OUP user menu

Post-mortem MRI-guided sampling of multiple sclerosis brain lesions
Increased yield of active demyelinating and (p)reactive lesions

C. J. A. De Groot, E. Bergers, W. Kamphorst, R. Ravid, C. H. Polman, F. Barkhof, P. van der Valk
DOI: http://dx.doi.org/10.1093/brain/124.8.1635 1635-1645 First published online: 1 August 2001

Summary

Macroscopic sampling of multiple sclerosis lesions in the brain tends to find chronic lesions. For a better understanding of the dynamics of the multiple sclerosis disease process, research into new and developing lesions is of great interest. As MRI in vivo effectively demonstrates lesions in multiple sclerosis patients, we have applied it to unfixed post-mortem brain slices to identify abnormalities, in order to obtain a higher yield of active lesions. The Netherlands Brain Bank organized the rapid autopsy of 29 multiple sclerosis patients. The brain was cut in 1 cm coronal slices. One or two slices were subjected to T1- and T2-weighted MRI, and then cut at the plane of the MRI scan into 5 mm thick opposing sections. Areas of interest were identified based on the MRI findings and excised. One half was fixed in 10% formalin and paraffin-embedded, and the corresponding area in the adjacent half was snap-frozen in liquid nitrogen. In total, 136 out of 174 brain tissue samples could be matched with the abnormalities seen on T2-weighted MRIs. The stage of lesional development was determined (immuno) histochemically. For 54 MRI-detectable samples, it was recorded whether they were macroscopically detectable, i.e. visible and/or palpable. Histopathological analysis revealed that 48% of the hyperintense areas seen on T2-weighted images represented active lesions, including lesions localized in the normal appearing white matter, without apparent loss of myelin but nevertheless showing a variable degree of oedema, small clusters of microglial cells with enhanced major histocompatibility complex class II antigen, CD45 and CD68 antigen expression and a variable number of perivascular lymphocytes around small blood vessels [designated as (p)reactive lesions]. From the macro-scopically not-visible/not-palpable MRI-detected abnormalities, 58% were (p)reactive lesions and 21% contained active demyelinating lesions. In contrast, visible and/or palpable brain tissue samples mainly contained chronic inactive lesions. We conclude that MRI-guided sampling of brain tissue increases the yield of active multiple sclerosis lesions, including active demyelinating and (p)reactive lesions.

  • brain tissue
  • immunohistochemistry
  • magnetic resonance imaging
  • microglia
  • multiple sclerosis
  • GFAP = glial fibrillary acidic protein
  • LCA = leucocyte common antigen
  • LFB = luxol fast blue
  • MHC = major histocompatibility complex
  • PAS = periodic acid–Schiff

Introduction

For obvious reasons, most pathology studies on multiple sclerosis rely on post-mortem human brain and spinal cord tissue (Adams, 1989a, b). As multiple sclerosis generally presents at autopsy as a chronic disease, most of the lesions found in the CNS are old (Ludwin, 2000). As a matter of course, neuropathological examination is largely restricted to the visible and palpable end-stage lesions. Study of lesions at an earlier stage, when the disease process is still active, is of particular interest for unravelling the pathogenesis of multiple sclerosis. Unfortunately, active demyelinating multiple sclerosis lesions are difficult to find, especially in unfixed brain tissue, unless samples are taken more or less blindly. This has forced investigators to sample as much brain tissue as possible, to find at least some early lesions, and this is a cumbersome and ineffective process.

The number of cerebral lesions on MRI scanning is usually much higher than clinically expected and shows dynamic changes over time (Isaacs et al., 1988; Koopmans et al., 1989). This encouraged us to develop a protocol to explore the benefit of T2- and T1-weighted MRI of post-mortem brain material in order to obtain more and earlier multiple sclerosis lesions. As all histopathological features of multiple sclerosis, such as inflammation, oedema, demyelination, axonal loss and gliosis are depicted as hyperintense lesions on T2-weighted images, a thorough characterization of the maturation and immunological activation stage of each sample was performed to establish the pathological specificity of the observed MRI abnormalities (Miller et al., 1998). We have used a modification of the staging system described by Bö and colleagues (et al., 1994) in order to include the discrete abnormalities often found in macroscopically normal appearing white matter (Van der Valk and De Groot, 2000). The procedure proved highly effective for selecting earlier lesions and yielded a high number of active demyelinating and (p)reactive lesions. Furthermore, almost half of the brain samples that were hyperintense on T2-weighted MRIs were neither visible nor palpable on macroscopic examination, even though they represented the full spectrum of lesion stages.

Methods

Patients

Brains from 29 multiple sclerosis patients (mean age 63 years, range 34–85 years; mean disease duration 25 years, range 7–54 years; mean post-mortem delay 7.5 h, range 3–11 h) were obtained at autopsy. The post-mortem specimens used in this study were obtained through the rapid autopsy system of the Netherlands Brain Bank, which has been approved by the Ethics Committee of the University Hospital Vrije Universiteit in Amsterdam. All patients and the next of kin had given written consent for autopsy, and for use of their brain and spinal cord tissue for scientific research. The clinical and autopsy data are shown in Table 1. Because almost all patients had been staying in a nursing home, no formal EDSS (Expanded Disability Status Scale) assessments had been performed during the last years of their life. Based on available medical documents, the multiple sclerosis disease type was determined retrospectively. All patients were in the progressive phase of the disease. In four patients, no data allowing a classification as either secondary progressive or primary progressive were available.

View this table:
Table 1

Autopsy and clinical descriptions

NBB no.Age (years)/sexType of MSDisease durationPM delay (h:min)Cause of death
Netherlands Brain Bank (NBB) number = combination of year of death (≥1996) and serial number (e.g. 025); disease duration = duration since first symptom (years); PM delay = post-mortem delay; F = female; M = male; SP = secondary progressive multiple sclerosis; PP = primary progressive multiple sclerosis; ND = not determined due to poor documentation.
96-02534/FSP10 6:50Heart failure
96-02669/FSP19 9:15Respiratory insufficiency
96-03674/FPP25 5:30Euthanasia
96-03957/FPP19 5:45Sepsis
96-04035/FSP10 5:45Cachexia
96-07440/FSP14 7:00Dehydration
96-07681/FSP or PP40 4:15Cachexia
96-10472/MSP22 4:45Carcinoma of the bladder/intestine
96-12153/FSP19 7:15Pneumonia
97-00662/FSP25 6:45Heart failure
97-05085/FPP54 4:15Cardio-pulmonary failure
97-07363/FSP32 6:45Cardiac arrest
97-07750/MSP or PP 7 5:40Pneumonia/cachexia
97-12346/MSP23 3:45Pneumonia
97-16040/FSP11 7:00Aspiration pneumonia
98-00970/FSP27 6:30Urinary tract infection
98-04477/MSP41 7:00Cachexia
98-06675/FSP or PP18 4:50Complications with bypass surgery
98-15876/FSP3414:15Respiratory insufficiency
98-17683/MPP42 7:00Pneumonia
98-17960/FSP36 8:50Cerebrovascular accident
98-18570/FPP19 8:50Heart failure
99-02564/FSP25 7:45Dehydration and pneumonia
99-05145/FSP1410:50Euthanasia
99-05458/FSP20 8:10Euthanasia
99-05672/MPP44 8:30Respiratory insufficiency
99-06279/FPP1210:00Cardiac arrest
99-07371/FND23 8:20Pneumonia
99-08671/FSP3110:20Respiratory insufficiency/cachexia

Tissue collection and MRI procedures

The brain was sliced into 1 cm thick coronal slices (Van Waesberghe et al., 1999). Subsequently, MRI of one or two brain slices was performed at 1.0 or 1.5 tesla using a surface coil and included T2-weighted spin echo [2200/20–80/1 (TR/TE/excitations where TR = repetition time and TE = echo time)] and T1-weighted spin echo (500/15/2). For T2 and T1 spin echo imaging, a single slice MRI was obtained in the centre of the brain slice. For all sequences, a 5 mm slice thickness, a field of view of 150 × 200 mm and a matrix size of 192 × 256 mm were used. Upon return of the scanned slices and MRIs to the autopsy room, the slices were cut at 5 mm using a 5 mm deep cutting device. The cut surface of the halved slice corresponded to the imaging plane. Guided by the T2-weighted MRI, several hyperintense lesions visible on the scan were dissected. Care was taken to include both isointense and hypointense lesions on T1-weighted images. A tissue block from one half was fixed in 10% neutral-buffered formalin (for paraffin embedding; ~2–5 cm2 in size), and the tissue block obtained from the adjacent half was snap-frozen in liquid nitrogen (~1–3 cm2 in size for technical reasons). The dissected brain tissue samples were both fixed and frozen in order to carry out additional pathogenetic studies for which unfixed material often is needed. All lesions were marked on the MRI scan, so the aspect of each lesion visible on the MRI scan could be matched afterwards with the corresponding (immuno)histologically stained sections derived from each bisected brain tissue sample. In this way, 174 brain samples were collected. In addition, from 10 multiple sclerosis cases (98-066, 98-158, 98-185, 99-025, 99-051, 99-054, 99-056, 99-062, 99-073 and 99-086) the macroscopic aspects of 54 MRI-sampled lesions, i.e. whether the lesion was visible and/or palpable, were recorded.

To confirm the clinical diagnosis, in addition to the MRI-detectable brain samples used in this study, many other brain and spinal cord sections were examined neuropathologically.

Histology and immunohistochemistry

For histopathological evaluation, 5 μm thick paraffin-embedded sections were stained with H&E (haematoxylin and eosin), Luxol fast blue (LFB) and with combined LFB–periodic acid–Schiff (PAS) reaction staining to delineate areas of myelin breakdown and the presence of material positive for LFB and PAS in phagocytic macrophages. In addition, 5 μm thick cryostat sections were stained with H&E and oil red O (to detect neutral lipids in phagocytic macrophages). For immunohistochemistry, both 5 μm thick paraffin-embedded and cryostat sections were processed and stained with a panel of antibodies directed against myelin, leucocyte- and glial-specific antigens (see Table 2) applicable to paraffin-embedded and frozen brain material, using the avidin–biotin–horseradish peroxidase method as described (De Groot et al., 1999). Conforming to the classification described by Bö and colleages and Van der Valk and De Groot (et al., 1994; Van der Valk and De Groot, 2000), different types of lesional stages in the MRI-sampled frozen and formalin-fixed brain material were identified.

View this table:
Table 2

Primary antibodies used for staining multiple sclerosis lesions

AntibodyIsotypeCellular specificityDilutionSource*
MBP = myelin basic protein. *Monosan, Uden, The Netherlands; Dako, Copenhagen, Denmark; Boehringer Mannheim, Germany. A gift of Dr J. H. Hilgers, Department of Obstetrics and Gynaecology, University Hospital Vrije Universteit, Amsterdam, The Netherlands.
GFAPIgG1Astrocytes1 : 10Monosan
CD68/KP1IgG1Macrophages/microglial cells1 : 600Dako
CD45/LCAIgG1Leucocytes1 : 50Dako
HLA-DR/LN3IgG1MHC class II antigens1 : 50J. H. Hilgers
MBPIgG1Myelin, myelin degradation products1 : 100Boehringer Mannheim

In the present study, the occurrence of so-called `shadow plaques' in the brain tissue samples was not investigated.

The LFB stainings of the paraffin sections were used to match slides and MRIs correctly. As previously described, on T1-weighted MRIs, the lesions were classified as either isointense with white matter, mildly hypointense when signal intensity was equal to or higher than grey matter, or severely hypointense when a lesion showed lower signal intensity than grey matter (van Walderveen et al., 1998; Van Waesberghe et al., 1999). The classification into mildly and severely hypointense was performed by one of the authors (E.B.) without knowledge of the histopathological changes.

Criteria used for defining different lesional stages

The different lesional stages in the collected material were determined on the basis of immunological parameters such as the presence of inflammatory cells (CD45-positive lymphocytes/macrophages and CD68-positive macrophages, containing phagocytosed myelin components), enhanced expression of major histocompatibility complex (MHC) class II antigen (HLA-DR) and CD45 expression on leucocytes (including lymphocytes) and resident microglial cells, the intensity of astrogliosis, using an antibody for glial fibrillary acidic protein (GFAP), as well as certain morphological criteria such as cellularity of the centre and rim of the lesion (Van Waesberghe et al., 1999; Van der Valk and De Groot, 2000). In Table 3, a detailed description is given of the five different stages that could be distinguished in the MRI-sampled brain tissues. We used this staging system as the (p)reactive stage can only be recognized at the tissue level by immunological means (i.e. enhanced protein expression for MHC class II and CD45 antigens on resident microglial cells and infiltrating lymphocytes). Determination of the type of lesional stage present in each sample was performed independently by at least two of the authors (W.K., C.J.A.D.G and P.v.d.V.).

View this table:
Table 3

Classification of different lesion types

Type of lesion
NormalLow expression of CD68 (KP1), CD45 (LCA) and HLA-DR (MHC class II antigens) on resident microglial cells. No inflammation, no loss of myelin.
(P)reactiveStrong expression for CD68, CD45 and HLA-DR on clusters of microglial cells localized in the close vicinity of microvessels in the normal appearing white matter. Peri- and intravascular accumulation of CD45- and HLA-DR-positive lymphocytes in microvessels. A variable degree of oedema is detectable, but there is no apparent loss of myelin.
ActiveDemyelinating lesion with perivascular and parenchymal macrophages that have a strong CD68, CD45-and HLA-DR expression. These phagocytes contain myelin basic protein-positive myelin components or oil red O/PAS-positive lipids. A variable number of strongly CD45-expressing lymphocytes are localized predominantly within the perivascular spaces of blood vessels. GFAP-positive reactive astrocytes with long processes are distributed throughout the demyelinating regions.
Chronic activeThe hypocellular demyelinated centre of the lesion contains a small number of CD68-positive macrophages with some residual oil red O/PAS-positive (neutral) lipids. Few CD45-positive lymphocytes are present in perivascular cuffs. The hypercellular rim contains perivascular and parenchymal CD68-positive macrophages with myelin basic protein-positive myelin components or oil red O/PAS-positive lipids. GFAP-positive reactive astrocytes are localized predominantly at the edge of the lesion centre and within the hypercellular rim.
Chronic inactiveThe hypocellular demyelinated lesion contains largely (isomorphic) gliosis with widened extracellular spaces. A very small number of residual CD68-positive macrophages are present in the brain parenchyma and CD45-positive lymphocytes within the perivascular spaces of blood vessels.

Results

Localization of MRI-detectable lesions and matching with LFB-stained brain tissue sections

It appeared that 22% (n = 38) of the LFB-stained slides could not be matched correctly with the T2-weighted MRIs due to the absence of topographical identification marks or incorrect positioning of the MRI. These brain samples were not used in the present study.

Neuropathological findings

The MRI-detectable brain tissue samples represented the complete range of histopathological lesion types as described in Table 3. Thirty-two were classified as (p)reactive, 30 as active, 25 as chronic active and 30 as chronic inactive. Thirteen samples did not contain multiple sclerosis-related histopathological characteristics (except for variable oedema).

As shown in Table 4, lesions were found in most of the dissected brain samples (n = 136). Both formalin-fixed and snap-frozen brain material derived from the opposite tissue block contained comparable lesional activity. Therefore, both frozen and fixed brain material could be used for staging of the lesions. In the paraffin-embedded material, a slightly higher number of different types of lesions was detected within an individual brain tissue sample. This resulted from the smaller size of the snap-frozen material (~1–3 cm2) that could still be handled for cryostat sections. Nevertheless, Table 4 shows that there was a considerable homogeneity in the type of lesions within an individual patient. In 14 out of 29 patients, all lesions were of the same pattern, and in at least seven additional patients the different lesion types appeared closely related, i.e. (p)reactive and active, chronic active and chronic inactive. In addition, the size and shape of different lesions obtained from one patient were often similar (not shown). Figures 1–3 demonstrate representative examples of the different lesion types that could be matched with the corresponding T2- and T1-weighted MRIs.

View this table:
Table 4

Histopathological findings of post-mortem T 2 -weighted MRI-detectable white matter brain samples

NBB no.Type of MSNo. of samplesNo. of lesionsLesional stage
Normal(P)reactiveActiveChronic activeChronic inactive
Lesional activity is assessed in brain tissue samples taken from one or two 1 cm2 thick brain slices. On T2-weighted MRI, regions of interest were sampled and divided into two equal parts. One half of the sample was frozen in liquid N2 and the other half fixed in 10% neutral-buffered formalin and routinely embedded in paraffin. Here the different lesional stages found in the paraffin-embedded brain samples are shown. ND = not determined; SP = secondary progressive; PP = primary progressive; RR = relapsing–remitting. *Demyelinated lesions were present in other brain areas or in the spinal cord allowing the diagnosis of multiple sclerosis.
96-025SP 3 3 3
96-026SP 3 3 3
96-036PP 3 2 1 2
96-039PP 5 5 1 4
96-040SP 4 5 1 3 1
96-074SP 8 8 4 4
96-076SP or PP 5 2 1 2
96-104SP 3 3 2 1
96-121SP 4 4 4
97-006SP 7 2 2
97-050PP 3 3 3
97-073SP 1 1 1
97-077SP or PP 3 3 2 1
97-123SP 7 7 7
97-160SP 5 5 5
98-009SP 6 5 1 4
98-044SP 2 0* 2
98-066SP or PP 4 0* 4
98-158SP 3 2 1 1 1
98-176PP 8 7 1 3 4
98-179SP 2 2 2
98-185PP 8 8 5 3
99-025SP 9 9 9
99-051SP 7 7 4 3
99-054SP 3 2 1 1 1
99-056PP 4 4 1 3
99-062PP 4 4 2 2
99-073ND 6 3 2 2 1
99-086RR-SP 6 8 1 2 5
Total1361171332302530
Fig. 1

Correlation of T1- and T2-weighted MRI and histopathology. Coronal slice of post-mortem T1- (A) and T2-weighted (B) MRI of multiple sclerosis case 96-074. Frames with the dotted lines are T2-weighted MRI-visible brain samples that were excised but not illustrated here. Frames 1 and 2 are depicted in C–F, showing that these are demyelinated lesions that are both severely hypointense on T1 MRI. (C) Photograph of an LFB-stained paraffin-embedded brain tissue section that could be matched with frame 1 on the T2-weighted MRI which contained an active demyelinating lesion in the white matter (asterisk, see D). (D) CD68 immunostaining was performed on a 5 μm thick tissue section cut from the frozen part of the excised brain sample shown in frame 1. Numerous parenchymal and perivascular foamy macrophages are distributed throughout the lesion (arrows). (E) Photograph of an LFB-stained paraffin-embedded brain tissue section that could be matched with frame 2 on the T2-weighted MRI which contained an active demyelinating lesion in the white matter (star, see F). (F) Microphotograph of a myelin basic protein immunostaining that was performed on a 5 μm thick paraffin-embedded tissue section cut from the formalin-fixed part of the excised brain sample shown in frame 2. Phagocytic macrophages containing myelin basic protein-positive myelin components are distributed through the lesion (arrows). Scale bars: D = 20 μm; F = 1 μm (haematoxylin staining of nuclei).

Fig. 2

Correlation of T1- and T2-weighted MRI and histopathology. Coronal slice of post-mortem T1- (A) and T2-weighted (B) MRI of multiple sclerosis case 98-185. Frames with the dotted lines are T2-weighted MRI-visible brain samples that were excised but not illustrated here. Frame 1 is depicted in CF, showing mildly hypointense T1-weighted MRI abnormalities which were macroscopically neither visible nor palpable. (C) Photograph of an LFB-stained paraffin-embedded brain tissue section that could be matched with frame 1 on the T2-weighted MRI which contained a small active demyelinating lesion (a) and two (p)reactive lesions (b and c) in the white matter. (D) Photomicrograph of an active demyelinating multiple sclerosis lesion with numerous phagocytic macrophages containing LFB-positive myelin components (arrows). (E) Region b contains a (p)reactive lesion that is characterized by accumulated CD45-immunoreactive lymphocytes (arrowheads) in a blood vessel. The vessel is surrounded by a cluster of microglial cells that are strongly immunoreactive for CD45 (asterisk; arrows). (F) In region c, clusters of microglial cells are present that display an intense immunoreactivity for HLA-DR (arrows). Scale bars: D = 1 μm; E and F = 20 μm (haematoxylin staining of nuclei).

Fig. 3

Correlation of T1- and T2-weighted MRI and histopathology. Coronal slice of post-mortem T1- (A) and T2-weighted (B) MRI of multiple sclerosis case 99-025. Frames with the dotted lines are T2-weighted MRI-visible brain samples that were excised but not illustrated here. Frame 1 is depicted in CF, showing a severely hypointense lesion that was macroscopically visible and palpable. (C) Photograph of an LFB-stained paraffin-embedded brain tissue section that could be matched with frame 1 on the T2-weighted MRI which contained a chronic inactive demyelinated lesion (asterisk, see D). (D) Photomicrograph of this region showing a hypocellular area which is totally demyelinated (LFB staining). (E) Only a few residual lymphocytes (arrows) and macrophages (arrowheads) are present. (F) The demyelinated lesion is filled up with the cellular processes of GFAP-immunoreactive astrocytes (arrows). Scale bars: D, E and F = 20 μm (haematoxylin staining of nuclei).

Correlation of lesion pathology with MRI findings

Table 5 shows the correlations between the different lesional stages and the different signal intensities that were seen on T1-weighted MRIs. Brain samples showed variable appearances within a given patient on the MRIs (isointense, mildly hypointense or severly hypointense on T1). Only two cases (96-040 and 99-025) showed severely hypointense abnormalities on T1 in all dissected brain samples; the lesions of the first case were classified as active and of the latter case as chronic inactive. Severely hypointense lesions strongly correlated with chronic active and chronic inactive lesions compared with normal and (p)reactive lesions (Pearson χ2, P < 0.01). The large majority of the (p)reactive lesions were mildly hypointense on T1-weighted MRIs (81%), whereas active lesions could also be severely hypointense (33%). No significant differences were found when (p)reactive lesions were compared with samples without multiple sclerosis-related abnormalities. Further- more, the group of brain samples that were severely hypointense contained more chronic lesions.

View this table:
Table 5

T 1 -weighted characteristics of the different types of multiple sclerosis lesions

Lesion stageNo. of formalin-fixed brain samples (n = 130)No. of T1 abnormalities (%)
IsointenseMildly hypointenseSeverely hypointense
Significance of difference from normal: *not significant; P < 0.001 (Pearson χ2 test); P < 0.001; §P < 0.001.
Normal136 (46) 7 (54) 0
(P)reactive324 (13)26 (81) 2 (6)*
Active302 (7)18 (60)10 (33)
Chronic active252 (8) 5 (20)18 (72)
Chronic inactive30016 (53)14 (47)§

In the course of the current protocol, we already observed that MRI-detectable brain samples contained many (p)reactive lesions. Therefore, in the second part of the study, we recorded whether the dissected samples were visible and/or palpable. Table 6 shows the correlation between the macroscopic aspects of the sampled brain tissue and the MRI findings. In this subgroup, 44% of the MRI-detectable lesions were macroscopically neither visible nor palpable. The majority of these T2-weighted MRI-detectable lesions were mildly hypointense on the T1-weighted image (33%). Histopathological characterization of the MRI-detectable but macroscopically not visible or palpable brain tissue samples revealed different stages of lesional development: three were classified as normal brain tissue, 14 as (p)reactive, five as active and two as chronic inactive. In contrast, 67% of the brain samples that were visible and/or palpable contained chronic inactive lesions (n = 20). The remaining samples within this group were classified as (p)reactive (n = 4), active (n =4) and chronic active (n = 2).

View this table:
Table 6

Correlation between macroscopic aspects and MRI abnormalities of post-mortem brain samples derived from 10 multiple sclerosis cases

MacroscopyNo. of samples [n = 54 (%)]No. of T1 abnormalites (%)Lesion stage
IsointenseMildly hypointenseSeverely hypointenseNormal(P)reactiveActiveChronic activeChronic inactive
Visible/palpable19 (35)0 8 (15)11 (20)0 01018
Visible/not palpable 8 (15)3 (5.5) 3 (5.5) 2 (4)0 132 2
Not visible/palpable 3 (6)0 3 (6) 00 300 0
Not visible/not palpable24 (44)5 (9)18 (33) 1 (2)31450 2

Discussion

This study shows that use of MRI-guided sampling of post-mortem brain tissue resulted in a high yield of inflammatory and demyelinating multiple sclerosis lesions derived from patients who were in the progressive phase of their disease at the time of autopsy. From 10 patients, the macroscopic aspect of the MRI-detectable brain sample was recorded, i.e. whether the lesion was visible and/or palpable. Forty-four per cent of the MRI-detectable abnormalities were macroscopically neither visible nor palpable. Interestingly, the majority of these brain samples contained lesions that were classified as either (p)reactive or active demyelinating. Thus, the procedure effectively detects inflammatory lesions in a group of patients with a long disease history. A disadvantage of the MRI-guided sampling method was that 22% of the initially sampled brain tissue blocks could not be matched properly with abnormalities seen on the T2-weighted MRIs. This can be explained partly by a lack of topographical landmarks in the tissue sections such as the presence of ventricular or cortical gyral structures. Thus, the sampling method could be improved by including recognizable landmarks.

An important finding in the present study was the observation that the lesional patterns, i.e. the size and shape of lesions in different brain samples derived from the same patient, were homogeneous. This suggests patient-specific lesion formation. In line with this, Lucchinetti and colleagues found a homogeneous pattern of demyelination within a patient and a heterogeneous pattern of demyelination between patients using active demyelinating multiple sclerosis lesions derived from biopsy and autopsy cases (Lucchinetti et al., 2000). Further studies are needed, including more cases, to conclude that multiple sclerosis might have multiple pathogenic mechanisms or aetiologies based on lesion heterogeneity found between patients (Ludwin, 2000).

The advent of MRI has altered patient management in multiple sclerosis considerably, and the present study shows a very effective sampling method to include immunologically active lesions in autopsy material. The use of T2-weighted MRI-guided sampling of post-mortem material enabled dissection of `non-old' lesions. However, T2-weighted MRIs are sensitive to all the pathological stages, and thus are not able to discriminate between different stages. Our data once more confirm that the pathology ranging from oedema and mild demyelination through completely destroyed lesions are seen as hyperintense lesions as described earlier (Van Waesberghe et al., 1999; van Walderveen et al., 1999).

In a previous study, using part of the brain samples (109 samples from 17 patients) described in the present study, T2 visible lesions were selected for measuring magnetization transfer ratios, T1 contrast ratio, myelin and axonal density. T1 hypointensity and magnetization transfer ratios were strongly associated with axonal density. The lowest magnetization transfer ratio values (0.15) and axonal densities (determined according to a visual ranking system from 0 to 100% on Bodian-stained tissue sections) were found in the severely hypointense T1 lesions (Van Waesberghe et al., 1999). In the present study, T1 hypointensity was strongly associated with a chronic stage of lesion development. Nevertheless, it was shown that within the same patient, a variable appearance of lesions on T1-weighted MRIs can still occur. Thus, it can be concluded that the degree of T1 hypointensity is affected by different histopathological abnormalities such as the extent of axonal loss, the degree of extracellular oedema and demyelination.

Besides the frequent occurrence of active demyelinating, chronic active and chronic inactive lesions, a large number of lesions showed discrete (immunological) changes, including oedema and an increased CD45, CD68 and MHC class II expression either on clusters or on individual microglial cells in the close vicinity of blood vessels containing a variable number of CD45-positive lymphocytes. However, no apparent loss of myelin was detected in these (p)reactive lesions (Van der Valk and De Groot, 2000). The findings suggest that these lesions are very early, i.e. they may precede active lesions. Further studies should provide additional evidence for this hypothesis. Immunological analyses (i.e. for chemokine/chemokine receptor expression patterns, transcription factors controlling MHC class I and class II expression and tumour necrosis factor-α) currently are being performed on freshly frozen brain tissue samples and might provide further information as to whether these (p)reactive stages represent the initial inflammatory phase of lesion development. Given the high frequence of (p)reactive lesions in these chronic multiple sclerosis patients, it presumably is unlikely that they all progress into full-blown chronic lesions.

At present, the source of different inflammatory mediators responsible for the local activation of microglial cells is not precisely known, but data have been published that show that microglial activation could be the result of the upregulation and induction of adhesion molecules, chemokines and a cytokine cascade. Chemoattractive mechanisms might play an important role in migration of (microglial) cells towards the (p)reactive and active lesions (Ransohoff, 1999; Boven et al., 2000; Zhang et al., 2000). In developing multiple sclerosis lesions, it is thought that there is an impaired function of the blood–brain barrier, which would allow infiltration of inflammatory cells into the parenchyma of the CNS. However, whether blood monocyte-derived phagocytes or cells originating from resident microglial cells are the most important source of the accumulated phagocytes in active demyelinating multiple sclerosis lesions is still a matter of debate (Ulvestad et al., 1994; Li et al., 1996; Carson and Sutcliffe, 1999; Trapp et al., 1999).

Brain samples also contained areas of reduced myelin density either in close proximity to or at the edges of a demyelinated lesion. Further characterization is necessary to ascertain whether these are possibly (partly) demyelinated or remyelinated lesions.

Our results indicate that our post-mortem tissue samples obtained from patients with definite and typical multiple sclerosis are representative of progressive multiple scerosis in vivo. Based on the distribution and density of inflammatory cells (lymphocytes and phagocytic macrophages) and activated microglial cells, demyelinated multiple sclerosis lesions were divided into four categories. Except for the (p)reactive lesion type, the other types of lesions have been described in many other studies which involve human multiple sclerosis material (et al., 1994; Trapp et al., 1999; Van der Valk and De Groot, 2000).

The present study clearly shows that guidance using MRI is more useful than macroscopy alone when sampling post-mortem lesioned brain tissue, as lesions that are visible and palpable in autopsy material have quite often become chronic. In spite of the logistical problems, combining standard macroscopic dissection with MRI scanning allows a representative sampling of multiple sclerosis lesions. The finding of many immunologically active lesions in patients with longstanding disease again underlines the ongoing nature of the pathological process in multiple sclerosis.

Acknowledgments

We wish to thank Michiel Kooreman, Anne Holtrop and José Wouda from the Netherlands Brain Bank for their excellent support during the autopsy procedure, Lisette Montagne and Marieke Pigmans for the histo- and immunohistochemical staining, Ellen Mommers for statistical analysis, and Sandra Hulshof for help with the figures. This study was funded by the Dutch Foundation `Vrienden Multiple Sclerosis Research'.

References

View Abstract