hypertension and neuronal degeneration in excised rat spinal … · diffusion-weighted imaging...
TRANSCRIPT
www.elsevier.com/locate/yexnr
Experimental Neurology 184 (2003) 726–736
Hypertension and neuronal degeneration in excised rat spinal cord studied
by high-b value q-space diffusion magnetic resonance imaging
Yaniv Assaf,a,b,* Adi Mayk,c Sarah Eliash,d Zipora Speiser,d and Yoram Cohena
aSchool of Chemistry, The Sackler Faculty of Exact Sciences, Tel Aviv University, Tel Aviv 69978, IsraelbDepartment of Radiology, Tel Aviv Sourasky Medical Center, Wohl Institute for Advanced Imaging, Tel Aviv 64239, Israel
cTEVA Pharmaceutical Industries Ltd. and Sackler School of Medicine, Tel Aviv University, Tel Aviv 69978, IsraeldDepartment of Physiology and Pharmacology, Sackler School of Medicine, Tel Aviv University, Tel Aviv 69978, Israel
Received 2 February 2003; revised 11 May 2003; accepted 19 May 2003
Abstract
Hypertension is one of the major risk factors of stroke and vascular dementia (VaD). We used stroke prone spontaneous hypertensive rats
(SPSHRs) as a model for neuronal degeneration frequently occurring in humans with vascular disease. Recently, high b value q-space
diffusion-weighted imaging (DWI) was shown to be very sensitive to the pathophysiological state of the white matter. We studied the spinal
cords of SPSHR rats ex vivo after the appearance of motor impairments using diffusion anisotropy and q-space diffusion imaging (measured
at a high b value of up to 1 � 105 s/mm2).
The diffusion anisotropy images computed from low b value data set (bmax approximately 2500 s/mm2) showed a small but statistically
significant decrease (approximately 12%, P < 0.05) in the diffusion anisotropy in the spinal cords of the SPSHR group as compared to control
rats. However, more significant changes were found in the high b value q-space diffusion images. The q-space displacement values in the
white matter of the SPSHR group were found to be higher by more than 70% (P < 0.002) than that of the control group. These observations
concurred with electron microscopy (EM) that showed significant demyelination in the spinal cords of the SPSHR group. These results seem
to indicate that high b value q-space DWI might be a sensitive method for following demyelination and axonal loss associated with vascular
insults.
D 2003 Elsevier Inc. All rights reserved.
Keywords: MRI; Stroke; Hypertension; Diffusion; q-Space DWI
Introduction matter lesions appear hyperintense in T -weighted images at
Hypertension is one of the major risk factors for ischemic
white matter lesions (also termed leukoaraiosis) (Liao et al.,
1996; Skoog, 1998; van Swieten et al., 1991). Indeed,
vascular dementia (VaD) is one of the three most common
causes of dementia among the elderly (Johansson, 1994; Lis
and Gaviria, 1997; van Gijn, 1998). It is also known that
stroke increases the risk of VaD by more than 9-fold
(Johansson, 1994; Lis and Gaviria, 1997; van Gijn, 1998).
Magnetic resonance imaging (MRI) is one of the most
valuable diagnostic tools of leukoaraiosis since the white
0014-4886/$ - see front matter D 2003 Elsevier Inc. All rights reserved.
doi:10.1016/S0014-4886(03)00274-7
* Corresponding author. Department of Radiology, Tel Aviv Sourasky
Medical Center, Wohl Institute for Advanced Imaging, Tel Aviv 64239,
Israel. Fax: +972-3-6973080.
E-mail address: [email protected] (Y. Assaf).
2
the chronic stages of the disease (Giubilei et al., 1997;
Wahlund, 1994). Histopathological studies of these lesions
revealed demyelination and axonal loss (Johansson, 1994;
Liao et al., 1996; Lis and Gaviria, 1997; Skoog, 1998; van
Gijn, 1998; van Swieten et al., 1991). The clinical signs of
VaD include deterioration in cognitive function, reduced
speed of mental processes, concentration, verbal and visual
memory (Johansson, 1994; Lis and Gaviria, 1997; van Gijn,
1998) that can be related to demyelination and axonal loss in
the white matter.
Diffusion-weighted imaging (DWI) and diffusion anisot-
ropy measurements are widely used today in studying white
matter in VaD (Choi et al., 2000; Ellis et al., 1999; Hanyu et
al., 1997; Jones et al., 1999a; Ono et al., 1997; Rose et al.,
2000; Werring et al., 2000). Several studies showed that T2
hyperintensity in the white matter of VaD patients is usually
Y. Assaf et al. / Experimental Neurology 184 (2003) 726–736 727
accompanied by an increase in the apparent diffusion coef-
ficient (ADC) (Choi et al., 2000; Helenius et al., 2002; Jones
et al., 1999b; O’Sullivan et al., 2001) and a decrease in the
diffusion anisotropy (Jones et al., 1999b; O’Sullivan et al.,
2001). Recently, it was shown that high b value diffusion
imaging produces images that are more specific to neuronal
water signals (Assaf et al., 2000, 2002a). The specificity of
the water signal at high b values to intra-axonal water was
demonstrated by diffusion spectroscopy and imaging on
several neuronal tissues (Assaf and Cohen, 1998, 2000; Assaf
et al., 2000, 2002a). In contrast to low b value DWI (b values
of less than 2000 s/mm2), the signal decay at high b value
diffusion experiments is not mono-exponential (Assaf and
Cohen, 1998, 2000; Assaf et al., 2000, 2002a; Niendorf et al.,
1996) and therefore should be analyzed in a way other than
low b value DWI. One approach is to analyze the non-mono-
exponential signal decay by the q-space analysis (Assaf and
Cohen, 1998, 2000; Assaf et al., 2000, 2002a; Cory and
Garroway, 1990; King et al., 1997). The q-space approach
enables the extraction of structural information of the sample
in cases of restricted diffusion (Callaghan et al., 1991; Cory
and Garroway, 1990). It should be noted that with the q-space
analysis, we directly relate the signal decay to displacement
space (through the q value) in contrast to conventional ADC
mapping which is related to the b value. This is an advantage
since the displacement space is the measured quantity in
diffusion measurements rather than the diffusion coefficient
that is extracted when analyzing the signal decay as a function
of the b values.
In contrast to other analysis methods that resort to
complicated mathematical models, the main advantage of
the q-space approach is that it is relatively model-free. The
q-space analysis is performed by Fourier transformation of
the signal decay with respect to the q value (defined as cdg/2p, where c is the gyro-magnetic ratio, d is gradient pulse
duration, and g is the pulsed gradient strength) to obtain the
displacement distribution profile of molecules in the system
according to Eq. (1) (Callaghan et al., 1991; Cory and
Garroway, 1990).
EDðqÞ ¼Z
P̄sðR;DÞexpði2pq � RÞdR ð1Þ
In this equation, ED( q) is the signal decay, P̄s(R, D) is thedisplacement distribution profile, D is the diffusion time,
and R is the net displacement. Using the q-space analysis, it
is possible to extract the displacement distribution profile
for water molecules in relatively complex systems. It was
shown that this function can be quantified using two
parameters: the displacement, calculated from the full-width
at half-height of the displacement profile, and the probabil-
ity for zero displacement, calculated from the intensity of
the displacement profile, for a certain diffusion time (Assaf
et al., 2000). Such q-space analyzed diffusion images of rat
spinal cord at different times after birth were found to be
very sensitive to myelination and the process of neuronal
maturation (Assaf et al., 2000). Recently, this approach was
found to be extremely sensitive for following white matter
degeneration in multiple sclerosis (MS) (Assaf et al.,
2002a).
Stroke prone spontaneous hypertensive rats (SPSHRs)
spontaneously develop hypertension, leading to multifocal
stroke lesions in the brain (Blezer et al., 1999; Takahashi et
al., 1993). The severity of the hypertension and stroke
lesions increases when the rats are subjected to a high-salt
diet. This chronic hypertension is accompanied by motor
and cognitive impairments, probably due, inter alia, to the
formation of small multiple ischemic brain lesions (Blezer et
al., 1999; Takahashi et al., 1993) which seems to cause
axonal degeneration and axonal loss. In this study, we used
SPSHRs that were nourished with a high-salt diet, which
developed severe motor impairment that in some cases
resulted in paralysis. We postulated that although the main
pathology in this model occurs in the brain, the degeneration
of neuronal pathways might also be observed in the spinal
cord. In the present study, we studied the spinal cords of
control and SPSHR, ex vivo, by comparing the results
obtained from high b value q-space diffusion imaging,
low b value diffusion anisotropy, T1-weighted imaging,
and electron microscopy (EM).
Materials and methods
Animal and tissue preparation
Male SPSHRs (n = 4) were purchased from Iffa Credo,
France. At 6–7 weeks of age, the rats were fed with
Japanese stroke-prone diet (Zeigler Bros., Gardiner, PA,
USA) together with drinking fluid of 1% NaCl. As the
control group, we used age-matched Wistar rats (n = 3)
which had normal blood pressure and were not subjected to
the high-salt diet. As early as 3 weeks after the onset of the
high-salt diet, the rats showed increased hypertension, and
were followed up to 2 months thereafter. During this time
period, the rats underwent behavioral and neurological
examinations and blood pressure measurements. At the
age of about 6 months, after the rats developed severe
motor dysfunction, they were sacrificed by an overdose of
pentobarbital (300 mg/kg); their spinal cords were excised
after injection of formalin solution through the aorta to
provide better fixation. This procedure lasted several
minutes. Afterwards, the spinal cord was exposed from
the brainstem and was cut at the upper edge of C1 and
below the thorax. We then cut the specimen at the level of
the cervical cord (C3–T1) and kept for 7–21 days in 10%
formalin solution until the MRI experiments. To avoid
damage to the spinal cord, it was excised together with
the vertebrae and placed in the MRI tube according to Fig.
1a. After the completion of the MRI experiments, the spinal
cords were carefully removed from the vertebrae and
prepared for EM.
Fig. 1. (a) Experimental setup showing the location of the spinal cord sample in the NMR tube and MRI slice location superimposed. (b) T1-weighted MR
image of excised, formalin-fixed, rat spinal cord. The two regions of interest (ROI) that were used throughout this study are depicted on the image. The ROI
were chosen to represent the white and the gray matter of the spinal cord.
Y. Assaf et al. / Experimental Neurology 184 (2003) 726–736728
Neurological and blood pressure examinations
Before the MRI experiment, the blood pressure and
behavioral parameters of the rats were evaluated weekly,
as previously reported (Eliash et al., 2001). Briefly, the
systolic blood pressure and heart rates were measured in the
animals by tail-cuff sphygmomanometry (Narco Biosystems
Inc., Houston, TX, USA). Neurological examinations con-
sisted of examinations of seizure, paralysis, loss of body
symmetry, loss of motor coordination, gait response, and
balance. Most parameters were rated as either 0 (normal) or
1 to 3 depending on severity of incapacitation, with a
maximum score of 12.
MRI experiments
MRI experiments were performed on an 8.4T spectrom-
eter (Bruker, Germany) equipped with a micro5 imaging
probe (Bruker). MRI was performed on three axial slices
chosen perpendicular to the long axis of the cord (3 mm
thickness with 2 mm gap between slices, FOV of 15 � 15
mm—see Fig. 1a). The q-space data set was acquired using
a stimulated echo DWI sequence with the following
parameters: TR/TE/D/d = 500/30/150/2 ms. The diffusion
gradients were incremented between 0 and 15 G mm�1 in
16 equal steps and applied perpendicular to the long axis of
the cord. The maximal b value (bmax) in these experiments
was 9.6 � 104 s/mm2 and the maximal q value ( qmax) was
127.7 mm�1. To evaluate the effect of diffusion anisotropy
at low b values, we acquired another set of diffusion
images with the following parameters: TR/TE/D/d = 500/
30/70/2 ms and maximal gradient strength of approximate-
ly 3.5 G mm�1 resulting in maximal b value of approxi-
mately 2500 s/mm2. The diffusion gradients were applied
twice: once parallel (represented by the z symbol) and
once perpendicular (represented by the ? symbol). The
diffusion anisotropy index was calculated using the follow-
ing equation:
r ¼ ðADCz � ACD?Þ=ðADCz þ ADC?Þ ð2Þ
where ADCz and ADC? are the apparent diffusion coef-
ficients measured from the signal decay up to b value of
2500 s/mm2 using the Stejskal–Tanner equation:
lnðE=E0Þ ¼ �c2d2g2ðD � d=3ÞADC ¼ �bADC ð3Þ
where c, d, g, and D have the above meanings, E and E0
are the signal intensity in the presence and absence of the
diffusion gradients, respectively. ADC is the apparent
diffusion coefficient and b represents the entire diffusion
weighting. The acquisition of the high b value data set
lasted 1 h and the low b value data set 40 min. The total time
of the MRI protocol (T1, conventional DWI and q-space
diffusion data) was 2 h.
q-Space image analysis
Image analysis of the q-space data set was performed as
described before using a MatlabR program (Assaf et al.,
2000). The q-space data set was extrapolated, using a
multi-exponential decay function, to 128 data points to
increase FT resolution and to avoid FT ringing effects.
Then the signal decay in each pixel was transformed into
displacement distribution profiles using Eq. (1). The Four-
ier transformation of the signal decay with respect to q
produced a displacement distribution profile for each of the
pixels in the image. Two parameters of the displacement
distribution profile, the displacement (calculated from the
Y. Assaf et al. / Experimental Neurology 184 (2003) 726–736 729
full-width at half-height) and the probability for zero
displacement (given by the height of the displacement
distribution profile) were then extracted for each pixel in
the image. Finally, two sub-images, displacement and
probability images, were constructed based on these two
parameters (Assaf et al., 2000).
Statistical data analysis
Fig. 1b shows a typical T1-weighted image of a rat spinal
cord with two regions of interest—one in the white matter
and one in the gray matter. All numerical data extracted for
the white matter and gray matter throughout this work
represent these ROI. The data were not analyzed in a
blinded manner because the image analysis was an auto-
mated computed procedure and ROI selection did not
require any subjective decision-making. For each rat, the
diffusion parameters (displacement, probability, FA, and
ADC) were extracted per slice for each of the two ROI.
The data used for statistics consisted, per each ROI and
diffusion parameter (displacement, probability, FA, and
ADC), four expectation values for the SPSHR group and
three expectation values for the control group. The two
Fig. 2. (a) T1-weighted images on which two pixels are presented, in the gray matte
pixels shown in (a). (c) Signal decay as a function of the q value of the pixels sh
Arrows represent the full-width at half-maximum and the peak of the displaceme
groups were tested for statistical significance difference
using a Student’s t test.
Results
All SPSHRs gradually developed hypertension starting
from a systolic blood pressure of 120 F 11 mm Hg at the
beginning of the experiment and reaching a value of 199 F36 mm Hg at 6 weeks post-initialization of the high-salt diet.
All rats that were studied developed reduced motor capac-
ities manifested in paralysis (two out of four rats), poor
walking on a beam (three out of four rats), and poor grasp
and balance on a beam (four out of four rats). Control group
rats did not develop any of these signs.
Fig. 2 shows an example for q-space analysis for a pixel
in the white matter and in the gray matter as shown in Fig.
2a. Fig. 2b depicts the signal decay as a function of the b
values showing the non-mono-exponential relation between
the decay and the b values, especially in the white matter
pixel. Fig. 2c shows the same data in Fig. 2b but as a
function of the q values showing Gaussian-shaped signal
decay. Fig. 2d shows the Fourier transformation of the
r and in the white matter. (b) Signal decay as a function of the b values of the
own in (a). (d) Displacement distribution profile of the data shown in (c).
nt profile.
Fig. 3. Complete MRI data set on two excised spinal cords of the control group. (a, e) T1-weighted MR image; (b, f) low b value diffusion anisotropy index (r)image; (c, g) high b value q-space displacement image; and (d, h) high b value q-space probability image.
Y. Assaf et al. / Experimental Neurology 184 (2003) 726–736730
signal decay in Fig. 2c. The vertical arrows represent the
full-width at half-maximum of the displacement profile from
which the displacement value is calculated. The horizontal
arrows represent the peak of the displacement profile from
which the probability for zero displacement is calculated.
Figs. 3 and 4 show the MRI data sets for the spinal cords
of two representative control and SPSHR rats. The MRI data
set consists of a T1-weighted image, low b value diffusion
anisotropy index image, and high b value q-space displace-
ment and probability images. The contrast in all images
looks similar giving good discrimination between gray and
white matter. Usually, in vivo T1-weighted images of rat
spinal cord show only a weak contrast between gray and
white matter. However, the spinal cords used in this study
were fixed in formalin which seems to shorten the T1 of the
gray matter more than the white matter (Tovi and Ericsson,
1992) and therefore causes the white matter areas to appear
Fig. 4. Complete MRI data set on two excised spinal cords of the SPSHR group. (a
image; (c, g) high b value q-space displacement image; and (d, h) high b value q
hypointense as compared to the gray matter in both groups
(Figs. 3a, e and 4a, e).
The low b value (bmax approximately 2500 s/mm2)
diffusion anisotropy index images of the spinal cords of
two control rats and two SPSHRs are shown in Figs. 3b, f
and 4b, f, respectively. For these particular spinal cords, the
anisotropy index seems to be similar both in the gray and
white matter. However, for the whole group, the differences
in the diffusion anisotropy index of the white matter were
found to be small but statistically significant (0.83 F 0.12
and 0.73 F 0.12, P < 0.05 for the control and the SPSHR
groups, respectively). In the gray matter, the values for the
two groups were found to be similar (0.45 F 0.12 and
0.40 F 0.13, nonsignificant, for the control and the
SPSHR groups, respectively). The observed differences
in the diffusion anisotropy in the white matter were found
to result mainly from the differences in the ADC extracted
, e) T1-weighted MR image; (b, f) low b value diffusion anisotropy index (r)-space probability image.
Fig. 5. (a) Diffusion anisotropy index values, (b) displacement values, (c) probability for zero displacement values, (d) ADCz values, and (e) ADC? values for
the gray and white matter of the control and SPSHR groups. The values are mean and standard deviations over the whole group. The values were extracted
from the ROI presented in Fig. 1. The symbols (*) and (**) represent significant changes of P < 0.05 and P < 0.002, respectively.
Fig. 6. Electron microscopy of a white matter section from a control rat spinal cord (a) (magnification, � 1500). (b, c) EM images of a white matter section from
SPSHR spinal cords (magnification, � 1500). The bars in a–c are of 5 Am. Arrows and asterisks represent damaged axons. (d) EM image of one axon of a
control rat spinal cord (magnification, � 5000). (e, f) EM images of two axons from SPSHR spinal cords (magnification, � 5000). The bars in d– f represent
1 Am.
Y. Assaf et al. / Experimental Neurology 184 (2003) 726–736 731
Fig. 7. (a) The signal decay from the low b value anisotropy experiments.
Squares and circles represent signal decay in the white matter of the control
group measured perpendicular and parallel to the spinal cord, respectively.
Up and down triangles represent signal decay in the white matter of the
SPSHR group measured perpendicular and parallel to the spinal cord,
respectively. (b) The signal decay from the high b value diffusion
experiments. In the high b value experiments, the diffusion images were
acquired perpendicular to the long axis of the spinal cord. Squares represent
the control group and circles represent the SPSHR group.
Y. Assaf et al. / Experimental Neurology 184 (2003) 726–736732
perpendicular to the long axis of the fibers. In the gray
matter, ADCz were found to be 3.3 F 0.8 and 4.1 F 0.7
(�10�6 cm2/s) for the control and SPSHR group, respec-
tively, while in the white matter, these values were found
to be 4.0 F 1.3 and 3.8 F 1.2 (�10�6 cm2/s) for the
control and SPSHR groups, respectively. In the gray
matter, ADC? were found to be 1.2 F 0.3 and 1.7 F0.3 (�10�6 cm2/s) for the control and SPSHR group,
respectively, while in the white matter, these values were
found to be 0.5 F 0.3 and 0.7 F 0.4 (�10�6 cm2/s) for
the control and SPSHR groups, respectively. These results
are graphically summarized in Figs. 5a, d, and e.
High b value, q-space analyzed displacement images are
shown in Figs. 3c, g and 4c, g for control and SPSHR spinal
cords, respectively. For these particular spines, the mean
displacement in the white matter is much higher for the
SPSHR than the control rat (Figs. 3c, g vs. Figs. 4c, g), while
in the gray matter, similar displacements are observed. The
mean displacement values for the control and SPSHR groups
were found to be 2.2 F 0.3 and 3.8 F 0.6 Am (P < 0.002),
respectively. In the gray matter, no significant differences
were found in the mean displacement between the two
groups (6.1 F 1.3 and 6.4 F 1.5 Am, NS, for the control
and SPSHR groups, respectively). These results are graph-
ically summarized in Fig. 5b.
Figs. 3d, h and 4d, h show high b value, q-space analyzed
probability images for control and SPSHR spinal cords. For
these particular spinal cords, the mean probability for zero
displacement for the SPSHR is much lower than that of the
control rat (Figs. 3d, h vs. Figs. 4d, h), while in the gray
matter, similar probabilities are observed. Mean probability
values in the white matter for the control and SPSHR spinal
cord groups were found to be 10.8 F 1.4 and 6.5 F 0.9 a.u.
(P < 0.001), respectively. In the gray matter, no significant
differences were found between the two groups for the
probability for zero displacement (4.1 F 0.6 Am and 4.1
F 0.9 a.u., for the control and SPSHR groups, respectively).
These results are graphically summarized in Fig. 5c.
Fig. 6 shows EM images taken from the white matter of
the control (Fig. 6a) and SPSHR (Figs. 6b, c) spinal cords.
The axons of the control spinal cord (Fig. 6a, zoomed at Fig.
6d) seem to be intact and the myelin wraps around the axon
are thicker and less damaged than the SPSHR spinal cord
where some axons seem to be disrupted (see, e.g., the
arrows in Figs. 6b, c). The myelin layers of most axons in
the SPSHR seem to be damaged to variable extents (see the
asterisks in Figs. 6b, c). The main pathologies observed are
the formation of myelin debris in the area of the axon (Fig.
6e) and the formation of large vacuoles between the axon
and the disrupted myelin layers (Fig. 6f). These pathologies
produce structural damage in the SPSHR spinal cord that
might be related to demyelination.
Fig. 7 shows signal decay as a function of the b values for
the low b value diffusion anisotropy experiment (Fig. 7a) and
the high b value experiment (Fig. 7b) extracted from the white
matter ROI and averaged over all rats. The low b value
diffusion signal decay (up to b of 2500 s/mm2) shows very
little change between the two groups. The differences become
much more apparent at high b values (b > 10,000 s/mm2).
Fig. 8 shows pixel-based histograms of the low b value
anisotropy and the high b value q-space displacement and
probability values. In contrast to the ROI analysis, here it is
possible to observe that the low b value anisotropy index
detects some differences between the two groups (Fig. 8a).
Some pixels that were at the high anisotropy region (r >
0.8) now seem to have lower anisotropy values of less than
0.7. However, much larger differences are observed for the
displacement and probability parameters extracted from the
high b value q-space analysis. The displacement values of
between 1 and 3 Am which represent normal white matter
displacement values seem to shift in the SPSHR histogram
Fig. 8. Histograms of the pixel distributions as a function of the low b value
diffusion anisotropy (a), the high b value displacement (b), and the high b
value probability for zero displacement (c).
Y. Assaf et al. / Experimental Neurology 184 (2003) 726–736 733
to much higher displacement values in the range of 3 to 6
Am (Fig. 8b). The probability for zero displacement gives
the best discrimination between gray and white matter in the
control group (Fig. 8c). The SPSHR histogram shown in
Fig. 8c shows that the probability for zero displacement
having values in the range of 8 to 14 (values that represent
normal white matter in the control group) shifted to lower
values in the range of 2 to 8 for the SPSHR group.
Discussion
Hypertension and chronic vascular diseases are usually
characterized by lesions in the white matter, which are seen
as hyperintense areas on T2-weighted images (also termed
leukoaraiosis) (Giubilei et al., 1997; Wahlund, 1994). These
lesions can be detected in severe stages of the disease and
are usually accompanied by a progressive decline of the
motor and/or cognitive functions of the patient (Giubilei et
al., 1997; Johansson, 1994; Liao et al., 1996; Lis and
Gaviria, 1997; Skoog, 1998; van Gijn, 1998; van Swieten
et al., 1991; Wahlund, 1994). It is believed that axonal loss
and demyelination after transient ischemic attacks are the
major mechanisms that lead to this pathology (Johansson,
1994; Liao et al., 1996; Lis and Gaviria, 1997; Skoog, 1998;
van Gijn, 1998; van Swieten et al., 1991). Yet, it has been
suggested that the damaged white matter areas (leukoaraio-
sis) might not represent the entire damage to the white
matter. Indeed, the absence of myelin may not be necessar-
ily detected in the conventional MR images (Rooney et al.,
1997). T1- and T2-weighted images, which are sensitive to
the relaxation times of the water protons, will change when
demyelination and axonal loss are accompanied by inflam-
mation or severe tissue loss. In cases of initial demyelination
or non-inflammatory damage, conventional MRI might not
detect the entire abnormalities.
In this study, we examined the spinal cords of SPSHRs,
ex vivo, to evaluate the ability of high b value q-space
diffusion MRI for characterization of axonal degeneration
following vascular insult. Although the primary neuronal
pathology of this model is in the brain, we chose to study
the spinal cord of these rats after the occurrence of motor
abnormalities for several reasons: firstly, we wanted to
verify if this model of chronic hypertension produces spinal
cord damage (axonal degeneration) which can be detected
by MRI. In addition, neuronal damage in the brain can result
in inflammation and other cellular degeneration processes
that could defy detection and isolation of axonal and/or
myelin degeneration. This study was performed ex vivo to
provide high-quality comparison between low and high b
value diffusion MR imaging. Because in vivo high b value
diffusion MRI is more sensitive to motion artifacts and has
inherent low signal-to-noise ratio, it requires long acquisi-
tion times. These technical problems make the acquisition of
such diffusion data on spinal cord of small animals, in vivo,
very challenging. Formalin fixation is known to change
relaxation times and therefore influence the measured signal
in MRI (Tovi and Ericsson, 1992). The effect of formalin on
diffusion measurements has not been studied. However,
previous examples have shown that formalin-fixated tissue
can be used to obtain structural features both before and
after damage development (Lester et al., 2000; Nossin-
Manor et al., 2002). Therefore, as no major structural
changes should occur in the tissue due to the fixation and
because of the large differences between the groups in the
high b value q-space diffusion data, we believe that ex vivo
Y. Assaf et al. / Experimental Neurology 184 (2003) 726–736734
measurements are relevant at least for probing the efficiency
of measuring diffusion to follow structural damage to the
tissue. The potential diagnostic power of the above meth-
odology in the in vivo situation of the spinal cord remains to
be explored.
Diffusion imaging is a well-established modality in
MRI. It is mainly used for diagnosis of brain pathologies
such as stroke (Darquie et al., 2001; Eis et al., 1995;
Hasegawa et al., 1995; Le Bihan, 1995; Moseley et al.,
1990a). Water diffusion in white matter, as measured by
MRI, was found to be anisotropic (Moseley et al., 1990b).
Several studies were devoted to evaluation of the influence
and relative contribution of the myelin membrane to this
anisotropy (Basser et al., 1994; Beaulieu et al., 1998;
Gulani et al., 2001; Nevo et al., 2001; Ono et al., 1995),
as the myelin membrane was believed to be the main cause
for the apparent restricted diffusion perpendicular to the
long axis of the axons. However, it was shown that
diffusion anisotropy could also be found in the non-
myelinated olfactory nerve of the garfish and that the
degree of anisotropy there was similar to that in the
myelinated trigeminal and optic nerves (Beaulieu et al.,
1998). Furthermore, a study on transgenic dysmyelinated
mice has shown that diffusion anisotropy is unaffected
when multiple myelin wraps are absent (Ono et al.,
1995). By contrast, other studies on animal models of
demyelination have shown that the diffusion anisotropy is
significantly reduced when myelin is disrupted (Gulani et
al., 2001; Nevo et al., 2001). This reduction was attributed
to a decrease in the ADC measured perpendicular to the
long axis of the axons.
Diffusion anisotropy measurements are usually done
with low b values in the range of 750 and 2000 s/mm2. It
has been suggested that anisotropy measured at such low b
values mainly represents the diffusion of water molecules in
the extracellular space (Kraemer et al., 1999). This could
explain the fact that myelin has a low influence on the
anisotropy values measured at low b values. In our measure-
ments, we detected a 12% reduction in the anisotropy values
between the control and the SPSHR group (Fig. 5a). This
reduction was not apparent from the signal decay curves
(Fig. 7a), but only after calculation of the anisotropy index
(r) and averaging all the spinal cords, the differentiation
between the two groups became significant (Fig. 8a).
Despite the major damage to the myelin membrane, as
revealed by EM (Fig. 6), only relatively small changes in
the diffusion anisotropy were observed, supporting the
notion that the myelin is not, as already discussed, the only
contributor to diffusion anisotropy when measured by low b
value diffusion imaging.
The signal decay at the low b value range can be
analyzed by the Stejskal–Tanner equation (Eq. (3)) (Le
Bihan, 1995) as a linear relationship is found between the
logarithm of the normalized signal decay and the b values
(Fig. 7a). This relationship holds in the case of free
diffusion (Le Bihan, 1995). However, the diffusion of
molecules in the neuronal tissues is far from being free.
Cellular organelles, intracellular cytoskeleton, connective
tissue, proteins, and cellular membranes cause the diffusion
to be hindered and restricted. In these cases, a deviation
from the linear relationship between the normalized loga-
rithm of signal decay and the b values should be observed
(Callaghan et al., 1991; Coy and Callaghan, 1994; Kuchel
et al., 1997). Indeed, when diffusion is measured in neuro-
nal tissues with b values higher than 2500 s/mm2, a non-
mono-exponential signal decay is observed (both in vivo
and ex vivo) (Assaf and Cohen, 1998, 2000; Assaf et al.,
2000, 2002a,b; Mulkern et al., 1999; Niendorf et al., 1996;
Peled et al., 1999; Stanisz et al., 1997; and also in this
study) (Fig. 7b). The magnitude of the apparent slow-
decaying component, extracted from the multi-exponential
signal decay, was found to be larger in areas of white matter
(Assaf and Cohen, 2000), and was also found to be much
larger when measured perpendicular to the long axis of the
neuronal fibers (Assaf and Cohen, 2000). Moreover, when
myelin was absent, the slow-decaying component was
found to be very small or reduced (Assaf et al., 2000) or
disrupted (Assaf et al., 2002b). Based on these observa-
tions, this slow-decaying component was tentatively
assigned to restricted diffusion of intra-axonal water mole-
cules (Assaf and Cohen, 2000). This component is more
apparent in white matter mainly due to the much slower
exchange rate between intra- and extra-axonal water in such
tissues. The slow exchange rate will cause the water motion
within the axon to be restricted during the diffusion time
used in this study. Indeed, in the model presented here,
much larger differences were observed between the two
groups at high b values (Fig. 7b). This supports the
assumption that the signal decay at high b values is more
sensitive to the pathophysiological state of myelin and the
axonal integrity than low b value DWI (compare Figs. 7a
and b).
In this study, we used the q-space approach to analyze
the diffusion data as it enables analysis without the need
to resort to any explicit model. The q-space analysis
produces the displacement distribution profile from the
signal decay using the Fourier relation given in Eq. (1)
(Assaf and Cohen, 2000; Assaf et al., 2000; Cory and
Garroway, 1990). This method was previously applied to
study ex vivo spinal cord maturation (Assaf et al., 2000)
and in vivo neurodegeneration in MS on humans using a
clinical MRI scanner (Assaf et al., 2002a). Using this
methodology, two sub-images (i.e., the displacement and
probability for zero displacement) are obtained from the
displacement distribution profile (Figs. 3c, g and d, h). As
this set of images takes into account the whole signal
decay (low and high b values) and emphasizes the high b
value range, a highly significant difference was observed
between the control and the SPSHR groups (Figs. 5b, c).
From the ROI analysis, an approximately 73% increase in
the displacement values was observed in the white matter,
suggesting that water molecules can translate larger dis-
Y. Assaf et al. / Experimental Neurology 184 (2003) 726–736 735
tances in the spinal cords of the SPSHR group, since
water diffusion is less restricted. This agrees with the EM
that shows the formation of large vacuoles between the
axons and the disrupted myelin layers. These structural
changes imply that the mean displacement of the intra-
axonal water molecules should increase. The probability
for zero displacement value showed a 40% reduction
between the control and SPSHR groups. As the probabil-
ity for zero displacement represents the probability of
water molecules to stay close to their point of origin
(which is high in cases of restricted diffusion), this
reduction is also expected with the loss of restricted
diffusion. In general, the differences observed in the q-
space parameters were much more significant than those
observed using the low b value diffusion anisotropy
images (compare Fig. 5a to Figs. 5b and c and Fig. 8a
to Figs. 8b and c).
The q-space analyzed diffusion images provided a
significant discrimination between the control and SPSHR
groups. Also, the changes observed in these images are
in line with the histological findings. High b value, q-
space analyzed diffusion MRI is believed to be more
sensitive to myelin disorders than low b value diffusion
anisotropy, and is indeed found to correlate better with
the EM findings. The use of the q-space method in vivo
faces significant problems because it requires the acqui-
sition of several images with very high b values in which
the signal-to-noise ratio is poor. Therefore, the long
acquisition times combined with respiratory motion
makes the in vivo application of this method complicated,
especially for spinal cord studies. Nevertheless, these
problems can be partially overcome with the use of fast
acquisition techniques such as diffusion-weighted echo-
planar imaging (DW-EPI). Indeed, the q-space approach
was recently applied to study demyelination in human
subjects with MS (Assaf et al., 2002a) using DW-EPI.
Significant changes were observed in the displacement
and probability values in areas of MS lesions and, more
importantly, in areas of normal appearing white matter
(NAWM). This result, and the findings in the present
study, imply that the high b value q-space analyzed
diffusion imaging can follow demyelinating processes
with high sensitivity. It has been demonstrated that this
method can be used for characterization of neuronal
damage following vascular insult occurring in this exper-
imental model.
Acknowledgments
The authors thank the Adams Super Center for Brain
Research of Tel Aviv University for financial support.
Support of this project by the German Federal Ministry of
Education and Research (BMBF) within the framework of
German–Israeli Project (DIP) is gratefully acknowledged.
References
Assaf, Y., Cohen, Y., 1998. Non-monoexponential attenuation of the water
and N-acetyl aspartate signals due to diffusion in brain tissue. J. Magn.
Reson. 131, 69–85.
Assaf, Y., Cohen, Y., 2000. Assignment of the water slow diffusing com-
ponent in CNS using q-space diffusion MRS: implications to fiber tract
imaging. Magn. Reson. Med. 43, 191–199.
Assaf, Y., Mayk, A., Cohen, Y., 2000. Displacement imaging of spinal
cord using q-space diffusion weighted MRI. Magn. Reson. Med. 44,
713–722.
Assaf, Y., Ben-Bashat, D., Chapman, J., Peled, S., Biton, I.E., Kafri, M.,
Segev, Y., Hendler, T., Korczyn, A.D., Graif, M., Cohen, Y., 2002a.
Evaluation of the physiological state of white matter by high b value
q-space analyzed diffusion-weighted imaging: application to multiple
sclerosis. Magn. Reson. Med. 47, 115–126.
Assaf, Y., Kafri, M., Shinar, H., Chapman, J., Korczyn, A.D., Navon, G.,
Cohen, Y., 2002b. Changes in axonal morphology in experimental al-
lergic neuritis as studied by q-space 1H and 2H DQF diffusion magnetic
resonance spectroscopy. Magn. Reson. Med. 48, 71–81.
Basser, P.J., Le-Bihan, D., Mattiello, J., 1994. MR diffusion tensor spec-
troscopy and imaging. Biophys. J. 66, 259–267.
Beaulieu, C., Fenrich, F.R., Allen, P.S., 1998. Multicomponent water
proton transverse relaxation and T2-discriminated water diffusion in
myelinated and nonmyelinated nerve. Magn. Reson. Imaging 16,
1201–1210.
Blezer, E.L., Nicolay, K., Viergever, M.A., Koomans, H.A., Joles, J.A.,
1999. MRI-based quantification of cerebral edema in individual SHRSP
rats using averaged criteria determined before the occurrence of edema.
Magn. Reson. Imaging 17, 903–907.
Callaghan, P.T., Coy, A., MacGowan, D., Packer, K.J., Zelaya, F.O., 1991.
Diffraction-like effects in NMR diffusion studies of fluids in porous
solids. Nature 351, 467–469.
Choi, S.H., Na, D.L., Chung, C.S., Lee, K.H., Na, D.G., Adair, J.C., 2000.
Diffusion-weighted MRI in vascular dementia. Neurology 54, 83–89.
Cory, D.G., Garroway, A.N., 1990. Measurement of translational displace-
ment probabilities by NMR: an indicator of compartmentation. Magn.
Reson. Med. 14, 435–444.
Coy, A., Callaghan, P.T., 1994. Pulsed gradient spin-echo NMR diffusive
diffraction experiments on water surrounding close-packed polymer
spheres. J. Colloid Interface Sci. 168, 373–379.
Darquie, A., Poline, J.B., Poupon, C., Saint-Jalmes, H., Le Bihan, D.,
2001. Transient decrease in water diffusion observed in human occipital
cortex during visual stimulation. Proc. Natl. Acad. Sci. U. S. A. 98,
9391–9395.
Eis, M., Els, T., Hoehn-Berlage, M., 1995. High resolution quantitative
relaxation and diffusion MRI of three different experimental brain tu-
mors in rat. Magn. Reson. Med. 34, 835–844.
Eliash, S., Speiser, Z., Cohen, S., 2001. Rasagiline and its (S) enantiomer
increase survival and prevent stroke in salt-loaded stroke-prone sponta-
neously hypertensive rats. J. Neural Transm. 108, 909–923.
Ellis, C.M., Simmons, A., Jones, D.K., Bland, J., Dawson, J.M., Horsfield,
M.A., Williams, S.C.R., Leigh, P.N., 1999. Diffusion tensor MRI as-
sesses corticospinal tract damage in ALS. Neurology 53, 1051–1058.
Giubilei, F., Bastianello, S., Paolillo, A., Gasperini, C., Tisei, P., Casini,
A.R., Gragnani, A., Bozzao, L., Fieschi, C., 1997. Quantitative mag-
netic resonance analysis in vascular dementia. J. Neurol. 244, 246–251.
Gulani, V., Webb, A.G., Duncan, I.D., Lauterbur, P.C., 2001. Apparent
diffusion tensor measurements in myelin-deficient rat spinal cords.
Magn. Reson. Med. 45, 191–195.
Hanyu, H., Shindo, H., Kakizaki, D., Abe, K., Iwamoto, T., Takasaki, M.,
1997. Increased water diffusion in cerebral white matter in Alzheimer’s
disease. Gerontology 43, 343–351.
Hasegawa, Y., Latour, L.L., Formato, J.E., Sotak, C.H., Fisher, M., 1995.
Spreading waves of a reduced diffusion coefficient of water in normal
and ischemic rat brain. J. Cereb. Blood Flow Metab. 15, 179–187.
Y. Assaf et al. / Experimental Neurology 184 (2003) 726–736736
Helenius, J., Soinne, L., Salonen, O., Kaste, M., Tatlisumak, T., 2002.
Leukoaraiosis, ischemic stroke, and normal white matter on diffusion-
weighted MRI. Stroke 33, 45–50.
Johansson, B.B., 1994. Pathogenesis of vascular dementia: the possible role
of hypertension. Dementia 5, 174–176.
Jones, D.K., Simmons, A., Williams, S.C.R., Horsfield, M.A., 1999a. Non-
invasive assessment of axonal fiber connectivity in the human brain via
diffusion tensor MRI. Magn. Reson. Med. 42, 37–41.
Jones, D.K., Lythgoe, D., Horsfield, M.A., Simmons, A., Williams, S.C.R.,
Markus, H.S., 1999b. Characterization of white matter damage in is-
chemic leukoaraiosis with diffusion tensor MRI. Stroke 30, 393–397.
King, M.D., Houseman, J., Gadian, D.G., Connelly, A., 1997. Localized
q-space imaging of the mouse brain. Magn. Reson. Med. 38, 930–937.
Kraemer, F., Darquie, A., Clark, C.A., Le Bihan, D., 1999. Separation of
two diffusion compartments in the human brain. Proc. Int. Soc. Magn.
Reson. Med. 7, 1808.
Kuchel, P.W., Coy, A., Stilbs, P., 1997. NMR ‘‘diffusion–diffraction’’ of
water revealing alignment of erythrocytes in a magnetic field and their
dimensions and membrane transport characteristics. Magn. Reson. Med.
37, 637–643.
Le Bihan, D., 1995. Diffusion and Perfusion Magnetic Resonance Imaging:
Application to Functional MRI, 1st ed. Raven Press, New York.
Lester, D.S., Pine, P.S., Delnomdedieu, M., Johannessen, J.N., Johnson,
G.A., 2000. Virtual neuropathology: three dimensional visualization of
lesion due to toxic insult. Toxicol. Pathol. 28, 100–104.
Liao, D., Cooper, L., Cai, J., Toole, J.F., Bryan, N.R., Hutchinson, R.G.,
Tyroler, H.A., 1996. Presence and severity of cerebral white matter
lesions and hypertension, its treatment, and its control. Stroke 27,
2262–2270.
Lis, C.G., Gaviria, M., 1997. Vascular dementia, hypertension, and the
brain. Neurol. Res. 19, 471–480.
Moseley, M.E., Cohen, Y., Kucharczyk, J., Mintorovitch, J., Asgari, H.S.,
Wendland, M.F., Tsuruda, J., Norman, D., 1990a. Diffusion-weighted
MR imaging of anisotropic water diffusion in cat central-nervous-sys-
tem. Radiology 176, 439–445.
Moseley, M.E., Cohen, Y., Mintorovitch, J., Chileuitt, L., Shimizu, H.,
Kucharczyk, J., Wendland, M.F., Weinstein, P.R., 1990b. Early detec-
tion of regional cerebral ischemia in cats: comparison of diffusion- and
T2-weighted MRI and spectroscopy. Magn. Reson. Med. 14, 330–346.
Mulkern, R.V., Gudbjartsson, H., Westin, C.F., Zengingonul, H.P., Gartner,
W., Guttmann, C.R.G., Robertson, R.L., Kyriakos, W., Schwartz, R.,
Holtzman, D., Jolesz, F.A., Maier, S.E., 1999. Multi-component appa-
rent diffusion coefficients in human brain. NMR Biomed. 12, 51–62.
Nevo, U., Hauben, E., Yoles, E., Agranov, E., Akselrod, S., Schwartz, M.,
Neeman, M., 2001. Diffusion anisotropy MRI for quantitative assess-
ment of recovery in injured rat spinal cord. Magn. Reson. Med. 45, 1–9.
Niendorf, T., Dijkhuizen, R.M., Norris, D.G., Campagne, M.V., Nicolay,
K., 1996. Biexponential diffusion attenuation in various states of brain
tissue: implications for diffusion-weighted imaging. Magn. Reson.
Med. 36, 847–857.
Nossin-Manor, R., Duvdevani, R., Cohen, Y., 2002. q-Space high b value
diffusion MRI of hemi-crush in rat spinal cord: evidence for sponta-
neous regeneration. Magn. Reson. Imaging 20, 231–241.
Ono, J., Harada, K., Takahashi, M., Maeda, M., Ikenaka, K., Sakurai, K.,
Sakai, N., Kagawa, T., Fritz-Zieroth, B., Nagai, T., Nihei, A., Hashi-
moto, S., Okada, S., 1995. Differentiation between dysmyelination and
demyelination using magnetic resonance diffusional anisotropy. Brain
Res. 671, 141–148.
Ono, J., Harada, K., Mano, T., Sakurai, K., Okada, S., 1997. Differentiation
of dys- and demyelination using diffusional anisotropy. Pediatr. Neurol.
16, 63–66.
O’Sullivan, M., Summers, P.E., Jones, D.K., Jarosz, J.M., Williams,
S.C.R., Markus, H.S., 2001. Normal-appearing white matter in ischemic
leukoaraiosis: a diffusion tensor MRI study. Neurology 57, 2307–2310.
Peled, S., Cory, D.G., Raymond, S.A., Kirschner, D.A., Jolesz, F.A., 1999.
Water diffusion, T2, and compartmentation in frog sciatic nerve. Magn.
Reson. Med. 42, 911–918.
Rooney, W.D., Goodkin, D.E., Schuff, N., Meyerhoff, D.J., Norman, D.,
Weiner, M.W., 1997. H-1 MRSI of normal appearing white matter in
multiple sclerosis. Mult. Scler. 3, 231–237.
Rose, S.E., Chen, F., Chalk, J.B., Zelaya, F.O., Strugnell, W.E., Benson,
M., Semple, J., Doddrell, D.M., 2000. Loss of connectivity in Alzheim-
er’s disease: an evaluation of white matter tract integrity with colour
coded MR diffusion tensor imaging. J. Neurol. Neurosurg. Psychiatry
69, 528–530.
Skoog, I., 1998. A review on blood pressure and ischaemic white matter
lesions. Dement. Geriatr. Cogn. Disord. 9 (Suppl. 1), 9–13.
Stanisz, G.J., Szafer, A., Wright, G.A., Henkelman, R.M., 1997. An ana-
lytical model of restricted diffusion in bovine optic nerve. Magn. Reson.
Med. 37, 103–111.
Takahashi, M., Fritz-Zieroth, B., Chikugo, T., Ogawa, H., 1993. Differ-
entiation of chronic lesions after stroke in stroke-prone spontaneously
hypertensive rats using diffusion weighted MRI. Magn. Reson. Med.
30, 485–488.
Tovi, M., Ericsson, A., 1992. Measurements of T1 and T2 over time in
formalin-fixed human whole-brain specimens. ACTA Radiol. 33,
400–404.
van Gijn, J., 1998. Leukoaraiosis and vascular dementia. Neurology 51
(Suppl. 3), S3–S8.
van Swieten, J.C., Geyskes, G.G., Derix, M.M., Peeck, B.M., Ramos,
L.M., van Latum, J.C., van Gijn, J., 1991. Hypertension in the elderly
is associated with white matter lesions and cognitive decline. Ann.
Neurol. 30, 825–830.
Wahlund, L.O., 1994. Brain imaging and vascular dementia. Dementia 5,
193–196.
Werring, D.J., Brassat, D., Droogan, A.G., Clark, C.A., Symms, M.R.,
Barker, G.J., MacManus, D.G., Thompson, A.J., Miller, D.H., 2000.
The pathogenesis of lesions and normal-appearing white matter
changes in multiple sclerosis—A serial diffusion MRI study. Brain
123, 1667–1676.