Sunteți pe pagina 1din 9

Neuroradiology (2010) 52:109–117

DOI 10.1007/s00234-009-0593-9

DIAGNOSTIC NEURORADIOLOGY

A longitudinal observational study of brain atrophy


rate reflecting four decades of multiple sclerosis:
a comparison of serial 1D, 2D, and volumetric
measurements from MRI images
Juha Martola & Jakob Bergström & Sten Fredrikson & Leszek Stawiarz & Jan Hillert &
Yi Zhang & Olof Flodmark & Anders Lilja & Anders Ekbom & Peter Aspelin &
Maria Kristoffersen Wiberg

Received: 26 June 2009 / Accepted: 24 August 2009 / Published online: 23 September 2009
# Springer-Verlag 2009

Abstract Methods Thirty-seven MS patients (age range 24–65 years


Introduction Multiple sclerosis (MS) has a variable with disease duration 1–33 years) were consecutively selected
progression with an early onset of atrophy. Individual and evaluated with MRI at baseline 1995 and in 1996. They
longitudinal radiological evaluations (over decades) are were followed up for a decade (mean of 9.25 years, range
difficult to perform due to the limited availability of 7.3–10 years) up to 2003–2005. Brain parenchymal volume
magnetic resonance imaging (MRI) in the past, patients and volumes of the supratentorial ventricles were analyzed
lost in follow-up, and the continuous updating of with semi-automated volumetric measurements at three time
scanners. We studied a cohort with widespread disease points (1995, 1996, and 2003–2005).
duration at baseline. The observed individual atrophy Results Volumetric differences were found over shorter
rates over time of 10 years represented four decades of periods of time (1–7 months); however, differences vanished
disease span. by the end of follow-up. A uniform longitudinal decrease in
brain volume and increase in ventricle volumes were found.
Frontal horn width (1D) correlated strongest to 3D measures.
J. Martola (*) : Y. Zhang : P. Aspelin : M. K. Wiberg No statistical differences of atrophy rates between MS
Division of Radiology, Department of Clinical Science,
Intervention, and Technology, Karolinska Institutet,
courses were found. Supratentorial ventricular volumes were
14186 Stockholm, Sweden associated with disability and this association persisted
e-mail: juha.martola@ki.se during follow-up.
Conclusion Despite variable clinical courses, the degen-
O. Flodmark : A. Lilja
erative effects of MS progression expressed in brain
Department of Neuroradiology,
Department of Clinical Neuroscience, atrophy seem to uniformly progress over longer periods
Karolinska University Hospital, of time. These volumetric changes can be detected using
Stockholm, Sweden 1D and 2D measurements performed on a routine PACS
S. Fredrikson : L. Stawiarz : J. Hillert
workstation.
Division of Neurology, Department of Clinical Neuroscience,
Karolinska Institutet, Keywords MRI . Multiple sclerosis . Brain atrophy .
Stockholm, Sweden Disability
A. Ekbom
Clinical Epidemiology Unit, Karolinska Institutet,
Stockholm, Sweden Introduction

J. Bergström
Multiple sclerosis (MS) affects the whole central nervous
The Medical Statistics Unit, Department of Learning, Informatics,
Management & Ethics (LIME), Karolinska Institutet, system causing early brain tissue destruction often followed
Stockholm, Sweden by persistent clinical deficit [1]. MS has a variable clinical
110 Neuroradiology (2010) 52:109–117

progression difficult to predict. Moreover, there is a poor between the rate of the longitudinal volumetric changes and
correlation between magnetic resonance imaging (MRI) MS course, age at MS onset, and disease duration. Finally,
findings and clinical course which is referred to as “the the fourth aim was to evaluate which longitudinal 1D or 2D
clinicoradiological paradox” [2]. Therefore, evaluation of measurement had the best correlation to the longitudinal 3D
the predictive factors of disease development is challenging. changes.
Longitudinal follow-ups over decades are almost impossible
to achieve due to practical reasons and the limited availability
of MRI in the past. Materials and methods
Early onset of brain atrophy is common in [1, 3, 4]. It is
usually detected visually by abnormal widening of cerebral Participants
sulci, cisterns, and compensatory enlargement of brain
ventricles due to loss of brain parenchyma. There are few The Karolinska Institutet ethics committee approved the
longitudinal studies [5, 6]; mostly are often performed in study. Informed consent was obtained from all of the
association with clinical trials. The studies are difficult to patients. This study was followed up over nearly 10 years
compare due to technical reasons—differences in MRI from 1995–1996 to 2003–2005 at the Department of
protocols and performed analysis. Few studies compare Neurology, Karolinska University Hospital, Sweden. The
relapsing–remitting MS (RRMS), secondary-progressive MS study base consisted of 45 consecutively selected voluntary
(SPMS), and primary progressive MS (PPMS) at the same patients who fulfilled the Poser criteria for clinically
time [7, 8]. Atrophy has been proposed to be a marker of definitive MS at baseline in 1995–1996 [20]. All the patients
disability [1]. Whole-brain atrophy has been reported to vary had also MS according to the revised McDonald criteria
between MS courses [9–14]. Asymmetries in supratentorial (evaluated retrospectively) [21]. The clinical diagnosis of
atrophy have recently been reported in shorter follow-ups MS was made by experienced neurologists. At the baseline
[15]. (1995–1996), three MS courses were represented (Table 1).
Most studies concerning MS brain volume atrophy are Clinical disability was monitored at the Department of
currently performed using automated or semi-automated Neurology, Karolinska University Hospital by using the
methods, which have limited accessibility in clinical EDSS [18]. EDSSs were additionally adjusted for disease
radiological use [13, 15, 16]. Serial 1D or 2D measures, duration and expressed as MSSSs [19]. The MSSS scale
which can be performed on any PACS workstation, might ends at 30-year disease duration. Since six patients had greater
be sufficient in revealing progression of the net destructive than 30 years disease duration at the study’s conclusion, only
inflammatory processes causing atrophy [13, 17]. To date, 31 of the 37 patients could be longitudinally analyzed using
longitudinal correlations between rates of volumetric the scale.
changes of the brain to 1D or 2D measures are poorly Only eight patients (of the original 45) were lost in
documented. follow-up. The remaining 37 patients were followed over
There are many questions which a longitudinal study 9.25 years (range 7.3–10). The 37 patients that were
could address, including: What is the course of the followed up and the eight lost were distributed evenly on
destructive changes over decades? Is the rate of brain both sides of the median MSSS development. Therefore,
atrophy accelerating or declining from disease onset? Are confounding by indication (selection bias) was ruled out for
there differences between MS courses? Does atrophy both the groups [22].
associate with disability? To answer these questions, we
used a patient cohort of variable disease duration at baseline Table 1 Cohort characteristics.
(from 1 to 33 years) and carried out individual follow-ups
Measure Group Summary
for nearly one decade. The patients were evaluated with
different MR scanners at entry versus the end. Despite of Gendera Females 70.27% (n=26)
that, the comparison between individual volume progression Males 29.73% (n=11)
rates was still possible. The individual brain atrophy rates Interferona Treatment 64.86% (n=24)
reflected a disease span of four decades. Nontreatment 35.14% (n=13)
There were four goals of our study. We first aimed to MS coursea RRMS 43.2% (n=16)
evaluate the rate of longitudinal volumetric changes of the SPMS 46% (n=17)
brain, the third ventricle, and lateral ventricles. The second PPMS 10.8% (n=4)
aim was to analyze the association between these volumetric Age at baselineb 42 (10.29, n=37)
measures and disability status expressed in expanded
a
disability status score (EDSS) [18] and MS severity score Percentage of total (sample size)
b
(MSSS) [19]. The third aim was to study the association Mean (1 SD, sample size)
Neuroradiology (2010) 52:109–117 111

Twenty-four of the 37 patients were treated with 1.5 mm gaps were added to 5 mm slices. Each imaging
interferon during the follow-up period (Table 1). There slice was then measured as it was 6.5 mm thick. This
was no randomized selection for treatment and information compensated a systematic error in volumetric analysis.
on compliance was not available. Six patients of the 37 Volumetric results were further compared with the
converted during the follow-up from RRMS to SPMS and results of regional atrophy measures (1D and 2D) as
were included in the SPMS group for evaluations of the described in details in our previous papers [22–24].
results. The second EDSS was not obtained for two patients Regional atrophy measures (1D and 2D) were performed
in 1996. A mean EDSS value from both first and second by a radiologist using a PACS workstation (Sectra, Sweden)
EDSSs in 1995 and 1996 was calculated for the whole with a distance tool. MR images were evaluated by a
cohort. radiologist blinded for clinical outcome data. The 1D
measures on T1-weighted sequences were defined as
MRI acquisition bifrontal distance, the third ventricle width, intercaudate
distance, bifrontal ratio, Evans ratio, bicaudate ratio, and
At study entry, we performed two (range 1–7 months) biuncal ratio. 2D corpus callosum midsagittal area measure-
T1-weighted MRI examinations on a 1.5-T Signa unit ments were performed on a T2-weighted sagittal sequence.
(GE, Milwaukee, WI, USA) using image matrix 256×256
(acquisition variables: repetition time, 640 ms; echo time, Statistical method
13 ms; number of excitations, 2; flip angle, 90°). The
postcontrast scans consisted of contiguous 5-mm axial slices. The statistical analyses were performed using the software
Field of view was 250×188 mm. SAS version 9.1.3 (SAS Institute Inc., Cary, NC, USA) and
At the end of the study (2003–2005), we performed LogXact 7.0 (Cytel Inc., Cambridge, MA, USA). Graphical
T1-weighted MRI examinations on a 1.5-T Vision unit presentations were produced using the Statistica 7.0
(Siemens, Erlangen, Germany) using image matrix 256×192 software (StatSoft Inc., Tulsa, OK, USA).
(acquisition variables: repetition time, 570 ms; echo time, The commonly used analysis of variance for repeated
14 ms; number of excitations, 2; flip angle, 90°). The measures assumes that all patients are measured on the
MRI examinations consisted of contiguous 5-mm axial same occasions. When duration defines time, the meas-
postcontrast slices and intersection gap of 1.5 mm. The urements will be inherently unbalanced (Fig. 1). The
field of view was 220×220 mm. statistical method of random coefficient models is well
suited for analyzing data with such characteristics. A
Image processing linear regression line is fitted for every patient with a slope
representing annual linear change of volume and an
Images were transferred to a HERMES workstation intercept representing the cohort brain volume at disease
(Nuclear Diagnostics, Stockholm, Sweden) where volumetric onset. The mean of the slopes and intercepts are then used
measurements were performed by a neuroradiological rater to make conclusions on the population of MS patients
(YZ) who was blinded to all clinical information. A regarding changes in annual brain volume measures.
semiautomatic tool in HERMES MultiModality, Region
Growing, was used in volumetric evaluation of lateral
ventricles, the third ventricle, and the brain parenchymal
volume on original axial T1-weighted images. Due to
individual variations of signal intensity distribution, a test
of growing effect was visually checked to set specific
individual parameters before each final growing volume
was produced. When a leakage of region of interest
happened after growing, manual masking was used before
growing. Intrarater reliability (ICC) was calculated from
ten randomly selected subjects with double measurements.
ICC of the lateral ventricles, third ventricle, and whole-
brain volume evaluated with HERMES semiautomatic tool
was 0.93.
MRI examinations of 1995–1996 were performed
without intersection gaps. The MRI examinations of
2003–2005 were performed with 1.5 mm gaps. HERMES Fig. 1 Mean predicted decrease in brain parenchymal volume over
tools automatically adjusted evaluations for the gaps as four decades (expressed in months)
112 Neuroradiology (2010) 52:109–117

Possible associations between the patient’s brain volume


at entry and annual brain volume change can be estimated
within the model.
The atrophy of brain volume outcomes were analyzed in
three separate random coefficient models. The effect of
gender, MS course, interferon treatment, and age at baseline
on the different brain volume outcomes was controlled for
in all the analyses. The difference between left and right
lateral ventricles was also estimated when performing the
lateral ventricle brain volume analysis. Accelerating or
declining rate of atrophy at the end of follow-up was tested
for all models using a second degree polynomial model
term (quadric relationship). Due to the low power to detect
any such effects, more weight was given to visual
inspection of the data. Kenward–Rogers adjustment for
Fig. 3 Mean predicted increase in left ventricle volume over four
degrees-of-freedom correction method was used to produce decades (expressed in months)
valid F and t statistics. A p value less than 0.05 was used as
the cutoff limit for statistical significance. Residual and Principal component analysis (PCA) was performed to
normal probability plots were used to evaluate the reveal atrophy rate patterns (clusters) between corpus
assumptions of equal variance, normally distributed residuals, callosum midsagittal area, normalized 1D ratios, supratento-
and nonexistence of outliers. rial ventricular widths, and brain volume measurements
The effects on disability measures, EDSS and MSSS, regarding linear annual change of brain volume. Pattern
were analyzed using exact logistic regression. Exact coefficients of 0.7 or above were used to decide which brain
logistic regression produces more conservative results measure variable formed clusters. Variables that formed a
than asymptotic logistic regression when sample size is cluster were regarded as having a high intrarelationship.
small, representation of cases is sparse, and/or if the data Spearman correlations were estimated between each
are highly imbalanced [25]. Patient’s annual brain volume brain volume measure and EDSS/MSSS for the year 1995
rate of change, estimated by the random coefficient and 2005, respectively. The results should be regarded as
models, and disease duration at baseline were used as descriptive statistics; hence, no p values are reported.
independent variables. For analysis of lateral ventricle
outcome, the mean of the left and right lateral ventricle
volumes was used as a change variable. In the “Result” Results
section, this variable is referred to as “mean lateral
ventricle”. Separate models were fitted for each atrophy In the three random coefficient models used, significant
rate of change variable to avoid multicolinearity. linear atrophy progression was demonstrated in all analyses.
Neither statistical analysis nor visual inspection of the data
(Figs. 1, 2, 3, and 4) did reveal an accelerating or declining

Fig. 2 Mean predicted increase in third ventricle volume over four Fig. 4 Mean predicted increase in right ventricle volume over four
decades (expressed in months) decades (expressed in months)
Neuroradiology (2010) 52:109–117 113

Table 2 Results of 3D measure-


ments outcome analyses. Effect Estimate SE 95% CI p value

Lower Upper

Third ventricle (cm3)


Disease duration (per year) 0.049 0.006 0.037 0.061 <0.0001
Brain parenchymal volume (cm3)
a
Results of male versus female Disease duration (per year)a −5.72 1.08 −7.92 −3.54 <0.0001
not shown Lateral ventricle (log-cm3)
b
Results of left side versus right Disease duration (per year)b 0.042 0.004 0.032 0.051 <0.0001
side not shown

rate of atrophy at the end of follow-up. Volumetric changes over four decades of follow-up. The estimated correlation
were found over shorter periods of time (1–7 months) at between brain volume at entry and annual change over
baseline (Fig. 1); however, this effect vanished at the end time was −0.41 (p=0.06, not shown). The minus sign
(Figs. 1, 2, 3, and 4). Gender was a significant predictor of indicates that patients with larger third ventricle volumes
brain volume throughout the follow-up. However, the (at baseline) had less increase in volume over time than
progression of atrophy did not differ between genders. those with smaller brain volumes (at baseline). Note that
the correlation is moderate. There was no statistical
Brain parenchymal volume evidence of an accelerating or declining rate of ventricular
enlargement at the end of follow-up (test of quadric
An annual decrease in brain volume of −5.72 cm3 (p<0.0001; relationship: p=0.540; Fig. 2)
see Table 2) was estimated in the applied random coefficient
model (Fig. 1). Males displayed a larger brain volume then Left and right lateral ventricles (log-transformed)
females with an observed difference of 143.85 cm3 (95%
confidence interval (CI) 38.01, 249.69; p=0.009). The An annual increase in left (Fig. 3) and right (Fig. 4) lateral
difference between genders was persistent during the four ventricle volume of 0.042 cm3 (p<0.001; see Table 2) was
decades of disease duration (test for interaction: duration by demonstrated in the performed analysis. A difference of
gender, p>0.223). The results suggested a linear change of 0.15 cm3 (95% CI 0.09, 0.20; p<0.0001) at entry between
parenchymal brain volume and that women had a smaller left and right ventricles was estimated. There were no
brain volume than men (Fig. 1). The correlation between indications that the difference between left and right lateral
brain volume at baseline and later atrophy was −0.5 ventricles would increase or decrease over disease duration
(p=0.02). There was no statistical evidence of an accelerating (test for interaction: duration by side, p>0.964). The results
or declining rate of atrophy at the end of follow-up (test of suggest positive linear atrophy progression with the
quadric relationship: p=0.993) left lateral ventricle displaying a larger volume than the
right lateral ventricle. The correlation between left and right
The third ventricle lateral ventricle volume change rates was fairly large, 0.84
(p=0.001). The positive sign and the magnitude of the
An annual change in third ventricle volume of 0.049 cm3 correlation indicated that the rate of change in volume was
(p<0.0001; see Table 2) was found (Fig. 2). This result the same in the left and right lateral ventricles. There was
imply that third ventricle brain volume increased linearly no statistical evidence of an accelerating or declining rate of
Table 3 Univariate results of
3D measurement analyses. Effect Odds ratio 95% CI p value

Lower Upper

EDSS
Brain parenchymal volumea 1.72 0.4 7.71 0.449
Third ventricleb 1.36 0.99 1.95 0.053
Mean lateral ventriclec 1.24 1.01 1.62 0.037
Results regarding duration at MSSS
entry not shown
a
Brain parenchymal volumea 0.99 0.82 1.19 0.945
Unit 0.1
b
Third ventricleb 1.52 1.01 2.57 0.044
Unit 0.001
c
Mean lateral ventriclec 1.46 1.09 2.21 0.006
Unit 0.01
114 Neuroradiology (2010) 52:109–117

ventricular enlargement at the end of follow-up (test of Volume measurements versus disability
quadric relationship: p=0.268; Figs. 3 and 4)
The associations between brain parenchymal volume or
Atrophy and disability supratentorial ventricular volume and EDSS were weak to
moderate at baseline and at the end of follow-up (lowest
The statistical analysis of EDSS and MSSS did not support an correlation, −0.17; largest correlation, 0.49; Table 5). The
effect of disease duration (at entry) with an odds ratio nearly associations between above-mentioned volumes and
equal to 1 (not shown). The displayed odds ratios of the MSSS were weak to moderate at baseline and at the end of
annual volume change rate of the third ventricle on EDSS and follow-up (lowest correlation, −0.06; largest correlation, 0.41;
MSSS were 1.36 (p=0.053; see Table 3) and 1.52 (p=0.044), Table 5). In general, the correlations were somewhat higher
respectively. The analyses of annual volume change rate of at the end for both EDSS and for MSSS. Positive signs of
mean lateral ventricle on EDSS and MSSS displayed similar correlation indicate that a larger volume yields a more
results with odds ratios of 1.24 (p=0.037) and 1.46 pronounced disability (for lateral and third ventricle
(p = 0.006), respectively. Brain parenchymal volume volumes). Negative signs of correlation indicate that a smaller
displayed an odds ratios near of 1.72 on EDSS (p=0.449), volume yields more disability (for brain parenchymal volume).
but an odds ratio near 1 on MSSS (p=0.945). However, many of the correlations were near zero (no
relationship) or low.
Multivariate evaluation of brain measures

The principal component analysis revealed a one cluster Discussion


solution, with the first component explaining 52.2% of the
variance in the data (Table 4). The variables included in the We found progressive uniform atrophy rate reflecting four
cluster were frontal horn width, bifrontal ratio, third decades of disease duration (Fig. 5). We analyzed the
ventricle width, Evans ratio, intercaudate distance, mean individual rate changes of the three supratentorial ventricles
lateral ventricle (log), the third ventricle, and bicaudate and brain parenchymal volume in patients for disease spans
distance. The correlations from the atrophy rate variables, covering these four decades. Longitudinal studies of
included in the first component, ranged from 0.71 to 0.86. atrophy in MS are few [5–15] and difficult to perform, as
The variables included in the cluster measured the same technical differences are unavoidable due to continuous
unobservable structure. Brain parenchymal volume (0.64) updating of MRI scanners. MS affects the entire central
and corpus callosum midsagittal area (0.62) formed a nervous system, including the spinal cord, leading to
cluster. Biuncal ratio did not follow any obvious pattern. problems interpreting results [12].
Several factors may affect atrophy evaluation in MS
and opponent processes can be present at the same time.
Brain volume reduction is caused by axonal loss,
Table 4 Results of correlations between rates of change (between
different brain measures). decrease of edema, resolution of inflammation, scarring,
demyelination, dehydration, and normal aging [26]. On
Disease rate Component the contrary, brain volume increases can be caused by
inflammation, edema, gliosis (tissue bulk), and remyeli-
1 2 3 4
nation [26]. These effects could explain the short-time
Frontal horn width (1D) 0.86 – – – volumetric differences we found (within 1–7 months after
Bifrontal ratio (1D) 0.85 – – – baseline), differences which vanished by the end of
Third ventricle width (1D) 0.85 – – – follow-up (Fig. 1).
Evans ratio (1D) 0.81 – – –
Mean lateral ventricle (log, 3D) 0.79 – – –
Table 5 Descriptive statistics of the relationship between brain
Intercaudate distance (1D) 0.73 – −0.67 –
volume change rate and 3D measurements at baseline and end.
Third ventricle (3D) 0.72 – – –
Bicaudate ratio (1D) 0.71 – −0.68 – EDSS MSSS
Biuncal ratio (1D) – – – −0.73
Baseline End Baseline End
Brain parenchymal volume (3D) – 0.64 – –
Corpus callosum (2D) – 0.62 – – Brain parenchymal volume −0.17 −0.26 −0.12 −0.21
Eigen values 5.75 1.5 1.21 0.9 Mean lateral ventricle 0.27 0.48 −0.06 0.41
Variance explained (%) 52.2 13.7 10.96 8.14 Third ventricle 0.41 0.49 0.1 0.38
Neuroradiology (2010) 52:109–117 115

Fig. 5 Atrophy progression during four decades of MS duration.„


Please note the widening of both sulci and lateral ventricles. The first
(a), second (b), third (c), and fourth (d) decade of MS progression (NB
Images refer to different MS patients)

Our results suggest that brain atrophy and compensatory


supratentorial ventricular enlargement are continuous
processes that exist over four decades of disease duration
without any particular time point altering atrophy rate. This
is in agreement with Fisher et al. who reported significant
whole-brain atrophy expressed in brain parenchymal
fraction in an 8-year (interferon beta-1a) follow-up study
of RRMS patients [5]. However, we could not find any sign
of reduction of the brain atrophy rate as Fisher et al. did [5].
Our findings do not conflict with reports of varying atrophy
rate and asymmetric atrophy distribution over shorter
periods [15]; instead, they suggest that these effects might
vanish over longer periods of time.
Callosal atrophy rate and brain atrophy were correlated
with each other as shown in PCA analyses, i.e., 2D
evaluation of corpus callosum midsagittal area revealed
local and global brain atrophy progress. 1D and 3D
measures of supratentorial ventricular enlargement correlated
with each other (Evans ratio, Biuncal ratio, the third ventricle
width, bifrontal ratio, interuncal distance, and bicaudate ratio).
The progression of ventricular enlargement was revealed in
1D measures. The biuncal ratio did not follow the other
atrophy patterns in PCA analyses.
We could not find differences in 3D atrophy patterns
between MS courses over longer periods of time. The
statistical weight is limited to some extent due to few
patients in the PPMS group. Results are in accordance with
two other MS studies [7, 8]. We did not find dependence of
MS course on corpus callosum and ventricular diameters in
our atrophy studies [22, 23]. Our results evinced the odds
of an increase in EDSS and MSSS increase as the atrophy
rates of the third ventricle and mean lateral ventricle
increase. The relationships between 3D measurements and
EDSS were weak to moderate at baseline and at the end of
follow-up. This is in accordance with our previous reports
[22, 23]. EDSS is known to be insensitive to clinical
changes, lacks linearity, and does not take disease duration
into account [27].
Our study is limited by a sparse representation of older
patients and a lack of normal controls. Normal controls are
a general dilemma in longitudinal follow-ups. In our
previous study, we found that predicted corpus callosum
area at symptom onset is representative of normal
population [22]. We also found there the effect of age at
baseline [22, 23]. Older patients had larger supratentorial
ventricles at entry but their rates of regional atrophy
progression did not significantly differ from younger
patients. Reports concerning normal aging of healthy
adults in the same age span as our cohort showed no
116 Neuroradiology (2010) 52:109–117

signs of elevated age-related atrophy [28–30]. However, 6. Horakova D, Dwyer MG, Havrdova E, Cox JL, Dolezal O,
the lack of normal controls prevented us from determining Bergsland N, Rimes B, Seidl Z, Vaneckova M, Zivadinov R
(2009) Gray matter atrophy and disability progression in
how much MS added to the annual brain atrophy rate patients with early relapsing–remitting multiple sclerosis: a
caused by normal aging (−0.5–1.4% for MS [9–14, 31, 32] 5-year longitudinal study. J Neurol Sci 282:112–119
and −0.3% for healthy controls [28]). 7. Fox NC, Jenkins R, Leary SM, Stevenson VL, Losseff NA, Crum
3D evaluations were affected by differences in MRI WR, Harvey RJ, Rossor MN, Miller DH, Thompson AJ (2000)
Progressive cerebral atrophy in MS: a serial study using
protocols including 1.5-mm interslice gaps in MR images at registered, volumetric MRI. Neurology 54:807–812
final MRI examinations 2003–2005. However, HERMES 8. Kalkers NF, Ameziane N, Bot JC, Minneboo A, Polman CH,
software compensated for this shortage as explained in Barkhof F (2002) Longitudinal brain volume measurement in
“Image processing” part. To some extent, this might bias multiple sclerosis: rate of brain atrophy is independent of the
disease subtype. Arch Neurol 59:1572–1576
the statistical volumetric results of “four decades overall 9. Bermel RA, Sharma J, Tjoa CW, Puli SR, Bakshi R (2003) A
atrophy rate”. Despite this slight uncertainty, the most semiautomated measure of whole-brain atrophy in multiple
important conclusion remains that individual 3D atrophy sclerosis. J Neurol Sci 208:57–65
rates did not significantly differ over four decades of 10. Chard DT, Griffin CM, Parker GJ, Kapoor R, Thompson AJ,
Miller DH (2002) Brain atrophy in clinically early relapsing–
disease. remitting multiple sclerosis. Brain 125:327–337
11. De Stefano N, Matthews PM, Filippi M, Agosta F, De Luca M,
Bartolozzi ML, Guidi L, Ghezzi A, Montanari E, Cifelli A,
Conclusions Federico A, Smith SM (2003) Evidence of early cortical atrophy
in MS: relevance to white matter changes and disability.
Neurology 60:1157–1162
We found out uniformly progressing atrophy representing 12. Lin X, Blumhardt LD, Constantinescu CS (2003) The relationship
four decades of MS duration, with 1D, 2D, and 3D of brain and cervical cord volume to disability in clinical subtypes
measures correlating with each other. We noticed short- of multiple sclerosis: a three-dimensional study. Acta Neurol
Scand 108:401–406
term volumetric changes disappearing over time. The 13. Turner B, Ramli N, Blumhardt LD, Jaspan T (2001) Ventricular
results indicated that the rate of atrophy might to be enlargement in multiple sclerosis: a comparison of three-
independent of MS course. Frontal horn width (1D), dimensional and linear MRI estimates. Neuroradiology 43:608–
which correlated strongest with volumetric changes, was 614
14. Stevenson VL, Miller DH, Leary SM, Rovaris M, Barkhof F,
found to be the best measure revealing atrophy progres- Brochet B, Dousset V, Filippi M, Hintzen R, Montalban X,
sion. Supratentorial ventricular enlargement correlated Polman CH, Rovira A, de Sa J, Thompson AJ (2000) One year
with both EDSS and MSSS. Serial 1D and 2D measures follow up study of primary and transitional progressive multiple
can be performed on any PACS workstation and may sclerosis. J Neurol Neurosurg Psychiatry 6:713–718
15. Pagani E, Rocca MA, Gallo A, Rovaris M, Martinelli V, Comi G,
contribute to routine evaluation of global irreversible Filippi M (2005) Regional brain atrophy evolves differently in
tissue damage in MS. patients with multiple sclerosis according to clinical phenotype.
Am J Neuroradiol 26:341–346
16. Sharma J, Sanfilipo MP, Benedict RH, Weinstock-Guttman B,
Munschauer FE 3rd, Bakshi R (2004) Whole-brain atrophy in
Conflict of interest statement We declare that we have no conflict multiple sclerosis measured by automated versus semiautomated
of interest. MR imaging segmentation. Am J Neuroradiol 25:985–996
17. Butzkueven H, Kolbe SC, Jolley DJ, Brown JY, Cook MJ, van der
Mei IA, Groom PS, Carey J, Eckholdt J, Rubio JP, Taylor BV,
Mitchell PJ, Egan GF, Kilpatrick TJ (2008) Validation of linear
References cerebral atrophy markers in multiple sclerosis. J Clin Neurosci
15:130–137
1. Bermel RA, Bakshi R (2006) The measurement and clinical 18. Kurtzke JF (1983) Rating neurologic impairment in multiple
relevance of brain atrophy in multiple sclerosis. Lancet Neurol sclerosis: an expanded disability status scale (EDSS). Neurology
5:158–170 33:1444–1452
2. Barkhof F (2002) The clinico-radiological paradox in multiple 19. Roxburgh RH, Seaman SR, Masterman T, Hensiek AE, Sawcer
sclerosis revisited. Curr Opin Neurol 15:239–245 SJ, Vukusic S, Achiti I, Confavreux C, Coustans M, le Page E,
3. Brex PA, Jenkins R, Fox NC, Crum WR, O’Riordan JI, Plant GT, Edan G, McDonnell GV, Hawkins S, Trojano M, Liquiri M,
Miller DH (2000) Detection of ventricular enlargement in patients Cocco E, Marruso MG, Tesser F, Leone MA, Weber A, Zipp F,
at earliest clinical stage of MS. Neurology 54:1689–1691 Miterski B, Epplen JT, Oturai A, Sørensen PS, Celius EG, Lara
4. Luks TL, Goodkin DE, Nelson SJ, Majumdar S, Bacchetti P, NT, Montalban X, Villoslada P, Silva AM, Marta M, Leite I,
Portnoy D, Sloan R (2000) A longitudinal study of ventricular Dubois B, Rubio J, Butzkueven H, Kilpatrick T, Mycko MP,
volume in early relapsing–remitting multiple sclerosis. Mult Scler Selmaj KW, Rio ME, Sá M, Salemi G, Savettieri G, Hillert J,
6:332 Compston DA (2005) Multiple sclerosis severity score: using
5. Fisher E, Rudick RA, Simon JH, Cutter G, Baier M, Lee JC, disability and disease duration to rate disease severity. Neurology
Miller D, Weinstock-Guttman B, Mass MK, Dougherty DS, 64:1144–1151
Simonian NA (2002) Eight-year follow-up study of brain atrophy 20. Poser CM, Paty DW, Scheinberg L, McDonald WI, Davis FA,
in patients with MS. Neurology 59:1412–1420 Ebers GC, Johnson KP, Sibley WA, Silberberg DH, Tourtellotte
Neuroradiology (2010) 52:109–117 117

WW (1983) New diagnostic criteria for multiple sclerosis: 27. Hobart J, Freeman J, Thompson A (2000) Kurtzke scales
guidelines for research protocols. Ann Neurol 13:227–231 revisited: the application of psychometric methods to clinical
21. Polman CH, Reingold SC, Edan G, Filippi M, HP Hartung, intuition. Brain 123:1027–1040
Kappos L, Lublin FD, Metz LM, McFarland HF, O’Connor PW, 28. Scahill RI, Frost C, Jenkins R, Whitwell JL, Rossor MN, Fox NC
Sandberg-Wollheim M, Thompson AJ, Weinshenker BG, Wolinsky (2003) A longitudinal study of brain volume changes in normal
JS (2005) Diagnostic criteria for multiple sclerosis: 2005 revisions to aging using serial registered magnetic resonance imaging. Arch
the “McDonald Criteria”. Ann Neurol 58:840–846 Neurol 60:989–994
22. Martola J, Stawiarz L, Fredrikson S, Hillert J, Bergström J, Flodmark 29. Bozalli M, Cercignani M, Caltagirone C (2008) Brain volumetrics to
O, Kristoffersen Wiberg M (2007) Progression of non-age-related investigate aging and the principal forms of degenerative cognitive
callosal brain atrophy in multiple sclerosis: a 9-year longitudinal decline: a brief report. Magn Reson Imaging 26:1065–1070
MRI study representing four decades of disease development. J 30. Kochunov P, Thompson PM, Coyle TR, Lancaster JL, Kochunov
Neurol Neurosurg Psychiatry 78:375–380 V, Royall D, Mangin JF, Rivière D, Fox PT (2008) Relationship
23. Martola J, Stawiarz L, Fredrikson S, Hillert J, Bergstrom J, among neuroimaging indices of cerebral health during normal
Flodmark O, Aspelin P, Kristoffersen Wiberg M (2008) Rate of aging. Hum Brain Mapp 29:36–45
ventricular enlargement in multiple sclerosis: a nine-year magnetic 31. Horsfield MA, Rovaris M, Rocca MA, Rossi P, Benedict RH,
resonance imaging follow-up study. Acta Radiol 49:570–579 Filippi M, Bakshi R (2003) Whole-brain atrophy in multiple
24. Martola J, Stawiarz L, Fredrikson S, Hillert J, Bergström J, Flodmark sclerosis measured by two segmentation processes from various
O, Aspelin P, Kristoffersen Wiberg M (2009) One-dimensional-ratio MRI sequences. J Neurol Sci 216:169–177
measures of atrophy progression in multiple sclerosis as evaluated by 32. Molyneux PD, Kappos L, Polman C, Pozzilli C, Barkhof F,
longitudinal magnetic resonance imaging. Acta Radiol 1–9 Filippi M, Yousry T, Hahn D, Wagner K, Ghazi M, Beckmann K,
25. Armatige P, Berry G, Matthews JNS (2002) Statistical methods in Dahlke F, Losseff N, Barker GJ, Thompson AJ, Miller DH (2000)
medical research, 4th edn. Blackwell, Oxford, p 496 The effect of interferon beta-1b treatment on MRI measures of
26. Anderson VM, Fox NC, Miller DH (2006) Magnetic resonance cerebral atrophy in secondary progressive multiple sclerosis.
imaging measures of brain atrophy in multiple sclerosis. J Magn European Study Group on interferon beta-1b in secondary
Res Imaging 23:605–618 progressive multiple sclerosis. Brain 123:2256–2263