SEARCH

SEARCH BY CITATION

Keywords:

  • Vigabatrin;
  • Magnetic resonance imaging abnormalities;
  • Infantile spasms;
  • Infants;
  • Seizures;
  • Epilepsy

Summary

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

Purpose: Vigabatrin used to treat infantile spasms (IS) has been associated with transient magnetic resonance imaging (MRI) abnormalities. We carried out a retrospective review to better characterize the frequency of those abnormalities in IS and in children and adults treated with vigabatrin for refractory complex partial seizures (CPS).

Methods: Medical records and 332 cranial MRIs from 205 infants (aged ≤24 months) with IS treated at 10 sites in the United States and Canada were collected. Similarly, 2,074 images from 668 children (aged 2–16 years) and adults (aged >16 years) with CPS were re-reviewed. Prespecified MRI abnormalities were defined as any hyperintensity on T2-weighted or fluid-attenuated inversion-recovery (FLAIR) sequences with or without diffusion restriction not readily explained by a radiographically well-characterized pathology. MRIs were read by two neuroradiologists blinded to treatment group. The incidence and prevalence of MRI abnormalities associated with vigabatrin were estimated.

Results: Among infants with IS, the prevalence of prespecified MRI abnormalities was significantly higher among vigabatrin-treated versus vigabatrin-naive subjects (22% vs. 4%; p < 0.001). Of nine subjects in the prevalence population with at least one subsequent determinate MRI, resolution of MRI abnormalities occurred in six (66.7%)—vigabatrin was discontinued in four. Among adults and children treated with vigabatrin for CPS, there was no statistically significant difference in the incidence or prevalence of prespecified MRI abnormalities between vigabatrin-exposed and vigabatrin-naive subjects.

Discussion: Vigabatrin is associated with transient, asymptomatic MRI abnormalities in infants treated for IS. The majority of these MRI abnormalities resolved, even in subjects who remained on vigabatrin therapy.

Vigabatrin (VGB, Sabril [Ovation Pharmaceuticals, Deerfield, IL, U.S.A.], γ-vinyl γ-aminobutyric acid (GABA)) is a selective and irreversible GABA transaminase inhibitor with antiepileptic activity (Lippert et al., 1977). Vigabatrin has been shown to be effective as add-on therapy for the treatment of refractory complex partial seizures (CPS) in children and adults (Sander et al., 1990; French et al., 1996; Bruni et al., 2000; Guberman & Bruni, 2000) and as primary therapy for infantile spasms (IS) (Chiron et al., 1990; Aicardi et al., 1996; Chiron et al., 1997; Wohlrab et al., 1998; Appleton et al., 1999; Elterman et al., 2001). Vigabatrin is approved in Europe, Canada, and many other countries outside the United States for these indications.

In animal toxicology studies, vigabatrin was associated with histopathologic abnormalities in rodents and dogs but not monkeys. Microvacuolization of glial cells was observed in specific regions of the brain, predominantly in the midline and deep hemispheric structures, such as the brainstem, cerebellum, basal ganglia, and white matter tracts, including the anterior commissure of rats and the fornix of dogs (Gibson et al., 1990; Yarrington et al., 1993). The observed microvacuoles were within myelin laminae; therefore, this pathology was termed intramyelinic edema (IME). Further study showed that with chronic vigabatrin administration, IME developed as early as 4 weeks in rodents and dogs, was completely reversible upon discontinuation of vigabatrin, and did not result in any permanent damage to the brain parenchyma (Peyster et al., 1995). Of importance to note, IME in animals correlated well with prolongation of evoked potential (EP) latencies and with high T2-weighted signal on magnetic resonance imaging (MRI) (Jackson et al., 1994; Weiss et al., 1994; Peyster et al., 1995; Preece et al., 2004). In response to these findings in animals, the effects of vigabatrin in humans have been investigated systematically. Several published reports examining human brain tissue obtained at autopsy or during surgery from vigabatrin-treated patients did not reveal changes suggestive of microvacuolization, myelin separation, or demyelination (Pedersen et al., 1987; Ben-Menachem et al., 1988,Paljarvi et al., 1990; Cannon et al., 1991a, 1991b; Hammond et al., 1992; Bryant et al., 1994). Moreover, a contemporaneous review of serial MRIs, neurologic examinations, and multimodality EPs from more than 400 adults and 200 children treated with vigabatrin for CPS in prospective clinical trials conducted between 1990 and 1998 provided no evidence of changes to suggest IME in these populations (Ovation Pharmaceuticals, data on file). Subsequently, a comprehensive review of animal and human data investigating the potential of vigabatrin to cause MRI abnormalities similar to those associated with IME in animals failed to identify any definite cases of such abnormalities in humans treated with vigabatrin (Cohen et al., 2000).

Recently, however, data have come to light indicating that the use of vigabatrin to treat IS may be associated with transient MRI abnormalities characterized by hyperintense T2-weighted signals within deep brain structures (i.e., basal ganglia, thalami, anterior commissure, corpus callosum, or midbrain). Pearl et al. (2006) originally reported three cases among 15 infants treated for IS, and 10 cases were subsequently reported in postmarketing surveillance. This led to further investigation of the possible link between the use of vigabatrin for IS and the occurrence of MRI abnormalities. Results from a retrospective review of 213 children with IS showed that 23 (11%) exhibited MRI abnormalities similar to those reported by Pearl and colleagues (Ovation Pharmaceuticals, data on file). Consistent with these findings, Desguerre et al. (2008) recently reported transient diffusion-weighted imaging (DWI) abnormalities in six infants (30%) treated with vigabatrin for IS. Four of these subjects experienced resolution of abnormalities while still being treated with vigabatrin. These studies support the hypothesis that vigabatrin may cause transient, asymptomatic MRI abnormalities in a subset of infants treated for IS. However, the shortcoming of these prior studies was the lack of a control group not treated with vigabatrin. This is important given that MRI signal abnormalities, specifically T2-weighted hyperintensities, have been observed in children with epilepsy and also are associated with a variety of neurometabolic disorders (Bekiesinska-Figatowska et al., 2001; Ziyeh et al., 2002; Desai et al., 2003; Oster et al., 2003; Pearl et al., 2003; Urbach, 2005; Horvath et al., 2006).

The goals of the present analyses and retrospective study were to test the hypothesis that vigabatrin is associated with transient MRI abnormalities in infants with IS and to estimate, using rigorous and conservative criteria, this prevalence against the background of MRI abnormalities known to occur in patients with epilepsy. Presented herein and described independently are: (1) Ovation study OV-1019, conducted in infants with IS; and (2) a repeat review of MRI scans obtained from 12 previously described prospective studies conducted between 1990 and 1998 in more than 600 adults and children with CPS.

Methods

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

Study OV-1019 in infants with IS

Subjects

Subjects were retrospectively identified at each site. Eligible subjects included those with IS (i.e., West syndrome) as a primary or secondary diagnosis [per International League Against Epilepsy (ILAE) criteria] who were aged ≤24 months at the time of diagnosis and who had cranial MRIs available for evaluation, including at least one cranial MRI with any T2-weighted, fluid-attenuated inversion-recovery (FLAIR), or DWI sequence that was performed at or before age 35 months and after the date of onset of IS. Eligible subjects must have been treated either with vigabatrin or another antiepileptic therapy [e.g., ketogenic diet, corticosteroids, or other antiepileptic drugs (AEDs)].

Subject records were solicited from 10 major referral centers in the United States and Canada. All participating sites received approval for this protocol from their institutional review board and completed case report forms for each subject whose records were provided for analysis.

Study design

The primary objective of study OV-1019 was to compare the incidence of prespecified abnormalities on cranial MRI between vigabatrin-exposed and vigabatrin-naive cohorts of pediatric subjects treated for IS. Secondary objectives included the following: comparison of the prevalence of prespecified abnormalities on MRI between vigabatrin-exposed and vigabatrin-naive subjects; assessment of the risk factors for prespecified abnormalities on cranial MRI; and understanding the time course and resolution of prespecified MRI signal abnormalities.

The data analysis accounted for subjects who were treated with other therapeutic modalities before starting vigabatrin; hence, a single subject could contribute data to both the vigabatrin-naive and vigabatrin-exposed categories. Treatment periods were, therefore, analyzed as follows:

Vigabatrin-initial—Subjects started on vigabatrin at the time of diagnosis of IS were analyzed in this treatment category for the duration of their follow-up in the study. They contributed data to the vigabatrin-exposed category.

Vigabatrin-subsequent—These subjects were treated with vigabatrin only after a period of treatment with other modalities. Hence, depending on the timing of their MRI examinations, they may have contributed data to both vigabatrin-naive and vigabatrin-exposed categories during different treatment periods.

Vigabatrin-never—This group comprised subjects never treated with vigabatrin, therefore, they contributed data to only the vigabatrin-naive category.

Vigabatrin-naive—This treatment category included subjects who never received vigabatrin plus the vigabatrin-subsequent therapy group before starting vigabatrin.

Vigabatrin-exposed—This treatment category comprised all subjects after their first exposure to vigabatrin, including the vigabatrin-initial group and vigabatrin-subsequent group after starting vigabatrin. Subjects still were considered vigabatrin-exposed after discontinuing vigabatrin.

Data also were collected on the extent of exposure to vigabatrin, including when treatment was initiated relative to diagnosis of IS, whether it was primary or secondary treatment, the duration of treatment, daily dose (low dose, <125 mg/kg/day; or high dose, ≥125 mg/kg/day), and cumulative dose of vigabatrin.

Data on vigabatrin-associated visual-field defects were not purposely collected or considered in this study.

Image processing and interpretation

Two assessments were performed: one clinical assessment directly from the site and one on centralized MRI readings. Image processing was conducted by Perceptive Informatics, Inc. (Waltham, MA, U.S.A.), a nationally recognized, independent, neuroimaging center. All MRIs were submitted to Perceptive Informatics either as digital DICOM files or as hard copy films, where they were scanned and converted to a proprietary digital format. Quality assurance checks were carried out, and identifying data were blinded. The images were then reviewed by two experienced pediatric neuroradiologists. Image interpretations were adjudicated by a third, senior pediatric neuroradiologist if the primary reviewers disagreed as to overall assessment. Reviewers were blinded to subject-identifying information and treatment.

Reviewers were instructed to report prespecified MRI abnormalities, defined as any hyperintensities on T2-weighted or FLAIR sequences with or without diffusion restriction not readily explained by a radiographically well-characterized pathology. Reviewers made an overall assessment of “within normal limits,”“indeterminate,”“abnormal with a radiographically well-characterized pathologic process (e.g., tuberous sclerosis, cerebral dysgenesis, cavernous angioma, or other),” or “abnormal with signal abnormalities present with uncertain pathological significance.”“Indeterminate” indicated the reviewer was unable to determine with reasonable certainty whether an abnormality was present. The category of “abnormal with signal abnormalities of uncertain pathological significance” included the prespecified abnormalities being sought. All abnormal, determinate imaging studies were then read in detail and abnormalities specified by region and imaging sequence. Magnetic resonance images lacking at least one adequate T2-weighted or FLAIR sequence could not be assessed for the presence of prespecified signal abnormalities and were excluded from the independent review process as “uninterpretable.”

Two estimates of incidence were calculated: one based on the incidence population, which required a baseline MRI free of prespecified abnormalities and at least one postbaseline MRI; and another based on the generalized incidence population, which required two MRIs at any time during the course of treatment, the first of which was free of prespecified abnormalities. The purpose of analyzing the generalized incidence population was to capture all cases for which the MRI changed from normal to abnormal. Prevalence was defined as the occurrence of at least one prespecified MRI signal abnormality on T2-weighted, FLAIR, and/or DWI during a treatment period. A baseline scan was not required. The prevalence population included all subjects, regardless of whether they had a baseline scan, who had determinate MRI examinations during or after treatment with vigabatrin.

Statistical analysis

Descriptive statistics for the vigabatrin-exposed and the vigabatrin-naive groups were summarized. The prevalence of prespecified MRI abnormalities was compared between subgroups using the Cochran-Mantel-Haenszel test, and the incidence was compared between subgroups using Fisher’s exact test. Attributable and relative risks were calculated with confidence intervals (CIs). Risk factors for prespecified cranial MRI signal abnormalities were assessed.

Children and adults with CPS

Subjects

Serial MRIs, neurologic examinations, and multimodality EPs from more than 400 adults and 200 children treated with vigabatrin for CPS in prospective clinical trials conducted between 1990 and 1998 were reanalyzed. All subjects had a diagnosis of CPS, and both children (aged 2–16 years) and adults (aged >16 years) were included. Subjects must have had at least one cranial MRI performed while on vigabatrin to assess the prevalence of prespecified signal abnormalities, and at least one baseline and one subsequent cranial MRI performed while on vigabatrin to assess the incidence of prespecified signal abnormalities. These subjects were enrolled in five pediatric and seven adult multicenter clinical trials conducted in the United States, Canada, and Europe, and many participated in more than one trial. In the majority of these trials, vigabatrin was being evaluated as adjunctive therapy in subjects with difficult-to-control CPS. One adult study investigated the efficacy and safety of vigabatrin monotherapy. Six were single-arm, open-label studies, and six were randomized, placebo-controlled studies.

Study design

The primary endpoint of the reanalysis was the incidence and prevalence of prespecified MRI abnormalities in subjects following treatment with vigabatrin (at the time or before the MRI was performed) compared with subjects not previously exposed to vigabatrin at the time the MRI was performed, which included subjects who received a placebo add-on to their existing AED therapy and who may later have been treated with vigabatrin in another study. All available MRIs were obtained from the prospective CPS trials conducted by the prior sponsor between 1990 and 1998. The MRI images reviewed herein already had been evaluated by the prior sponsor; however, in this repeat review, prespecified MRI abnormalities were conservatively defined as any hyperintensity on T2-weighted or FLAIR sequences with or without diffuse restriction not readily explained by a radiographically well-characterized pathology. Similar to study OV-1019, subjects were grouped into several nonexclusive analysis populations to reflect MRI findings relative to the timing of subjects’ MRIs in relationship to use of vigabatrin as follows: vigabatrin-exposed at any time prior to the MRI, placebo-only, vigabatrin-only, and placebo-vigabatrin. Subjects initially receiving placebo could have been included in the vigabatrin-naive population for analysis of MRIs performed before subsequent treatment with vigabatrin. Subjects also were grouped for analysis based on age (2–11 years, 12–16 years, and adults). Upon entry into the prior clinical trials, all eligible subjects were to have received a neurologic examination.

Data on vigabatrin-associated visual field defects were not purposely collected or considered in this analysis.

Image processing and interpretation

Similar to study OV-1019, image processing and interpretation was carried out by Perceptive Informatics, Inc. All images were reviewed in a blinded fashion by two primary neuroradiologists, with adjudication by a third in the event the assessment of the primary reviewers was discordant. All reviewers were blinded to subject identity and treatment. The definition of prespecified MRI abnormalities and assessment categories were identical to those described previously for study OV-1019. In total, 11 neuroradiologists participated in this study, and a formal assessment of variability was conducted by having all 11 reviewers rate an identical set of 17 MRI examinations to determine percent agreement.

Statistical analysis

The prevalence of prespecified MRI abnormalities was compared between subgroups using the Cochran–Mantel–Haenszel test, and the incidence was compared between subgroups using Fisher’s exact test.

Results

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

Study OV-1019 in infants with IS

Subjects

Images from 205 infants treated at 10 sites in the United States and Canada were reviewed. Subjects’ baseline characteristics are presented in Table 1. A total of 93 subjects received vigabatrin either as initial therapy (n = 38) or as subsequent therapy (n = 55), and 112 subjects never received vigabatrin (vigabatrin-never group). Median age at first treatment for IS was similar in the vigabatrin-initial group (6.2 months), vigabatrin-subsequent group (7.5 months), and vigabatrin-never group (6.4 months). The most common etiologies of IS were cryptogenic (46% of all subjects), cerebral dysgenesis (25%), and tuberous sclerosis (14%). Generally, these etiologies were balanced between groups; however, tuberous sclerosis was significantly more frequent among vigabatrin-treated subjects (p = 0.003), and hemorrhage was significantly more frequent in the never-vigabatrin group (p = 0.027).

Table 1.   Demographics and etiology of infantile spasms (eligible subjects)
Baseline characteristicsVigabatrinNever-vigabatrin (n = 112)All subjects (N = 205)p-valuea
Initial (n = 38)Subsequent (n = 55)All (n = 93)
  1. CNS, central nervous system; IS, infantile spasms.

  2. ap-values comparing three mutually exclusive cohorts (vigabatrin-initial, vigabatrin-subsequent, and never-vigabatrin) are based on chi-square tests for categorical variables and F-tests for continuous variables. Chi-square tests include “missing/unknown,”“not available,” and “unknown” categories.

Gender, n (%)
 Male23 (60.5)31 (56.4)54 (58.1)59 (52.7)113 (55.1)0.677
 Female15 (39.5)24 (43.6)39 (41.9)53 (47.3)92 (44.9)
Ethnicity, n (%)
 Hispanic or Latino2 (5.3)8 (14.5)10 (10.8)5 (4.5)15 (7.3)0.093
 Not hispanic or Latino32 (84.2)36 (65.5)68 (73.1)85 (75.9)153 (74.6)
 Unknown4 (10.5)11 (20.0)15 (16.1)22 (19.6)37 (18.0)
Race, n (%)
 Caucasian31 (81.6)45 (81.8)76 (81.7)51 (45.5)127 (62.0)<0.001
 Non-Caucasian5 (13.2)6 (10.9)11 (11.8)45 (40.2)56 (27.3)
 Unknown2 (5.3)4 (7.3)6 (6.5)16 (14.3)22 (10.7)
Median age at diagnosis of IS, months
 6.26.06.07.16.40.261
IS etiology
 Trauma0 (0.0)0 (0.0)0 (0.0)5 (4.5)5 (2.4)0.119
 Hemorrhage0 (0.0)1 (1.8)1 (1.1)11 (9.8)12 (5.9)0.027
 Hypoxic-ischemic encephalopathy 3 (7.9)2 (3.6)5 (5.4)13 (11.6)18 (8.8)0.226
 Tuberous sclerosis8 (21.1)13 (23.6)21 (22.6)7 (6.3)28 (13.7)0.003
 Cryptogenic17 (44.7)23 (41.8)40 (43.0)54 (48.2)94 (45.9)0.729
 Toxic0 (0.0)1 (1.8)1 (1.1)0 (0.0)1 (0.5)0.254
 Chromosomal abnormality2 (5.3)4 (7.3)6 (6.5)12 (10.7)18 (8.8)0.531
 Cerebral dysgenesis6 (15.8)15 (27.3)21 (22.6)31 (27.7)52 (25.4)0.323
 CNS infection0 (0.0)2 (3.6)2 (2.2)1 (0.9)3 (1.5)0.270
 Inborn error of metabolism0 (0.0)1 (1.8)1 (1.1)1 (0.9)2 (1.0)0.675

Seizure history is summarized in Table 2. The majority of subjects had a history of prolonged seizures, and most had either partial seizures with or without secondary generalization (type IA, IB, or IC) or generalized seizures at onset (type II). Partial seizures without secondary generalization were significantly more common in vigabatrin-treated subjects (p = 0.002). Fifteen subjects screened for the study who had toxic or metabolic disorders associated with MRI abnormalities were excluded.

Table 2.   Seizure history (eligible subjects)
 VigabatrinNever-vigabatrinAll subjectsp-valuea
Initial (n = 38), n (%)Subsequent (n = 55), n (%)All (n = 93), n (%)(n = 112), n (%)(N = 205), n (%)
  1. ER, emergency room; NC, not calculated.

  2. NC occurs because of zero cells. Seizure types are not mutually exclusive. Indicators for prolonged seizures also are not mutually exclusive. Subjects may appear in more than one row.

  3. ap-values comparing three mutually exclusive cohorts (vigabatrin-initial, vigabatrin-subsequent, and never-vigabatrin) are based on chi-square tests.

  4. bIncludes subjects who had other seizures indicated but type of seizure was not available.

Other seizure types
 Partial [I]1 (2.6)6 (10.9)7 (7.5)5 (4.5)12 (5.9)0.161
 Partial without secondary generalization [IA, IB]14 (36.8)17 (30.9)31 (33.3)15 (13.4)46 (22.4)0.002
 Partial with secondary generalization [IC]4 (10.5)9 (16.4)13 (14.0)8 (7.1)21 (10.2)0.181
 Generalized at onset [II]4 (10.5)16 (29.1)20 (21.5)33 (29.5)53 (25.9)0.057
 Unclassified [III]1 (2.6)4 (7.3)5 (5.4)9 (8.0)14 (6.8)0.515
 Unknownb0 (0.0)0 (0.0)0 (0.0)1 (0.9)1 (0.5)NC
History of prolonged seizures as evidenced by
 ER or hospital admission8 (21.1)18 (32.7)26 (28.0)44 (39.3)70 (34.1)0.119
 Need for rescue medication6 (15.8)22 (40.0)28 (30.1)35 (31.3)63 (30.7)0.045

Twenty of 38 subjects (53%) in the vigabatrin-initial group and nine of 55 subjects (16%) in the vigabatrin-subsequent group received vigabatrin monotherapy. The median maximum vigabatrin dose was 139 mg/kg/day in the vigabatrin-initial group and 140 mg/kg/day in the vigabatrin-subsequent group. The median average daily dose was 99 mg/kg/day in the vigabatrin-initial group and 101 mg/kg/day in the vigabatrin-subsequent group. In the prevalence population, 32 subjects were treated with low daily doses of vigabatrin (<125 mg/kg/day), and 44 subjects were treated with high daily doses (≥125 mg/kg/day). Overall, 68% of subjects received other therapies, including corticosteroids (60% of all subjects), benzodiazepine derivatives (35%), barbiturates (35%), and valproic acid (24%). All other AEDs were used in <20% of subjects. The majority of subjects (60%) were receiving at least three AEDs, 23% were receiving two AEDs, and 17% were receiving one AED. Twelve of the 205 total subjects (6%) were treated with a ketogenic diet. The median duration of treatment for IS at the last determinate MRI was 3.3 months in both the vigabatrin-subsequent and vigabatrin-initial groups.

Incidence and prevalence

The background rate of prespecified MRI-signal abnormalities was calculated from the 70 eligible subjects with determinate MRIs prior to treatment of IS; 2 of these subjects (2.9%) had prespecified signal abnormalities on their baseline MRIs. Neither of these subjects received vigabatrin.

The incidence and prevalence of prespecified MRI abnormalities in each subgroup are shown in Table 3. Among subjects with a determinate baseline MRI free of prespecified abnormalities (i.e., incidence population), at least one abnormality was seen postbaseline in 4 of 12 subjects (33.3%) in the low-dose vigabatrin group, 5 of 12 subjects (41.7%) in the high-dose vigabatrin group, and one of 17 subjects (5.9%) in the vigabatrin-naive group. The observed difference between all vigabatrin-treated subjects and vigabatrin-naive subjects with respect to incidence of MRI signal abnormalities was statistically significant (36% vs. 6%; p = 0.031), and the attributable risk was 30.1% (95% CI, 8.2–52.0%).

Table 3.   Incidence and prevalence of MRI-signal abnormalities in infants (eligible subjects in the prevalence and incidence populations)
Incidence (Incidence population)Subjects, n
Vigabatrin-exposedVigabatrin-naive
Low-doseHigh-doseAll
(n = 12)(n = 12)(n = 25)(n = 17)
  1. MRI, magnetic resonance imaging; CI, confidence interval.

  2. Prevalence is defined as the occurrence of at least one prespecified MRI-signal abnormality seen in a treatment period on T2-weighted, FLAIR, and/or DWI. There is no requirement of having a baseline MRI.

  3. A subject who received vigabatrin subsequent to nonvigabatrin initial therapy for IS was counted as at risk for a prespecified MRI signal abnormality in any of the initial non–vigabatrin or subsequent vigabatrin treatment periods in which a determinate MRI was obtained. The sum of the counts for vigabatrin-exposed and vigabatrin-naive groups may be greater than the total number of subjects. The dose level for some vigabatrin subjects was unknown; therefore, the count of all vigabatrin subjects may be greater than the sum of the counts of low- and high-dose vigabatrin subjects.

 Prespecified MRI abnormality4591
 No prespecified MRI abnormality871616
 Incidence (%) (95% CI)33.3 (9.9–65.1)41.7 (15.2–72.3) 36.0 (18.0–57.5)  5.9 (0.1–28.7)
Prevalence (Prevalence population)(n = 32)(n = 44)(n = 79)(n = 98)
 MRI abnormality present413174
 MRI abnormality absent28316293
 Prevalence (%) (95% CI) 12.5 (3.5–29.0) 29.5 (16.8–45.2) 21.5 (13.1–32.2)  4.1 (1.1–10.1)

Among subjects who had MRI examinations during or after treatment for IS (i.e., prevalence population), at least one abnormality was seen in 4 of 32 subjects (12.5%) in the low-dose vigabatrin group, 13 of 44 subjects (29.5%) in the high-dose vigabatrin group, and 4 of 98 subjects (4.1%) in the vigabatrin-naive group. The observed difference in prevalence of MRI-signal abnormalities between all vigabatrin-treated subjects and nonexposed subjects was statistically significant (22% vs. 4%; p < 0.001), and the attributable risk was 13.6% (95% CI 3.8–23.4%). There also was some evidence of a dose effect among subjects treated with vigabatrin, although this effect did not reach statistical significance (p = 0.099).

The typical pattern of abnormality attributed to vigabatrin exposure by Pearl et al. (2006) and more recently reported by Desguerre et al. (2008) in vigabatrin-treated infants is illustrated by one vigabatrin-exposed subject in study OV-1019 (Fig. 1.) The T2-weighted signal in the globus pallidus, thalamus, dorsal midbrain, and roof nuclei of the cerebellum was symmetrically increased. Of the 17 vigabatrin-exposed subjects with the prespecified T2 abnormality, 13 had findings consistent with this pattern—although varying greatly in degree. The other four subjects had either focal or multifocal abnormalities that did not resemble this pattern and presumably represented processes other than drug toxicity. Of the four cases with abnormal T2-weighted signal but without vigabatrin exposure, one had bilateral, symmetric high T2-weighted signal and diffusion restriction in the globus pallidus, but the temporal evolution and appearance was highly suggestive of hypoxic-ischemic damage. The other three vigabatrin-naive subjects had other patterns of abnormality that could not be confused with that shown in Fig. 1 and did not suggest drug toxicity.

image

Figure 1.  T2-weighted coronal magnetic resonance imaging (MRI) of a vigabatrin-treated subject demonstrating increased signal in the thalamus, globus pallidus, and midbrain. At the time of the MRI, the subject had been taking vigabatrin for 36 days. This MRI represents the most marked T2 abnormalities seen in this study.

Download figure to PowerPoint

Other assessments

Radiographic resolution of MRI-signal abnormalities after discontinuation of vigabatrin treatment was assessed for the prevalence and incidence populations. Of the 93 subjects in the prevalence population exposed to vigabatrin, 16 (17.2%) had a prespecified abnormality while on vigabatrin, of which 9 had at least one subsequent determinate MRI examination. Of these nine subjects, the MRI abnormalities resolved in six (66.7%)—in two (33.3%) during vigabatrin treatment and in four (66.7%) after discontinuing vigabatrin. Of the 25 subjects in the incidence population exposed to vigabatrin, four (16.0%) had a prespecified abnormality while on vigabatrin and at least one subsequent determinate MRI examination. The MRI abnormalities resolved in three of these four subjects (75.0%)—in one (33.3%) during vigabatrin therapy and in two (66.7%) after discontinuing vigabatrin. A sequence of MRI images showing the appearance of T2-weighted and DWI hyperintensities after initiating vigabatrin and their subsequent resolution after discontinuing vigabatrin is shown in Fig. 2.

image

Figure 2.  (A, C, and E) T2-weighted and (B, D, and F) diffusion-weighted MRI of a subject before beginning treatment with vigabatrin (A and B), 4 months after starting vigabatrin (C and D), and 4 months after discontinuing vigabatrin (E and F). Although taking vigabatrin (C and D), the subject displayed moderately increased T2-weighted signal and marked diffusion restriction in the thalami bilaterally (arrows). These abnormalities resolved completely 4 months after discontinuing vigabatrin (E and F).

Download figure to PowerPoint

Median time from first treatment for IS to first MRI-signal abnormality could not be accurately determined because of the irregular frequency of MRI examinations; however, among subjects treated with vigabatrin, median time to detection of the first MRI-signal abnormality was 11 months in the high-dose group and 24 months in the low-dose group. In a detailed analysis of possible risk factors for developing MRI abnormalities, which included all demographic and disease characteristics, no subject characteristic, other than vigabatrin exposure, emerged as a risk factor.

Results from a post hoc analysis comparing prevalence of MRI-signal abnormality between the vigabatrin-initial and vigabatrin-subsequent groups showed that 4 of 37 subjects (10.8%) in the former group and 13 of 42 subjects (31%) in the latter group had at least one abnormality. This difference approached statistical significance (p = 0.053).

Children and adults with CPS

Subjects

A total of 668 subjects were eligible for inclusion in the analysis. Of these, 656 subjects were exposed to vigabatrin at some point (vigabatrin-exposed group), and 12 subjects had received placebo as add-on to their existing AED therapy (placebo-only group). Among the 656 subjects exposed to vigabatrin, 421 were treated initially with vigabatrin (vigabatrin-only group), and 235 initially were randomized to placebo and subsequently received vigabatrin during another clinical study (placebo-vigabatrin group).

Among pediatric subjects (aged 2–11 years), the median age at first dose of vigabatrin was 7 years. The median maximum daily dose in this group was 2,761 mg/day (range: 285–6,857 mg/day), and the median cumulative dose was 604 g until the last determinant MRI. For adolescent subjects (aged 12–16 years), the median age at first dose of vigabatrin was 14 years, ranging from 18 weeks to >10 years. In this group, the median maximum daily dose was 4,000 mg/day (range: 1,200–15,507 mg/day), and the median cumulative dose was 1,057 g. Among adult subjects, the median age at first dose of vigabatrin was 34 years (range: 18–69 years). In this group, the median maximum daily dose was 5,500 mg/day (range: 0–9,000 mg/day), and the median cumulative dose was 1,896 g.

Among all eligible subjects, carbamazepine was the most frequently used concurrent AED (62%), followed by “other” AEDs (48%), benzodiazepines (28%), valproic acid (27%), hydantoins (16%), and barbiturates (10%) (Table 4). Subjects most frequently enrolled while taking two other AEDs (50%); 30% of subjects were taking one other AED, and 20% of subjects were taking three or more AEDs. There were no statistically significant differences between groups regarding either drug category or number of concomitant AEDs.

Table 4.   Concurrent AEDs and other epilepsy therapies used in pediatric and adult subjects with complex partial seizures (all eligible subjects)
 Vigabatrin-only (n = 421)Placebo-only (n = 12)Placebo-vigabatrin (n = 235)Vigabatrin-exposeda (n = 656)Vigabatrin- naiveb (n = 242)All (n = 668) p-valuec
  1. AED, antiepileptic drug.

  2. aFor subjects who received placebo and vigabatrin, includes only AEDs taken during studies when subject received vigabatrin.

  3. bFor subjects who received placebo and vigabatrin, includes only AEDs taken during studies when subject received placebo.

  4. cP values comparing three mutually exclusive cohorts (vigabatrin-only, placebo-only, and placebo-vigabatrin) are based on chi-square test.

  5. dCounts reflect the number of types of AEDs used by a subject. If a subject used multiple AEDs within the same type (i.e., multiple benzodiazepines), this will be reflected as a single AED.

Other AEDs, n (%)
 Benzodiazepines119 (28.3)3 (25.0)68 (28.9)180 (27.4)55 (22.7)190 (28.4)0.499
 Barbiturates41 (9.7)0 (0.0)23 (9.8)64 (9.8)13 (5.4)64 (9.6) 
 Carbamazepine254 (60.3)7 (58.3)154 (65.5)407 (62.0)153 (63.2)415 (62.1) 
 Hydantoins74 (17.6)0 (0.0)36 (15.3)106 (16.2)29 (12.0)110 (16.5) 
 Valproic acid118 (28.0)6 (50.0)59 (25.1)177 (27.0)61 (25.2)183 (27.4) 
 Other AED215 (51.1)4 (33.3)104 (44.3)314 (47.9)92 (38.0)323 (48.4) 
Number of types of  other AEDs, n (%)d
 1119 (28.3)4 (33.3)79 (33.6)198 (30.2)81 (33.5)202 (30.2)0.261
 2213 (50.6)8 (66.7)111 (47.2)324 (49.4)118 (48.8)332 (49.7) 
 ≥389 (21.1)0 (0.0)45 (19.1)134 (20.4)43 (17.8)134 (20.1) 
Cranial MRI examinations

Of 2,074 available scans from the original studies, 2,037 were interpretable and 2,024 were determinant.

Incidence and prevalence

The baseline rate of prespecified MRI abnormalities is summarized for each age group in Table 5. Among all eligible subjects who had a determinate MRI before vigabatrin exposure (n = 609), 67 (11%) had prespecified signal abnormalities at baseline (95% CI, 8.6–13.8%). The baseline rate was highest among adults.

Table 5.   Background rate of MRI-signal abnormalities in pediatric and adult subjects with complex partial seizures
 Determinate MRI, nPrespecified abnormalities at baseline, n (%) 95% CI
  1. CI, confidence interval; MRI, magnetic resonance imaging.

Determinate MRI without vigabatrin exposure
Eligible subjects60967 (11.0)8.6–13.8
 Ages 2 through 11 years1258 (6.4)2.8–12.2
 Ages 12 through 16 years1037 (6.8)2.8–13.5
 Adults38152 (13.6)10.4–17.5

Of the 515 subjects in the incidence population, 51 of 474 vigabatrin-exposed subjects (10.8%; 95% CI, 8.1–13.9%) developed a prespecified abnormality compared with 12 of 150 vigabatrin-naive subjects (8.0%; 95% CI, 4.2–13.6%) (Table 6). Overall, there were no statistically significant differences in prespecified MRI abnormality incidence between vigabatrin-exposed and vigabatrin-naive subjects (p = 0.437) or among any age group. Similarly, analysis of the prevalence population (n = 635) demonstrated that 84 of 592 vigabatrin-exposed subjects (14.2%; 95% CI, 11.5–17.3%) had a prespecified MRI signal abnormality compared with 25 of 191 vigabatrin-naive subjects (13.1%; 95% CI, 8.7–18.7%) (Table 6). This difference also was not statistically significant (p = 0.579).

Table 6.   Incidence and prevalence of MRI-signal abnormalities among pediatric and adult subjects with complex partial seizures
 Vigabatrin-exposeda (n = 474)Vigabatrin-naive (n = 150)
Incidence, %95% CIIncidence, %95% CI
  1. CI, confidence interval; MRI, magnetic resonance imaging.

  2. aSome patients are included in both groups.

Incidence (Incidence population; N = 515)
 Eligible subjects10.88.1–13.98.04.2–13.6
 Ages 2 through 11 years8.03.5–15.25.60.7–18.7
 Ages 12 through 16 years6.321.–14.011.11.4–34.7
 Adults12.99.3–17.38.33.7–15.8
 Vigabatrin-exposeda (n = 592)Vigabatrin-naivea (n = 191)
Incidence, %95% CIIncidence, %95% CI
Prevalence (Prevalence population; N = 635)
 Eligible subjects14.211.5–17.313.18.7–18.7
 Ages 2 through 11 years8.54.3–14.712.24.6–24.8
 Ages 12 through 16 years7.83.4–14.915.44.4–34.9
 Adults18.014.2–22.412.97.4–20.4

The incidence of MRI abnormalities by anatomic region was analyzed for all eligible subjects in this protocol. For composite anatomic regions, the incidence of abnormalities involving the basal ganglia, thalamus, brainstem, or cerebellum was 2.3% in vigabatrin-exposed subjects and 3.6% in vigabatrin-naive subjects (p = 0.403). However, the incidence of MRI abnormalities in other regions, mainly in the cerebral hemispheric white matter, was higher in both vigabatrin-exposed subjects (9.5%) and vigabatrin-naive subjects (5.8%; p = 0.188).

Discussion

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

The data presented herein demonstrate that vigabatrin is associated with a statistically significant increased incidence and prevalence of MRI abnormalities in infants receiving treatment for IS. Using conservative criteria, the estimated incidence was significantly higher among vigabatrin-treated subjects compared with vigabatrin-naive subjects (36% vs. 6%; p = 0.031). Likewise, the estimated prevalence was higher in vigabatrin-treated subjects compared with vigabatrin-naive subjects (22% vs. 4%; p < 0.001). On the basis of these estimates, the relative risk of developing prespecified MRI abnormalities after vigabatrin exposure is 5- to 6-fold higher than that associated with exposure to other IS treatments. (It should be noted, however, that the CIs around these estimates were broad.) These findings are consistent with previous reports and case reviews from single institutions suggesting that vigabatrin-associated MRI abnormalities occur in 10–20% of treated infants (Pearl et al., 2006). The observed anatomic location of vigabatrin-associated MRI abnormalities (i.e., the brainstem, cerebellum, basal ganglia, and thalamus) also was consistent with that in previous reports, suggesting these are common sites of MRI abnormalities in patients treated with vigabatrin. Moreover, these anatomic locations also are common sites of MRI abnormalities reported in inborn errors of metabolism (Desguerre et al., 2003). Considering the high proportion of IS of unknown etiology (i.e., cryptogenic spasms), one cannot exclude the fact that some of these transient MRI abnormalities may have a yet unidentified metabolic cause.

These MRI changes are best appreciated on T2-weighted, FLAIR, and DWI images, and could have been underestimated before DWI was more systematically performed. These changes appear to be related to the vigabatrin daily dose. Daily vigabatrin doses ≥125 mg/kg/day were associated with a nonsignificant trend toward a higher incidence of MRI abnormalities. Moreover, treatment with vigabatrin after other AEDs was associated with a nonsignificant trend toward a higher prevalence of MRI abnormalities. Results of this study and others also indicate that these MRI abnormalities are transient, generally resolve completely (even in subjects that continued to take vigabatrin), and do not appear to be associated with any clinical sequelae (Desguerre et al., 2008). However, data from Desguerre et al. (2008) and unpublished reports (Ovation Pharmaceuticals, data on file) raise the possibility of a transient movement disorder occurring in association with the MRI abnormalities.

Pearl et al. (2006) reported the first three cases of MRI abnormalities in infants aged 13 months or younger who were treated with high doses of vigabatrin (130–160 mg/kg/day for 3–4 months) at Children’s National Medical Center. These MRI findings consisted of T2-weighted hyperintensities in the basal ganglia, thalami, corpus callosum, anterior commissure, brainstem, globus pallidi, and/or dentate nuclei. Baseline MRIs showed no abnormalities, and in all three cases the observed T2-weighted hyperintensity resolved completely following vigabatrin discontinuation. These results are consistent with those reported by Desguerre et al. (2008) in that resolution of abnormalities was observed after approximately 12 months in six cases, four of which showed resolution while still being treated with vigabatrin. These seminal observations were instructive and established an important clinical benchmark. However, it was necessary to perform study OV-1019 using a robust protocol for review and for defining prespecified MRI abnormalities to determine the true prevalence relative to the untreated population, which appears to be <20%.

Currently, there is no evidence to suggest that this MRI abnormality occurs in children or adults treated with vigabatrin for CPS. The retrospective, masked, repeat review of adults and children treated with vigabatrin for CPS in prospective clinical trials failed to demonstrate an association between vigabatrin exposure and MRI abnormalities. Moreover, the prespecified abnormalities detected in this analysis were mostly nonspecific, cerebral hemispheric white matter T2-weighted or FLAIR hyperintensities, which is not consistent with the anatomic regions predominantly affected in infants treated with vigabatrin for IS (i.e., basal ganglia, thalamus, brainstem, or cerebellum). Using inclusive criteria, there was no statistically significant difference in the incidence or prevalence of prespecified MRI abnormalities between vigabatrin-exposed and vigabatrin-naive subjects. Notably, prespecified MRI abnormalities included high T2-weighted or FLAIR signal or diffusion restriction not explainable by a pathologic process readily diagnosed radiographically (e.g., prior ischemia). Therefore, many cases with nonspecific, age-related “small vessel ischemic” white matter disease were included in the definition of a prespecified abnormality. This was a conservative approach meant to capture all potential abnormalities. Although it is not clear why these abnormalities are observed only in infants and not in children and adults, one could speculate based on these findings that it may be related to either developmental changes in brain myelination in infants or a possible underlying metabolic condition that predisposes these infants to such an MRI abnormality.

In conclusion, vigabatrin is associated with asymptomatic MRI abnormalities characterized by increased T2-weighted signal in infants with IS, whereas children, adolescents, and adults treated with vigabatrin do not appear to be at risk of developing MRI abnormalities. These abnormalities, if they occur, are transient and appear to be dose dependent, and the majority resolve even if treatment with vigabatrin is continued. Although there is no evidence that these MRI changes cause harm, it is important to be vigilant and to err on the side of caution. As part of the diagnostic evaluation of IS, an initial MRI followed by repeat MRIs approximately one and 3 months after starting treatment will have a high probability of detecting any vigabatrin-associated changes based on the known time course. However, MRI examination requires sedation and hence carries some risk (Cote & Wilson, 2006). Therefore, routine MRI surveillance of this population is not recommended, because the long-term clinical sequelae of vigabatrin-associated MRI changes are unknown. Moreover, if MRI changes are discovered, we suggest that healthcare providers balance the benefits of continued vigabatrin use against the risk of MRI surveillance. Although this study clarifies the incidence of these vigabatrin-associated MRI abnormalities, it does not inform the clinician of an appropriate response. A prospective study in children with IS treated with vigabatrin, as well as with other therapies, is warranted to provide additional data on potential long-term clinical consequences of MRI-signal abnormalities.

Acknowledgments

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

Financial support for medical editorial assistance was provided by Ovation Pharmaceuticals. We thank Marithea Goberville, PhD and Angela Fracasso for medical editorial assistance, and Sarah Torri and Sandy Bialek-Smith for expert technical assistance. We also thank the following coinvestigators, study personnel, and neuroradiologists who contributed to the research presented herein: Susana Camposano, MD, Karen Butler, Misha Durmeier, MD; Michelle Ellis; Milagros Salas-Prato, PhD; Rima Nabbout, MD; Olivier, Dulac, MD, PhD; Angela Perez; Michelle Zabala; Jennifer Howell; Michele Johnson, MD; Victoria Silvera, MD; Richard Bronen, MD; Douglas Burd, MD; Stephen Sweriduk, MD; Lyle Gesner, MD; Michael Duhaney, MD; Gerald Ross, MD; Stephen Willing, MD; Gul Moonis, MD; Asim Mian, MD; Jimmy Wang, MD; Glenn Barest, MD; Rafeeque Bhadelia, MD; and Yair Safriel, MD.

We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.

Disclosures of conflicts of interest: JW, CC, JC, RE, and ET have served as paid consultants to Ovation Pharmaceuticals. JW, RE, DS, and ET have received grants from Ovation Pharmaceuticals for other research or activities not reported in this article. The remaining authors have no conflicts of interest.

References

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References
  • Aicardi J, Mumford J, Dumas C, Wood S. (1996) Vigabatrin as initial therapy for infantile spasms: a European retrospective survey. Sabril IS Investigator and Peer Review Groups. Epilepsia. 37:638642.
  • Appleton R, Peters AC, Mumford JP, Shaw DE. (1999) Randomised, placebo-controlled study of vigabatrin as first-line treatment of infantile spasms. Epilepsia. 40:16271633.
  • Bekiesinska-Figatowska M, Walecki J, Wagiel K, Jurkiewicz E. (2001) MRI in the diagnosis of congenital white matter diseases and other neurodegenerative diseases. Neurol Neurochir Pol 35:605619.
  • Ben-Menachem E, Nordborg C, Hedstrom A, Augustinsson LE, Silvenius H. (1988) Case report of surgical brain sample after 2.5 years of vigabatrin therapy [abstract]. Epilepsia. 29(suppl 5):699.
  • Bruni J, Guberman A, Vachon L, Desforges C. (2000) Vigabatrin as add-on therapy for adult complex partial seizures: a double-blind, placebo-controlled multicentre study. The Canadian Vigabatrin Study Group. Seizure. 9:224232.
  • Bryant A, Lopes MB, Bebin EM. (1994) Absence of intramyelinic edema in three patients treated with vigabatrin [abstract]. Epilepsia. 35(suppl 8):115.
  • Cannon D, Weller RO, Mumford JP. (1991a) No vigabatrin-induced microvacuolization in human brains: further neuropathological investigations [abstract]. Epilepsia. 32(suppl 1):12.
  • Cannon DJ, Butler WH, Mumford JP, Lewis PJ. (1991b) Neuropathologic findings in patients receiving long-term vigabatrin therapy for chronic intractable epilepsy. J Child Neurol (suppl 2):S17S24.
  • Chiron C, Dulac O, Luna D, Palacios L, Mondragon S, Beaumont D, Mumford JP. (1990) Vigabatrin in infantile spasms. Lancet 335:363364.
  • Chiron C, Dumas C, Jambaque I, Mumford J, Dulac O. (1997) Randomized trial comparing vigabatrin and hydrocortisone in infantile spasms due to tuberous sclerosis. Epilepsy Res 26:389395.
  • Cohen JA, Fisher RS, Brigell MG, Peyster RG, Sze G. (2000) The potential for vigabatrin-induced intramyelinic edema in humans. Epilepsia 41:148157.
  • Cote CJ, Wilson S. (2006) Guidelines for monitoring and management of pediatric patients during and after sedation for diagnostic and therapeutic procedures: an update. Pediatrics 118:25872602.
  • Desai NK, Runge VM, Crisp DE, Crisp MB, Naul LG. (2003) Magnetic resonance imaging of the brain in glutaric acidemia type I: a review of the literature and a report of four new cases with attention to the basal ganglia and imaging technique. Invest Radiol 38:489496.
  • Desguerre I, Pinton F, Nabbout R, Moutard ML, N’Guyen S, Marsac C, Ponsot G, Dulac O. (2003) Infantile spasms with basal ganglia MRI hypersignal may reveal mitochondrial disorder due to T8993G MT DNA mutation. Neuropediatrics 34:265269.
  • Desguerre I, Marti I, Valayannopoulos V, Bahi-Buisson N, Dulac O, Plouin P, Delonlay P, Hertz-Pannier L, Boddaert N. (2008) Transient magnetic resonance diffusion abnormalities in West syndrome: the radiological expression of non-convulsive status epilepticus? Dev Med Child Neurol 50:112116.
  • Elterman RD, Shields WD, Mansfield KA, Nakagawa J. (2001) Randomized trial of vigabatrin in patients with infantile spasms. Neurology 57:14161421.
  • French JA, Mosier M, Walker S, Sommerville K, Sussman N. (1996) A double-blind, placebo-controlled study of vigabatrin three g/day in patients with uncontrolled complex partial seizures. Vigabatrin Protocol 024 Investigative Cohort. Neurology 46:5461.
  • Gibson JP, Yarrington JT, Loudy DE, Gerbig CG, Hurst GH, Newberne JW. (1990) Chronic toxicity studies with vigabatrin, a GABA-transaminase inhibitor. Toxicol Pathol 18:225238.
  • Guberman A, Bruni J. (2000) Long-term open multicentre, add-on trial of vigabatrin in adult resistant partial epilepsy. The Canadian Vigabatrin Study Group. Seizure 9:112118.
  • Hammond EJ, Ballinger WE Jr, Lu L, Wilder BJ, Uthman BM, Reid SA. (1992) Absence of cortical white matter changes in three patients undergoing long-term vigabatrin therapy. Epilepsy Res 12:261265.
  • Horvath R, Abicht A, Holinski-Feder E, Laner A, Gempel K, Prokisch H, Lochmuller H, Klopstock T, Jaksch M. (2006) Leigh syndrome caused by mutations in the flavoprotein (Fp) subunit of succinate dehydrogenase (SDHA). J Neurol Neurosurg Psychiatry 77:7476.
  • Jackson GD, Williams SR, Weller RO, Van Bruggen N, Preece NE, Williams SC, Butler WH, Duncan JS. (1994) Vigabatrin-induced lesions in the rat brain demonstrated by quantitative magnetic resonance imaging. Epilepsy Res 18:5766.
  • Lippert B, Metcalf BW, Jung MJ, Casara P. (1977) 4-amino-hex-5-enoic acid, a selective catalytic inhibitor of 4-aminobutyric-acid aminotransferase in mammalian brain. Eur J Biochem 74:441445.
  • Oster J, Doherty C, Grant PE, Simon M, Cole AJ. (2003) Diffusion-weighted imaging abnormalities in the splenium after seizures. Epilepsia 44:852854.
  • Paljarvi L, Vapalahti M, Sivenius J, Riekkinen P. (1990) Neuropathological findings in 5 patients with vigabatrin treatment [abstract]. Neurology 40(suppl 1):157.
  • Pearl PL, Novotny EJ, Acosta MT, Jakobs C, Gibson KM. (2003) Succinic semialdehyde dehydrogenase deficiency in children and adults. Ann Neurol 54(suppl 6):S73S80.
  • Pearl P, Molloy-Wells E, McClintock WM, Vezina LG, Conry JA, Elling NJ, Tsuchida T, Heath C, Cushner-Weinstein S, Kolodgie M, Weinstein SL, Gaillard WD. (2006) MRI abnormalities associated with vigabatrin therapy: higher risk in infants? Epilepsia 47(suppl 4):14.
  • Pedersen B, Hojgaard K, Dam M. (1987) Vigabatrin: no microvacuoles in a human brain. Epilepsy Res 1:7476.
  • Peyster RG, Sussman NM, Hershey BL, Heydorn WE, Meyerson LR, Yarrington JT, Gibson JP. (1995) Use of ex vivo magnetic resonance imaging to detect onset of vigabatrin-induced intramyelinic edema in canine brain. Epilepsia 36:93100.
  • Preece NE, Houseman J, King MD, Weller RO, Williams SR. (2004) Development of vigabatrin-induced lesions in the rat brain studied by magnetic resonance imaging, histology, and immunocytochemistry. Synapse 53:3643.
  • Sander JW, Trevisol-Bittencourt PC, Hart YM, Shorvon SD. (1990) Evaluation of vigabatrin as an add-on drug in the management of severe epilepsy. J Neurol Neurosurg Psychiatry 53:10081010.
  • Urbach H. (2005) Imaging of the epilepsies. Eur Radiol 15:494500.
  • Weiss KL, Schroeder CE, Kastin SJ, Gibson JP, Yarrington JT, Heydorn WE, McBride RG, Sussman NM, Arezzo JC. (1994) MRI monitoring of vigabatrin-induced intramyelinic edema in dogs. Neurology 44:19441949.
  • Wohlrab G, Boltshauser E, Schmitt B. (1998) Vigabatrin as a first-line drug in West syndrome: clinical and electroencephalographic outcome. Neuropediatrics 29:133136.
  • Yarrington JT, Gibson JP, Dillberger JE, Hurst G, Lippert B, Sussman NM, Heydorn WE, Marler RJ. (1993) Sequential neuropathology of dogs treated with vigabatrin, a GABA-transaminase inhibitor. Toxicol Pathol 21:480489.
  • Ziyeh S, Berlis A, Korinthenberg R, Spreer J, Schumacher M. (2002) Selective involvement of the globus pallidus and dentate nucleus in succinic semialdehyde dehydrogenase deficiency. Pediatr Radiol 32:598600.