Objective Although vigabatrin irreversibly constricts the visual field, it remains a potent therapy for infantile spasms and a third-line drug for refractory epilepsies. In albino animals, this drug induces a reduction in retinal cell function, retinal disorganisation and cone photoreceptor damage. The objective of this study was to investigate the light dependence of the vigabatrin-elicited retinal toxicity and to screen for molecules preventing this secondary effect of vigabatrin. Methods Rats and mice were treated daily with vigabatrin 40mg and 3mg, respectively. Retinal cell lesions were demonstrated by assessing cell function with electroretinogram measurements, and quantifying retinal disorganization, gliosis and cone cell densities. Results Vigabatrin-elicited retinal lesions were prevented by maintaining animals in darkness during treatment. Different mechanisms including taurine deficiency were reported to produce such phototoxicity; we therefore measured amino acid plasma levels in vigabatrin-treated animals. Taurine levels were 67% lower in vigabatrin-treated animals than in control animals. Taurine supplementation reduced all components of retinal lesions in both rats and mice. Among 6 vigabatrin-treated infants, the taurine plasma level was found to be below normal in three patients and undetectable in two patients. Interpretation These results indicate that vigabatrin generates a taurine deficiency responsible for its retinal phototoxicity. Future studies will investigate whether co-treatment with taurine and vigabatrin can limit epileptic seizures without inducing the constriction of the visual field. Patients on vigabatrin could gain immediate benefit from reduced light exposures and dietetic advice on taurine-rich foods.

Epileptic patients experienced an irreversible loss of their peripheral visual field upon treatment with vigabatrin (gamma-vinyl GABA), an inhibitor of the GABA degrading enzyme, GABA transaminase. Subsequently, central visual function was reported to also be irreversibly altered. This visual loss is associated with a decrease in the electroretinogram measurement localizing the deficit to the retina. To investigate its cellular origin, we treated rats daily with vigabatrin for 45 days. Two days after arresting this treatment, rats exhibited an irreversible decrease in the photopic electroretinogram, the flicker response, and the oscillatory potentials. These functional alterations were associated with a peripheral disorganization of the outer retina. However, photoreceptor damage was not limited to these disorganized areas, but cone inner and outer segments were severely injured in more central areas and their numbers were irreversibly decreased by 17 to 20%. Ultrastructural examination of the retina confirmed the presence of major photoreceptor damages, which were further supported by terminal deoxynucleotidyltransferase-mediated dUTP nick end labeling (TUNEL) and caspase-3 activation both indicative of photoreceptor apoptosis. This study suggests that the visual field loss in vigabatrin-treated epileptic patients may result from a sequence of events starting from cone cell injury to a more severe disorganization of the photoreceptor layer.

In different animal models, photoreceptor degeneration was correlated to an abnormal increase in cGMP concentration. The cGMP-induced photoreceptor toxicity was demonstrated by applying the phosphodiesterase inhibitor 3-isobutyl-1-methylxanthine on retinal explants. To assess the role of cGMP-gated channels in this cGMP toxicity, the Ca(2+) channel blockers verapamil and L- and D-diltiazem, which block cGMP-gated channels with different efficacies, were applied to in vitro animal models of photoreceptor degeneration. These models included: (i) adult rat retinal explants incubated with zaprinast, a more specific inhibitor of the rod phosphodiesterase than 3-isobutyl-1-methylxanthine and (ii) rd mouse retinal explants. Photoreceptor apoptosis was assessed by terminal dUTP nick end labelling and caspase 3 activation. Effects of the blockers on the synaptic rod Ca(2+) channels were measured by patch-clamp recording. In the zaprinast-induced photoreceptor degeneration model, both diltiazem isomers rescued photoreceptors whereas verapamil had no influence. Their neuroprotective efficacy was correlated to their inhibition of cGMP-gated channels (l-diltiazem>d-diltiazem>verapamil=0). In contrast, all three Ca(2+) channel blockers suppressed rod Ca(2+) channel currents similarly. This suppression of the currents by the diltiazem isomers was very weak (16.5%) at the neuroprotective concentration (10 microm). In rd retinal explants, both diltiazem isomers also slowed down rod degeneration in contrast to verapamil. L-diltiazem exhibited this effect at concentrations ranging from 1 to 20 microm. This study further supports the photoreceptor neuroprotection by diltiazem particularly in the rd mouse retina, whereas the absence of neuroprotection by verapamil further suggests the role of cGMP-gated channel activation in the induction of photoreceptor degeneration.