| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 All authors: Department of Neuroradiology, Duke University Medical Center, Box 3808, Erwin Rd., Durham, NC 27710.
Received November 17, 2003;
accepted after revision February 19, 2004.
Address correspondence to S. M. Kealey.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. Thirty-six white matter (WM) plaques were identified in 20 patients with MS. Plaque FA was measured by placing regions of interest (ROIs) on plaques on diffusion tensor images. We compared FA values in identical mirror-image ROIs placed on normal-appearing WM in the contralateral hemisphere. This comparison showed a mean decrease in FA of 41% in plaques, serving as the threshold for outlining abnormal regions in normal-appearing WM surrounding plaques. ROIs were placed around each plaque and FA values were compared with those in the mirror-image ROIs. Combined areas of perilesional normal-appearing WM with 40% or more FA reduction plus plaque were compared with the areas of abnormality on T2-weighted images using a paired Student's t test. A p value of 0.05 or less was considered significant.
RESULTS. Mean plaque area was 60 mm2 (range, 15-103 mm2), mean plaque FA was 0.251 (range, 0.133-0.436), and mean FA of contralateral normal-appearing WM was 0.429 (range, 0.204-0.712). Applying a threshold of 40% FA reduction, mean combined area of abnormal WM (including plaque seen on T2-weighted sequences) was 87 mm2 (range, 30-251 mm2) or 145% of the mean plaque area that was seen on T2-weighted images (p < 0.001).
CONCLUSION. Using an operator-defined threshold of abnormal FA values based on plaque anisotropy characteristics, we saw a statistically significant increase in plaque size.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Clinical trials of therapeutic agents are heavily dependent on assessment of disease burden as defined on MRI. Specifically, these studies measure plaque burden by measuring volume of individual plaques on T2-weighted images and by providing a disease-severity score. Recent evidence has indicated that the extent of white matter (WM) involvement in the demyelinating process is greater than is evident using standard MRI techniques; both diffusion tensor imaging and MR spectroscopy have shown involvement of normal-appearing cerebral WM in patients with MS [1, 4-8], suggesting that these methods may be a useful addition to the imaging assessment of disease burden in demyelinating conditions. More important, if plaque burden is underestimated on conventional T2-weighted images, then the results of clinical trials based on T2-weighted imaging may misrepresent actual disease response.
Previous studies have shown that fractional anisotropy (FA) values in normal-appearing WM adjacent to MS plaques are lowered, which introduces the possibility that the size of plaques on diffusion tensor imaging differs from the size of plaques as seen on T2-weighted images [4]. The purpose of this study was to compare the size of MS plaques with lesion size as defined by a thresholding technique on FA maps. We hypothesized that substantial differences in size would be found between plaques as seen on T2-weighted images and lesions as seen on FA maps.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Conventional MRI sequences acquired included axial T2-weighted images with these parameters: TR/TE, 2,800/100; field of view, 22 cm2; matrix, 256 x 192; 5-mm section thickness; and a 2.5-mm interslice gap. Axial T1-weighted images were also acquired before and after administration of gadopentetate dimeglumine (Magnevist, Berlex Laboratories). The presence of lesions representative of MS plaques on the T2-weighted images was determined. Our criteria for deciding whether a lesion on T2-weighted images represented an MS plaque were that it be oval in appearance, oriented toward or abutting the lateral ventricles, and lacking associated restricted diffusion. At least one plaque was chosen for analysis in each patient. The criteria for choosing a plaque for analysis were that it have a minimum size of 15 mm2 with normal-appearing WM in both the mirror-image site in the contralateral hemisphere and immediately adjacent to the mirror-image site. Plaques were excluded if the contralateral WM appeared abnormal on T2-weighted images because that finding would preclude comparison of FA values in the plaque with those in contralateral normal-appearing WM. A minimum plaque size of 15 mm2 was chosen because the signal-to-noise ratio in smaller plaques would be unacceptably low. The presence of plaque contrast enhancement and hypointense appearance on T1-weighted images were recorded.
Diffusion tensor MR images of the entire brain were acquired in six
orthogonal directions with a 5-mm slice thickness and 2.5-mm gap on a 1.5-T MR
scanner (Signa, GE Healthcare) using a standard head coil (field of view, 40
x 20 cm; matrix, 128 x 64; 4 excitations). The diffusion tensor
MRI protocol consisted of a single-shot spin-echo echoplanar sequence with
TR/TE, 12,000/107; inversion time, 2,200 msec; and 1 excitation.
Diffusion-sensitizing gradient encoding was applied on separate images in six
directions using a diffusion-weighted factor b of 1,000
sec/mm2. The raw diffusion tensor data were transferred to an
independent workstation (Advantage Windows, GE Healthcare) and processed using
Functool software (GE Healthcare) to generate maps of FA. The six independent
elements of the diffusion tensor Dxx, Dyy, Dzz, Dxy, Dxz, and
Dyz were statistically calculated for each voxel using a method
previously described
[9-12]
based on the equation
 |
where bij is the component of the i row and
j column of the diffusion gradient matrix b, A(b)
is the resulting echo intensity for a gradient sequence with directions and
magnitudes of the resulting diffusion-sensitizing gradients described by the
b matrix, A(b = 0) is the echo intensity when
b is the zero matrix (no diffusion gradient), and
Dij is the corresponding component of the diffusion tensor
D [9,
10]. Once the elements of the
diffusion tensor were obtained, its eigenvalues were calculated via
diagonalization of the tensor matrix. FA was then calculated according to the
equation
 |
where E1, E2, and E3 are the three eigenvalues and d is [(E1 + E2 + E3) / 3].
We used FA as the index of anisotropy because FA is generally considered to be a robust measure of anisotropy and is also the most widely used anisotropy index, permitting comparison with data from other groups. FA also has the advantages of good gray-WM contrast and a high contrast-to-noise ratio [13]. FA represents the anisotropic portion of total diffusion and values range from 0 to 1, where 0 represents isotropic diffusion and 1 represents extremely anisotropic diffusion. The FA value is unitless because it represents a ratio of diffusion coefficients. The calculations for FA were performed for each voxel and displayed as an anisotropy map that was scaled appropriately for display.
Using the postprocessed FA map, oval ROIs conforming to the site of MS plaques on FA maps were placed by one trained radiologist to measure FA changes in plaques. The ROIs were drawn around the boundary of plaques; subsequently, ROI size varied according to plaque size, with a range of 15-103 mm2. Careful attention was paid to ensure that the ROIs conformed to the boundaries of the plaque as seen on T2-weighted images, which were viewed side by side with FA maps. These FA values in regions conforming to the site of the plaque were compared with FA values in identical mirror-image ROIs placed at analogous regions to plaques on normal-appearing WM in the contralateral cerebral hemisphere (Fig. 1A, 1B, 1C, 1D, 1E). A trained neuroradiologist confirmed ROI positioning.
|
|
|
|
|
A mean decrease in FA values (range, 28-70%) in regions conforming to the sites of plaques of 41% was measured. On the basis of this finding, we set a threshold of 40% decrease in FA as the criterion for defining abnormal-appearing WM on FA maps. We then measured FA decreases in WM regions adjacent to plaques as follows: We placed standard ROIs of between 40 and 55 mm2 around the central ROI in a contiguous manner. We considered that ROIs smaller than this would have a poor signal-to-noise ratio and result in less accurate FA measurements. These peripheral ROIs were placed immediately adjacent to the central ROI to encompass as much of the peripheral WM as possible, and their FA values were recorded. Then ROIs were placed in the equivalent mirror-image locations in the contralateral hemisphere. Those ROIs that had at least a 40% reduction in FA compared with their mirror images were recorded, and the other ROIs that did not meet this threshold were deleted. The normal-appearing WM was systematically investigated in a centrifugal manner by contiguously placing further ROIs adjacent to those peripheral ROIs that had met the 40% threshold to plot the full extent of normal-appearing WM defined as abnormal by this criterion. The area of all ROIs having at least a 40% decrease in FA values was added to the area of the original plaque to give a total area of abnormal-appearing WM on FA maps for that plaque. The size of resultant regions of abnormal-appearing WM was compared with the size of the corresponding plaque seen on T2-weighted images using a paired Student's t test. A p value of 0.05 or less was considered significant. These steps are graphically depicted in Figures 1A, 1B, 1C, 1D, 1E and 2A, 2B, 2C, 2D, 2E.
|
|
|
|
|
In addition to our primary comparison, the percentage of decrease of FA values in regions on FA maps corresponding to plaques was compared with plaque size on T2-weighted images using Pearson's correlation coefficient. As a subset analysis, we compared the FA values in regions on FA maps corresponding to enhancing plaques with FA values in regions on FA maps corresponding to un-enhancing plaques, using the Student's t test.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
0.001). Plaque size and FA parameters are shown in
Table 1.
|
In 12 plaques in nine patients, none of the WM regions adjacent to sites corresponding to plaques showed an FA decrease of 40%. In these 12 plaques, FA values in WM regions adjacent to sites corresponding to plaques were between 24% and 39% of FA values in analogous mirror-image sites. Comparison of these 12 plaques with the other plaques studied showed no statistically significant difference in mean plaque size (p = 0.86), mean plaque FA value (p = 0.67), or mean FA of contralateral normal-appearing WM (p = 0.07). Plaque growth is depicted in Figure 3.
|
No significant relationship was seen between plaque area on T2-weighted images and mean FA value in regions on FA maps corresponding to plaques (p = 0.13, Pearson's correlation coefficient = -0.26). No significant difference appeared between the FA values in regions corresponding to enhancing plaques and regions corresponding to nonenhancing plaques (p = 0.22).
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Although MS is primarily characterized by myelin loss, recent histologic assessments have indicated that both axonal injury and neuronal loss also play a role in pathogenesis [15-18]. Postmortem studies have also confirmed significant axonal loss in the normal-appearing WM of MS patients [19]. Proton MR spectroscopy has shown reduced levels of the neuronal marker N-acetyl aspartate in the brains of patients with MS, suggesting the presence of widespread axonal injury [20]. Both myelin loss and direct disruption of the nerve axons would result in greater random motion of water molecules across fiber tract pathways, resulting in reduced FA values in areas of damaged tissue.
We found significantly reduced anisotropy values both in regions on FA maps corresponding to MS plaques that were evident on T2-weighted images and in the immediately adjacent normal-appearing WM regions, indicating that the true plaque size as determined on diffusion tensor imaging is often substantially greater than that seen on standard MRI sequences. These values suggest that current MRI sequences underestimate the amount of diseased WM in MS and support findings from histologic studies also described in this report. Previous studies have shown that WM in MS patients may be more diffusely affected than is seen on T2-weighted images [4]. We found that the regions of focal abnormality generally referred to as plaques are, in fact, often much larger than T2-weighted studies indicate. It remains to be determined whether diffusion tensor imaging findings will more closely correlate with the clinical manifestations of disease activity. If such a positive correlation exists between the anisotropy-defined plaque boundaries and clinical manifestations of disease, this fact would be expected to have major implications for the design and interpretation of trials of therapeutic agents in MS.
The FA reduction in regions on FA maps corresponding to MS plaques is likely to be due to a complex process with varying contributions of perivascular inflammation, proliferation of astrocytes, loss of myelin, and destruction of axons [21, 22]. For instance, MS plaques that are hypointense on T1-weighted MR images have been correlated with the presence of matrix destruction and loss of axons [23]. Some studies have shown an increase in the water diffusibility and reduction in FA in such lesions in MS [21, 24]. Our study included only one plaque that was hypointense on T1-weighted images, precluding any meaningful comparison of FA values in such plaques with those without associated T1 signal abnormality.
It is likely that the FA reductions we found in this study in normal-appearing WM adjacent to plaques are related to either less advanced or less chronic areas of tissue damage. The periplaque WM may comprise damaged tissue with a predominantly inflammatory infiltrate and associated astrocyte proliferation rather than marked axonal and myelin loss. Another possible explanation for decreased anisotropy in normal-appearing WM adjacent to regions corresponding to plaques can be found in the method by which plaques are typically formed. MS plaques characteristically develop in a circumferential manner around a small vein and extend outward in a centrifugal manner. Therefore, a greater degree of tissue damage and axonal loss with consequently lower values of FA would be expected to occur in the center of the plaque. Less severe demyelination and axonal disruption at the periphery of a centrifugally expanding plaque may be sufficient to result in measurable reductions in anisotropy but yet not be sufficiently advanced to result in signal abnormality on T2-weighted images. These regions may represent either areas of plaque regression (in which some myelin repair or resolution of inflammation has occurred) or simply areas that have not yet sustained sufficient injury to reflect abnormality on T2-weighted images.
Wallerian degeneration has also been suggested as a possible mechanism of reductions in FA in normal-appearing WM distant from MS plaques [25]. Wallerian degeneration would cause progressively decreased FA in an ante-grade fashion along the projected fiber tract affected by the MS plaque. Therefore, it is possible that plaques situated on slices at a more cephalad level than the ones on which we measured FA values could also have contributed to the measured FA reductions in the selected periplaque regions. This hypothesis cannot be assessed in our study because one cannot accurately determine most specific WM pathways on FA maps. However, fiber tract mapping techniques continue to evolve, and this hypothesis could be tested by determining the relationship of normal-appearing WM regions with decreased anisotropy values to WM tracts coursing through plaques.
In one third of plaques in our study, the FA values of the normal-appearing WM regions adjacent to plaques were reduced by less than 40%, so the total lesion area that we defined on FA maps was the same as that of the area seen on T2-weighted images. However, the measured anisotropy values in these periplaque regions were still uniformly substantially reduced (by 24-39%) when compared with analogous normal-appearing WM sites in the contralateral hemisphere. This finding reflects the fact that the threshold we chose was deliberately high, being essentially equivalent to the FA reduction shown in plaques themselves. Setting different thresholds will result in regions of abnormal-appearing WM of different sizes on FA maps. Thus, the actual areas of abnormal-appearing WM having substantial FA decreases are likely to be larger than those defined by our thresholding technique. The imaging characteristics of this subset of plaques were not significantly different from those seen in the other plaques studied. No previous MRI studies were available in these cases to permit a correlation with plaque chronicity.
We found no correlation between plaque contrast enhancement and the degree of reduction of FA values either in the plaque or in the adjacent normal-appearing WM regions in our study. Given the small number (4/36) of enhancing plaques included in this study, however, these results should be interpreted with some caution. One previously published study [24] reported statistically significant FA reductions in ring-enhancing, but not densely enhancing, MS plaques compared with normal-appearing WM in control subjects. The differences between results in their report and in our study may reflect the relatively small number of enhancing plaques in our sample. Plaque contrast enhancement is reported to generally correlate with plaque activity and the presence of acute inflammation [26]. Therefore, if more severe demyelination and axonal loss is seen in long-standing plaques, one would actually expect decreases in FA values in contrast-enhancing plaques to be less marked than in plaques in which longstanding axonal and myelin destruction appears, as we found in our small sample.
Some limitations to this study are evident. First, our reference value for FA was normal-appearing WM in the contralateral hemisphere of each patient. It has been previously reported that the disease process globally affects the WM in MS, leading to reductions in anisotropy in regions distant from plaques evident on standard MRI sequences [4]. This fact would suggest that the analogous (mirror image) site we used as a reference of normality in this study might not have represented truly normal WM. However, if this were the case, our results might actually underestimate the degree to which FA is reduced in regions on FA maps corresponding to plaques and adjacent WM regions because findings in our patients would be even more profound when measured against normal WM in control subjects. We intend to address this issue in future studies with reference data from age-matched healthy control subjects.
Second, we did not correlate FA decreases with clinical symptoms. It will be important to determine whether these two features are related, and we will address that relation in future studies. The method we describe of calculating plaque size on the basis of changes in anisotropy is obviously dependent on the threshold of FA reduction chosen by the operator. We deliberately chose a high threshold of anisotropy reduction, almost as great as the FA reduction in plaques as seen on T2-weighted images, to robustly test our initial hypothesis. Future studies must address the question of an appropriate threshold for the definition of abnormal-appearing WM. This issue could be addressed by correlation of plaque size on diffusion tensor imaging at different thresholds of FA reduction with clinical disability and disease course. It will also be important to delineate any relationship of FA values to plaque chronicity. FA values should evolve with time as tissue destruction progresses; longitudinal studies will be required to determine whether FA reductions can predict the eventual extent of T2 signal abnormalities in MS.
Finally, at the time of this study, fiber tract-mapping software was not available to us. Future analysis of mean FA values in fiber tracts along important WM pathways might allow further insight into the role of diffusion tensor imaging in assessment of MS patients and in determining what role, if any, wallerian degeneration plays in decreased FA values in normal-appearing WM.
In conclusion, this study indicates that the focal regions of abnormal-appearing WM seen as hyperintense regions on T2-weighted images in MS patients are often much smaller than the corresponding abnormality on diffusion tensor imaging. This fact may have important ramifications for interpretation of results of MS pharmacologic trials that use T2-weighted MRI studies to assess therapeutic effect. Further study is required to determine the relationship of disturbances in WM anisotropy to clinical parameters of disease activity and possibly prognosis.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  Y. Assaf, L. Ben-Sira, S. Constantini, L.C. Chang, and L. Beni-Adani Diffusion tensor imaging in hydrocephalus: initial experience. AJNR Am. J. Neuroradiol., September 1, 2006; 27(8): 1717 - 1724. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. Ni, V. Kavcic, T. Zhu, S. Ekholm, and J. Zhong Effects of number of diffusion gradient directions on derived diffusion tensor imaging indices in human brain. AJNR Am. J. Neuroradiol., September 1, 2006; 27(8): 1776 - 1781. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. M. Provenzale, S. Mukundan, and D. P. Barboriak Diffusion-weighted and Perfusion MR Imaging for Brain Tumor Characterization and Assessment of Treatment Response. Radiology, June 1, 2006; 239(3): 632 - 649. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Kim, S.-K. Lee, J. D. Lee, Y. W. Kim, and D. I. Kim Decreased Fractional Anisotropy of Middle Cerebellar Peduncle in Crossed Cerebellar Diaschisis: Diffusion-Tensor Imaging-Positron-Emission Tomography Correlation Study AJNR Am. J. Neuroradiol., October 1, 2005; 26(9): 2224 - 2228. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
