Histopathologic correlate of hypointense lesions on T1-weighted spin-echo MRI in multiple sclerosis

In multiple sclerosis (MS), a discrepancy exists between clinical disability and abnormalities on conventional T2-weighted spin-echo (SE) MRI. Several explanations have been put forward,¹ including inadequate methods to determine disability. The MR technique, on the other hand, has several limitations, including the inability of T2-weighted MR images to identify the pathophysiologic heterogeneity of lesions, contributing to this clinicoradiologic paradox.² Early lesions, characterized by edema, inflammation, and mild demyelination, as well as chronic lesions, characterized by severe demyelination, gliosis, and axonal loss, alter tissue water concentration and distribution, and subsequently increase signal intensity (SI) on T2-weighted images. However, only severe demyelination and axonal loss are likely to contribute to persistent conduction impairment and deficit. Further, the spinal cord is a frequent site of symptomatic lesions, but is often not routinely examined in patients with MS.

To improve the correlation between abnormalities on MR images and persistent deficit, specific MR markers of demyelination or axonal loss are needed. Apart from strongly reduced magnetization transfer (MT) ratios and reduced N-acetyl aspartate peaks using proton MR spectroscopy, hypointensity on T1-weighted SE images has been proposed as an MR marker more specific for severe tissue destruction in MS.^3-7

Hypointense lesions ("black holes") have low SI compared with surrounding normal-appearing white matter (NAWM) on T1-weighted (short TR/short TE) SE images, and appear bright on T2-weighted SE MR images. Several studies suggest that hypointense lesions are the MR equivalent of severe demyelination or axonal loss. Firstly, hypointense lesions are characterized by low magnetization transfer ratios, indicative of considerable structural loss.⁸ Secondly, a strong inverse correlation exists between T1 relaxation time and loss of MT.⁹ Furthermore, increase in hypointense lesion load correlates strongly with increase in disability in two preliminary follow-up studies.^6,7

Postmortem MRI is a reliable method to investigate the pathologic correlate of white matter changes. The presence and location of MR abnormalities on T2-weighted images correlate well with pathologic findings in formalin-fixed tissue.^10,11 Unfortunately, fixation of brain tissue alters T1 relaxation time,¹² for which reason postmortem T1-weighted images of formalin-fixed tissue cannot be used to investigate hypointense lesions. In unfixed brain tissue T1 and T2 relaxation times will remain intact up to 24 hours after death,¹² and therefore postmortem images of unfixed brain material obtained within this time period do allow analysis of hypointense lesions.

To evaluate the concept of using hypointense lesions as a putative marker for severe tissue destruction, we investigated the histopathologic correlate of MS lesions on postmortem MR images of fresh brain material.

Methods. Five brains of patients with clinically definite MS were obtained from the Netherlands Brain Bank within 12 hours after death. The unfixed brains were scanned directly after removal, in a coronal plane at 0.6 (Technicare, Solon, OH) or 1.5 T (SP63, Siemens AG, Erlangen, Germany), with the brains placed in a standard head coil. MRI included a T2-weighted SE pulse sequence (0.6 T: TR/TE/number of excitations [NEX], 2,000/48/2); 1.5 T: 2,800/45/1) and a T1-weighted SE (0.6 T: 585/28/4; 1.5 T: 400/15/3) sequence. For all sequences a slice thickness of 5 mm, a field of view of 180 mm, and a matrix size of 256 × 256 mm was used.

After the imaging procedure, a neuropathologist (blinded to the MR findings) cut the brains according to the MRI plane at 5-mm intervals. The brains were examined macroscopically and tissue blocks were excised, formalin fixed, and paraffin embedded. Sections were stained with hematoxylin-eosin(HE), Klüver-Barrera, and Bodian stains, and immunostained for glial fibrillary acid protein (GFAP).

The postmortem brain MR images were reviewed by two readers (M.v.W., F.B.), who were not involved in the subsequent histopathologic analysis. Only those lesions were selected for analysis for which both readers agreed in conference on the coregistration of the plaques seen on MR images and at histology. Contour and shape of the lesions, as well as the location in the brain, were taken into account during the coregistration. After this, the T2-weighted images were compared with the T1-weighted images to locate the lesions on the T1-weighted images. Degree of hypointensity (score, 0 to 2) of the selected lesions was estimated on the T1-weighted images in comparison with NAWM. Score 0 was given when lesions had the same SI as NAWM (not visible) on T1-weighted images (isointense lesions), score 1 represents mild hypointensity compared with NAWM ("gray"), and score 2 was given when a lesion was severely hypointense compared with NAWM ("black;"figure 1). Further, computer-assisted analysis of the selected lesions was performed, which included calculation of the contrast ratio on T1-weighted images. The contrast ratio was defined as the SI of a lesion divided by the SI of NAWM, which was analyzed in the same slice as the lesions were visible. SI of NAWM was measured as the mean of two rectangular regions of interest surrounding the lesion, each having an area of at least 10 mm². For isointense lesions (not visible on T1-weighted images), the T2-weighted images were used to locate precisely the lesions on the corresponding T1-weighted images to allow for SI measurement.

Table 1 summarizes the histopathologic variables that were analyzed. To assess plaque activity, the cellularity of plaque center and border, number of vessels with clearly visible perivascular infiltrate, presence of (thinly) myelinated axons, and presence of reactive astrocytes within the plaques and in surrounding white matter were analyzed. The plaques were classified, according to their stage, into inactive chronic (a hypocellular MS plaque), active chronic (a hypocellular MS plaque with a hypercellular edge, compared with the cellularity of NAWM), and acute (a hypercellular MS plaque).¹³ Furthermore, shadow plaques were distinguished as MS plaques in which the number or the intensity of the myelin staining of the myelinated axons is decreased.¹⁴ Degree of matrix destruction (loss of density of the neuropil or enlargement of size of the meshes of neuropil) and the percentage of residual axons of a lesion were analyzed in comparison with an area with NAWM. For this purpose, the HE and Bodian sections of the selected lesions were photographed(magnification, × 400) and ordered according to a visual ranking system from 1 to 10 (degree of matrix destruction) and 0 to 100% (residual axons; see table 1 and figures 2 and 3). The number of reactive astrocytes was counted in plaques and surrounding tissue to estimate the amount of gliosis.

Correlations were examined by calculating the Spearman rank correlation coefficient (R). p Values less than 0.05 were considered statistically significant.

Results. Descriptive. The clinical characteristics of the patients are listed in table 2. All patients had chronic progressive MS (four patients had secondary progressive [SP] MS and one patient had primary progressive [PP] MS). Disease duration ranged from 8 to 30 years (mean, 20 years). All patients were admitted to a nursing home until their death.

Twenty-eight tissue blocks were available for coregistration with postmortem MR images. Nineteen lesions could be coregistered with certainty, and one to eight lesions were identified per patient. The distribution of the lesions and the histopathologic stage within patients is shown intable 3. Of the 19 lesions, six were isointense to NAWM, seven were mildly hypointense, and six were severely hypointense. The median contrast ratio of the isointense lesions was 0.98 (range, 0.95 to 1.0), 0.87 for the mildly hypointense lesions (range, 0.77 to 0.98), and 0.70 for the severely hypointense lesions (range, 0.58 to 0.77). The histopathologic characteristics of these lesions are summarized intable 4.

Of the isointense lesions, three contained a hypocellular center (two chronic active, one chronic inactive), in two lesions the center was hypercellular (acute), and one lesion was classified as shadow plaque. Of the mildly hypointense lesions, five contained a hypocellular center (four chronic active, one chronic inactive), and in two lesions the center was hypercellular (acute). All severely hypointense lesions contained a hypocellular center (five chronic active, one chronic inactive).

Eighteen lesions were almost completely demyelinated and one was a shadow plaque. Thinly myelinated axons were only absent in the rim of two lesions, both graded as mildly hypointense.

In all lesions at least some degree of matrix destruction was present. The density of the neuropil was more decreased in hypointense lesions (mild and severe) compared with isointense lesions (seetable 4).

Reactive astrocytes were present in all lesions, as well as in the surrounding NAWM. In 15 plaques, the number of reactive astrocytes within the lesion was higher compared with the number in the surrounding white matter. In three isointense lesion and in one severely hypointense lesion the number of reactive astrocytes in the surrounding white matter was higher.

In one isointense and one mildly hypointense lesion no axonal loss was present compared with the NAWM, and all other lesions showed axonal loss, varying from only 10% in isointense lesions to a complete loss of axons in severe hypointense lesions (see figure 4).

Statistics. Contrast ratios of the lesions were highly correlated (R = -0.90; p < 0.01), with degree of hypointensity scored semiquantitatively (figure 5). Degree of hypointensity on T1-weighted MR images correlated significantly with the percentage of residual axons (R = -0.71; p = 0.001). A correlation between degree of matrix destruction and degree of hypointensity was present, that did not reach statistical significance (R = 0.45; p = 0.052). Degree of matrix destruction showed a significant inverse correlation with percentage of residual axons (R = -0.65; p = 0.003). No correlations were observed between degree of hypointensity on T1-weighted MR images and histopathologic variables of plaque activity. Degree of hypointensity was inversely correlated with the number of reactive astrocytes in surrounding tissue (R = -0.45; p = 0.051), but not with number of reactive astrocytes within the plaque (R = -0.10; p = 0.69). No other histopathologic parameter correlated with degree of hypointensity on T1-weighted SE images.

Discussion. In this postmortem study we analyzed the histopathologic correlate of MS lesions, which on T1-weighted (short TR/short TE) MR images are represented as regions of low SI (black holes) compared with NAWM. Nineteen lesions could be coregistered with certainty in five MS patients. This was only a small part of the total number of lesions visible on postmortem MRI. The difficulty in coregistering more lesions was a consequence of the scanning method that had been used in the early years of postmortem MRI. After imaging, the whole fresh brain was cut and it appeared difficult to obtain the same slice position as in the MRI. Since then we have adapted our strategy. Two fresh brain slices are excised (frontal and occipital) and imaged. According to abnormalities on the MR hard copies, tissue blocks are excised and formalin fixed. The advantage of this new protocol is twofold: it is less time consuming than imaging the whole brain, and coregistration of lesions is facilitated.

The composition of the material had some drawbacks that may hinder general applicability of our results. First, the majority of the lesions were chronic MS plaques (11 chronic active, three chronic inactive) showing hypocellularity and absence of myelinated axons with reactive astrocytes and gliosis. This finding may be inherent to the use of postmortem material: in our study the mean disease duration was 20 years and all patients had entered the progressive phase. Second, five of six severely hypointense lesions and eight of 11 chronic active lesions occurred in one patient. Theoretically, this may indicate that the correlation found between axonal loss and hypointense lesions is an individual trait of Patient 2, who may have been suffering from an aggressive form of MS. However, we also found one isointense and two mildly hypointense lesions in this patient, making this assumption unlikely. Third, 15 of 19 lesions occurred in two of five patients, implying that a larger study on a more varied cohort is still desirable.

Although demyelination with preservation of axons is the pathologic hallmark of MS, 17 lesions (87.5%) in our study showed axonal loss varying from 10% in isointense lesions to a complete loss of axons in severely hypointense lesions. This is in concordance with the study of Lassmann et al.,¹⁵ who found that in late chronic MS, secondary tract degeneration and brain atrophy are common features.

Degree of hypointensity on T1-weighted MR images correlated significantly with percentage of residual axons in our study. This finding might explain the good correlation found between increase in hypointense lesion load and increase in disability in MS patients.^6,7 In the study by Truyen et al.⁷ this correlation was only found in SP MS patients and could not be established for patients in the relapsing-remitting phase of the disease, suggesting that hypointense lesions, and thus axonal loss, develop as a consequence of "failure of remission."

Earlier studies have shown that axonal loss coincides with expansion of the extracellular space^15,16 and that MR characteristics of an expanded extracellular space are marked prolongation of T1 and biexponential T2 relaxation times.¹⁷ Other factors, like edema, also enlarge the extracellular space, indicating that axonal loss is not the only histopathologic feature that might prolong T1 relaxation time in MS lesions and cause hypointensity on T1-weighted images. One preliminary study¹⁸ showed that the majority of new enhancing (acute) lesions in MS appears hypointense (at least to some degree) on corresponding unenhanced T1-weighted images, but most return to isointensity within 3 months after enhancement, although they often will remain visible on T2-weighted images. Further, severe vascular lesions, such as lacunar infarcts, could also cause low SI on T1-weighted images.¹⁹ Relative to the occurrence of hypointense MS lesions, this will be an infrequent finding. Histopathologically we have not observed findings such as central necrosis and surrounding isomorphic gliosis.²⁰ More generally, further studies on the occurrence of hypointense lesions in disorders other than MS and on their natural history is needed before hypointense lesions can be definitely accepted as a surrogate marker for fixed deficit in MS.

The number of reactive astrocytes within the lesions did not correlate with degree of hypointensity of the lesions. The number of astrocytes in the surrounding NAWM showed an inverse correlation with degree of hypointensity of the lesions. Most of these astrocytes appeared reactive. An increase in astrocytes and their fibrillary processes frequently is not restricted to the plaque area itself but may extend to the surrounding NAWM. Remyelination and repair of preexisting lesions or activation of astrocytes in response to increased permeability and inflammatory edema around the acute lesion might explain this feature.²¹

The presence of thinly myelinated axons in the rim (in contrast to the center of the lesion) may indicate attempts at remyelination.^14,22 According to Prineas et al.,²² typical chronic MS plaques found at autopsy in patients with long-standing disease are not the outcome of a single episode of acute demyelination, but are residua of multiple episodes of acute demyelination centered on the same blood vessel or group of vessels. As a consequence of ongoing inflammatory activity at the edge of a lesion or recurrent demyelination of the same lesion, remyelination fails and subsequently shadow plaques (representing areas of thinly myelinated axons resulting from remyelination) will convert into the classic fully demyelinated plaques.²² In our study, one lesion was classified as shadow plaque. This lesion was visible on the T2-weighted image, but not on the corresponding T1-weighted image (isointense lesion). This indicates that, although repair of myelin sheaths does occur in MS, residual signal alterations on T2-weighted images persist, which may contribute to the weak correlation found between clinical disability and T2 lesion load. An alternative explanation may be that this plaque represented a"nascent" plaque, without inflammatory reaction or loss of oligodendrocytes.

In conclusion, our postmortem study suggests that hypointense lesions seen on T1-weighted SE MR images in progressive MS patients, at least in part, are associated histopathologically with severe tissue destruction, including axonal loss. However, more extensive and more varied patient material has to be examined to substantiate the hypothesis that chronic hypointense lesions can serve as a surrogate marker of persistent deficit in MS.

Acknowledgments

The MR Center for MS Research is supported by the Stichting Vrienden MS Research, the Academic Hospital Vrije Universiteit, and the Medical Faculty of the Vrije Universiteit.