng/L). These levels were significantly higher than those of RRMS in remission and SPMS (p < 0.001). No differences were found between NFL levels of RRMS patients in remission and patients during secondary progression. CSF NFL levels above the detection limit of the method, that is, >125 ng/L, were recorded in 21 of 23 (91%) of the RRMS-rel group, 8 of 18 (44%) of the RRMS-rem group, 12 of 25 (48%) of the SPMS patients, and 4 of 50 (8%) of control subjects.
Figure. (A) Box plot indicating the concentration of the neurofilament light protein (NFL) in CSF of relapsing–remitting multiple sclerosis patients during relapse (RRMS rel) and in remission (RRMS rem), in patients with secondary progression (SPMS), and in healthy control subjects. Boxes include median, 25th, and 75th percentiles; bars indicate 10th and 90th percentiles, and triangles indicate individual values. N = number of subjects. (B) Logarithmic line plot indicating concentrations of NFL in CSF of individual relapsing–remitting MS patients measured after the onset of acute relapses. Squares indicate the concentration at each lumbar spinal tap. (C) Scattergram of the concentration of glial fibrillary acidic protein (GFAP) in CSF against neurologic deficit, scored with the Expanded Disability Statue Scale (EDSS) from relapsing-remitting multiple sclerosis patients in remission (RRMS rem), and with secondary progression (SPMS) (r = 0.521, p < 0.001). Triangles indicate individual values of patients in remission and in secondary progression. Lines indicate means and 95% CI.
To explore the clinical usefulness of NFL as a marker for relapses, we applied a discriminatory level of 674 ng/L, that is, 2 SD above the mean CSF NFL level of RRMS-rem patients. At this level, the specificity was 94% and the sensitivity was 70% to identify relapses in RRMS.
Thirteen patients of the RRMS-rel group who participated in the prospective study were included between 5 and 34 days (median 16 days) after their relapse onset. Two patients from this group dropped out after the second lumbar spinal tap: one due to postpuncture headache and the second due to a new exacerbation. The NFL levels increased between the first and the second lumbar spinal tap, performed 3 weeks later (p = 0.05). Between the second and third lumbar punctures, a moderate drop in the CSF NFL level was recorded (p < 0.05; table 2).
Table 2 Concentrations of neurofilament light protein in CSF of 13 relapsing–remitting MS patients prospectively followed after onset of previous relapse
The temporal change of the CSF NFL levels after relapse onset was further analyzed on an individual basis (see the figure, B). A large interindividual range of NFL levels was revealed. With two exceptions, NFL values were lower at the last spinal tap and significantly lower when CSF samples were obtained 60 days or later from relapse onset (p = 0.03).
We found no relationship between CSF NFL levels and age, disease duration, or gender, nor was a relationship found with EDSS, and no association was found between the NFL level and the type of clinical syndrome of the relapse. The lack of relationship with neurologic deficits was also true for the prospective part of the study. Neither could any influence on the CSF NFL level be discerned as a dysfunction of the blood–brain barrier, according to the albumin ratio.
GFAP.
MS patients had higher GFAP levels in CSF (mean 611 ng/L, SD 308 ng/L) compared with control subjects (mean 435 ng/L, SD 160 ng/L; p < 0.001). This difference was most pronounced for the SPMS group, but it was significant also for the RRMS-rel group (while not for the RRMS-rem group) (table 3). Between the different MS groups, we observed no inconsistency of significance.
Table 3 Concentrations of GFAP in CSF of patients and healthy control subjects
In the prospective RRMS-rel group, the GFAP levels were essentially unchanged during the study period. At the first lumbar puncture, the mean GFAP level was 558 ng/L (SD 328 ng/L), 3 weeks later 589 ng/L (SD 322 ng/L), and at 3 months, the CSF GFAP mean level was 587 ng/L (SD 252 ng/L).
A correlation between GFAP levels and EDSS scores was found in MS patients (r = 0.473, p < 0.0001). Whereas this dependence was most pronounced in the SPMS group (r = 0.732, p < 0,001), it was significant also for RRMS-rem patients (r = 0.627, p < 0.005) but not for the RRMS-rel group (see the figure, C). Except for a weak correlation with age (r = 0.28, p = 0.02), no relationship was revealed between the CSF GFAP levels of patients and disease duration or gender. However, in control subjects, age correlated with the CSF GFAP levels (r = 0.37, p = 0.009). Partial correlation, controlling for age, still showed correlations with EDSS in the combined group of RRMS-rem and SPMS (r = 0.48, p < 0.001) and for SPMS (r = 0.74, p < 0.0001).
S100B and NSE.
The CSF levels of S100B and NSE were not different between MS patients and control subjects, and no association of these CNS-derived proteins to demographic or clinical data seemed discernible.
Discussion.
With the introduction of immunomodulatory drugs for MS treatment, the identification of objective and reliable methods that will allow quantification of disease activity as well as recognition of different inflammatory and neuropathologic processes has become increasingly important. We demonstrate in the current report that two CNS-derived proteins may be clinically useful as biologic markers in MS. Our results show that high NFL levels, indicating considerable axonal damage, are associated with relapses and that increased levels of GFAP, a marker of astrogliosis, correlate with increased neurologic disability.
Although axonal damage is extensive in chronic MS lesions, recent studies based on immunohistochemical methods report axonal damage also in acute lesions,3,4⇓ and MR investigations present evidence that axonal damage is an early event in MS2,16⇓ and that it might even be present already at the clinical onset of the disease.17 In this study, we found that increased NFL level is a general finding in all stages of MS. Similar increases of CSF NFL levels have been reported in progressive MS9 as well as in RRMS10 but could not be repeated recently,18 probably owing to differences in the sensitivity of the immunoassay and in the selection of patients. We confirmed the strong association between the NFL levels and relapses in the current study and also prospectively for the first time. The highest levels appeared approximately 3 weeks after relapse onset. When a discriminatory level was applied to identify relapses, the specificity of the assay was 94% and the sensitivity was 70%. Thus, it seems possible to use CSF NFL determination as an objective mean to support the clinical examination of relapses. The increase of CSF NFL levels was not confined to a specific clinical syndrome of the relapse, and we could not find any correlation between CSF NFL levels and neurologic deficits scored with EDSS. Presumably, this might be explained by delay of the peak efflux of NFL during the relapses.
The temporal correlation between high release of NFL to CSF and clinical exacerbations suggests a pathophysiologic connection between relapses and axonal damage and NFL. The fact that active MS lesions are 10-fold more frequent than the number of clinical exacerbations19 has previously been considered, indicating that a relatively low number of lesions in eloquent areas are involved and that only some of the clinically active lesions are located in the spinal cord. Other plausible mechanisms relevant in development of relapses might be de- and remyelination or humoral factors such as certain cytokines and nitric oxide affecting the conduction of nerve impulses or interfering with synaptic transmission. Our results would indicate that axonal damage may be a most important background to clinical exacerbations as essentially all high levels of NFL in CSF were recorded during relapses and only moderate or low release of NFL was observed in CSF during the clinically more stable stages of the disease. Previously, we reported that high NFL levels dropped approximately 100 days after relapse onset.10 Although we confirm this decrease in the prospectively followed relapse group, it remained considerably higher in some patients compared with patients in the remission group. They had not had a relapse for at least 3 months and were clinically stable. We have no clinical indication that this observation was due to a more aggressive disease. The NFL level probably reflects the extension of the axonal damage during the relapse. However, it could also be caused by subclinical disease activity.
The cause of axonal loss in MS is still poorly understood. Probably, there are both inflammation-induced and neurodegenerative causes. The axonal loss is most conspicuous in areas of an active inflammation.3,4⇓ In accordance, the association found by us between relapses and high levels of NFL in CSF might suggest influence of inflammatory components. The role of NFL as a target for immune attacks has recently also attracted much attention. Elevated IgG anti-NFL antibodies were recorded in CSF assays of MS.18,20⇓ This finding was further supported by immunocytochemistry observations indicating that CSF and serum from a number of patients with MS seem to bind to axonal cytoskeletal structures and that CSF oligoclonal IgG bands are formed in conjunction with NFL.20 The possible immunopathogenic role of NFL was also demonstrated by correlations between intrathecal NFL antibody levels and MRI measures of lesion load and cerebral atrophy.18 As pointed out previously,21 these data corroborate the notion that early axonal damage contributes to cerebral atrophy and clinical disease progression and provides further rationale for early intervention with immunomodulatory MS therapy, which seems to reduce also the rate of axonal loss.22
We have previously found augmented concentrations of CSF GFAP in RRMS patients and the levels increased further during a 2-year observation period.23 In the current study, we report a significant elevation of CSF GFAP concentrations in all subgroups except for the RRMS-rem group. This is in contrast with others who only found elevated GFAP levels in MS patients who were severely disabled.24–26⇓⇓ These discrepancies are probably due to methodologic differences and differences in the selection of the control group. The sensitivity of the ELISA must be questioned in one of these studies as all samples from 32 patients except 2 were below the detection limit of their method.24 The control groups of the two other studies consisted of a variety of conditions, many of which apparently involved the nervous system.25,26⇓ In contrast, our control subjects consisted of 50 healthy blood donors, none with neurologic symptoms or signs. As GFAP has been found to be elevated also in other CNS diseases, for example, dementia,27 brain infarction,28 and neuroborreliosis,29 it should be regarded as a nonspecific biomarker of CNS tissue injury. Thus, their reference levels might have been inappropriately high with potential underestimation of the number of MS cases with pathologic GFAP levels. Proteinases and their inhibitors are present in the CSF of MS patients with the net proteolytic activity dependent on the balance. Theoretically, this might cause changed levels of GFAP and NFL in MS patients. However, both proteins are intermediate filament proteins subjected to similar proteolytic systems. The lack of correlation between the CSF levels of GFAP and NFL indicates that this proteolytic activity should not be of major importance with regard to the conclusions of this study.
We report a relationship between the release of GFAP in CSF of MS patients and the neurologic deficits, scored with EDSS (r = 0.47, p < 0.001). The correlation was most evident during secondary progression (r = 0.732, p < 0,001) and was also observed during remission (r = 0.627, p < 0.005) but could not be detected in RRMS during relapses. Although age was found to influence the CSF GFAP levels, statistical adjustments for this variable had essentially no effect on the relationship between GFAP levels and EDSS scores of MS patients.
The strong correlation we found between CSF GFAP concentrations and EDSS during RRMS in remission and especially in SPMS is of the same magnitude as the best surrogate markers of clinical progression found with MR techniques.30,31⇓ GFAP seems to be one of the best biomarkers in CSF for clinical disability. It probably indicates the gradual change toward a more chronic neurodegenerative stage of MS during the course of the disease in which astrogliosis prevails.
It is important to emphasize that our results are based on group comparisons with considerable overlap between the groups. The usefulness to follow GFAP levels in individual patients should be further explored in long-term prospective studies.
We were unable to demonstrate increased S100B or NSE concentrations in CSF in our studied MS patients. Although a recent report has indicated that S100B could be useful to differentiate between RRMS and progressive MS,26 the value of S100B as a marker of disease activity in MS has not been established and the results from several studies are contradictory.25,26,32,33⇓⇓⇓ NSE has previously attracted most interest as a diagnostic and neurodegenerative bio-marker of prion diseases and dementia.34,35⇓ Recent reports on gray matter involvement in MS36,37⇓ suggest damage to neural somas. Our results do not support this hypothesis. No elevations of CSF NSE levels were recorded, a finding that is in line with some previous reports.32,38⇓
Acknowledgments
Supported by grant from NHR Göteborg and Edit Jacobsson Foundation.
- Received March 31, 2003.
- Accepted August 20, 2003.
References
Letters: Rapid online correspondence
- Neurofilament light protein and glial fibrillary acidic protein as biological markers in MS
- Reply to Avasarala
REQUIREMENTS
If you are uploading a letter concerning an article:
You must have updated your disclosures within six months: http://submit.neurology.org
Your co-authors must send a completed Publishing Agreement Form to Neurology Staff (not necessary for the lead/corresponding author as the form below will suffice) before you upload your comment.
If you are responding to a comment that was written about an article you originally authored:
You (and co-authors) do not need to fill out forms or check disclosures as author forms are still valid
and apply to letter.
Submission specifications:
- Submissions must be < 200 words with < 5 references. Reference 1 must be the article on which you are commenting.
- Submissions should not have more than 5 authors. (Exception: original author replies can include all original authors of the article)
- Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- Submitted comments are subject to editing and editor review prior to posting.
You May Also be Interested in
Topics Discussed
Alert Me
Recommended articles
Neurofilament and glial fibrillary acidic protein in multiple sclerosis
Serum GFAP and neurofilament light as biomarkers of disease activity and disability in NMOSD
Serum biomarkers in myelin oligodendrocyte glycoprotein antibody–associated disease
High serum neurofilament light chain normalizes after hematopoietic stem cell transplantation for MS