Chronic administration of vigabatrin (gamma-vinyl GABA) in dogs produc
es reversible microvacuolation (intramyelinic edema) in discrete brain
regions. Histologic changes are most notable in the columns of the fo
rnix and regions of the hypothalamus, thalamus, optic tract, and hippo
campus. In an attempt to image these changes in vivo, we performed hig
h-field MRI on seven treated and four control dogs at baseline and aft
er 15 weeks of dosing with vigabatrin (300 mg/kg/d). All dogs underwen
t parallel electrophysiologic assessment to determine the effects of v
igabatrin on afferent conduction. At 15 weeks, all treated dogs showed
increased T-2. and decreased T-1-weighted signals, with changes from
baseline most prominent in the columns of the fornix and to a lesser d
egree in the surrounding hypothalamus and thalamus. MRIs performed on
control dogs were unremarkable. We then perfused a random selection of
four treated and two control dogs and imaged their brains ex vivo pri
or to sectioning. Ex vivo imaging confirmed the in vivo findings and s
trongly correlated with both electrophysiologic and subsequent histopa
thologic findings. Imaging was repeated in the surviving dogs 5 and 12
weeks after discontinuation of dosing. Signal abnormalities in the tr
eated dogs progressively diminished during recovery, paralleling the e
lectrophysiologic and histopathologic results. These findings demonstr
ate that MRI can detect signal changes anatomically congruent with vig
abatrin-induced intramyelinic edema and suggest that MRI may provide a
useful noninvasive tool for monitoring patients during clinical trial
s.