Post-mortem high-resolution MRI of the spinal cord in multiple sclerosis -A correlative study with conventional MRI, histopathology and clinical phenotype
Gjla. Nijeholt et al., Post-mortem high-resolution MRI of the spinal cord in multiple sclerosis -A correlative study with conventional MRI, histopathology and clinical phenotype, BRAIN, 124, 2001, pp. 154-166
We used high-resolution MRI to study the post-mortem appearance of spinal c
ord multiple sclerosis in relation to histopathology and low-resolution ima
ges. Fifty-nine 3 cm long formalin-fixed spinal cord specimens from 19 mult
iple sclerosis patients and three controls were studied. Clinical character
istics of each patient were reviewed. High-field MRI consisted of proton-de
nsity weighted spin-echo imaging,vith an in-plane resolution of 80 mum. Spe
cimens were also imaged at 1.0 T, with 1 mm pixel resolution. After MRI, th
e specimens were cut at 5 mm intervals and stained for myelin (Luxol fast b
lue/cresyl violet) and axons (Bodian method). Two observers scored the MRIs
for abnormalities and divided them into (i) well-delineated areas of high
signal intensity (SI) and (ii) poorly defined areas of mildly increased SI.
Abnormalities were scored semiquantitatively, white matter and grey matter
separately. In 81 sections the total area of abnormalities per section was
measured on both histopathology sections and on matched high-field MRIs, A
bnormalities ranged from just a few abnormal areas to complete involvement
of the spinal cord specimen. Patients with an aggressive disease course had
more abnormalities than patients with a mild or intermediate disease cours
e. Areas of mildly increased SI were seen in all specimens, and were often
found around focal high-SI lesions. However, in six patients, areas of mild
ly increased SI were the predominant finding on the MRIs, correlating with
a primary progressive disease course. Histopathologically, high-SI areas co
rrelated with complete demyelination, while mildly increased SI corresponde
d with partial demyelination. All areas scored as abnormal by the neuropath
ologist were also found on the MRIs, and sizes measured using both methods
correlated well (r = 0.85, P < 0.01). On conventional MRIs, abnormalities c
ould be recognized fairly well. However, better differentiation could be ma
de between high-SI and mildly increased SI abnormalities on the 4.7 T image
s. In conclusion, high-resolution MRI revealed a great range of abnormaliti
es in spinal cord multiple sclerosis, which related to disease course durin
g life. Furthermore, we found very good correlation between the extent of a
bnormalities shown by histopathology and the SI changes on proton-density M
RIs, mainly relating to demyelination revealed histopathologically.