Diffusely Abnormal White Matter in Chronic Multiple Sclerosis:  Imaging and Histopathologic Analysis FREE

Multiple sclerosis (MS) is a chronic, inflammatory, demyelinating disease of the central nervous system typically characterized by focal lesions in the white matter (WM). Magnetic resonance imaging (MRI) is the most sensitive imaging technique for detecting MS lesions in vivo, and lesion load measurements based on conventional T2-weighted MRI are widely used to monitor treatment effects in therapeutic trials.¹ However, there is only a modest correlation between the lesion load on conventional MRI and the clinical disability of patients with MS, a phenomenon referred to as clinicoradiologic dissociation.²

To explain this dissociation, it was suggested that factors other than focal WM lesions on MRI might contribute to the development of disability, that is, invisible pathologic abnormalities in the so-called normal-appearing white matter (NAWM) and normal-appearing gray matter.^3- 6Quantitative MRI techniques (eg, magnetization transfer imaging, diffusion tensor imaging (DTI), and T1 and T2 relaxation time measurements), and high-field MRI have proved to be more sensitive in detecting subtle abnormalities,^7- 10 overcoming some of the limitations of conventional MRI.^11- 13

Apart from this invisible pathology on conventional T2-weighted imaging, there is also pathology that is visible but unaccounted for in standard MS disease burden ratings. These diffuse and subtle signal hyperintensities in the WM are commonly referred to as diffusely abnormal WM (DAWM) or dirty WM, and the signal intensity is higher in these areas than in the surrounding NAWM but lower than in focal WM lesions.¹⁴ Although not rated in a standard clinical and research setting, DAWM may be expected to contribute to the clinicoradiologic dissociation if considerable pathologic abnormalities are present in these regions.

So far, no systematic studies exist, to our knowledge, that have characterized the pathologic features of DAWM in MS; hence, it is unclear whether DAWM represents an early stage of inflammation and lesion development, a more chronic ongoing pathologic abnormality, or even a reparative process (remyelination). Assessment of underlying pathologic abnormalities and a clear definition of DAWM should provide a more accurate estimation of disease burden and disease progression in MS.

We investigated fixed postmortem brain tissue of patients with chronic MS using qualitative MRI at 1.5 and 4.7 T and quantitative MRI (magnetization transfer imaging, diffusion tensor imaging and T1 and T2 relaxation time mapping) at 1.5 T to answer the following questions: (1) What are the underlying qualitative MRI and histopathologic characteristics of DAWM compared with NAWM and focal WM lesions? (2) Do quantitative MRI measures at standard field strength (1.5 T) specifically reflect the extent and nature of the underlying pathologic changes in DAWM?

Seventeen coronally cut, 10-mm-thick hemispheric brain slices of 10 patients with chronic MS (7 women; mean age, 66.8 years) were selected after rapid autopsy (mean postmortem delay, 8.5 hours) and were formalin fixed for several weeks. Table 1 provides demographic and neuropathologic details of the patients. Clinical courses were determined by means of retrospective medical record reviews according to established criteria.¹⁵ Ethics approval was obtained from the VU University Medical Center, and all the donors gave informed consent before their death for the use of their tissue and medical records.

Imaging at 1.5 T was performed using a Magnetom Vision scanner (Siemens AG, Erlangen, Germany).¹⁶ Conventional measurements consisted of the following: (1) dual-echo T2-weighted spin-echo images (repetition time [TR]/echo time [TE] 1/TE2/number of signals acquired, 2755 milliseconds/45 milliseconds/90 milliseconds/2; field of view (FOV), 80 × 128 mm; matrix size, 160 × 256; and section thickness, 3 mm); and (2) multislab 3-dimensional fluid-attenuated inversion recovery¹⁷ images (TR/TE/inversion time/number of signals acquired, 6500 milliseconds/120 milliseconds /2200 milliseconds/1; echo train length, 27; 8 partitions per slab; partition thickness, 1.25 mm; FOV, 125 × 200 mm; and matrix size, 160 × 256).

Quantitative imaging at 1.5 T consisted of the following: (1) T1 relaxation time mapping: 6 sets of 3-dimensional fast low-angle shot (3D-FLASH) images were acquired (TR/TE/number of signals acquired, 15 milliseconds/4 milliseconds/4; partition thickness, 3 mm; FOV, 80 × 128 mm; and matrix size, 80 × 128), with nominal flip angles of 2° to 25°. B₁ maps were generated from 5 additional sets of 3D-FLASH images (TR/TE/number of signals acquired, 15 milliseconds/5 milliseconds/4; partition thickness, 3 mm; FOV, 80 × 128 mm; and matrix size, 80 × 128), with nominal flip angles varying between 140° and 220°.⁹ (2) T2 relaxation time mapping: a multi-echo Carr-Purcell-Meiboom-Gill sequence (TR/number of signals acquired, 2500 milliseconds/1; FOV, 80 × 128 mm; matrix size, 80 × 128 mm, and section thickness, 3 mm), with alternating 180° pulses and 16 equidistant echoes, starting from 20.5 milliseconds, was applied. (3) Magnetization transfer imaging maps: a 3D-FLASH sequence (TR/TE/number of signals acquired, 27 milliseconds/4 milliseconds/1; partition thickness, 5 mm; FOV, 128 mm rectangular; matrix size, 128 mm; and flip angle, 20°) was acquired, 1 with a gaussian magnetization transfer prepulse and 1 without. (4) Diffusion tensor imaging: a diffusion-weighted single-shot stimulated echo acquisition mode sequence (TR/TE/number of signals acquired, 6000 milliseconds/65 milliseconds/84; FOV, 80 × 128 mm; matrix size, 40 × 64; flip angle, 11°; and section thickness, 8 mm) was used.¹⁰ Seven volumes were acquired: 6 with different noncollinear directions and 1 without diffusion weighting. These volumes were used to calculate the apparent diffusion coefficient (ADC) and fractional anisotropy (FA) maps. Examples of different quantitative MRI maps and image quality are seen in Figure 1.

Measurements at high field (4.7 T) were performed using an experimental animal scanner (Varian Inc, Palo Alto, California) (horizontal bore) and consisted of a mainly proton density (PD)–weighted sequence (3-dimensional fast spin echo) (TR/TE/number of signals acquired, 4000 milliseconds/9 milliseconds/2; echo train length, 8; partition thickness, 0.39 mm; 64 partitions per slab; FOV, 100 × 100 mm; matrix size, 256 × 256; and acquisition time, 4 hours 55 minutes).

After MRI, the brain slices were cut in half to reveal the center of the imaged plane and were embedded in paraffin. Stainings and immunohistochemical analyses were performed on 10-μm-thick sections. To assess tissue morphologic features and quality and myelin and axonal density, hematoxylin-eosin, Luxol fast blue (LFB)–periodic acid–Schiff (PAS), and Bodian silver stains were performed.

Immunohistochemical analysis was performed on adjacent sections with antibodies against the following targets: glial fibrillary acidic protein (GFAP) (Dako; Glostrup, Denmark), antigen-presenting cells and microglia (HLA-DR, courtesy of J Hilgers, PhD, Amsterdam), proteolipid protein (PLP) (AbD Serotec; Oxford, England), β-amyloid precursor protein (APP) (Zymed Laboratories Inc; South San Francisco, California), and fibrinogen (Dako). Bound primary antibody was detected using the EnVision method (Dako).

The DAWM and focal WM lesions were defined using 1.5-T PD images; NAWM was defined using 4.7-T PD images to avoid inclusion of focal WM lesions undetected on the 1.5-T images (Figure 2).

Regions of interest were placed in selected representative DAWM areas and in focal WM lesions and NAWM. Quantitative MRI values were measured in the regions of interest for comparisons among DAWM, NAWM, and focal WM lesions.

The DAWM was defined as a uniform, nonfocal area of signal increase on the PD-weighted sequence at 1.5 T, with a subtly increased signal intensity compared with the signal intensity of focal WM lesions. The DAWM signal hyperintensity usually tapered off toward the NAWM, leading to a relatively unsharply defined border of DAWM areas, again compared with focal WM lesions. Classification of NAWM and DAWM was performed blinded to histopathologic features.

After MRI-to-histopathology matching^16- 17 (Figure 2), regions of interest were placed onto the corresponding areas in the tissue sections. Histologic scorings were performed blinded to MRI data.

Staining intensity was measured by means of light transmittance on digital images of histologic sections using ImageJ_1.37 (http://rsbweb.nih.gov/ij). High values (increased light transmittance) correspond with low staining intensity. Transmittance of Bodian (Tm_Bodian, PLP-(Tm_plp), LFB-PAS–(TM_LFB-PAS), GFAP-(Tm_GFAP), and fibrinogen-(Tm_fibrinogen) stained sections, was used to quantify axonal density, myelin density, gliosis, and blood-brain barrier leakage, respectively. In addition, axons were counted in 3 random areas per region of interest using a morphometric grid (10 mm², original magnificaton ×1250) and were averaged per region of interest (Count_Axon).

To assess the extent of astrogliosis, glial filament staining and the morphologic aspect of cell bodies were scored separately because gliosis may show different features depending on the stage of astrocytic activation (Figure 3).¹⁸ The GFAP-positive cell bodies were scored as 0 (not enlarged) or 1 (enlarged). Increase of GFAP staining in glial filaments (F_SQ GFAP) was assessed in 3 steps: 0, normal appearance; 1, increased glial processes; and 2, severely increased glial processes.

Acute axonal injury was scored as 0 (no APP positivity) or 1 (APP-positive beads or end bulbs).¹⁹ The antigen-presenting cells were scored as follows: 0, no increase; 1, mild increase (10-15 APCs); or 2, severe increase (>15 APCs). The (p)reactive lesions were noted separately as present or absent. Remyelination was assessed by evaluating the absence or presence of shadow plaques.²⁰ Lesion activation stage was evaluated in the regions of interest as described previously,²¹ and care was taken not to mistake DAWM areas for confluent (chronic) WM lesions. Cortical lesions were counted and classified into 4 lesion subtypes as described elsewhere.²²

Data analysis was performed using a statistical software program (SPSS 14.0 for Windows; SPSS Inc, Chicago, Illinois). Quantitative MRI measurements and quantitative neuropathologic data were compared using a general linear mixed-model analysis, accounting for a nested design. Pairwise comparisons were performed between tissue types (DAWM and NAWM, DAWM and focal WM lesions, and focal WM lesions and NAWM). Bonferroni-corrected P < .05 was considered to be statistically significant.

Results of semiquantitative histopathologic scorings were compared using the Mann-Whitney test. Spearman rank correlation coefficient was used to assess correlations between histopathologic findings and quantitative MRI variables. Significance was accepted at the level of P < .05.

A total of 42 regions of interest in 17 tissue blocks from 10 patients with chronic MS were analyzed. Of these 42 regions of interest, 16 were placed in DAWM, 15 in NAWM, and 11 in focal WM lesions. The MRI-to-histopathology matching was successful in almost all cases; only 8% of 630 measurements had to be excluded from further analysis.

Mostly, DAWM was seen in periventricular WM or in the centrum semiovale of frontal and parietal slices and was found in direct proximity of focal WM lesions and in areas with no visible focal abnormalities. The DAWM usually extended over large parts of the WM (Figure 2). At both field strengths, DAWM displayed uniform, nonpatchy signal hyperintensity. The DAWM is characterized by significant axonal loss, myelin pallor, and chronic fibrillary gliosis.

The Tm_Bodian and Tm_LFB-PAS differed significantly among DAWM, NAWM, and focal WM lesions (Figure 3). The Tm_PLP, however, showed no difference between DAWM and NAWM (Table 2). The Tm_Fibrinogen did not differ among the 3 tissue types, whereas Tm_GFAP showed equal values for focal WM lesions and NAWM and significantly lower values (corresponding to a higher staining intensity of GFAP) for DAWM.

The Count_Axon also differed significantly among the investigated areas. The number of axons was reduced by 40% in DAWM and by 67% in focal WM lesions compared with the average number of axons in NAWM.

Abnormally enlarged astrocytic cell bodies were seen in 37 of 39 analyzed regions of interest (95%) (Table 2). Semiquantitative scores for glial processes (F_SQ GFAP) were highest in focal WM lesions, followed by DAWM and NAWM.

Acute axonal pathology and remyelination were absent in DAWM. Activated antigen-presenting cells were found in all areas but were significantly higher in DAWM than in NAWM. Five of 11 focal WM lesions were classified as chronic active and 6 as chronic inactive. No lesions ([p]reactive or otherwise) were found in DAWM. A total of 133 cortical lesions were counted in the brain sections, 5 of which were classified as type I, 22 as type II, 101 as type III, and 5 as type IV. No spatial relation was detected between the extent of DAWM and the occurrence of gray matter lesions.

The DAWM can be distinguished from NAWM and focal WM lesions using quantitative MRI. Quantitative MRI values in DAWM were intermediate to those of focal WM lesions and NAWM (Table 2). T1 and T2 relaxation times were significantly different among NAWM, DAWM, and focal WM lesions. For FA and magnetic transfer image (MTR) measurements, no significant difference was detected between DAWM and focal WM lesions. The ADC measurements differed significantly only between focal WM lesions and NAWM.

Higher T1 and T2 relaxation times and lower FA values correlated with axonal loss (lower Count_Axon and higher Tm_Bodian) and with reduced myelin density (higher Tm_LFB-PAS and Tm_PLP) (Table 3 and Figure 4). In addition, T1 and T2 relaxation times correlated with increased fibrillary gliosis (higher F_SQ GFAP) but not with Tm_GFAP. Furthermore, a correlation was found between decreased MTR and axonal (Tm_Bodian and Count_Axon) and myelin lipid (Tm_LFB-PAS) loss. The ADC correlated with Count_Axon.

Both Count_Axon and Tm_Bodian showed significant differences among the 3 tissue types and correlated with each other (r = −0.81, P < .001). In contrast, Tm_GFAP and semiquantitative counts of fiber density were not correlated (r = −0.29, P = .08). As glial cell bodies were rated as abnormal in almost all cases, no correlations between semiquantitative scores of glial cell bodies and Tm_GFAP were calculated (Figure 4).

This study shows that DAWM in chronic MS has distinct radiologic and histopathologic characteristics that are significantly different from those of focal WM lesions and NAWM. Higher magnetic field strengths may be expected to improve the detection of inflammatory brain lesions in the MS WM.^12- 13 Specifically, one might expect to see multiple smaller or confluent lesions in DAWM when using high-field MRI, as DAWM or dirty WM^11,14,23 was discussed to represent areas with newly developing, so-called (p)reactive, lesions.²⁴ In the present study, high-field MRI of DAWM did not support this, and DAWM on 1.5 T was also diffusely abnormal on 4.7 T. This seems to indicate that DAWM represents a pathologic process distinct from (developing) focal WM lesions. This finding was confirmed in these histopathologic results.

So far, histopathologic studies of DAWM are scarce, partly because the definition of DAWM has so far been equally “diffuse”: until now, it was unclear whether DAWM consists of (early) inflammatory lesions or of more chronic pathologic abnormalities.

The DAWM areas histopathologically showed reduced myelin density and axonal loss and chronic fibrillary gliosis. The chronic character of this process was supported by the absence of acute axonal pathology, (p)reactive lesions, and blood-brain barrier leakage. The chronic gliosis was characterized by small glial cell bodies and a dense glial fiber meshwork (Figure 2). Shadow plaques, indicative of remyelination, were absent in DAWM. The observed mild activation of antigen-presenting cells in the WM is consistent with the recently described findings of widespread microglial activation⁵ and was more pronounced in DAWM than in NAWM.

Injury to axons and myelin in the central nervous system can occur through at least 2 mechanisms: as a consequence of direct injury (eg, by acute inflammatory pathology)²⁵ or as a result of degeneration distant to focal lesions (eg, wallerian degeneration).²⁶ In the present study, histopathologic analysis confirmed the chronic character of DAWM, which leads to the suspicion that its pathologic condition results from a secondary degenerative process remote from acute, focal damage. It is well established that secondary axonal degeneration proceeds through various stages, each with characteristic histologic and MRI features.²⁷ In the chronic phase, it appears hyperintense on T2-weighted images,²⁸ and histopathologic studies on wallerian degeneration²⁹ are consistent with the present histopathologic findings in DAWM.

The DAWM appeared slightly more frequent in the centrum semiovale in this study, which may be due to the fact that this WM region harbors many crossing fibers and inflammatory lesions, affecting many descending and ascending WM tracts. This “projecting pathology” was suggested previously in an MRI study showing secondary axonal degeneration in the corticospinal tracts of patients with MS with isolated neurologic syndromes.²⁸ Therefore, DAWM may be found in the direct surroundings of focal WM lesions but may also be observed in areas not directly related to focal WM lesions. Whether distinct effects of axonal transection in focal WM lesions alone or whether gray matter damage additionally accounts for the full extent of pathologic features in DAWM could not be shown in this study.

Axonal counts and Tm_Bodian showed good correlation. However, semiquantitative ratings of gliosis and Tm_GFAP did not correlate. Mean Tm_GFAP was equal in NAWM and focal WM lesions, whereas semiquantitative scores of glial fiber density showed a significant increase in DAWM compared with NAWM and also in focal WM lesions compared with DAWM. This heterogeneity of staining patterns may be responsible for the fact that the semiquantitative ratings of gliosis (taking the different stages of gliosis into account) correlated well with T1 and T2 relaxation time measurements in the present study, whereas Tm_GFAP showed no correlation with any of the quantitative MRI measurements. These findings now indicate that using transmittance for measuring gliosis may yield insufficiently precise information.

Of all quantitative MRI measurements, T1 and T2 relaxation time measurements showed highly significant differences among DAWM, NAWM, and focal WM lesions and produced the strongest correlations with axonal and myelin density and gliosis. Therefore, they should be considered most specific in detecting any tissue abnormalities. However, the clinical implementation of these sequences is technically more challenging, and, consequently, they are less commonly distributed. On the other hand, diffusion tensor imaging and MTR measures are more widely used in MS research and clinical trials, and in this study, FA (and to a lesser extent MTR) enabled distinction between DAWM and NAWM and between DAWM and focal WM lesions. Moreover, FA and MTR correlated strongly with axonal loss and reduced myelin density. The ADC showed few differences between regions of interest and weaker correlations with histopathologic findings. This finding can be explained by the use of formalin-fixed material, which is known to affect MRI measures (eg, shortening of relaxation times), thus affecting direct comparisons with in vivo measurements.^30- 31 Nevertheless, fixed material can still be suitable for providing clinically relevant conclusions.³¹ However, interpretation of MRI techniques, such as ADC (and MTR), which depend on free mobility of water molecules throughout the brain parenchyma, may be more challenging.³²

This study characterized DAWM in chronic MS and showed that it differs from focal WM lesions and NAWM in pathologic and imaging terms. Areas of DAWM consist of reduced myelin density, extensive axonal loss, and chronic fibrillary gliosis, differing from focal WM lesions and NAWM pathologic findings in that they seem to reflect the secondary, cumulative effects of ongoing (focal) pathologic abnormalities in the MS brain instead of focal inflammation. From these data, we conclude that tissue damage in DAWM, although mostly more subtle than that of focal WM lesions, is substantial. Clinical effects of DAWM could not be investigated in the present postmortem study and should be investigated in future studies focusing on DAWM in the in vivo situation using a validated scoring system for DAWM. Given its extent in these brain samples and its discriminate pathologic findings, inclusion of DAWM in lesion load assessments should be expected to result in a more accurate estimation of total disease burden than evaluation of the focal WM lesion load alone. Furthermore, it can potentially be of use as a new disease marker in future clinical trials that focus on the neurodegenerative aspects of MS.

Correspondence: Alexandra Seewann, MD, Department of Neurology, VU University Medical Center, De Boelelaan 1117, 1081 HV Amsterdam, the Netherlands (a.seewann@vumc.nl).

Accepted for Publication: October 3, 2008.

Author Contributions: All authors had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Vrenken, Castelijns, Polman, Barkhof, and Geurts. Acquisition of data: Seewann, Vrenken, van der Valk, Blezer, Pouwels, and Geurts. Analysis and interpretation of data: Seewann, Vrenken, van der Valk, Knol, Pouwels, Barkhof, and Geurts. Drafting of the manuscript: Seewann, Vrenken, Knol, and Geurts. Critical revision of the manuscript for important intellectual content: Vrenken, van der Valk, Blezer, Knol, Castelijns, Polman, Pouwels, Barkhof, and Geurts. Statistical analysis: Knol. Obtained funding: Geurts. Administrative, technical, and material support: Seewann, van der Valk, Blezer, and Castelijns. Study supervision: Vrenken, van der Valk, Knol, Castelijns, Polman, Pouwels, Barkhof, and Geurts.

Financial Disclosure: None reported.

Funding/Support: The MS Center Amsterdam received program grant 05-358c from the Dutch MS Research Foundation. This work was also supported by project grant 98-371 from the Dutch MS Research Foundation.

Additional Contributions: Christa van den Berg provided expert assistance with (immuno)histochemical stainings; Annette van der Toorn, PhD, helped with sequence programming at 4.7 T; and Lars Bö, PhD provided advice and scientific input.