ORIGINAL ARTICLE

Axonal Damage in Relapsing MultipleSclerosis is Markedly Reduced

by Natalizumab

Martin Gunnarsson, MD, PhD,1 Clas Malmestrom, MD, PhD,2

Markus Axelsson, MD,2 Peter Sundstrom, MD, PhD,3 Charlotte Dahle, MD, PhD,4,5

Magnus Vrethem, MD, PhD,4 Tomas Olsson, MD, PhD,6 Fredrik Piehl, MD, PhD,6

Niklas Norgren, PhD,7 Lars Rosengren, MD, PhD,2 Anders Svenningsson, MD, PhD,3

and Jan Lycke, MD, PhD2

Objective: The impact of present disease-modifying treatments (DMTs) in multiple sclerosis (MS) on nerve injury andreactive astrogliosis is still unclear. Therefore, we studied the effect of natalizumab treatment on the release of 2brain-specific tissue damage markers into cerebrospinal fluid (CSF) in MS patients.Methods: CSF samples from 92 patients with relapsing forms of MS were collected in a prospective manner prior tonatalizumab treatment and after 6 or 12 months. In 86 cases, natalizumab was used as second-line DMT due tobreakthrough of disease activity. The levels of neurofilament light (NFL) and glial fibrillary acidic protein (GFAP) weredetermined using highly sensitive in-house developed enzyme-linked immunosorbent assays.Results: Natalizumab treatment led to a 3-fold reduction of NFL levels, from a mean value of 1,300 (standarddeviation [SD], 2,200) to 400 (SD, 270) ng/l (p < 0.001). The later value was not significantly different from thatfound in healthy control subjects (350ng/l; SD, 170; n ¼ 28). Subgroup analysis revealed a consistent effect on NFLrelease, regardless of previous DMT or whether patients had relapses or were in remission within 3 months prior tonatalizumab treatment. No differences between pre- and post-treatment levels of GFAP were detected.Interpretation: Our data demonstrate that natalizumab treatment reduces the accumulation of nerve injury inrelapsing forms of MS. It is anticipated that highly effective anti-inflammatory treatment can reduce axonal loss,thereby preventing development of permanent neurological disability.

ANN NEUROL 2011;69:83–89

Multiple sclerosis (MS) is a chronic inflammatory

demyelinating disease of the central nervous system

(CNS). Axonal damage occurs both within MS lesions and

in the normal-appearing white matter already at early

stages of the disease, as demonstrated by neuropathological

studies,1 magnetic resonance imaging (MRI) techniques,2–4

and by use of biomarkers.5–9 Numerous findings suggest

that axonal loss followed by astrogliosis is ultimately re-

sponsible for the development of irreversible neurological

deficits.8–18 The role of immune system-mediated mecha-

nisms in causing such processes, however, has not yet been

clarified. Disease-modifying treatments (DMTs) in MS

have so far been designed to modulate the immune system

and are almost exclusively effective in disease phases in

which the inflammatory component is prominent. Conse-

quently, DMTs in current use reduce the number of relap-

ses and appearance of T2 as well as gadolinium-enhanced

MRI lesions in relapsing-remitting MS (RRMS) patients.

Natalizumab, a monoclonal antibody preventing lympho-

cyte migration across the blood-brain barrier, is a highly

efficient agent in this respect.19–23 Whether early immuno-

modulatory intervention in fact retains axonal integrity and

View this article online at wileyonlinelibrary.com. DOI: 10.1002/ana.22247

Received May 13, 2010, and in revised form Jul 26, 2010. Accepted for publication Aug 20, 2010.

Address correspondence to Dr Svenningsson, Department of Neurology, Norrlands University Hospital, S-901 85 Umea, Sweden.

E-mail: anders.svenningsson@me.com

From the 1Department of Neurology, Orebro University Hospital, Orebro; 2Department of Neurology, Sahlgrenska University Hospital, Goteborg;3Department of Neurology, Norrlands University Hospital, Umea; 4Department of Neurology and Neurophysiology, Linkoping University Hospital,

Linkoping; 5Clinical Immunology Unit, Department of Clinical and Experimental Medicine, Linkoping University, Linkoping; 6Neuroimmunology Unit, Center

for Molecular Medicine, Department of Clinical Neuroscience, Karolinska Institute, Stockholm; and 7UmanDiagnostics, Umea, Sweden.

VC 2011 American Neurological Association 83

prevents astrogliosis deserves further investigation. Detec-

tion of axonal and astroglial cytoskeletal proteins, such as

neurofilament light (NFL) and glial fibrillary acidic protein

(GFAP), in the cerebrospinal fluid (CSF) reflects these

processes in vivo.8,9 To study the effects of intense immune

modulation within the CNS on axonal damage and astro-

gliosis, we quantified NFL and GFAP in the CSF from 92

MS patients undergoing 6 to 12 months of treatment with

natalizumab.

Subjects and Methods

Patients and CSF CollectionFollowing acquisition of informed consent, 92 MS

patients (83 RRMS and 9 secondary progressive MS

with relapses [SPRMS]) were included consecutively

using a multicenter approach. Patients fulfilled the re-

vised McDonald criteria24 and were scheduled to start

treatment with 300mg natalizumab intravenously once

monthly according to Swedish guidelines. All patients

presented either a highly active disease course de novo or

breakthrough disease activity in terms of relapses in the

presence of other DMTs. The mean annual relapse rate

was 1.2 (standard deviation [SD], 1.1) during the year

prior to natalizumab treatment. Among the 92 patients

included, 6 were treatment naive, whereas 86 had been

using other DMTs, of which 67 were treated with inter-

feron-b, 4 with glatiramer acetate, 8 with mitoxantrone,

and 7 with other treatments. Thirty patients had experi-

enced a relapse within 3 months prior to natalizumab

treatment, whereas 62 patients were in remission during

this period. Patient demographics are shown in Table 1.

Medical history and additional clinical data such as

relapse count during the study period were recorded pro-

spectively in the medical record and in the Swedish MS

registry. CSF was collected consecutively by lumbar spi-

nal taps prior to treatment with natalizumab and after 6

(n ¼ 8) or 12 (n ¼ 84) months. Clinical examination

using the Expanded Disability Status Scale (EDSS)25 was

performed by trained neurologists no more than 1 week

apart from the lumbar puncture. Ten milliliters of CSF

was centrifuged at 1,300 to 1,800g for 10 minutes. Ali-

quots of supernatant (0.5–1.0ml) were immediately snap-

frozen to �80�C and not thawed until analysis. All

patients underwent routine CSF examinations including

cell counts, and protein electrophoresis and determina-

tion of immunoglobulin (Ig)G-index were performed in

59 patients. CSF samples from 28 healthy volunteers,

recruited among blood donors and university students,

were used as reference.

All NFL and GFAP analysis were performed simul-

taneously at the Section of Neurochemistry, Sahlgrenska

University Hospital, Molndal. CSF samples were blinded

for laboratory personnel during all stages of analysis. The

study was approved by the local ethics committee at

Uppsala University, Sweden.

NFL Enzyme-Linked ImmunoassayThe UmanDiagnostics NF-light enzyme-linked immuno-

assay (ELISA), developed by Norgren and coworkers,

detects 1 of the low molecular weight chains of neuronal

intracellular intermediate filaments.9 The assay is based

on 2 highly specific monoclonal antibodies and a biotin-

streptavidin horseradish peroxidase (HRP) system. Analy-

sis was performed at room temperature. Cerebrospinal

fluid samples were diluted 1:1 with sample dilution

buffer to a total volume of 100ll and incubated with

agitation (800rpm) for 1 hour in precoated anti-NFL

ELISA plates. Thereafter, a 100ll solution of tracer anti-

body (biotin anti-NFL) was added to each well and incu-

bated for 45 minutes. Washing cycles were performed af-

ter all incubations. Detection was performed using 100llstreptavidin-HRP incubated for 30 minutes, followed by

another incubation with 100ll 3,30,5,50-tetramethylben-

zidine for 15 minutes;. A volume of 50ll stop solution

(8% v/v sulphuric acid) was added to each well, and ab-

sorbance was read at k490nm. The sensitivity of the

NFL assay was 31ng/l.

GFAP ELISAMeasurements of GFAP were performed using the ELISA

procedure previously developed by Rosengren and co-

workers.17 In brief, ELISA plates were coated with hen

TABLE 1: Demographics of Multiple SclerosisPatients and HCs

Characteristic MS, n 5 92 HCs,n 5 28

Relapse,n 5 30a

Remission,n 5 62b

Gender,F/M

20/10 34/28 8/20

Age, yr 36 (19–59),p ¼ 0.020vs HCc

38 (14–56),p ¼ 0.034vs HCc

43(27–62)

MSduration,yr

9.0 (0.5–28) 10 (2–26) N/A

Values for age and MS duration are given as mean (range).aDenotes MS patients with relapses within 3 months priorto natalizumab treatment onset.bDenotes MS patients in remission during this period.ct test.MS ¼ multiple sclerosis; HC ¼ healthy control; N/A ¼not applicable.

ANNALS of Neurology

84 Volume 69, No. 1

anti-GFAP IgG followed by incubation with CSF sam-

ples at room temperature for 2 hours. Rabbit anti-GFAP

IgG was added as secondary antibody and incubated for

1 hour at room temperature. Washing cycles were per-

formed after each incubation step. Detection was per-

formed with peroxidase-conjugated donkey antirabbit

IgG and 3,30,5,50-tetramethylbenzidine substrate. Absorb-

ance was read at k490 nm, and the sensitivity of the

GFAP assay was 16ng/l.

Statistical AnalysisWilcoxon signed ranks test and Mann-Whitney U test

were used for analysis of nonparametric data. Analysis of

parametric data, paired and unpaired, was performed using

t test. Test for age influence on the NFL levels in MS

patients as compared to healthy controls was done via lin-

ear regression analysis. Statistical calculations were per-

formed in SPSS 16.0 software (SPSS Inc., Chicago, IL).

Results

Clinical Data and CSF Inflammatory ParametersThe annual relapse rate during the natalizumab treatment

period was reduced from a mean value of 1.2 (SD, 1.1;

year prior to treatment) to 0.15 (SD, 0.36; p < 0.001).

The median pretreatment EDSS score was 4.0 (2.5–5.5,

25th–75th percentile) as compared to 3.5 (2.0–5.0,

25th–75th percentile) post-treatment (p < 0.001). More-

over, CSF inflammatory parameters were significantly

decreased in terms of cell count from a mean of 5.4

(SD, 6.9) to 1.6 (SD, 1.3) � 106 cells/l (p < 0.001) and

IgG-index from a mean of 1.1 (SD, 0.60) to 0.94 (SD,

FIGURE 1: Neurofilament light in cerebrospinal fluid (CSF)following natalizumab treatment. Intravenous infusions of300mg natalizumab were performed monthly in 92 multiplesclerosis patients. Neurofilament light (NFL) levels in theCSF were compared to levels obtained in 28 healthyindividuals. Pretreatment mean value was significantlydifferent from that obtained after treatment (p < 0.001) andin healthy controls (p < 0.001). No significant difference wasfound between mean values after treatment and in healthycontrols, respectively. SE 5 standard error.

TABLE 2: Clinical and Cerebrospinal Fluid Parameters following Natalizumab Treatment

Parameter Relapse, n 530a Remission, n 5 62b Statistics

Pretreatment After 6–12MonthsTreatment

Pretreatment After6–12 MonthsTreatment

Relapses/yr 2.0 (1.1)c 0.10 (0.31)c 0.84 (0.83)c 0.18 (0.39)c p < 0.001c

p < 0.001c

EDSS 3.5 (2.5–4.5)d 2.75(1.375–4.0)d

4.0 (2.25–6.0)d 4.0(2.375–5.625)d

p < 0.001d

p ¼ 0.023d

Cell count, 106 cells/l 5.8 (6.0)c 1.4 (1.2)c 5.2 (7.3)c 1.6 (1.4)c p < 0.001c

p < 0.001c

IgG index 0.93 (0.32)c,e 0.83 (0.29)c,e 1.1 (0.68)c,f 1.0 (0.56)c,f p ¼ 0.033c

p ¼ 0.001c

NFL, ng/l 2300 (3600)c 350 (170)c 860 (780)c 430 (310)c p < 0.005c

p < 0.001c

Values are given as mean (standard deviation) except for EDSS, which is presented as median (25th–75th percentile).aDenotes MS patients with relapses within 3 months prior to natalizumab treatment.bDenotes MS patients in remission during this period.cParied sample t test.dMann-Whitney U test.en¼ 17.fn ¼ 42.EDSS ¼ Expanded Disability Status Scale; IgG ¼ immunoglobulin G; NFL ¼ neurofilament light; MS ¼ multiple sclerosis.

Gunnarsson et al: Relapsing MS and Natalizumab

January 2011 85

0.50; p < 0.001). Stratification of the cohort into

patients with relapses within 3 months prior to natalizu-

mab treatment and patients in remission during this pe-

riod is shown in Table 2.

NFLPrior to natalizumab treatment, the mean NFL concen-

tration in CSF from the 92 MS patients studied was

1,300ng/l (SD, 2200), whereas healthy controls displayed

a mean value of 350ng/l (SD, 170; p < 0.001; Fig 1).

After treatment, however, the mean NFL level was signif-

icantly reduced to 400 (SD, 270) ng/l (p < 0.001).

Notably, the post-treatment value did not differ signifi-

cantly from that found in healthy control subjects.

We have previously shown that NFL levels are

increased in the 3-month period after a relapse.7–9 There-

fore, we made a separate analysis of the groups of

patients having a relapse versus being in remission during

this time period before lumbar puncture. Thus, the

mean NFL concentration in CSF was determined as

2,300ng/l (SD, 3,600) in MS patients with relapses (n ¼30), as compared to 860ng/l (SD, 780) in patients dis-

playing clinical remission for at least 3 months (n ¼ 62)

(p < 0.038; equal variance not assumed). When exclu-

sively analyzing MS patients in remission, NFL levels

were still significantly reduced following natalizumab

treatment (p < 0.001; see Table 2). Both patients with

and those without ongoing relapses during the 3-month

period prior to natalizumab treatment displayed post-

treatment NFL levels not significantly different from

those obtained in healthy controls (mean, 350ng/l; SD,

170; see Table 2).

Pretreatment levels of NFL in treatment-naive MS

patients (n ¼ 6) were higher (3,300ng/l; SD, 6,600) but

not significantly different from those obtained in patients

subjected to DMTs prior to natalizumab (1,200ng/l; SD,

1,600; n ¼ 86). An adequate comparison regarding dif-

ferent types of immune modulation was not possible due

to different treatment regimes and uneven patient distri-

bution. After being on natalizumab treatment for 6 to 12

months, both treatment-naive MS patients and those

who had previously undergone immune modulation pre-

sented NFL levels similar to values obtained in healthy

controls (data not shown).

Subgroup analysis was also performed regarding

NFL levels in patients with SPRMS (n ¼ 9). These

patients presented a significantly higher mean post-treat-

ment NFL concentration (630ng/l; SD, 420) as com-

pared to the 83 RRMS patients (380ng/l; SD, 240; p <

0.001) and healthy controls (350ng/l; SD, 170; p <

0.001; Fig 2). Pretreatment NFL levels, however, were

not higher in the SPMRS group as compared to RRMS

patients. The SPRMS patients still presented a highly sig-

nificant decrease in mean NFL concentration following

natalizumab treatment (p < 0.001). In terms of neuro-

logical deficits, the SPRMS patients had a median EDSS

score of 6.5 (6.0–7.25, 25th–75th percentile), whereas

the corresponding value in RRMS patients was 3.5

(2.375–5.0, 25th–75th percentile; p < 0.001). Disease

duration and age distribution were similar (data not

shown).

Because data in the control group indicated an age-

dependent variance in NFL concentration (data not

shown), we tested for a significant contribution of this

parameter in comparisons between values from controls

and MS patients after natalizumab treatment. Using

covariate analysis, we found no significant age-dependent

influence on the results presented (data not shown).

GFAPIn contrast to NFL, determination of GFAP levels in

CSF prior to natalizumab treatment of the 92 MS

patients did not yield any significant difference as com-

pared to values obtained in healthy controls (data not

shown). Subgroup analysis of the 9 SPRMS patients with

relapses, however, revealed a mean pretreatment GFAP

concentration that was significantly higher than that

found in the control group (790 [SD, 430] vs 400 [SD,

162] ng/l; p < 0.001; Fig 3). On the other hand, GFAP

FIGURE 2: Neurofilament light (NFL) in cerebrospinal fluidin different disease courses of multiple sclerosis (MS).Before intravenous treatment with 300mg natalizumabmonthly, mean NFL concentrations in secondary progressiveMS with relapses (SPRMS) and relapsing-remitting MS(RRMS) patients were not significantly different. In bothgroups of patients, values obtained after treatment weresignificantly decreased (p < 0.001). After treatment, themean NFL level in SPRMS was significantly higher than thatdetermined in RRMS patients (p < 0.001) and healthycontrols (p < 0.001). SE 5 standard error.

ANNALS of Neurology

86 Volume 69, No. 1

levels were similar in healthy controls and RRMS

patients (mean, 510ng/l; SD, 460; n ¼ 83).

Natalizumab treatment for 6 to 12 months had no

significant influence on the GFAP levels in CSF, neither

in the entire 92 MS patient population nor in cases of

subgroup analysis (see Fig 3).

Discussion

Irreversible neurological deficits in MS likely emerge as a

consequence of accumulating nerve injury starting al-

ready in early phases of the disease.8–18 Due to the long

lag phase before clinical disability can be detected and a

lack of reliable biomarkers, the effect of current immuno-

modulatory treatments on tissue damage has been diffi-

cult to assess. Previously, we have characterized NFL as a

marker of ongoing axonal damage.7–9 In this study, we

demonstrate that elevated NFL levels in CSF from 92

patients with relapsing forms of MS are markedly

decreased following natalizumab treatment for 6 to 12

months. Post-treatment NFL levels were not significantly

different from values obtained in healthy individuals.

Moreover, the effect was consistent regardless of whether

the patients were treated with first-line DMT at the time

of natalizumab treatment onset or not.

The natalizumab mechanism of action and its CNS

anti-inflammatory potential has been well characterized.

In the pivotal natalizumab trial, a highly significant

reduction in relapse rate and MRI lesion formation was

obtained.19 These results have been confirmed in open

follow-up studies of patients with treatment failure on

first-line DMTs and in highly active MS.20,21 The anti-

inflammatory effect is likely to be achieved by interfer-

ence with lymphocyte migration into the CNS, reflected

among others by a reduced expression of proinflamma-

tory cytokines and chemokines in the CSF compart-

ment.26–28 In the present investigation, both clinical pa-

rameters, that is, relapse rate and EDSS score, and CSF

inflammatory markers in terms of inflammatory cell

counts in CSF and IgG-index were decreased following

natalizumab treatment. These data are well in line with

the results presented in the pivotal natalizumab trial and

several other studies.19–23 Therefore, it is of certain inter-

est that we concomitantly demonstrate a normalization

of NFL levels in CSF from this cohort of MS patients.

In contrast to the marked effect on NFL release, no

changes in astrogliosis, as determined by GFAP quantifi-

cation, were seen following natalizumab treatment. This

may be due to the characteristics of the patient cohort,

because earlier data suggest that GFAP is elevated in sec-

ondary progressive but not RRMS.8,9,12,17 Such a notion

is supported by the fact that increased GFAP levels, as

compared to healthy controls, were found exclusively in

patients presenting an underlying progressive disease

course and more pronounced neurological deficits in

terms of EDSS score.

Accurate identification of MS patients with insuffi-

cient treatment effect and at high risk for development of

future neurological disability is becoming increasingly im-

portant. Immune modulation in MS partly prevents devel-

opment of sustained disability if initiated during the relaps-

ing phase of the disease, but the evidence for long-term

treatment efficacy has not yet been clarified. Furthermore,

the risk-benefit profiles of current DMTs differ consider-

ably, and difficulties in accurate assessment of treatment ef-

ficacy at the individual level in a short time perspective

present a clinical dilemma. Except for MRI parameters

reflecting CNS inflammation, there are presently no surro-

gate markers of disease activity in practical use. The Uman-

Diagnostics NF-light ELISA is based on 2 monoclonal

antibodies highly specific and sensitive for detection of

NFL in body fluids.9 By use of this assay and other immu-

nochemical setups, it has been shown that neurofilament

levels are increased in CSF during MS exacerbations.7–9,29

Moreover, high NFL levels determined at diagnostic lum-

bar punctures, especially in the presence of relapses, corre-

late with a worse prognosis as determined using EDSS and

multiple sclerosis severity score (MSSS) at clinical follow-

ups after 5 and 14 years, respectively.9,18 Notably, we have

earlier demonstrated significantly higher NFL levels also in

phases of clinical remission as compared to control sub-

jects.8,9 This likely reflects an ongoing clinically silent

FIGURE 3: Glial fibrillary acidic protein (GFAP) incerebrospinal fluid in different disease courses of multiplesclerosis (MS). The mean pre-treatment GFAP concentrationwas significantly higher in secondary progressive MS withrelapses (SPRMS) but not relapsing-remitting MS (RRMS)patients as compared to healthy controls (p < 0.001). Nosignificant differences were found after intravenoustreatment with 300 mg natalizumab monthly. SE 5 standarderror.

Gunnarsson et al: Relapsing MS and Natalizumab

January 2011 87

axonal damage that, according to data presented here, is

amendable to more intense immunomodulatory treatment.

NFL levels in this study were considerably increased, in both

the presence and absence of relapses. Furthermore, treat-

ment-naive patients as well as patients previously subjected

to first-line DMTs displayed increased NFL levels in CSF.

These results further strengthen the rationale for NFL deter-

mination as a measure of subclinical disease activity related

to ongoing nerve injury. We believe that our data also sup-

port NFL quantification in CSF as a putative outcome pa-

rameter in clinical trials of new drugs in MS. GFAP on the

other hand, does not seem to be a useful biomarker in this

aspect, at least not in relapsing-remitting phases of the dis-

ease. Further studies focused on correlations between levels

of NFL in CSF and atrophy measurements by emerging

MRI techniques may provide valuable additional data.

In conclusion, our data support the notion that

inflammatory activity in RRMS is associated with axonal

damage. Effective immunomodulatory treatment is likely

to be of utmost importance in preventing axonal loss

and development of irreversible CNS injury. It is antici-

pated that highly active control of CNS inflammation in

MS may have profound effects on the long-term course

of the disease, although this remains to be shown in

future studies. In this context, analysis of NFL levels in

the CSF have the potential to provide vital information

regarding treatment efficacy in the short-term perspective

and to facilitate treatment decisions.

Acknowledgment

Research performed in this project was supported by

grants from Orebro Society of Medicine (M.G.), Swedish

Society of the Neurologically Disabled (J.L.), Gothen-

burg Multiple Sclerosis Society (J.L.), Edith Jacobsson

Foundation (J.L.), Swedish Reasearch Council (T.O.),

Bibbi and Niels Jensen Foundation (T.O.), Soderbergs

Foundation (T.O.), Montell Williams Foundation, EU

fp6 (T.O.), Neuropromise (LSHM-CT-2005-018637)

(T.O.), and Biogen Idec (J.L.).

We thank S. Fridlund for skillful technical assistance.

Authorship

M.G. and C.M. contributed equally.

Potential Conflicts of Interest

M.G. has served on educational boards for Biogen Idec,

BayerScheringPharma, and Merck; has received speakers

honoraria from Biogen Idec, Merck, BayerScheringPharma,

and SanofiAventis; has received payment for development of

educational presentations from Biogen Idec; and has had

travel expenses reimbursed by Biogen Idec and Merck. C.M.

has received lecture honoraria from Biogen, and has had

travel expenses partially reimbursed by Biogen Idec. M.A. has

had travel expenses reimbursed by Biogen Idec and

MerckSerono. P.S. has served on an advisory board for

Novartis; has received lecture fees from Biogen Idec; and has

had travel expenses reimbursed by Biogen Idec. C.D. has

received personal compensation for activities with Biogen

Idec, Merck, BayerScheringPharma, and SanofiAventis as a

speaker and consultant. M.V. has a grant pending from

Biogen Idec and has received lecture honoraria from Biogen

Idec and Merck Serono. F.P. has a grant pending from Biogen

Idec; has received payment for development of educational

presentations from Biogen Idec, Novartis, and MerckSerono;

and has had travel expenses reimbursed by Biogen Idec and

Novartis. N.N. is employed by UmanDiagnostics AB.A.S.

has served on an advisory board for MerckSerono; has

received lecture honoraria from Biogen Idec and Baxter; and

has had travel expenses partially reimbursed by Biogen Idec,

MerckSerono, and Baxter. J.L. has served on advisory boards

for Biogen Idec and Merck Serono; has a grant pending from

Biogen Idec; has received speakers honoraria from Biogen

Idec and Merck Serono; and has had travel expenses

reimbursed by Biogen Idec. T.O. has received consultancy

fees from Biogen Idec, SanofiAventis, Merck, and Novartis;

and has grants pending from Biogen Idec, Merck,

SanofiAventis, and Bayer.

References1. Trapp BD, Peterson J, Ransohoff RM, et al. Axonal transection in

the lesions of multiple sclerosis. N Engl J Med 1998;338:278–285.

2. Pascual AM, Martinez-Bisbal MC, Bosca I, et al. Axonal loss is pro-gressive and partly dissociated from lesion load in early multiplesclerosis. Neurology 2007;69:63–67.

3. Filippi M, Bozzali M, Rovaris M, et al. Evidence for widespreadaxonal damage at the earliest clinical stage of multiple sclerosis.Brain 2003;126:433–437.

4. De Stefano N, Narayanan S, Francis SJ, et al. Diffuse axonaland tissue injury in patients with multiple sclerosis with low cere-bral lesion load and no disability. Arch Neurol 2002;59:1565–1571.

5. Teunissen CE, Iacobaeus E, Khademi M, et al. Combination ofCSF N-acetylaspartate and neurofilaments in multiple sclerosis.Neurology 2009;72:1322–1329.

6. Brettschneider J, Petzold A, Junker A, Tumani H. Axonal damagemarkers in the cerebrospinal fluid of patients with clinically iso-lated syndrome improve predicting conversion to definite multiplesclerosis. Mult Scler 2006;12:143–148.

7. Lycke JN, Karlsson JE, Andersen O, Rosengren LE. Neurofilamentprotein in cerebrospinal fluid: a potential marker of activity in mul-tiple sclerosis. J Neurol Neurosurg Psychiatry 1998;64:402–404.

8. Malmestrom C, Haghighi S, Rosengren L, et al. Neurofilamentlight protein and glial fibrillary acidic protein as biological markersin MS. Neurology 2003;61:1720–1725.

9. Norgren N, Sundstrom P, Svenningsson A, et al. Neurofilamentand glial fibrillary acidic protein in multiple sclerosis. Neurology2004;63:1586–1590.

ANNALS of Neurology

88 Volume 69, No. 1

10. Lim ET, Sellebjerg F, Jensen CV, et al. Acute axonal damage pre-dicts clinical outcome in patients with multiple sclerosis. Mult Scler2005;11:532–536.

11. Petzold A, Eikelenboom MJ, Keir G, et al. Axonal damage accu-mulates in the progressive phase of multiple sclerosis: three yearfollow up study. J Neurol Neurosurg Psychiatry 2005;76:206–211.

12. Petzold A, Eikelenboom MJ, Gveric D, et al. Markers for differentglial cell responses in multiple sclerosis: clinical and pathologicalcorrelations. Brain 2002;125:1462–1473.

13. Semra YK, Seidi OA, Sharief MK. Heightened intrathecal releaseof axonal cytoskeletal proteins in multiple sclerosis is associatedwith progressive disease and clinical disability. J Neuroimmunol2002;122:132–139.

14. De Stefano N, Narayanan S, Francis GS, et al. Evidence of axonaldamage in the early stages of multiple sclerosis and its relevanceto disability. Arch Neurol 2001;58:65–70.

15. Bjartmar C, Kidd G, Mork S, et al. Neurological disability corre-lates with spinal cord axonal loss and reduced N-acetyl aspartatein chronic multiple sclerosis patients. Ann Neurol 2000;48:893–901.

16. De Stefano N, Matthews PM, Fu L, et al. Axonal damage corre-lates with disability in patients with relapsing-remitting multiplesclerosis. Results of a longitudinal magnetic resonance spectros-copy study. Brain 1998;121(pt 8):1469–1477.

17. Rosengren LE, Lycke J, Andersen O. Glial fibrillary acidic proteinin CSF of multiple sclerosis patients: relation to neurological defi-cit. J Neurol Sci 1995;133:61–65.

18. Salzer J, Svenningsson A, Sundstrom P. Neurofilament light as aprognostic marker in multiple sclerosis. Mult Scler 2010;16:287–292.

19. Polman CH, O’Connor PW, Havrdova E, et al. A randomized, pla-cebo-controlled trial of natalizumab for relapsing multiple sclero-sis. N Engl J Med 2006;354:899–910.

20. Putzki N, Yaldizli O, Buhler R, et al. Natalizumab reduces clinicaland MRI activity in multiple sclerosis patients with high disease ac-tivity: results from a multicenter study in Switzerland. Eur Neurol2010;63:101–106.

21. Putzki N, Kollia K, Woods S, et al. Natalizumab is effective as sec-ond line therapy in the treatment of relapsing remitting multiplesclerosis. Eur J Neurol 2009;16:424–426.

22. Havrdova E, Galetta S, Hutchinson M, et al. Effect of natalizumabon clinical and radiological disease activity in multiple sclerosis: aretrospective analysis of the Natalizumab Safety and Efficacy inRelapsing-Remitting Multiple Sclerosis (AFFIRM) study. LancetNeurol 2009;8:254–260.

23. Miller DH, Soon D, Fernando KT, et al. MRI outcomes in a pla-cebo-controlled trial of natalizumab in relapsing MS. Neurology2007;68:1390–1401.

24. Polman CH, Reingold SC, Edan G, et al. Diagnostic criteria formultiple sclerosis: 2005 revisions to the ‘‘McDonald Criteria.’’ AnnNeurol 2005;58:840–846.

25. Kurtzke JF. Rating neurologic impairment in multiple sclerosis: anexpanded disability status scale (EDSS). Neurology 1983;33:1444–1452.

26. Mellergard J, Edstrom M, Vrethem M, et al. Natalizumab treat-ment in multiple sclerosis: marked decline of chemokines andcytokines in cerebrospinal fluid. Mult Scler 2010;16:208–217.

27. Sellebjerg F, Bornsen L, Khademi M, et al. Increased cerebrospi-nal fluid concentrations of the chemokine CXCL13 in active MS.Neurology 2009;73:2003–2010.

28. Khademi M, Bornsen L, Rafatnia F, et al. The effects of natalizu-mab on inflammatory mediators in multiple sclerosis: prospectsfor treatment-sensitive biomarkers. Eur J Neurol 2009;16:528–536.

29. Rejdak K, Petzold A, Stelmasiak Z, Giovannoni G. Cerebrospinalfluid brain specific proteins in relation to nitric oxide metabolitesduring relapse of multiple sclerosis. Mult Scler 2008;14:59–66.

Gunnarsson et al: Relapsing MS and Natalizumab

January 2011 89