Customize your JAMA Network experience by selecting one or more topics from the list below.
Anticonvulsants are known to cause different kinds of visual disturbances.1 Recently, vigabatrin has received attention because of considerable bilateral concentric field defects observed in a few patients receiving this drug as add-on therapy.2 These field constrictions appear to be irreversible in most cases. Surprisingly, only subtle changes of retinal function have been reported, most often alterations of the oscillatory potentials in the electroretinogram (ERG). Findings from histopathological studies in rats, mice, and dogs but not in monkeys and humans show a microvacuolation in myelin sheaths of the white matter when exposed to vigabatrin.3 However, all these alterations cannot sufficiently explain a severe concentric visual field constriction.
Report of a Case
A 17-year-old boy suffered from focal epilepsy of unknown origin since the age of 13. Following unsuccessful treatment with carbamazepine and valproate as monotherapy and polytherapy, he was put on a combination regimen of valproate and vigabatrin 18 months prior to our first examination (2 g of vigabatrin and 2.4 g of valproate per day). One year later, the patient became aware of a constriction of his visual field.
Figure 1, A, shows the visual field constriction (Goldmann kinetic perimetry). The results of the static perimetry were comparable (Figure 1, C). Findings for visual acuity (20/20 OU), motility, eye position, anterior segments, and fundus were completely normal. Color vision tested by the desaturated Panel D-15 test, absolute rod threshold after 30 minutes of dark adaptation (Hartinger dark adaptometer, Carl Zeiss, Jena, Germany), and pattern evoked visual potentials were also unremarkable. Baseline and light peak of the electro-oculogram (ISCEV standard) also rendered normal results (Arden ratio, 2.1 OD, 1.9 OS).
Goldmann kinetic visual fields of the patient taking vigabatrin (A) and a patient with retinitis pigmentosa (B). In part B, an additional isopter (V-4e) is displayed to illustrate that the difference between the sensitivity for bigger (V-4e) and smaller (I-4e) stimuli is rather small in the patient receiving treatment with vigabatrin. C, A 30° static perimetry of the patient receiving vigabatrin (Octopus perimeter 500 EZ, Interzeag, Schlieren, Switzerland), program 36; the missing blind spot in the visual field of the right eye is due to the rather coarse pattern of test stimuli (6°).
The Ganzfeld electroretinogram (ERG) performed according to the International Society for Clinical Electrophysiology of Vision (ISCEV) standard was basically normal. The b/a ratio was low (1.2; normal ratio, 1.39 (SD, 0.12) and the amplitude of the second oscillatory potential was 31 µV compared with a normal value of 76 µV (SD, 18). While the rod and maximal b-wave amplitudes fell within the center of the normal range, the cone and 30-Hz flicker responses were at the lower border. In the multifocal ERG there were local responses of normal amplitude (ie, the difference between positive peak and negative trough) and implicit time even in areas of missing perception for the Goldmann target III/4 (Figure 2, C). In contrast, the multifocal ERG of the patient with retinitis pigmentosa who had a comparable visual field constriction shows the typical decrease of amplitude and increase of implicit time in the affected regions (Figure 2, B and D).4 However, the wave form in the affected areas of the patient receiving treatment with vigabatrin was different in that the positive peak was low compared with the preceding negative trough (Figure 2, A, arrows).
The multifocal electroretinogram of the patient receiving treatment with vigabatrin and the patient with retinitis pigmentosa (left eyes) shown in Figure 1. A, Trace array of the patient receiving treatment with vigabatrin showing the 61 local responses representing the cone activity within the central 50° of visual angle. The change of wave form is marked by arrows (see "Report of a Case" section for further details). B, Trace array of the patient with retinitis pigmentosa showing the typical decrease of amplitude and increase of implicit time in the affected regions. C, A 3-D representation (scalar product) of the data from part A. The central peak corresponds to the high cone density of the fovea; the blind spot is visible on the left. D, A 3-D representation of the data for part B. Note the peripheral loss of response density.
The ERG-off response (red light, 200-millisecond stimulus duration) was present, but the implicit time was late (230 milliseconds; normal response time, 225.6 milliseconds (SD,1.95 milliseconds). The scotopic threshold response was normal. Readings from the transient and steady state pattern ERGs according to the ISCEV guidelines were normal for the central visual field. In contrast, stimulating the temporal retina where 50% of the pattern stimulus was displayed in the visual field loss area, the N95 component of the transient pattern ERG became smaller than the P50 component (about 70% of the P50 component) that never occurred in a normal observer.
Treatment with vigabatrin was immediately discontinued when the visual field defect was diagnosed. At the follow-up examinations the patient sometimes indicated a subjective improvement, but even 6 months after the first examination, the visual field contraction tested by kinetic and static perimetry remained the same.
In the case reported herein, the concentric visual field loss is probably due to the intake of vigabatrin because (1) the intake of carbamazepine which is known to interact with retinal function was cessated 12 months prior to the first symptoms of visual field constriction; (2) we are unaware of retinal toxic side effects associated with valproate treatment; and (3) the frequency and severity of seizures were low and therefore probably not affecting visual function. Comparing the multifocal ERG to that of a patient with retinitis pigmentosa shows that the outer retina is probably not the site of disease action. As in other reports, the oscillatory potentials of the Ganzfeld ERG were attenuated in our patient. Together with the selective reduction of the N95 component in the pattern ERG, there is evidence of an alteration of inner retinal function.
Vigabatrin is an aminobutyrate-aminotransferase antagonist and acts as a GABA (γ-aminobutyric acid) analog.5 GABA is a transmitter at different sites of the postreceptoral retina.6 It plays a role in the regulation of horizontal cell coupling, and an accumulation of GABA can be found in amacrine cells. The visual field constriction might result from a block of transmission between bipolar cells and amacrine and/or ganglion cells. Another possibility might be a loss of ganglion cell function. If so, the high redundancy of information-processing in the central retina may be the reason for sparing of the central visual field. Another explanation could be that vigabatrin is eventually more toxic to peripheral than to central ganglion cells. In any case, ganglion cell damage may lead to optic atrophy in later stages of the disease which has already been reported in other cases of visual field constriction associated with vigabatrin medication.
Corresponding author: Klaus Ruether, MD, Charité-Virchow Augenklinik, Augustenburger Platz 1, 13353 Berlin, Germany (e-mail: firstname.lastname@example.org).
Ruether K, Pung T, Kellner U, Schmitz B, Hartmann C, Seeliger M. Electrophysiologic Evaluation of a Patient With Peripheral Visual Field Contraction Associated With Vigabatrin. Arch Ophthalmol. 1998;116(6):817–819. doi:
Monkeypox Resource Center