An investigation into the relationship between vigabatrin, movement disorders, and brain magnetic resonance imaging abnormalities in children with infantile spasms




We aimed to investigate the relationship between movement disorders, changes on brain magnetic resonance imaging (MRI), and vigabatrin therapy in children with infantile spasms.


Retrospective review and brain MRI analysis of children enrolled in the International Collaborative Infantile Spasms Study (ICISS) who developed a movement disorder on vigabatrin therapy. Comparisons were made with controls within ICISS who had no movement disorder.


Ten of 124 infants had a movement disorder and in eight it had developed on vigabatrin therapy. Two had a movement disorder that resolved on dose-reduction of vigabatrin, one had improvement on withdrawing vigabatrin, two had resolution without any dose change, and in three it persisted despite vigabatrin withdrawal. The typical brain MRI changes associated with vigabatrin therapy were noted in two infants. Ten control infants were identified. Typical MRI changes noted with vigabatrin were noted in three controls.


It is possible that in two out of eight cases, vigabatrin was associated with the development of a movement disorder. In six out of eight cases a causal relationship was less plausible. The majority of infants treated with vigabatrin did not develop a movement disorder. MRI changes associated with vigabatrin do not appear to be specifically related to the movement disorder.

What this paper adds

  • Evidence for a direct causal link between vigabatrin and movement disorder is weak.
  • The vast majority of infants treated with vigabatrin do not develop a coincidental movement disorder.
  • Vigabatrin is associated with MRI changes in the basal ganglia.
  • MRI changes in the basal ganglia do not appear to be directly related to movement disorders.

Infantile spasms are an age-related epileptic encephalopathy that occurs in children, usually between the ages of 3 months and 18 months. The seizures manifest typically as clusters of either flexor or extensor epileptic spasms, which are often coincident with an arrest or regression of neurodevelopment. The electroencephalograph characteristically, but not always, shows a chaotic high voltage interictal pattern known as hypsarrhythmia. Many underlying aetiologies, such as tuberous sclerosis complex, trisomy 21, neuronal migration disorders, and hypoxic–ischaemic encephalopathy, have been associated with infantile spasms but in a significant minority of cases no cause is found. There has been much debate about the most effective treatment modality for these patients but there appears to be a relatively wide consensus that the two most effective therapies are either the GABA-ergic anticonvulsant, vigabatrin, or hormonal therapies (prednisolone, adrenocorticotrophin hormone [ACTH], or synthetic ACTH).[1]

It has been known since the 1980s that vigabatrin usage in animals was associated with the development of intramyelinic oedema in the central nervous system, most notably in the cerebellum, reticular formation and optic tracts in rats, and columns of the fornix and the optic tracts in dogs.[2, 3] Recently reports have surfaced in the literature linking vigabatrin usage in humans with characteristic magnetic resonance imaging (MRI) changes in the globus pallidus, thalamus, brain stem, and dentate nuclei of the cerebellum.[4, 5] These changes appear to be dose-dependent (i.e. more frequent with high-dose therapy) and spontaneously reversible.[6, 7] It is unclear what duration of therapy is necessary to induce these changes but it appears there is a peak incidence after 3 to 6 months of treatment. The changes are estimated to occur in between 22% and 32% of patients.[4, 5] There has been no consistent or convincing evidence that the MRI changes correlate to any clinical symptoms.

In 2009 in the United Kingdom, the Medicines and Healthcare Products Regulatory Agency (MHRA) produced a public assessment report looking not only at the issue of vigabatrin use and brain MRI abnormalities but also at a possible association with movement disorders in infancy. The report detailed that there had been 18 reports of movement disorders identified in clinical trials with vigabatrin before the drug was licensed and marketed and that there had been 58 worldwide reports of movement disorders identified after vigabatrin was licensed. These movement disorders occurred in individuals of all ages and were of multiple different types. The report provided no evidence that the MRI changes were linked to the movement disorders.[8] We are not aware of reports of movement disorder in children with infantile spasms but they are known to have occurred prior to the use of vigabatrin (B Neville, personal communications 2009).

In view of the mounting evidence of MRI brain changes and movement disorders that were being reported as being possibly associated with vigabatrin use, we decided to investigate this issue in the context of a large ongoing clinical trial of treatment modalities in infantile spasms, the International Collaborative Infantile Spasms Study (ICISS). ICISS is a pragmatic parallel group randomized trial comparing hormonal treatment combined with vigabatrin to hormonal treatment alone in the treatment of infantile spasm. We were interested to analyse how often movement disorders were reported as a problem in the trial, whether the movement disorders could in any way be attributed to vigabatrin treatment and whether there was any relationship between movement disorders and the MRI abnormalities that have been reported with vigabatrin use.


The study population included all children (124 children) with infantile spasms enrolled into the International Collaboration Infantile Spasms Study (ICISS) trial between January 2007 and August 2010 for treatment of IS. Criteria for inclusion into this analysis were children who were reported to have movement disorder from their primary clinician after enrolment into the ICISS trial. Exclusion criteria for examining the possible attribution of movement disorder to vigabatrin included children who had pre-existing movement disorder prior to enrolment into the ICISS cohort.

A retrospective clinical review and brain MRI analysis was performed on all patients who met the inclusion criteria to analyse the likely underlying cause of the movement disorder. This included detailed clinical information from the ICISS trial database and results of investigations performed. In addition, a pre-designed pro forma was sent to each primary clinician requesting further information on the patient's movement disorder (including the provision of any available video recordings of the movement disorder), antiepileptic drug therapy, and clinical status at latest follow-up.

Control brain MRIs from infants within the study without movement disorder were also randomly allocated for each index case, matched for age at randomization (SD 30d), treatment allocated and having an MRI available. All MRIs were retrospectively interpreted by an independent paediatric neuroradiologist (ML), who was blinded to the patient's clinical profile of whether movement disorder was present or not and whether the patient had been treated with vigabatrin. The neuroradiologist was also instructed to analyse and comment on the presence of any pre-defined MRI changes that have been previously reported in association with vigabatrin therapy.


Ten of 124 infants were reported to have had a movement disorder. Two of these infants had a movement disorder that preceded the onset of infantile spasms. Therefore eight infants were confirmed to have movement disorder after enrolment into the ICISS trial and were further studied: six were male and two were female (details are given in Table 1). Their age of presentation with infantile spasms ranged between 4 months and 7 months. Five infants had a plausible proven aetiology for their infantile spasms: two had chromosomal abnormalities (trisomy 21, microdeletion of 16p13), one had focal cortical scarring, one had a presumed neurometabolic disorder with an abnormal cerebrospinal fluid neurotransmitter profile, and one had Menkes disease (ATP7A gene mutation). All eight infants received vigabatrin treatment for their infantile spasms. They either received vigabatrin as combination therapy with steroid at enrolment (five children) or vigabatrin was added adjunctively (three children) at a later stage due to recurrence of infantile spasms.

Table 1. Clinical and radiological details of movement disorder cases
Case number (age in months at initiation of treatment)Initial treatment for spasmsAge at onset of MDType of MDOutcome of MDVigabatrin changes on brain MRIUnderlying aetiology of ISPossible causal link between vigabatrin and MD
  1. MD, movement disorder; MRI, magnetic resonance imaging; IS, infantile spasms.

1 (6)Prednisolone (stopped day 30) – vigabatrin for relapse age 8mo9moGeneralized complex myoclonusResolved within 72h on discontinuation of vigabatrinNoMicrodeletion 16p13.11Possibly
2 (7)Prednisolone (stopped day 30) with vigabatrin7.5moMild generalized choreoathetosisResolved spontaneously age 8.5mo without stopping vigabatrinNoFocal cortical scarringUnlikely
3 (7)Prednisolone (stopped day 30) vigabatrin added for non-response day 158moSignificant generalized choreoathetosisResolved within 1wk after vigabatrin dose reducedBilateral globus pallidus changes with restricted diffusion day 32Trisomy 21Possibly
4 (6)Prednisolone (stopped day 16) with vigabatrin (stopped age 9mo)6mo (MD started on day 9 of treatment)Head bobbing from age 3mo – orofacial dyskinesia and chorea of face and limbsBrief period choreoathetosis lasting 2wks – age 15mo, ongoing dyskinesia of head with fine intermittent tremor despite stopping vigabatrin age 9moNoUnderlying neurometabolic disorder (abnormal cerebrospinal fluid neurotransmitter)Unlikely
5 (4)Prednisolone (stopped day 30) with vigabatrin stopped day 84mo (MD started on day 5 of treatment)Choreoathetosis of head and hands, dyskinesia of tonguePersist with ongoing choreoathetosis of limbsNoNo aetiology identifiedPossibly
6 (7)Tetracosactide depot for 2wks, then prednisolone for 15d – vigabatrin started age 11mo for relapse7mo (MD started on day 8 of treatment but settled with clonazepam – recurred at 11mo)Initially limb dystonia and chorea – recurrence on starting vigabatrin (also coincided with IS recurrence, febrile illness and being weaned off clonazepam) with florid orofacial/truncal spasms and limb dystoniaOngoing episodes of dystonia and choreiform movementsNo, but MRI done before vigabatrin begunNo aetiology identifiedUnlikely
7 (7)Tetracosactide depot for 2mo, with vigabatrin stopped day 107mo (MD started on day 8 of treatment. Note: also had preceding rotavirus infection on day 7)Sudden onset of asymmetrical four limb movements initially similar to spasms – resolved with sleep and sedation – no electroencephalograph correlateMD improved with withdrawing vigabatrin; however, ongoing intermittent muscle spasmsNoMenkes disease (ATP7A gene mutation)Unlikely
8 (5)Tetracosactide depot for 2wks then prednisolone for 15d, with vigabatrin to age 9moDay 10Orofacial dystonic posturingResolved by day 34 with no changes to vigabatrin doseIncreased signal globus pallidus, thalamus, hypothalamus, brain stem, pons, and dentate nucleiNo identified aetiologyUnlikely

Table 1 summarizes the chronology of treatment received for infantile spasms, type of movement disorder, clinical course of movement disorder, underlying aetiology, clinical status at latest follow-up, and MRI results of the index cases. Of the eight infants, two had a movement disorder that resolved on withdrawal/dose reduction of vigabatrin, one had improvement on withdrawing vigabatrin, two had a movement disorder that resolved without any dose change of vigabatrin, and three had a movement disorder persisting in spite of vigabatrin withdrawal. Three infants received vigabatrin for relapse of infantile spasms. Seven index cases had brain MRI after initiation of vigabatrin treatment (the timing of the scan ranging from 5d–3mo post-initiation of vigabatrin). In two cases there were the characteristic changes in the globus pallidus, thalamus, hypothalamus, pons, and dentate nuclei that have been described as being associated with vigabatrin therapy (see Fig. 1).

Figure 1.

Increased signal in globus pallidus bilaterally in case number 8 (see Table 1) with movement disorder.

Ten control infants were identified. All control infants were receiving vigabatrin at the time of their MRI (the timing of the scan ranged from 1 day to 2.5 months post-initiation of vigabatrin treatment). The typical changes associated with vigabatrin, as described above, were noted in three of the control cases on MRI (see Fig. 2).

Figure 2.

Increase signal in globus pallidus bilaterally in control case without movement disorder.


A movement disorder was present in 10 out of 124 children (8%) with infantile spasms and had developed in 8 out of 124 (6.5%) children with infantile spasms after the initiation of anticonvulsant therapy, including vigabatrin. After analysis of the data, there appeared to be a close temporal relationship between vigabatrin therapy and the movement disorder in two cases (1.6%). In these cases (cases one and three) the movement disorder began after the initiation of vigabatrin therapy and resolved soon after vigabatrin was withdrawn. Both these cases had a proven underlying aetiology, trisomy 21 and 16p13 microdeletion, but neither of these aetiologies has separately been associated with a movement disorder. In four cases (cases four to seven) the movement disorder has persisted, improving in one, despite withdrawal of the vigabatrin. It is possible in these four cases that vigabatrin unmasked an underlying propensity for a movement disorder but the persistence of the movement disorder after the withdrawal of the agent makes a causal role for vigabatrin less likely. In two cases (cases two and eight) the movement disorder resolved without any dose change in vigabatrin. Again a causal role for vigabatrin in these cases is unlikely unless the postulated movement disorder caused by vigabatrin is transitory and self-resolving. It should also be noted that seven out of eight cases were also receiving hormonal therapy at the time of the onset of the movement disorder.

It would perhaps be surprising if vigabatrin were to cause a movement disorder given that it is a drug that has been designed to potentiate the actions of the inhibitory neurotransmitter GABA. Deficiencies in central nervous system levels of GABA are implicated in many movement disorders and in the development of spasticity. Indeed early work on vigabatrin suggested that it may have a role in the treatment of movement disorders and spasticity. It was seen to have a beneficial effect on tardive dyskinesias induced by neuroleptics, and there was conflicting evidence as to whether it ameliorated ataxia in degenerative cerebellar diseases.[9-13] There was one report of it having a positive effect on spasticity in metachromatic leukodystrophy.[14] There have been equivocal results in its use as a treatment of dyskinesias in Parkinson disease and spasticity in multiple sclerosis.[15, 16] In none of these studies was vigabatrin found to provoke or exacerbate movement disorder.

One of the reasons why concern about vigabatrin causing a movement disorder has arisen has been because of the discovery of vigabatrin-associated MRI changes in the basal ganglia and cerebellum. These changes have been seen in 22% to 32% of children with infantile spasms treated with vigabatrin. They appear to be related to both the length of treatment and the dose of therapy. The MRI changes are most prevalent after 3 to 6 months' treatment and after high-dose therapy.[4, 5] Because the changes involve areas of the brain that are concerned with control of movement, it is not unreasonable to suspect that vigabatrin might also be responsible for causing a movement disorder. However, in our study, it would appear that the MRI changes are equally common in children who do not develop a movement disorder. Three out of ten control cases had vigabatrin-associated signal change in the basal ganglia as compared with two out of eight index cases.

The MRI changes associated with vigabatrin use do not appear to occur in older children or adults. The reasons for this are unclear. It may be that infants are more susceptible to these changes due to developmental changes in myelination that take place during infancy. Others have speculated that the characteristic distribution of MRI changes in the basal ganglia, dentate nuclei, and brain stem reflect regional variation in GABA metabolism and that the restricted diffusion seen on diffusion-weighted images suggests a toxic effect caused by failure of the Na/K ATP pump or transient vacuolation. However, at present, the reason for these MRI changes is not definitively understood and it is unclear why the majority of infants do not exhibit them.

By definition, all the children who develop a movement disorder in the context of infantile spasms are already neurologically abnormal. It is possible that the movement disorder that is seen in some of these children is a manifestation of their underlying neurological abnormality even if a precise aetiology has not been defined. The combination of infantile spasms and complex movement disorder has been described in association with mutations in the ARX gene.[17, 18] It is possible that there are other aetiologies, as yet undescribed, that will account for some of the cases of infantile spasms and movement disorder.

This was not a prospective study set up to investigate MRI abnormalities and movement disorder associated with vigabatrin use. Rather it is an analysis of cases of movement disorder that have occurred within the context of a clinical trial. ICISS is a pragmatic clinical trial and does not specify an imaging protocol or the time at which imaging should occur. Consequently there was wide variation in the time at which imaging occurred with respect to initiation of vigabatrin therapy. It is possible that the prevalence of MRI changes was underestimated because imaging was performed before such changes would be expected to be manifest. However, this would have applied equally to index and control cases. Interestingly, the MRI changes were visible early in treatment in the cases described in this study (see Table 1). In one case, the changes were visible within 17 days of starting treatment.

In conclusion, this analysis confirms that vigabatrin is associated with changes in the basal ganglia, brain stem, and cerebellum on MRI. These changes were visible early in treatment. The MRI changes were as common in the control group as they were in the cases of movement disorder. From these data there did not appear to be a direct link between the imaging changes and the movement disorder. On the basis of the published data and the data in this study we would urge caution in attributing movement disorder in children with infantile spasms to vigabatrin therapy. In only two out of the eight cases of movement disorder described here was there a clear temporal relationship between the onset and cessation of movement disorder and vigabatrin therapy. It remains possible in those two cases that vigabatrin was responsible for the abnormal movements. However, in the other six cases it seems less likely that vigabatrin was responsible. If vigabatrin is associated with a movement disorder in patients with infantile spasms, then the relationship is weak: the vast majority of patients treated with the drug for infantile spasms do not develop a movement disorder. The relationship, at the moment, is also inconsistent: there are no reported cases, for example, of movement disorder in tuberous sclerosis patients treated with vigabatrin for infantile spasms. Any apparent relationship may also often be confounded by the underlying abnormal neurology in the patient: the movement disorder may well be the product of the underlying condition rather than the vigabatrin, which will be commonly used to treat their infantile spasms. Whilst it is possible that vigabatrin may be responsible for a small number of cases of movement disorder in these children, it is prudent to be careful to avoid ascertainment bias by attributing all movement disorders we now see in this population to vigabatrin therapy.


We acknowledge the assistance of Dr Peter Baxter (Sheffield Children's Hospital) who has supplied us with essential clinical information for this paper.