Intramyelinic edema (IME) was initially reported in toxicology studies in rats and dogs given high dosages of vigabatrin (38, 39). Consequently, clinical trials in the United States were temporarily suspended in 1983 (25). Microvacuoles, typically 10–100 μm in diameter, were found on histopathologic analysis in specific brain regions, most commonly in the cerebellum, reticular formation, and optic tract of rats and in the hypothalamus, thalamus, optic tract, and fornix columns of dogs (38–40). Of note, IME was not identified in vigabatrin-treated monkeys, even after dosages of 50 or 100 mg/kg/day for 6 years (39). Moreover, lesions were not identified in the spinal cord or peripheral nervous system of any species (39, 40).
Electron microscopic examination revealed that the microvacuolation resulted from fluid accumulation and separation of the outer lamellar sheaths of myelin along the intraperiod line (38), and immunohistochemical analyses indicated that the IME was generally associated with reactive astrocytosis and microglial activation (40). IME developed over a period of several weeks, even at high dosages, and plateaued over a 6- to 12-month period with continued exposure (39, 40). It then disappeared within weeks after vigabatrin was stopped, without any residual effects in dogs, although rats still had swollen axons and microscopic mineralized bodies in the cerebellum (39, 41). IME did not appear to progress to demyelination in any species (39). However, delayed myelination was observed in immature rats administered vigabatrin 50 mg/kg/day for up to 9 weeks (42).
Subsequently, IME observed in rats and dogs was correlated with delays in evoked potential (EP) (43, 44) and with increased T2-weighted signals on magnetic resonance imaging (MRI) (45, 46). Mild damage to central myelinated pathways may lead to delays in EP latency, whereas more significant damage would be expected to block impulse conduction, leading to reductions in EP amplitude (40). In dogs given vigabatrin 300 mg/kg/day, increases in the latencies of the cortical visual EP and somatosensory EP were evident at 6 weeks and reached statistical significance at 8 and 10 weeks, respectively (44). Both parameters returned to baseline by 5 weeks after vigabatrin was ceased. No changes in brainstem auditory EP or in peripheral or spinal conduction were observed.
The correlation between IME and MRI is illustrated by a study conducted in dogs given vigabatrin 300 mg/kg/day (47). T2 intensity was increased in the hypothalamus after 4 weeks and in the thalamus and fornix after 7 weeks. Both T2 intensity and microvacuolation continued to increase during the 12-week treatment course, and then both decreased after vigabatrin was discontinued. By 16 weeks after vigabatrin was halted, histopathology had returned to normal, and there was a marked trend for reversal of the T2 intensity changes. A study conducted several years later employed both T2-weighted and diffusion-weighted MRIs to evaluate changes in rat brain during exposure to vigabatrin 275 mg/kg/day for 12 weeks (48). Significant increases in T2 intensity were observed in the frontal and occipital cortices and in the cerebellar white matter, with the latter lesions more clearly distinguishable on diffusion-weighted images than by T2 contrast alone. After vigabatrin was stopped, the hyperintensity in the cerebellar white matter greatly decreased over a 12-week period.
Magnetic resonance imaging scans conducted during the first randomized controlled US trial of adjunctive vigabatrin for rCPS did not show any clinically important changes nor any evidence suggestive of IME (12). Prolongations ≥15% in either visual or somatosensory EP latency were reported for six patients (6.5%) receiving vigabatrin compared with nine patients (10.0%) receiving placebo, but none of the latency prolongations were associated with symptoms suggestive of IME. Neuropathologic examinations did not detect IME in 62 patients with refractory epilepsy who received vigabatrin for a mean of 28 months either before undergoing neurosurgery for epilepsy or before death (49). Most patients in this series received vigabatrin at dosages of 2–4 g/day. Similarly, case reports or small case series did not identify changes suggestive of microvacuolization, myelin separation, or demyelination in brain tissue obtained at surgery or at autopsy from patients treated with vigabatrin (50–52). A review of more than 400 adults and more than 200 children treated with vigabatrin for CPS, which involved serial MRIs, multimodality EPs, and neurologic examinations, found no evidence suggestive of IME (Lundbeck Inc., data on file).
A subsequent comprehensive review was conducted to identify adolescent and adult patients who had developed clinical abnormalities that may have been related to IME (40). The review included the global clinical trials database, worldwide post-marketing surveillance, and all published literature covering a period of more than 15 years through March 1997. Medical consultants unrelated to the corporate sponsor reviewed items relevant to potential IME and asked patients meeting selection criteria to undergo a 2-day follow-up reassessment that included EP, MRI, and neurologic examination. All information was also reviewed by an additional group of independent medical consultants. The databases comprised an estimated 350,000 patient-years of exposure to vigabatrin (i.e., approximately 175,000 patients exposed for 2 years at an average dosage of 2 g/day). The consultants found no trends in EP latency over time suggestive of IME, no evidence of IME on T2-weighted axial and coronal MRI, and no trends in clinical neurologic findings nor any associations with abnormal EP or MRI changes suggestive of IME. In summary, no cases of IME were identified in this comprehensive analysis. On the basis of these results, it was concluded that if IME occurs in humans, it is an extremely rare and reversible event, which does not result in clinically significant sequelae.
Since the comprehensive review in adolescents and adults was conducted, several reports have been published suggesting that use of vigabatrin to treat IS may produce transient MRI abnormalities consistent with IME (Fig. 1) (53–55). After identifying an index case of a 13-month-old infant who developed new T2-weighted hyperintensity while receiving vigabatrin for IS, Pearl et al. (53) retrospectively evaluated databases at two children’s hospitals to identify patients who underwent MRI testing while receiving vigabatrin. Including the index case, increased T2 intensity was identified in eight of 23 patients (34.8%) treated with vigabatrin for IS. The T2-weighted images from the index case before vigabatrin, during treatment, and following discontinuation are shown in Fig. 2. Locations of T2 hyperintensity included the thalamus (n = 7), midbrain (n = 7), globus pallidus (n = 6), cerebellar dentate nucleus (n = 6), dorsal medulla (n = 4), medial longitudinal fasciculi (n = 4), and corpus callosum (n = 1). The median age at time of MRI was 11 months (range, 9–18 months), median dosage of vigabatrin was 170 mg/kg/day (range, 83–220 mg/kg/day), and median duration of vigabatrin therapy was 3 months (range, 1–11 months). The T2 abnormalities were accompanied by increased signal on diffusion-weighted images, with apparent diffusion coefficient maps consistent with restricted diffusion. The MRI abnormalities completely resolved following withdrawal of vigabatrin in all patients except for minimal residual pallidal brightness in one patient. No MRI abnormalities were found in 56 other patients with IS who were not treated with vigabatrin.
Figure 1. Incidence of MRI abnormalities with vigabatrin for infantile spasms. aMedian value. MRI, magnetic resonance imaging.
Download figure to PowerPoint
Figure 2. Index case of MRI abnormalities with vigabatrin in a 13-month-old boy. Panel A: normal T2-weighted image prior to vigabatrin. Panel B: Increased signal intensity in thalami (arrows) after patient was on vigabatrin (83 mg/kg/day) for 4 months. Panel C: Apparent diffusion coefficient map shows restricted diffusion while on vigabatrin. Panel D: Resolution at 4 months following discontinuation of vigabatrin. MRI, magnetic resonance imaging. Cerebral MRI abnormalities associated with vigabatrin therapy, Pearl P, et al. Copyright © 2009 Epilepsia. Reproduced with permission of John Wiley & Sons, Inc.
Download figure to PowerPoint
Wheless and coauthors (54) retrospectively reviewed the medical records and cranial MRIs of 205 infants who received treatment for IS. MRI abnormalities — predefined as any hyperintensity on T2-weighted or fluid-attenuated inversion recovery sequences not readily explained by a well-characterized pathology — were identified significantly more frequently for infants treated with vigabatrin than in those who had not received vigabatrin (22% vs 4%; P < 0.001). MRI abnormalities were observed for four of 32 patients (12.5%) treated with low-dosage vigabatrin (<125 mg/kg/day) and for 13 of 44 patients (29.5%) treated with high-dosage vigabatrin (≥125 mg/kg/day) (Fig. 1). The pattern of MRI abnormalities was consistent with the observations made by Pearl et al. (53) for 13 patients. The remaining four patients had either focal or multifocal abnormalities that differed from this pattern and presumably reflected processes other than drug toxicity. Nine patients had at least one additional MRI evaluation, and MRI abnormalities resolved for six of these nine patients (66.7%), including two who continued to receive vigabatrin. In this study, MRI images from 668 children and adults with CPS were re-reviewed, but no difference in MRI abnormalities was detected between vigabatrin-exposed and vigabatrin-naïve patients.
Milh et al. (56) retrospectively evaluated MRI images from 22 infants with epilepsy (including 13 with IS) who were treated with vigabatrin. MRI hyperintensity on T2- and diffusion-weighted images increased transiently beginning 1 month after the start of treatment, peaked at 3–6 months, and then returned to low values after 12 months of treatment. The number of obvious hyperintense areas paralleled this time course and mostly involved the globi pallidi, posterior part of the pons, and mesencephalum. The duration or type of epilepsy, and the presence of seizures were not associated with the MRI hyperintensity.
Dracopoulos et al. (55) evaluated 107 patients with IS, who were grouped based on when they had MRI scans and whether they received therapy with vigabatrin ≥120 mg/kg/day. Group I patients (n = 81) had MRI scans during but not before vigabatrin therapy. Of these, 25 (30.9%) had MRI signal abnormalities, mostly in the globus pallidus (n = 24), brainstem (n = 14), dentate nucleus (n = 8), or thalamus (n = 6). Fifteen Group I patients with MRI abnormalities had follow-up MRIs. Resolution of the abnormalities was observed for 13 patients, including 11 patients who discontinued vigabatrin and two who continued to receive the drug. The two patients who still had abnormalities had discontinued vigabatrin for 3–4 weeks before the second MRI was conducted. Group II patients (n = 14) had MRI scans before and during vigabatrin therapy. All had normal MRIs before vigabatrin was started, and four (28.6%) patients demonstrated new MRI signal intensity changes after starting vigabatrin. Reversal of MRI abnormalities was observed for two patients who discontinued vigabatrin and one patient who continued to receive the drug. The other patients still had abnormalities while being weaned off of vigabatrin. Group III patients (n = 12) did not receive vigabatrin and had no detectable MRI abnormalities (Fig. 1).
Recently, diffusion tensor imaging was conducted for six patients with IS who were treated with vigabatrin and had abnormal T2- and/or diffusion-weighted images (57). The results suggested that axonal abnormalities play a greater role in the abnormal MRI findings than do changes in myelination. Together, these studies suggest that asymptomatic MRI abnormalities consistent with IME occur in 22–35% of infants treated with vigabatrin for IS. The MRI changes are more common at greater dosages of vigabatrin and appear to be transient in most cases. Reversal of MRI abnormalities was observed for most patients following discontinuation of vigabatrin, and, in some cases, even with continuation of the drug. These changes are likely to be age-dependent, as they are not evident for older children or adults. One case report has been published describing an infant with preexisting white matter abnormalities who received vigabatrin for IS at 9 months of age in combination with topiramate and then immediately had severe deterioration in neurologic function and died 3 weeks later (58). At autopsy, neuropathologic examination revealed white matter vacuolation and myelin splitting at the intraperiod line, consistent with the abnormalities observed in rats and mice. Imaging was not performed during the period of toxicity. This is the first reported case of histologically confirmed, vigabatrin-induced IME in humans.
In a recently published review, Iyer et al. (59) noted that vigabatrin-induced neurotoxicity in infants preferentially affects deep CNS structures (e.g., thalami, anterior commissure, globus pallidi, dentate nuclei, brainstem, corpus callosum), which is characterized by restricted diffusion and T2/fluid-attenuated inversion recovery high signal in these structures. Because bilateral thalamic involvement was often a prominent feature, they suggested that it raises suspicion for venous thrombosis, acute disseminated encephalomyelitis (ADEM), and tumor. However, they also recognized that patients with vigabatrin-induced changes are often asymptomatic with their seizures under control, mitigating against ischemia or ADEM, and that signal abnormalities were usually reversible on follow-up MRI (53–55, 60). They concluded that the clinical significance of these changes is unknown as they are often discovered incidentally and may resolve spontaneously even if patients remain on vigabatrin (54, 60).