The contribution of demyelination to axonal loss in multiple sclerosis K. Williams,

doi:10.1093/brain/awl074
Brain (2006), 129, 1507–1516
The contribution of demyelination to axonal
loss in multiple sclerosis
G. C. DeLuca,1 K. Williams,1 N. Evangelou,3 G. C. Ebers1 and M. M. Esiri1,2
1
Department of Clinical Neurology, University of Oxford, 2Department of Neuropathology, Oxford Radcliffe NHS Trust,
Oxford and 3Department of Neurology, Queen’s Medical Centre, University Hospital NHS Trust, Nottingham, UK
Correspondence to: Prof. M. M. Esiri, Neuropathology Department, Radcliffe Infirmary, Oxford OX2 6HE, UK
E-mail: [email protected]
Keywords: demyelination; axon; axonal degeneration; multiple sclerosis; neuropathology
Abbreviation: NAWM = normal appearing white matter
Received September 19, 2005. Revised February 24, 2006. Accepted March 3, 2006. Advance Access publication April 5, 2006
Introduction
Multiple sclerosis has long been classified as an inflammatory
demyelinating disease, associated with the formation of discrete or confluent foci of myelin loss and relative preservation
of axons. It is unambiguous that plaque accumulation and
progression of disability are temporally linked. However, the
belief that the acquisition of progressive disability in multiple
sclerosis results from the cumulative effect of plaques is an
assumption for which there has been little direct evidence.
The rediscovery of axonal damage and loss in plaques has
sharpened the focus on these demyelinated regions (Ferguson
et al., 1997; Trapp et al., 1998; Lovas et al., 2000). The plaquecentred view of disease progression has come under scrutiny,
as the traditional demyelinating paradigm in which plaques
determine outcome is challenged by a substantial number of
clinical observations.
A number of the clinical features of multiple sclerosis
may be considered to challenge the concept that
myelin-related pathology accounts for progressive disability.
These include the following: (i) individual plaques as
#
measured by relapses are not crucial to long-term clinical
outcome (Kremenchutzky et al., 2006; Confavreux et al.,
2000); (ii) the location of the first plaque and the extent
to which a patient recovers from that first plaque have no
influence on long-term outcome (Confavreux et al., 2003;
Kremenchutzky et al., 2006); (iii) only early relapse frequency
has been shown to associate with outcome (Weinshenker
et al., 1989b), and although relapses impact time to onset
of progression, they do not have an appreciable effect on the
rate of progression once a progressive phase begins
(Kremenchutzky et al., 1999) or a certain level of disability
is reached (Confavreux et al., 2000; Pittock et al., 2004); (iv)
treatment with interferon suppresses plaque load (T2 lesion
burden) without affecting long-term MRI measured atrophy
(Gasperini et al., 2002; Horsfield and Filippi, 2003; Lin et al.,
2003) and impact on disability remains uncertain.
From a clinical perspective, it could be argued that the role
of demyelination in axonal loss is most relevantly examined
in the corticospinal tracts. These are typically the first to
The Author (2006). Published by Oxford University Press on behalf of the Guarantors of Brain. All rights reserved. For Permissions, please email: [email protected]
Downloaded from by guest on October 28, 2014
The traditional notion that multiple sclerosis is a primary demyelinating disease has led to a plaque-centred view
of both aetiology and the pathogenesis of disease progression. The presence of axonal loss has received increasing recognition. However, the relative roles of demyelination and axonal loss have not been fully clarified in
multiple sclerosis nor have their possible interrelationships been elucidated. Post-mortem material from the
cerebrum, brainstem and spinal cord of 55 multiple sclerosis patients (29 males) with an age range of 25–83 years
(mean = 57.5 years) and length of disease history ranging from 2 to 43 years (mean = 17.1 years) was stained
for myelin. Plaque load was calculated by summing the relative proportion of plaque area compared with total
white matter area of the corticospinal and sensory tracts at each level. This was related to estimates of axonal
density and of total axon number in these tracts in the spinal cord. Our results indicate that plaque load did not
correlate with brain weight. Unexpectedly, after adjusting for sex, age and duration of disease, correlations
between total plaque load and axonal loss in both the corticospinal tract and sensory tracts were weak or absent
at each level investigated. Since there was little correlation between plaque load and axonal loss, the possibility
that demyelination is not the primary determinant of spinal cord axonal loss warrants consideration.
1508
Brain (2006), 129, 1507–1516
Pathological preparation
For each of the multiple sclerosis cases, formalin-fixed paraffinembedded sections were taken from the cerebrum (coronal section
at internal capsule including the posterior limb), brainstem (transverse sections from each of the mid-brain, mid-pons and midmedulla) and spinal cord (transverse sections at the high and low
cervical, high and low thoracic, and lumbar levels). Adjacent myelinstained and axon-stained sections from each level were obtained
using the Luxol fast blue cresyl violet and Palmgren silver stains,
respectively. Plaque load measures were related to estimates of
axonal density and total axon numbers reported in our previous
study, reproduced with permission in Fig. 1 (DeLuca et al., 2004a).
To assess more thoroughly the extent of plaque load impinging on
the corticospinal and sensory tracts throughout the length of the
spinal cord, spinal cords from a subset of 10 cases were sectioned at
5 mm intervals from cervical to lumbar cord levels. The subset of
spinal cords comprised five cases, each lying at the opposite ends of
initial total plaque load estimates obtained from the nine levels as
previously outlined. A total of 1355 discrete spinal cord sections
from these cases were stained for myelin so that total plaque load
measures could be obtained and subsequently related to estimates of
axonal density and total axon numbers as described above.
Quantification of axonal loss
Sections stained with Palmgren silver were used to demonstrate
axons. The axons in the corticospinal tract and sensory tract of
each side were examined via microscopy (·400). Five fields from
each side of the corticospinal tracts and six fields from the sensory
tracts on each side were digitized and then transferred to the NIH
Image software program, where the fibre counts were performed
automatically after setting of a threshold. To ensure that the oedema
and high cellular infiltration of active lesions would not confound
axonal counts, measures of axonal density were obtained in normal
appearing white matter (NAWM).
To validate our automated image analysis procedure, several
different automated settings (i.e. filters, thresholds, etc.) were
applied to the images previously counted manually by two observers
so that the axon numbers for each method could be compared. A
threshold was established that gave a strong correlation between the
final automated counts and manual counts (r = 0.956, P < 0.001).
To determine the distribution of axonal sizes observed in each of
the tracts, histograms were constructed. The distribution of axonal
sizes was used to categorize axonal fibres as either large or small
using a size range of >3 mm2 for the large fibres and 3 mm2 for the
small ones.
In order to obtain an estimate of the total number of axons in the
corticospinal and sensory tracts, we took the product of the axonal
density measures (axons/mm2) and cross-sectional areas (mm2) of
each respective tract.
Material and methods
Clinical material
Quantification of plaque load
Human autopsy material of 55 pathologically confirmed cases of
multiple sclerosis (29 males and 26 females) with an age range of
25–83 years (mean = 57.5 years) was studied. The length of disease
history ranged from 2 to 43 years (mean = 17.1 years). The postmortem material was derived from the autopsy brain and spinal cord
archive from the Neuropathology Department, Oxford Radcliffe
NHS Trust, and was obtained with consent from the next-of-kin
for use of tissue from research. The study was approved by the local
research ethics committee.
To quantify the extent of demyelination affecting the corticospinal
and sensory tracts in the multiple sclerosis cases, the cross-sectional
area of the multiple sclerosis lesions impinging on each tract were
identified and then measured. For the corticospinal tracts, plaque
load was quantified for cerebral and brainstem sections by a pointcounting method (Fig. 2). Plaques affecting the corticospinal and
sensory tracts in the spinal cord were quantified using a computerized image analysis system running NIH image software. Measures
of relative plaque load were obtained by estimating the area of
Downloaded from by guest on October 28, 2014
manifest the onset of the chronic progressive phase and they
play the largest role in disability as measured by the
ambulation-dependent Extended Disability Status Scale
(EDSS). Furthermore, they are discrete at several levels of
the neuraxis, allowing serial anatomical measurement in a
centripetal fashion. It is possible that the long processes
are selectively vulnerable by virtue of being more likely to
be affected by plaques than are shorter ones. The corticospinal axons, owing to their extraordinary length, are potentially vulnerable to degenerative processes. However, there
are undoubtedly many other influences on the topography
of axonal degeneration. It is not known why the combined
degeneration of both the corticospinal and sensory tracts is a
common pattern in several metabolic and degenerative
disorders such as sub-acute combined degeneration of the
spinal cord, Friedreich’s ataxia and hereditary spastic paraplegia (HSP) (Behan and Maia, 1974; Said et al., 1986;
Jitpimolmard et al., 1993; Hemmer et al., 1998). Even
when a gene mutation is known and its expression is characterized, the site-specific axonal degeneration observed in
these conditions cannot be easily explained. Nevertheless,
selective tract vulnerability appears to be a common feature
of processes that might otherwise be considered to be generalized or diffuse.
We have previously shown that there is a symmetric, sizedependent selective loss of axons in the corticospinal and
posterior column tracts in multiple sclerosis, a pattern implying differential sensitivity of the axons within these tracts.
Accordingly, it was anticipated that axonal loss might not be
easily attributable to the anatomical location of plaques alone
(Ganter et al., 1999). In the case of the corticospinal tracts in
multiple sclerosis, axonal loss is actually more selective than
in HSP (DeLuca et al., 2004b).
The present study utilizes what may be one of the largest
reported pathological cohorts of multiple sclerosis cases to
examine relationships among atrophy, plaque load and
axonal density/loss using quantitative techniques. By examining the population of axons in the corticospinal and sensory
tracts throughout the length of the spinal cord (as described
in our previous study) (DeLuca et al., 2004a) and plaque load
impinging on the respective tracts, the following questions
were addressed: (i) what is the nature and extent of the axonal
loss? and (ii) what is the relationship between total plaque
and axonal loss?
G. C. DeLuca et al.
Demyelination and axonal loss in multiple sclerosis
B
Corticospinal Tract Axonal Density
25,000
20,000
*
15,000
*
*
*
*
*
Control
MS
10,000
5,000
0
Medulla
Up C
Low C
Up T
LowT
Axonal Density (axons/sq. mm)
Axonal Density (axons/sq. mm)
A
Brain (2006), 129, 1507–1516
Sensory Tract Axonal Density
25,000
20,000
15,000
*
Control
MS
10,000
5,000
0
Lumbar
Up C
Low C
Level of Neuraxis
Corticospinal Tract Total Axon Number
1,000,000
900,000
800,000
700,000
600,000
Up T
Low T
Level of Neuraxis
Lumbar
Sensory Tract Total Axon Number
D
600,000
*
Control
MS
500,000
400,000
300,000
200,000
100,000
00,000
*
*
*
*
Medulla
Up C
Low C
Up T
Low T
*
Lumbar
Level of Neuraxis
Estimated TotalAxon Number
Estimated Total Axon Number
C
1509
500,000
400,000
*
*
Control
MS
*
300,000
200,000
100,000
0
Up C
Low C
Up T
Level of Neuraxis
Low T
Lumbar
Fig. 2 Coronal section of the cerebrum at the level of the internal capsule stained for myelin with Luxol fast blue cresyl violet (·10). The
area of white matter (outlined in red) and the area of demyelination (yellow shading) were quantified by a point-counting method. Measures
of relative plaque load were obtained by estimating the area of demyelination in relation to the total white matter area.
demyelination in relation to the total white matter area at each level
investigated. Lesions that were demyelinated (both active and inactive) were included in the calculation of plaque load measures, with
the majority of cases (52 out of 55) containing only chronic lesions.
Of the three active cases, only a small proportion of lesions (7 out
of 161) could be classified as active as defined by macrophages
labelled with PG-M1 6 Luxol fast blue/cresyl violet inclusions.
Fewer than 5% of the lesions showed evidence of remyelination
in all of the cases studied. When present, remyelination was often
confined to a narrow ribbon bordering the edges of chronic
plaques. As areas of remyelination contributed marginally to global
measures of plaque load, they were not included in global measures
of plaque load.
Measures of plaque load were estimated by summing the relative
plaque load measures of the cerebrum, brainstem and spinal cord for
the corticospinal tracts and by summing relative plaque load
measures of the spinal cord for the sensory tracts. To account for
the variation in the number of levels available in each case, total
plaque load measures were divided by the number of levels sampled.
One-tailed Pearson correlations between total plaque load and each
of axonal density, cross-sectional area and total axonal number were
calculated for each tract, respectively (under the assumption that
Downloaded from by guest on October 28, 2014
Fig. 1 Bar graphs of axonal density and total axon number of the corticospinal tract (A, C) and the posterior sensory tract (B, D)
in controls and multiple sclerosis cases [modified from DeLuca et al. (2004a)].
1510
Brain (2006), 129, 1507–1516
increases in area and total number of axons were not possible in the
patient group). As the total number of axons in the sensory tract was
significantly reduced only in the cervical and upper thoracic spinal
cord, total measures of plaque load were correlated with axonal loss
only at these levels. The relationship between relative changes in
lesion areas and regional axonal counts was tested by multiple
regression analysis. For the subset of cases (n = 10) in which the
spinal cords were sampled exhaustively, stepwise regression analysis
was performed using total plaque load as the first variable. Other
variables utilized in the regression analyses include sex, age and
duration of disease. All statistical calculations were performed
with SPSS for Windows version 11.5. P-values < 0.05 were deemed
statistically significant.
To evaluate axonal loss in plaques both within and between cases,
measures of axonal density within chronic plaques were compared
with measures of axonal density in NAWM. To control for variations
of axonal number at different levels of the neuraxis, comparisons of
axonal density inside and outside of plaques were restricted to levels
wherein axonal counts within a chronic plaque involving the entire
area of the lateral column on one side could be compared with the
NAWM of the other lateral column with no evidence of plaque
pathology. A total of 20 plaques from 16 cases met the specified
criteria.
To evaluate more thoroughly the extent of cerebral plaque load in
the cerebrum, the cerebral hemispheres from six of the cases studied
were sectioned coronally every 1 cm, as reported in a previous study
(Evangelou et al., 2000b). These cases were selected because they
showed a clear correlation between global measures of cerebral
plaque load and axonal loss in the corpus callosum. All multiple
sclerosis lesions were identified macroscopically and the crosssectional area of each lesion was measured using the stereological
technique of point counting (Howard et al., 1998). The volume of
the lesions was calculated by multiplying the surface area by the
depth of the section (1 cm). Lesion volumes for each case were
normalized by dividing the total volumes by the brain weight
reported at autopsy. Normalized cerebral lesion volumes were
related to estimates of axonal density and total axon numbers in
the corticospinal tracts. Cerebral lesions were not restricted to the
corticospinal tracts, in order to provide an estimate of the effect of
global cerebral plaque load on axonal loss.
Results
Corticospinal tracts
Axonal loss
In the corticospinal tract, it was found that there was a significant reduction in axonal density and total axonal loss
throughout all levels of the neuraxis in multiple sclerosis
cases when compared with controls. The results are summarized in Fig. 1.
In the crossed corticospinal tracts in multiple sclerosis,
fibres of smaller diameter (3 mm2) were preferentially
lost in contrast to the larger diameter fibres (>3 mm2),
which were relatively preserved (data not shown). These findings are described in more detail in DeLuca et al. (2004a).
Plaque load
Plaque load impinging on the corticospinal tracts was
variable among cases. The mean of the proportion of total
white matter in the lateral column affected by demyelination
of the nine levels initially studied was 0.1616 (range =
0.0197–0.5362). Relative plaque load measures from the
cerebrum (mean = 0.1101, range = 0.001–0.340), brainstem
(mean = 0.1236, range = 0–0.4848) and spinal cord (mean =
0.1949, range = 0–0.80) contributed to total plaque load
estimates. Plaque load measures were similar for both
sexes. Measures of total plaque load did not correlate well
with brain weight (r = 0.069, P = 0.323) nor with duration
of disease (r = 0.145, P = 0.186).
Comprehensive assessment of total spinal cord plaque load
of the subset of 10 cases revealed that the mean of the
proportion of total white matter in the lateral column affected
by demyelination was 0.0148 (range = 0–0.0534). The lower
cervical cord was the level of the neuraxis most significantly
burdened with plaque pathology affecting the corticospinal
tracts. Measures of total spinal cord plaque load did not
correlate well with brain weight (r = 0.138, P = 0.373)
nor with length of disease history (r = 0.306, P = 0.230).
For the subset of cases selected for detailed examination of
cerebral plaque load, the average lesion volume was 13.92 cm3
(range = 0.24–30.76 cm3). Measures of total cerebral plaque
load did not correlate well with brain weight (r = 0.567,
P = 0.319) nor with duration of disease (r = 0.021, P = 0.968).
Sensory tracts
Axonal loss
In the posterior white matter columns, there was a reduction
in axonal density and total axonal number in multiple sclerosis cases compared with controls throughout the length of the
spinal cord (Fig. 1). However, it was only at the cervical and
high thoracic cord levels that the reduction in total axonal
number reached statistical significance. There was a statistically significant reduction in axonal density in the smaller
diameter fibres but the larger diameter fibres showed no
significant reduction. These findings are described in more
detail in DeLuca et al. (2004 a).
Plaque load
Plaque load affecting the sensory tracts in the spinal cord
varied substantially among multiple sclerosis cases, with
the mean of the proportion of the total white matter area
in the posterior columns affected by plaque of the five levels
initially investigated being 0.07619 (range = 0–0.4737). The
cervical level of the spinal cord was the most frequent site of
plaque load affecting the sensory tracts. There was no significant difference between sexes in total measures of plaque
load affecting the sensory tracts. No significant correlations
were found between plaque load and brain weight (r = 0.066,
P = 0.333). There was no relationship identified between total
plaque load and duration of disease (r = 0.042, P = 0.398).
Downloaded from by guest on October 28, 2014
Cerebral hemispheres
G. C. DeLuca et al.
Demyelination and axonal loss in multiple sclerosis
For the subset of cases surveyed for total spinal cord plaque
load in detail, the average lesion load was 0.0282 (range =
0–0.1269). Measures of total spinal cord plaque load affecting
the sensory tracts did not correlate well with brain weight
(r = 0.193, P = 0.323). No significant relationship between
total spinal cord plaque load and duration of disease was
noted (r = 0.327, P = 0.214).
Correlations of plaque load with
axonal loss
Corticospinal tracts
Sensory tracts
Measures of plaque load did not correlate well with axonal
density in the upper cervical cord. In contrast, plaque load
correlated well with total axon number in the upper cervical
cord. These results are summarized in Table 4.
Upon comprehensive examination of total spinal cord
plaque load from the subset of 10 cases, measures of plaque
load did not correlate well with axonal density nor with total
axon number at the levels investigated. The results are
presented in Table 5. Upon consideration of the plaque subtype, chronic and active plaques each yielded similarly poor
correlations with axonal counts at each level investigated
(data not shown).
Axonal counts within chronic plaques compared
with NAWM
Measures of axonal density within chronic plaques did not
differ significantly from those obtained in NAWM both within
cases and between cases [axonal density (axons/mm2) (i) within
1511
Table 1 Corticospinal tract: simple and *partial
correlations of total plaque load and axonal loss in
multiple sclerosis
Level of neuraxis
Correlations
Simple (r)
Upper cervical
Density
Total
Lower cervical
Density
Total
Upper thoracic
Density
Total
Lower thoracic
Density
Total
Lumbar
Density
Total
P-value
Partial (r)
P-value
0.417
0.364
0.03
0.063
0.299
0.247
0.228
0.355
0.17
0.262
0.237
0.147
0.062
0.281
0.814
0.31
0.288
0.464
0.065
0.007
0.299
0.519
0.138
0.009
0.262
0.224
0.067
0.113
0.275
0.249
0.135
0.201
0.135
0.307
0.239
0.053
0.022
0.233
0.914
0.251
Density = axonal density (axons/mm2); total = total number of
axons. *Adjusted to account for sex, age and duration of disease.
Table 2 Corticospinal tract: correlations of total spinal
cord plaque load and axonal loss in multiple sclerosis
Level of neuraxis
Upper cervical
Density
Total
Lower cervical
Density
Total
Upper thoracic
Density
Total
Lower thoracic
Density
Total
Lumbar
Density
Total
Total
Correlations
Plaque load
R
P-value
0.001
0.491
0.761
0.201
0.067
0.413
0.518
0.245
0.185
0.037
0.045
0.472
0.462
0.586
0.435
0.083
0.165
0.318
0.514
0.269
0.149
0.069
0.013
0.002
0.005
Density = axonal density (axons/mm2); total = total number
of axons.
plaques = 14 050 (9150–24 625); and (ii) NAWM = 15 450
(5775–31 925), P = 0.614)].
Regression analyses
Upon consideration of the covariates sex, age and duration of
disease when comparing the effect of plaque load on axonal
density and total axonal number by multiple regression
models, the partial correlations for each tract were calculated.
The partial correlations for the corticospinal tracts and for the
Downloaded from by guest on October 28, 2014
Estimates of plaque load of the nine levels of the neuraxis
(from cerebrum to lumbar spinal cord) did not correlate well
with axonal density measures or with estimates of total axon
number at each level of the spinal cord surveyed. The results
are summarized in Table 1. The only level of the cord at which
the correlation between plaque load and total axonal loss was
significant was the upper thoracic.
Detailed measurement of total spinal cord plaque load
from the subset of cases sampled every 5 mm did not correlate
well with axonal density measures or with estimates of total
axon number at each level of the spinal cord surveyed. The
results are summarized in Table 2. Upon consideration of
the plaque subtype, chronic and active plaques each yielded
similarly poor correlations with axonal counts at each level
investigated (data not shown).
Upon more thorough evaluation of the plaque load in
the cerebral hemispheres in the subset of cases examined,
we found that cerebral lesion volume did not correlate
with axonal density measures nor with estimates of total
axonal number at each level of the cord investigated. The
results are summarized in Table 3.
Brain (2006), 129, 1507–1516
1512
Brain (2006), 129, 1507–1516
G. C. DeLuca et al.
Table 3 Corticospinal tract: simple correlations between
total cerebral plaque load and axonal loss in multiple
sclerosis
Level of neuraxis
Upper cervical
Density
Total
Lower cervical
Density
Total
Upper thoracic
Density
Total
Lower thoracic
Density
Total
Lumbar
Density
Total
Total
Correlations
Plaque load
R
0.288
0.269
0.356
0.365
0.273
0.525
0.364
0.238
0.057
0.111
0.464
0.429
0.571
0.243
0.158
0.347
0.246
0.352
0.345
0.281
Correlations
P-value
Partial (r)
P-value
0.468
0.496
0.01
0.009
0.429
0.278
0.052
0.248
0.218
0.322
0.123
0.044
0.146
0.362
0.467
0.069
0.16
0.08
0.218
0.355
0.174
0.082
0.427
0.722
Density = axonal density (axons/mm2); total = total number of
axons *Adjusted to account for sex, age and duration of disease.
sensory tracts are presented in Tables 1 and 4, respectively.
They showed no significant correlations between plaque load
and axonal number. For the subset of cases selected for
detailed quantification of total spinal cord load, stepwise
regression was carried out, with total spinal cord plaque
load being the first variable. No significant correlations
between total spinal cord plaque load and axonal density
and total axon number were found. Sex, age and duration
of disease had no appreciable effect on the correlations
between total spinal cord plaque load and axonal number.
Discussion
Traditionally, multiple sclerosis has been viewed as a primary
demyelinating disease (Lumsden, 1971; McAlpine et al., 1972;
Upper cervical
Density
Total
Lower cervical
Density
Total
Upper thoracic
Density
Total
0.033
0.578
0.351
0.115
0.248
0.187
0.233
0.361
0.329
0.017
0.071
0.492
0.478
0.032
0.041
Density = axonal density (axons/mm2); total = total number of
axons.
Poser, 1986). Accordingly, much attention has focused on the
role episodic inflammatory insults play in the formation of
plaques throughout the CNS. The observations that axonal
injury is associated with acute plaques and that extensive
axonal loss can occur in gliotic lesions in chronic multiple
sclerosis have engendered support for the concept that axonal
pathology is the consequence of inflammation-induced
demyelination (Trapp et al., 1998). Detailed neuropathological reports in combination with neuroimaging and
magnetic resonance spectroscopy studies in life validate the
existence of such axonal damage and loss, quantify the extent
to which it occurs and postulate how it may contribute to
clinical disability (Davie et al., 1995; Filippi et al., 1995;
Losseff et al., 1996; Ferguson et al., 1997; Ganter et al.,
1999; Trapp et al., 1999; Lovas et al., 2000; DeLuca et al.,
2004a). However, recent evidence that substantial loss of
axons occurs in areas apparently not affected by demyelination in the disease (Bjartmar et al., 2001), and that correlations between MRI measures of plaque load and axonal
damage are poor, challenge the plaque-centred view of axonal
degeneration (Iannucci et al., 1999; Mainero et al., 2001;
Bergers et al., 2002). The relative importance of axonal
loss attributable to acute plaques, the axonal loss occurring
in tracts without plaques and the Wallerian (or ‘dying back’)
nature of this axonal loss and its widely believed relationship
to plaque presence all remain unclear. As a result, the purpose
of the present study was to examine the distribution and
extent of axonal loss in the corticospinal and sensory tracts
and its relationship to plaque load throughout the length of
the neuraxis to gain better insight into the nature of axonal
damage and loss in multiple sclerosis.
The examination of such a large cohort of multiple sclerosis cases provides valuable insight into the nature of demyelination in the disease. Measures of plaque load impinging on
the corticospinal and sensory tracts varied significantly
between cases. The wide range of plaque load measures
between cases in these tracts reflects the variability of the
disease process and, more important to this study, enables
Downloaded from by guest on October 28, 2014
Table 4 Sensory tract: simple and *partial correlations
of total plaque load and axonal loss in multiple sclerosis
Simple (r)
P-value
P-value
Density = axonal density (axons/mm2); total = total number
of axons.
Upper cervical
Density
Total
Lower cervical
Density
Total
Upper thoracic
Density
Total
Level of neuraxis
Correlations
R
Level of neuraxis
Table 5 Sensory tract: correlations of total spinal cord
plaque load and axonal loss in multiple sclerosis
Demyelination and axonal loss in multiple sclerosis
1513
patients and in different tracts, reflecting the marked pathological heterogeneity encountered in the disease. This is also
the view supported by Kutzelnigg et al. (2005) in a study in
which cortical pathology was thought to contribute to diffuse
white matter injury.
We have found that axonal loss in the corticospinal tract
is symmetric and size-selective (small fibres 3 mm2 are
preferentially lost), and occurs throughout the length of
the neuraxis. In the corticospinal tract, plaque load did
not correlate well with decreases in axonal density or with
total number of axons at any of the levels investigated.
Consideration of the covariates sex, age and length of disease
duration resulted in similarly poor correlations between
plaque load and axonal loss. The comprehensive survey of
total plaque load in a subset of spinal cords also did not reveal
any significant correlation between plaque load and axonal
density nor total axon number. Both plaque subtypes (active
and inactive) each yielded similarly poor correlations with
axonal loss at each level investigated. To ensure that the effect
of cerebral plaque load on axonal loss was not underestimated, the cerebral plaque load from a subset of cases
was evaluated in detail. The cerebral plaque load in these
selected cases did not correlate with decreases in axonal
density or with total number of axons at any of the levels
investigated. These findings suggest that inflammationinduced demyelination does not determine fully the extent
to which axons are lost in this tract. Even though it has been
demonstrated that axonal injury is associated with acute plaques, it cannot be assumed that the formation of lesions is
necessarily causative of such axonal pathology and that all
axonal injury in these lesions is irreversible. The observations
that axonal loss in this tract is symmetric and size-selective
despite the fact that lesions involve the central white matter in
a disseminated fashion (Fog, 1950; Ikuta and Zimmerman,
1976; Oppenheimer, 1978), and that multiple sclerosis
patients rarely develop isolated progressive hemiparesis,
lend support to an underlying neurodegenerative process
(Coles et al., 1999). It is feasible that discrete changes in
the axonal microenvironment secondary to an underlying
axonopathy may elicit an immune response targeted at the
axon itself and/or surrounding myelin sheath that lead to the
formation of the demyelinating lesion (Silber and Sharief,
1999). It is also possible that lesions impinging on this
tract in sites not investigated could play a role in the axonal
pathology observed. As a result, the data cannot exclude the
possibility that inflammation and demyelination may
contribute to the axonal loss observed in the corticospinal
tracts in multiple sclerosis. However, plaques affecting the
corticospinal tracts were sampled at more levels than in any
other neuropathological study. Nevertheless, it demonstrates
that the widely accepted plaque-centred view of the disease
may not be sufficient to explain the axonal loss in this tract
and that alternative explanations for axonal loss need to be
considered.
In contrast to the corticospinal tracts, the sensory tracts
showed a significant, symmetric and size-selective reduction
Downloaded from by guest on October 28, 2014
the thorough examination of the extent to which total
measures of plaque load influence axonal loss in these tracts.
As supported by epidemiological data, no significant differences were found in plaque load measures between the
sexes (Weinshenker et al., 1989a, 1991). Plaque load did
not correlate well with brain weight, suggesting that
pathological mechanisms of tissue loss independent of
demyelination may be important in determining the extent
of atrophy in the brain, including axonal loss (McGavern
et al., 2000). Upon consideration of the relationship between
plaque load and duration of disease, it was observed that
the extent of demyelination in the corticospinal and sensory
tracts did not correlate with the length of disease history.
However, it must be considered that a cohort of multiple
sclerosis cases selected at autopsy is likely to have more
aggressive disease and/or increased disability than the
average. Therefore, the poor relationship between plaque
load and disease duration in the present study may reflect
(i) acute multiple sclerosis cases wherein death occurs early in
the course of their disease; or (ii) multiple sclerosis cases with
long-standing disease with chronic, ‘burnt out’ plaques that
do not contribute to an increased number of plaques
observed despite longer durations of disease. Fewer than
5% of plaques in the cohort examined showed evidence of
remyelination, which confirms the documented observation
that the extent to which remyelination occurs is highly dependent on duration of disease, with cases with increasing lengths
of disease history showing increasingly less evidence of
remyelination (Ozawa et al., 1994). The relative paucity of
remyelination in the spinal cord lesions in our cohort contrasts the degree of remyelination reported to occur in the
cerebral hemispheres (Prineas et al., 1993). Although it must
be considered that the identification of remyelination in the
spinal cord may be more difficult than in brain lesions and
that some areas classified as NAWM may represent lesions
with extensive remyelination, it is more likely that the question of remyelination is a possible confound for some lesions
but seems unlikely to account for the overall observations
presented in this study.
It has been proposed that the pathogenesis of axonal loss in
NAWM may occur via two distinct mechanisms. On the one
hand, axonal loss may result from a Wallerian degenerative
process in plaques, where inflammation and demyelination
are periodically exacerbated. In this circumstance, it would be
predicted that the distribution and extent of axonal damage is
determined by the distribution and extent of demyelinated
plaques and may help explain relapse-related disability. On
the other hand, the nature of the process may be one of
neurodegeneration wherein an underlying diffuse axonopathy, independent of inflammatory demyelination, contributes
significantly to axonal loss and results in the unrelenting
accumulation of disability often seen in the progressive
phase of the disease. In the latter scenario, it would be
predicted that the relationship between plaque load and
axonal loss would not be strong. It remains plausible that
both processes may operate to varying extents in different
Brain (2006), 129, 1507–1516
1514
Brain (2006), 129, 1507–1516
anatomically and functionally distinct tracts lends support
to the idea that a neurodegenerative process is operative.
There are some limitations in this study. All of the ascertainment biases common to autopsy studies surely will apply.
Although plaque load was examined in an exhaustive fashion
throughout the neuraxis, especially within the spinal cord, the
assessment of plaque load may not have determined the true
extent of demyelination in the multiple sclerosis cases. Cerebral, brainstem and spinal cord plaque load may be weak
reciprocal surrogates. Unfortunately, imaging information
was not available on the subjects studied to supplement
the pathological observations. However, the number of levels
investigated is substantial and widespread, extending from
the cerebrum to the lumbar spinal cord for the corticospinal
tracts. The comprehensive assessment of total spinal cord
plaque load in a subset of cases failed to reveal any significant
correlation between plaque load and axonal loss in either the
corticospinal or sensory tracts. To study the contribution
of cerebral plaque load to axonal loss in more detail, we
comprehensively examined the plaque load of the cerebral
hemispheres from a subset of cases. These cases had
previously shown a strong correlation between plaque load
and corpus callosum axonal loss (Evangelou et al., 2000b).
However, in contrast to the corpus callosum findings, no such
relationship could be demonstrated in the corticospinal tract.
While it should be acknowledged that the extent of axonal
loss may vary between different plaques of the same patient as
well as between lesions of different patients, it could be
argued that if a clear relationship between plaque load and
axonal loss in the corticospinal and sensory tracts were to
exist, then it should have been more readily apparent in the
current study. Despite such considerations, the fact remains
that a good correlation was found between cerebral plaque
load and corpus callosum axon loss in the study by Evangelou
et al. (2000b). As neuropathological studies can only provide
a static picture of a dynamic process, it is not possible to
prove definitively that axonal loss in the long tracts of the
multiple sclerosis spinal cord occurs independently of demyelination affecting these tracts. The present study, however,
provides several lines of evidence to suggest that axonal
loss in the multiple sclerosis spinal cord may occur, in
part, independently of inflammation-induced demyelination,
providing support to the claim that an underlying neurodegenerative process may be operative in the disease. Aside from
the difficulties with acquiring pathological material, the lack
of retrospective clinical information made it difficult to relate
the observed pathological findings to established clinical
scales, such as the EDSS.
Nevertheless, the current study may be the largest pathological cohort of multiple sclerosis sampled for plaque load
and tract-specific axonal loss in a comprehensive fashion. The
sample size has allowed the documentation of a number of
findings not described elsewhere. It has been observed that
(i) no close relationship between plaque load and axonal
loss exists; and (ii) axonal loss appears to be symmetric,
size-selective and tract-specific. These findings suggest that
Downloaded from by guest on October 28, 2014
in axonal density and total axonal number only in the upper
part of the spinal cord. Upon analysis of the relationship
between plaque load and axonal loss, there was a significant,
although modest, correlation between plaque load and total
axonal loss in the sensory tract in the upper cervical cord, the
region of most marked axonal loss. When the covariates sex,
age and duration of disease are taken into account, any
significant relationship between plaque load and axonal
loss in the sensory tract disappeared. Further supporting
the dissociation of the relationship between plaque load
and axonal loss, the extensive sampling of the subset of spinal
cords at 5 mm intervals revealed surprisingly weak correlations between total plaque load and total axon number. When
considered separately, active and inactive plaques each
showed similarly weak correlations between measures of
plaque load and axonal loss. As the sensory tract is largely
contained within the spinal cord with its axons terminating at
the gracile and cuneatus nuclei of the medulla, this finding is
particularly interesting for two reasons: first, plaque load was
surveyed meticulously throughout the length of the spinal
cord, thereby providing ample opportunity to discover a
possible relationship between plaque load and axonal loss
in this tract; and second, plaque load from the cerebrum,
a common site of plaque pathology, does not directly
influence the axonal viability of these sensory axons. These
observations in the sensory tract mirror the poor correlations
between plaque load and axonal loss found in the corticospinal tract. In short, the lack of impressive correlations
between plaque load and axonal loss in the long nerve tracts
in multiple sclerosis implies that mechanisms other than
inflammation-induced demyelination may play an important
role in axonal loss in multiple sclerosis, particularly in the
spinal cord.
Pathologically, axonal loss in multiple sclerosis has been
observed in a number of tracts such as the anterior optic
pathways, corpus callosum, corticospinal tracts and sensory
tracts (Ganter et al., 1999; Evangelou et al., 2000a, 2001;
Lovas et al., 2000; DeLuca et al., 2004a). Correlations between
plaque load and axonal loss in these tracts are variable. As
axonal loss correlated well with the plaque load in the cerebral
hemispheres in the corpus callosum, it was expected that a
similar relationship between plaque load and axonal loss
would be seen in the long tracts in multiple sclerosis.
However, the results of the present study were very different
in that they highlighted the limited role plaque load appears
to play in axonal loss in the corticospinal and sensory
tracts. The observation that axonal counts did not differ
significantly between chronic plaques and NAWM has
been reported elsewhere (Lovas et al., 2000), and further
underlines the poor relationship between plaque pathology
and axonal loss. Even though good correlations between plaque load and axonal loss do not signify causality, it remains
plausible that plaque pathology may affect axonal viability
to varying extents in different nerve fibre tracts perhaps
reflecting differential genetic influences at different sites.
The unifying feature of size-selective axonal loss in these
G. C. DeLuca et al.
Demyelination and axonal loss in multiple sclerosis
the complex pathogenesis of multiple sclerosis may encompass parallel neurodegenerative and Wallerian mechanisms of
axonal loss that affect different tracts, and even fibres within
these tracts, to varying extents. Recognition of such pathological heterogeneity within axonal bundles may not only
help explain the marked clinical variability seen in patients
but also may help elucidate key aetiological factors that drive
the course in multiple sclerosis.
Acknowledgements
We would like to thank all of the families who gave permission to allow the brains and spinal cords to be used for
research. This work was supported by the Multiple Sclerosis
Society of Canada, the Multiple Sclerosis Society of Great
Britain and Northern Ireland, and a Clarendon Scholarship
(G.C.D.), University of Oxford, UK.
1515
Ganter P, Prince C, Esiri MM. Spinal cord axonal loss in multiple sclerosis: a
post-mortem study. Neuropathol Appl Neurobiol 1999; 25: 459–67.
Gasperini C, Paolillo A, Giugni E, Galgani S, Bagnato F, Mainero C, et al. MRI
brain volume changes in relapsing-remitting multiple sclerosis patients
treated with interferon beta-1a. Mult Scler 2002; 8: 119–23.
Hemmer B, Glocker FX, Schumacher M, Deuschl G, Lucking CH. Subacute
combined degeneration: clinical, electrophysiological, and magnetic resonance imaging findings. [see comment]. J Neurol Neurosurg Psychiatry
1998; 65: 822–7.
Horsfield MA, Filippi M. Spinal cord atrophy and disability in multiple
sclerosis over four years. J Neurol Neurosurg Psychiatry 2003; 74: 1014–5.
Howard V, Reed MG. Royal Microscopical Society (Great Britain). Unbiased
stereology: three dimensional measurement in microscopy. Oxford: BIOS
Scientific in association with the Royal Microscopical Society; 1998.
Iannucci G, Minicucci L, Rodegher M, Sormani MP, Comi G, Filippi M.
Correlations between clinical and MRI involvement in multiple sclerosis:
assessment using T(1), T(2) and MT histograms. J Neurol Sci 1999;
171: 121–9.
Ikuta F, Zimmerman HM. Distribution of plaques in seventy autopsy cases of
multiple sclerosis in the United States. Neurology 1976; 26: 26–8.
Jitpimolmard S, Small J, King RH, Geddes J, Misra P, McLaughlin J, et al. The
sensory neuropathy of Friedreich’s ataxia: an autopsy study of a case with
prolonged survival. Acta Neuropathol (Berl) 1993; 86: 29–35.
Kremenchutzky M, Cottrell D, Rice G, Hader W, Baskerville J, Koopman W,
et al. The natural history of multiple sclerosis: a geographically based study.
7. Progressive-relapsing and relapsing-progressive multiple sclerosis: a
re-evaluation. Brain 1999; 122: 1941–50.
Kremenchutzky M, Rice GPA, Baskerville J, Wingerchuk DM, Ebers GC. The
natural history of multiple sclerosis: a geographically based study IX.
Observations on the progressive phase of the disease. Brain 2006; 129:
584–94.
Kutzelnigg A, Lucchinetti CF, Stadelmann C, Bruck W, Rauschka H,
Bergmann M, et al. Cortical demyelination and diffuse white matter injury
in multiple sclerosis. Brain 2005; 128: 2705–12.
Lin X, Tench CR, Turner B, Blumhardt LD, Constantinescu CS. Spinal cord
atrophy and disability in multiple sclerosis over four years: application of a
reproducible automated technique in monitoring disease progression in a
cohort of the interferon beta-1a (Rebif ) treatment trial. J Neurol Neurosurg Psychiatry 2003; 74: 1090–4.
Losseff NA, Webb SL, O’Riordan JI, Page R, Wang L, Barker GJ, et al. Spinal
cord atrophy and disability in multiple sclerosis. A new reproducible and
sensitive MRI method with potential to monitor disease progression. Brain
1996; 119: 701–8.
Lovas G, Szilagyi N, Majtenyi K, Palkovits M, Komoly S. Axonal changes in
chronic demyelinated cervical spinal cord plaques. Brain 2000; 123: 308–17.
Lumsden CE. The immunogenesis of the multiple sclerosis plaque. Brain
Research 1971; 28: 365–90.
Mainero C, De Stefano N, Iannucci G, Sormani MP, Guidi L, Federico A, et al.
Correlates of multiple sclerosis disability assessed in vivo using aggregates
of MR quantities. Neurology 2001; 56: 1331–4.
McAlpine D, Lumsden CE, Acheson ED. Multiple sclerosis: a reappraisal.
Edinburgh: Churchill Livingstone; 1972.
McGavern DB, Murray PD, Rivera-Quinones C, Schmelzer JD, Low PA,
Rodriguez M. Axonal loss results in spinal cord atrophy, electrophysiological abnormalities and neurological deficits following demyelination in a
chronic inflammatory model of multiple sclerosis. Brain 2000; 123: 519–31.
Oppenheimer DR. The cervical cord in multiple sclerosis. Neuropathol Appl
Neurobiol 1978; 4: 151–62.
Ozawa K, Suchanek G, Breitschopf H, Bruck W, Budka H, Jellinger K, et al.
Patterns of oligodendroglia pathology in multiple sclerosis. Brain 1994;
117: 1311–22.
Pittock SJ, Mayr WT, McClelland RL, Jorgensen NW, Weigand SD, Noseworthy JH, et al. Disability profile of multiple sclerosis did not change over
10 years in a population-based prevalence cohort. Neurology 2004; 62:
601–6.
Poser CM. Pathogenesis of multiple sclerosis. A critical reappraisal. Acta
Neuropathol (Berl) 1986; 71: 1–10.
Downloaded from by guest on October 28, 2014
References
Behan WM, Maia M. Strumpell’s familial spastic paraplegia: genetics and
neuropathology. J Neurol Neurosurg Psychiatry 1974; 37: 8–20.
Bergers E, Bot JC, De Groot CJ, Polman CH, Lycklama a Nijeholt GJ,
Castelijns JA, et al. Axonal damage in the spinal cord of multiple
sclerosis patients occurs largely independent of T2 MRI lesions. Neurology
2002; 59: 1766–71.
Bjartmar C, Kinkel RP, Kidd G, Rudick RA, Trapp BD. Axonal loss in
normal-appearing white matter in a patient with acute multiple sclerosis.
Neurology 2001; 57: 1248–52.
Coles AJ, Wing MG, Molyneux P, Paolillo A, Davie CM, Hale G, et al.
Monoclonal antibody treatment exposes three mechanisms underlying
the clinical course of multiple sclerosis. Ann Neurol 1999; 46: 296–304.
Confavreux C, Vukusic S, Moreau T, Adeleine P. Relapses and progression
of disability in multiple sclerosis. N Engl J Med 2000; 343: 1430–8.
Confavreux C, Vukusic S, Adeleine P. Early clinical predictors and progression of irreversible disability in multiple sclerosis: an amnesic process. Brain
2003; 126: 770–82.
Davie CA, Barker GJ, Webb S, Tofts PS, Thompson AJ, Harding AE, et al.
Persistent functional deficit in multiple sclerosis and autosomal dominant
cerebellar ataxia is associated with axon loss. Brain 1995; 118: 1583–92.
DeLuca GC, Ebers GC, Esiri MM. Axonal loss in multiple sclerosis: a pathological survey of the corticospinal and sensory tracts. Brain 2004a; 127:
1009–18.
DeLuca GC, Ebers GC, Esiri MM. The extent of axonal loss in the long tracts
in hereditary spastic paraplegia. Neuropathol Appl Neurobiol 2004b; 30:
576–84.
Evangelou N, Esiri MM, Smith S, Palace J, Matthews PM. Quantitative
pathological evidence for axonal loss in normal appearing white matter
in multiple sclerosis. Ann Neurol 2000a; 47: 391–5.
Evangelou N, Konz D, Esiri MM, Smith S, Palace J, Matthews PM. Regional
axonal loss in the corpus callosum correlates with cerebral white matter
lesion volume and distribution in multiple sclerosis. Brain 2000b; 123:
1845–9.
Evangelou N, Konz D, Esiri MM, Smith S, Palace J, Matthews PM. Sizeselective neuronal changes in the anterior optic pathways suggest a differential susceptibility to injury in multiple sclerosis. Brain 2001; 124:
1813–20.
Ferguson B, Matyszak MK, Esiri MM, Perry VH. Axonal damage in acute
multiple sclerosis lesions. Brain 1997; 120: 393–9.
Filippi M, Paty DW, Kappos L, Barkhof F, Compston DA, Thompson AJ, et al.
Correlations between changes in disability and T2-weighted brain
MRI activity in multiple sclerosis: a follow-up study. Neurology 1995;
45: 255–60.
Fog T. Topographical distribution of plaques in the spinal cord in multiple
sclerosis. Arch Neurol Psychiat 1950; 63: 382–414.
Brain (2006), 129, 1507–1516
1516
Brain (2006), 129, 1507–1516
Prineas JW, Barnard RO, Kwon EE, Sharer LR, Cho ES. Multiple sclerosis:
remyelination of nascent lesions. Ann Neurol 1993; 33: 137–51.
Said G, Marion MH, Selva J, Jamet C. Hypotrophic and dying-back nerve
fibers in Friedreich’s ataxia. Neurology 1986; 36: 1292–9.
Silber E, Sharief MK. Axonal degeneration in the pathogenesis of multiple
sclerosis. J Neurol Sci 1999; 170: 11–8.
Trapp BD, Peterson J, Ransohoff RM, Rudick R, Mork S, Bo L. Axonal
transection in the lesions of multiple sclerosis. N Engl J Med 1998; 338:
278–85.
Trapp BD, Ransohoff R, Rudick R. Axonal pathology in multiple sclerosis:
relationship to neurologic disability. Curr Opin Neurol 1999; 12: 295–302.
G. C. DeLuca et al.
Weinshenker BG, Bass B, Rice GP, Noseworthy J, Carriere W, Baskerville J,
et al. The natural history of multiple sclerosis: a geographically based study.
I. Clinical course and disability. Brain 1989a; 112: 133–46.
Weinshenker BG, Bass B, Rice GP, Noseworthy J, Carriere W, Baskerville J,
et al. The natural history of multiple sclerosis: a geographically based
study. 2. Predictive value of the early clinical course. Brain 1989b; 112:
1419–28.
Weinshenker BG, Rice GP, Noseworthy JH, Carriere W, Baskerville J,
Ebers GC. The natural history of multiple sclerosis: a geographically
based study. 3. Multivariate analysis of predictive factors and models of
outcome. Brain 1991; 114: 1045–56.
Downloaded from by guest on October 28, 2014