Article Text

Download PDFPDF

Marked increase in cerebrospinal fluid glial fibrillar acidic protein in neuromyelitis optica: an astrocytic damage marker
  1. T Misu1,2,
  2. R Takano1,
  3. K Fujihara1,2,
  4. T Takahashi3,
  5. S Sato4,
  6. Y Itoyama1
  1. 1
    Department of Neurology, Tohoku University Graduate School of Medicine, Sendai, Japan
  2. 2
    Department of Multiple Sclerosis Therapeutics, Tohoku University Graduate School of Medicine, Sendai, Japan
  3. 3
    Department of Neurology, Yonezawa National Hospital, Yonezawa, Japan
  4. 4
    Department of Neurology, Kohnan Hospital, Sendai, Japan
  1. Dr T Misu, Department of Neurology, Tohoku University Graduate School of Medicine, 1-1 Seiryomachi, Aobaku, Sendai 980-8574, Japan; misu{at}


Background: Neuromyelitis optica (NMO) is a neurological inflammatory disease associated with autoimmunity to aquaporin 4, predominantly localised in astrocytic foot processes. Recent studies have revealed that loss of aquaporin 4 and glial fibrillar acidic protein (GFAP) is a prominent feature of NMO lesions, suggesting astrocytic impairment.

Objective: To reveal a useful clinical biomarker of NMO.

Methods: Enzyme-linked immunosorbent assays were carried out for astrocytic markers GFAP and S100B in CSFs, obtained from the patients with NMO (n = 10) and multiple sclerosis (MS) (n = 10) manifesting acute myelitis, acute disseminated encephalomyelitis (ADEM) (n = 3), spinal infarction (n = 3), and other neurological diseases (OND) (n = 5).

Results: The CSF-GFAP levels during relapse in NMO (7666.0 (SD 15 266.5) ng/ml) were significantly over several thousand times higher than those in MS (0.7 (1.5)) or OND (0.6 (0.3)), and considerably higher than those in spinal infarction (354.7 (459.0)) and ADEM (0.4 (0.2)). They returned close to normal levels along with clinical improvement soon after corticosteroid therapy in NMO. There were strong correlations between the CSF-GFAP or S100B levels and expanded disability status scales or spinal lesion length in NMO (r>0.9).

Conclusions: CSF-GFAP and S100B may be clinically useful biomarkers in NMO, and astrocytic damage is strongly suggested in the acute phase of NMO.

Statistics from

Neuromyelitis optica (NMO) is an inflammatory disease of the central nervous system.1 There has been considerable controversy about whether NMO is a variant of multiple sclerosis (MS) or a distinct disease.2 3 Recently, a disease-specific autoantibody, NMO-IgG, was found in the sera from patients with NMO.4 Subsequently, it was found that NMO-IgG bound selectively to aquaporin (AQP) 4 water channel,5 which is densely expressed in astrocytic foot processes. In immunopathological studies of autopsied cases of NMO, AQP4 immunoreactivity was definitely lost in the perivascular areas in the acute inflammatory lesions of NMO, but not in MS lesions.68 We also showed that the staining of glial fibrillar acidic protein (GFAP) was lost in the NMO lesions lacking AQP4 immunoreactivity.6 7 These findings suggest that a dysfunction or damage of astrocytes probably related to AQP4 may be involved in the pathomechanism of NMO.6 7 However, there have been no studies of a clinically useful biomarker for astrocytic damage in NMO.


We analysed cerebrospinal fluid (CSF) of consecutive inpatients with NMO (n = 10), MS (n = 10), acute disseminated encephalomyelitis (ADEM) (n = 3), spinal cord infarction (n = 3) and those with other neurological disease controls (OND) (three with headache, two with conversion disorder) seen from 2003 to 2007 (table 1). NMO was diagnosed by Wingerchuk’s criteria,9 and MS was diagnosed by International Panel’s criteria.10 All cases of ADEM were diagnosed as postinfectious monophasic encephalomyelitis with diffuse or multifocal white-matter lesions by brain MRI, which fulfilled the proposed diagnostic criteria of ADEM.11 AQP4-antibody was assayed with AQP4-transfected HEK293 cells as we have previously described.12 All patients with NMO were seropositive for AQP4-antibody, and the other neurological controls were seronegative. All samples except OND were collected during the acute phase. All cases but one ADEM (multiple cerebral lesions) had acute spinal cord MRI lesions. We obtained informed consent from each patient, and the present study conformed to the guidelines of the institutional review board.

Table 1 Clinical profiles of neuromyelitis optica (NMO) and other neurological disease controls (OND)

Enzyme-linked immunosorbent assay (ELISA)

ELISA assays for CSF-GFAP (A05188, SPI bio, France) and S100B (BioVendor, Czech Republic) are commercially available, and we performed the assay according to the manufacturer’s protocol. For example, in the GFAP assay, duplicate CSF samples (100 ul) were applied to anti-GFAP antibody- precoated 96 well plates and incubated for 2 h. Duplicate standards and two blanks were also set. After three washes, biotin-labelled monoclonal anti-human GFAP antibody was added and incubated for 1 h. After washing, streptavidin-horseradish peroxidase tracer was applied for 1 h and then washed three times. Then, hydrogen peroxide/TAB substrate was added. The reaction was stopped by sulfuric acid solution, and the absorbance measured at 450 nm by an analyser (SPECTRA max 340PC, Molecular Devices, Japan). All samples over the maximum level of the standard curve were re-examined after appropriate dilution. The detection limit was 0.04 ng/ml in GFAP and 20 pg/ml in S100B.

Statistical analysis

We used the Mann–Whitney U test and Kruskal–Wallis rank test, where applicable. For correlations, we used the Spearman correlation coefficient by rank.


The CSF-GFAP levels in OND ranged from 0.3 to 1.0 (0.6 (0.3) ng/ml). However, the CSF-GFAP levels in NMO during acute exacerbation were much higher (7666.0 (15 266.5) ng/ml) than in the other groups (p<0.0003) (fig 1). The CSF-GFAP values in NMO were extraordinarily higher than those in MS (0.7 (1.5) ng/ml) (p<0.0004) or OND (p<0.0022) and much higher than those of ADEM (0.4 (0.2)) or spinal cord infarction (354.7 (459.0)). In three NMO patients whose CSF were collected again after corticosteroid therapy, the CSF-GFAP levels remarkably decreased to near the control levels after the treatment (17 826.6 (26 142.2) → 1.8 (1.5)) (fig 1). The CSF-GFAP levels strongly correlated with EDSS (r = 0.914) and the length of the spinal lesions (r = 0.944) in NMO.

Figure 1

Cerebrospinal fluid glial fibrillar acidic protein (CSF-GFAP) levels in neuromyelitis optica (NMO), multiple sclerosis (MS), acute disseminated encephalomyelitis (ADEM), spinal cord infarction and other neurological disease (OND). The GFAP levels are shown by a logarithmic scale, and the levels in relapse (rel) of NMO were much higher than those in the other groups including NMO remission (rem), MS, ADEM, spinal cord infarction (S.Inf) and OND (p<0.0003). In three NMO cases, the levels in relapse were extremely high, but they were nearly normal after methylprednisolone therapy.

The CSF-S100B levels in NMO (10 498.2 (18 591.8) pg/ml) in acute exacerbation were also significantly higher than those in MS (137.7 (51.04) pg/ml) (p = 0.0025) or OND (107.1 (63.0) pg/ml) (p = 0.0048), higher than those of ADEM (74.2 (32.0)) or spinal cord infarction (940.4 (944.6)), and the values decreased after corticosteroid therapy (132.3 (64.5)). The CSF-S100B levels strongly correlated with the length of spinal cord lesions (r = 0.944) in NMO.


We previously reported that GFAP immunoreactivity was widely lost in the spinal cord lesions of so-called optic-spinal MS.13 Those lesions were filled with a massive Schwann cell remyelination, probably due to astrocytic damage with the disruption of glia limitans.7 13 Recently, we demonstrated that NMO lesions consistently lost GFAP in addition to loss of AQP4 in the acute perivascular lesions, but that myelinated fibres were relatively well preserved.6 7 Therefore, we proposed that astrocytic dysfunction or damage via autoimmunity to AQP4 is a unique pathological feature of NMO.6 7 However, correlating pathological events to clinical severity has been difficult because of the lack of clinically useful biomarkers in NMO. Accordingly, we studied two astrocytic marker proteins in NMO patients.

In the present study, the CSF-GFAP and -S100B levels were greatly elevated during the acute phase of NMO as compared with those in MS and ADEM (about 10 000-fold for GFAP on average). It is interesting that such great increases of CSF-GFAP or -S100B in NMO correlated well with the clinical severity and the length of spinal cord lesions. Moreover, with clinical improvement soon after corticosteroid therapy, the CSF-GFAP levels dramatically returned to nearly normal levels. In contrast, the elevated CSF-GFAP levels in spinal infarction were relatively mild for the poor prognosis. We recently reported that neurofilament heavy chain concentrations were mildly increased in the CSF of patients with NMO compared with MS (seven- to eightfold on average), but this difference was much less than that of the CSF-astrocytic protein levels between NMO and MS.14 These findings suggest that the marked increase in CSF-GFAP or S100B is in parallel with the clinical severity and lesion expansion in NMO, and may not simply reflect a non-specific tissue damage.

The increases of GFAP in CSF have been reported in several disorders. In MS where GFAP is very rich in the plaques with astrogliosis,15 very mild elevations (less than three times the control values) of CSF-GFAP were reported, as seen in the present study.16 17 Such a mild CSF-GFAP elevation probably resulted from the release of GFAP from the chronic lesions with reactive astrogliosis into the CSF.18 In contrast, high levels of CSF-GFAP were observed in acute and destructive neurological diseases.18 19 In previous reports of CSF-GFAP with similar control values (200–500 pg/ml) and sensitivity (>32 pg/ml) to those in the present study, marked increases of CSF-GFAP were observed in cerebral infarction (78 216 (62 859) pg/ml) and herpetic encephalitis (28 250 (20 240) pg/ml),18 but the levels were far lower than those of NMO (7 666 000 (15 266 500) pg/ml) in the present study. Taken together, it is strongly suggested that remarkably profound astrocytic damage might occur in acute NMO lesions, and that NMO could be clearly discriminated from MS.

A recent in vitro study demonstrated that AQP4-antibody caused endocytosis/degradation of AQP4 that could lead to cytotoxicity to AQP4-expressing cells,20 which could explain the loss of AQP4 in astrocytes.68 These findings clearly support the fact that the significant elevations of CSF-GFAP or -S100B in NMO are related to anti-AQP4 autoimmunity affecting astrocytes and not just a reflection of non-specific tissue damage. Thus, measuring astrocytic markers, especially CSF-GFAP, would be useful in assessing astrocytopathy and clinical severity of NMO. Larger-scale studies with various myelopathies and controls will be needed to further clarify the unique astrocytopathy in NMO.


The authors thank B Bell for reading the manuscript.



  • Funding: This work was supported by research grants from the Ministry of Education, Science and Technology, and the Ministry of Health, Labor and Welfare of Japan.

  • Competing interests: None.

  • Ethics approval: Ethics approval was provided by the Medical Ethics Committee of Tohoku University School of Medicine.

  • Patient consent: Obtained.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.