Using the electroretinogram to detect and monitor the retinal toxicity of anticonvulsants (editorial)

Abstract

It is now clear that the majority of patients taking the GABAergic anti-epileptic drug vigabatrin develop visual dysfunction, particularly concentric constriction of the visual field. The visual disturbances may be mild or severe, often asymptomatic, with electroretinograms (ERG) confirming retinal dysfunction, especially of the cone system. In the first of these articles (Johnson et al), the reversibility of visual function loss from vigabatrin was assessed. Measurements of static and kinetic visual fields, visual acuity, color vision, and the ERG were recorded while patients were taking vigabatrin and again in 13 patients who had discontinued the drug because of lack of efficacy or reductions in visual field. Most of the patients had been off the drug for three to six months, although two patients had been drug-free for almost one year. Although ERG cone implicit time improved, most of the patients did not show improvement in either clinical measures of visual function or in ERG amplitudes. However, several patients who showed minimal visual field loss while on the drug had substantial recovery of ERG amplitudes. There was no statistical association between recovery of function and either duration of treatment or cumulative dosage. The multifocal ERG showed a diffuse loss of function that was not isolated to the periphery. Although the visual deficits in patients taking vigabatrin tend to be mild, most patients do not show improvement after they stop taking the drug, and visual field loss resulting from vigabatrin was not reversible. In the second article (Harding et al), the specific characteristics of the retinal electrophysiologic response in patients taking vigabatrin are delineated. A cohort of eight previous and 18 current vigabatrin users and eight never vigabatrin-treated patients with epilepsy receiving other anticonvulsants underwent ERG, electro-oculography (EOG), and automated static threshold perimetry. An additional 22 normal subjects underwent ERG. Of the 26 patients exposed to vigabatrin, 18 exhibited visual field loss that was not seen in the control subjects. The presence and severity of the field loss was significantly associated with the latency and amplitude of the ERG cone function. The amplitude of the cone flicker response was the strongest predictor of visual field loss, with a sensitivity of 100% and a specificity of 75%. The EOG, the photopic and scotopic ERG, and the latency of the ERG second oscillatory potential were not significantly related to the presence of visual field loss. Vigabatrin therapy was significantly associated with the photopic amplitude, the scotopic a-wave latency, and the latency of the second oscillatory potential. The authors conclude that in patients who cannot reliably perform perimetry, the cone-specific ERG flicker amplitude provides the best screening method for detecting vigabatrin-associated visual field loss. In the accompanying editorial, Miller emphasizes that many of those patients most likely to benefit from vigabatrin and related anticonvulsants are infants and young children who cannot perform perimetry or adults with such severe seizures and cognitive impairment that they either cannot undergo perimetry or cannot provide reliable, reproducible responses. Hence, the clinician caring for a patient in whom vigabatrin is thought to be the drug of choice has with the ERG a reliable method of monitoring retinal function (and thus visual field) during therapy. Miller recommends such monitoring be performed every six months. Should worsening in the amplitude of the cone flicker response become evident, a discussion can be held with the patient or the primary care giver regarding whether to continue therapy.—Nancy J. Newman