- Top of page
Summary: Purpose: To investigate visual function in the central 10 degrees in patients who have undergone vigabatrin (VGB) antiepileptic drug (AED) therapy with the aim of identifying a clinical regimen for assessing central visual function.
Methods: The sample comprised 12 epilepsy patients (mean age, 38.6 ± 11.7 years) who had been treated with VGB (either as monotherapy or polytherapy). A number of central visual-function tests were carried out for each eye, including high-contrast LogMAR visual acuity, short-wavelength automated perimetry (SWAP 10-2), spatial contrast sensitivity (CSV-1000), and Farnsworth-Munsell (FM) 100-hue colour discrimination.
Results: The group mean cumulative VGB dose was 5,086 ± 3,245 g. The average SWAP 10-2 mean deviation (MD) for the group was –3.24 ± 3.23 dB; 14 eyes of eight patients showed defects (range, –1.62 to –9.46 dB). The square root of the group mean total error score for FM 100-hue was 7.42 ± 3.84; nine eyes of five patients were classified as abnormal with an unsolved colour axis suggestive of complex drug interactions. For contrast sensitivity, 15 eyes of eight patients yielded abnormal results in one or more spatial frequencies. Defects were more prominent at higher spatial frequencies. Overall, four patients had defects in all three visual-function tests, six patients had mixed defects, and two patients were normal.
Conclusions: Visual-function deficits in epilepsy patients treated with VGB are present in the central 10 degrees of the retina. We recommend a battery of investigations, including SWAP 10-2 and spatial contrast sensitivity testing, to assess central visual function.
Previous studies, using perimetry and electrophysiologic techniques, have identified visual field abnormalities in a proportion of epilepsy patients treated with vigabatrin (VGB; γ−vinyl-aminobutyric acid), a GABA-transaminase inhibitor (1–6). Although the link between VGB and visual field loss is not conclusive, there is substantial evidence to suggest a strong association between the two. Characteristically, visual field defects associated with VGB are bilateral and are seen as concentric peripheral loss that is most severe nasally, with relative sparing of the temporal field. In many cases, the visual field constriction is asymptomatic and is recognised only when gross loss has occurred. Visual field loss in VGB-treated patients has been demonstrated with kinetic perimetry (7,8), static white–white perimetry (4,9,10), and short-wavelength automated perimetry (SWAP) (2).
Abnormal visual electrophysiologic findings have long been identified in patients with VGB-attributed visual-field defects. Typically, electroretinogram (ERG) abnormalities that have been associated with peripheral visual deficits include increased latency and decreased amplitude of the b-wave and reduced oscillatory potentials (11,12). Although the defective ERG responses often correlate with the peripheral locus of visual-field loss, some studies have shown that the defects extend beyond the sites of visual loss, indicating more widespread damage. Interestingly, the most consistent electrophysiologic markers that have been found to predict VGB-associated visual-field defects are abnormal ERG 30-Hz flicker and reduced oscillatory potentials (5,13–15). Abnormal 30-Hz flicker is indicative of retinal cone dysfunction, which anatomically is suggestive of damage in the central retina. This finding, taken together with isolated reports of abnormal visual evoked potentials (2,13) and multifocal ERG loss, the site of which has not always been isolated to the periphery, suggests an irregularity of the central visual field, possibly at the retinal level (11).
Morphologic change in the retina is evident both peripherally and centrally in VGB-treated patients. A range of abnormalities have been observed including narrowing of the retinal arterioles (13), atrophy and tessellation of the peripheral retina (1,9), surface wrinkling retinopathy of the macula (13), abnormalities of the retinal pigment epithelium, and pallor of the optic nerve (1,9,16). Krauss et al. (13) were among the first to identify clinical ophthalmic abnormalities in the central retina including reduced visual acuity, irregularities of the macula reflex, and some subtle colour-vision defects. More recently, Nousiainen et al. (17,18) reported that VGB impairs contrast sensitivity in patients with previously constricted visual fields, and that patients treated with carbamazepine (CBZ) and VGB monotherapy exhibit abnormal colour perception, with a higher incidence being noted in the VGB-treated group.
Psychophysical, ophthalmologic, and morphologic findings collectively suggest that central visual deficits are present in the macula region of VGB-treated patients. Despite this, routine clinical tests are still largely limited to examining the peripheral visual field. The current ophthalmic testing regimen often uses standard white–white perimetry that simultaneously assesses all visual pathways. If subtle damage occurs predominantly, or more readily, in a single processing stream, then it is likely to be masked by the responses of other pathways. Given this, testing strategies that are tuned to sample specific neural subgroups selectively may be more sensitive for the detection of subtle change in visual function.
Visual-field tests that assess the peripheral retina show greater variability than those that examine the central retina (19). It is our experience that patients diagnosed with epilepsy have a tendency to be more variable in their responses than do other patient groups. For these reasons, any test that can reliably and rapidly identify central visual defects will be of benefit. In this investigation, a battery of visual tests was applied to a group of epilepsy patients who have received VGB with the primary objective of suggesting an examination strategy that encompasses central visual function. For the purposes of this study, central tests refer to those that measure the central 10 degrees of the retina only, thus corresponding to macula and foveal function.
- Top of page
Of the subjects included, four had simple and complex partial seizures, two had simple and complex partial seizures that secondarily generalised, and six had complex partial seizures with secondary generalisation [International League Against Epilepsy (ILAE) classification]. The mean duration of the epilepsy for the group was 26.5 ± 12.3 years. A complete AED history was taken for each patient retrospectively. The maximal daily VGB dose received ranged from 500 to 3,500 mg/day. The average time period over which the drug was administered was 94 ± 36 months. The total cumulative VGB dose ranged from 1,277 to 16,425 g (mean, 5,086 ± 3,245 g). Eight of the 12 patients were receiving VGB at the time of the study. Only two of the subjects had received VGB as monotherapy; for the remaining subjects, concomitant AEDs included CBZ, sodium valproate (VPA), lamotrigine (LTG), phenytoin (PHT), and clobazam (CLB; Table 1).
Table 1. Central-function test results
|No.||AEDs received at time of testing||SWAP||CS||Color defect (Vingry's)||Color discrimination||Overall impression|
|3||CBZ, VPA, VGB||A||A||N||2||**|
|5||CBZ, CLB, LTG||A||N||N||2||**|
|8||CBZ, LTG, GBP, VGB||A||N||N||2||**|
All 12 subjects undertook high-contrast LogMAR visual acuity testing, SWAP 10-2 visual-field analysis, and spatial-contrast sensitivity testing. One subject failed to complete the FM 100-hue test for both eyes, and therefore data for 23 eyes are presented. Of the 12 patients, two were visually symptomatic (patients 10 and 12; see Table 1). In each case, the visually symptomatic patients were abnormal for all three central visual-function tests. The average high-contrast LogMAR visual acuity for the group was 0.0 ± 0.10, which is consistent with 20/20 vision. The average SWAP MD for the group was –3.24 ± 3.23 dB, and of these, 14 eyes of eight patients showed defects (range, –1.62 to –9.46 dB). The group mean total error score for FM 100-hue (expressed as a square root) was 7.423 ± 3.84. The FM 100-hue software classified six eyes of four patients with low colour discrimination, 13 eyes of seven patients with average colour discrimination, and four eyes of two patients with superior color discrimination. Vingry's analysis revealed that nine eyes of five patients were defective, and in each case, the color-discrimination axis was unresolved. The average spatial contrast-sensitivity values (expressed in logarithmic units) for the four spatial frequencies are given in Table 2. A comparison of contrast-sensitivity data with age-stratified normal data (22,23) revealed that 15 eyes of eight patients yielded abnormal contrast-sensitivity function in at least one spatial frequency. Defects were less prominent at the lower spatial frequencies. Table 1 summarises the results for each patient. Of the 12 epilepsy subjects included in this study, with the exception of high-contrast visual acuity, some form of central function deficit was identified in 10. Four patients demonstrated a defect in all tests (Fig. 1), and two patients had normal results in all three tests (Table 1). Of the two patients that were visually symptomatic, central function defects were identified in all tests performed.
Table 2. Contrast-sensitivity results
| ||Contrast sensitivity spatial frequency|
| ||3 cpd||6 cpd||12 cpd||18 cpd|
|Eyes showing abnormality [number (%)]||5 (21%)||4 (17%)||11 (46%)||11 (46%)|
|Patients showing abnormality [number (%)]||3 (25%)||3 (25%)||8 (66%)||7 (58%)|
Figure 1. Example of the contrast sensitivity, short-wavelength automated perimetry and Farnsworth–Munsell 100-hue results for a visually symptomatic vigabatrin-treated epilepsy patient exhibiting abnormalities in all three tests.
Download figure to PowerPoint
The data presented in Table 3 demonstrate the interaction between the FM 100-hue results and SWAP MD (Fig. 2) and FM 100-hue results and contrast sensitivity (Fig. 3A–D) for right and left eyes. In the majority of cases, a decrease in color discrimination correlated with a decrease in contrast sensitivity or SWAP MD. An investigation of VGB cumulative dosage and SWAP MD for right and left eyes separately revealed no significant difference between the two; however, when all eyes were considered together, a trend toward a negative correlation (r = –0.449; p = 0.028) in which SWAP MD decreased (showed more loss) with increasing VGB cumulative dose was apparent (Fig. 4). No other relation was found for VGB cumulative dose or duration with any other parameters tested (p > 0.05).
Table 3. Group-wise correlations between FM100-hue color discrimination and SWAP MD and contrast sensitivity at 3, 6, 12, and 18 cycles per degrees
| ||FM 100-hue color discrimination score|
| ||r Value||p Value|
|CS (3 cpd)||−0.634||−0.862||0.0267||0.0006|
|CS (6 cpd)||−0.831||−0.862||0.0008||0.0428|
|CS (12 cpd)||−0.851||−0.394||0.0004||0.2312|
|CS (18 cpd)||−0.192||0.315||0.5492||0.3460|
Figure 2. Scatterplot showing short-wavelength automated perimetry mean defect as a function of Farnsworth–Munsell 100-hue total error score. Right eyes are represented by circles, and left eyes, by triangles.
Download figure to PowerPoint
Figure 4. Scatterplot showing short-wavelength automated perimetry mean deviation against vigabatrin cumulative dosage. Right eyes are represented by circles, and left eyes, by triangles.
Download figure to PowerPoint
- Top of page
The battery of tests used in this investigation demonstrate that visual deficits in the central 10 degrees of the visual field are present in epilepsy patients who have previously, or who are currently, undergoing treatment with VGB. The majority of studies assessing visual function in VGB-treated epilepsy patients have focused on the peripheral retina, and consequently the results from investigations that focus on the central retina are of particular interest. Several clinical methods were used in this study including high-contrast LogMAR visual acuity, spatial contrast sensitivity, SWAP, and color discrimination with the FM 100-hue. Each technique offered unique information about the central retina that will serve to improve on the existing knowledge of epilepsy patients treated with AEDs. The modest sample size for this study was not dissimilar to that of previous studies on similar patient groups (2,5,24) and reflects the limited number of epilepsy patients that have been treated with VGB, are willing to participate in the study, and fulfil the study criteria.
Spatial contrast sensitivity
Contrast sensitivity is increasingly being used as a routine clinical tool (25), and the CSV-1000 provides an elegant method to determine contrast-sensitivity levels rapidly in even the most clinically demanding of patients. The test–retest variability of CSV-1000 has been reported as favorable (22), and the technique has proved sensitive enough to detect subtle differences in contrast sensitivity after refractive surgery (26), treatment changes in glaucoma (22), and retinopathy levels in diabetes (27).
In this study, 15 eyes of eight subjects exhibited an abnormality in contrast-sensitivity function, predominantly at higher frequencies. Nousiainen et al. (17) investigated contrast sensitivity by using a different method and reported no significant correlation with VGB drug duration or cumulative dosage, a finding that is in agreement with this study. Interestingly, Nousiainen et al. noted a correlation between contrast sensitivity and white–white peripheral visual field loss. Peripheral visual field testing was not performed in this study; however, no correlation was apparent between 10-2 SWAP global indices and contrast-sensitivity measures. This incongruity may have occurred for several reasons. First, the anatomic locations that are tested with the two techniques differ; contrast sensitivity tests at the fovea, whereas 10-2 SWAP tests extrafoveally in the macula region. Second, SWAP preferentially stimulates the parvocellular short-wavelength–sensitive (SWS) pathway, unlike the standard white–white perimetric test used by Nousiainen et al. (17). Finally, there is a difference in the spatial-frequency content of the blue stimulus compared with the white stimulus, which is smaller. An investigation of FM 100-hue color discrimination and spatial-contrast sensitivity, which are both tests of foveal function, revealed a highly significant relation between the two; a decrease in contrast correlated with an increase in the color-discrimination error score for most spatial frequencies (Fig. 3A–D).
The FM 100-hue test provides a relatively simple method for the assessment of chromatic function at the fovea. This test not only enables the separation of individuals with normal color vision into classes of superior, average, and low color discrimination, but also allows the measurement of the zones of color confusion in color-defective persons. In total, six eyes of four patients exhibited low color discrimination, and nine eyes of five patients were classified as defective by using Vingry's analysis. In each case, the zone of color confusion was unresolved. This indicates a diffuse or generalised loss that is consistent with toxic damage affecting all chromatic pathways equally.
An investigation of VGB drug history and color deficiency failed to identify any significant correlations between the FM 100-hue test results and VGB cumulative dosage or duration of treatment; a finding concordant with a previous study (18). Nousiainen et al. (18) investigated two epilepsy patient groups undergoing CBZ and VGB monotherapy and suggested that both AEDs impair color discrimination. Moreover, Bayer et al. (28) reported reduced colour discrimination in both CBZ- and PHT-treated patients.
Only two of the epilepsy patients included in this study had received VGB as monotherapy. Clinically, VGB is more commonly used as a second-line therapy, and it is unusual to find patients receiving it in isolation. Of the remaining subjects in this investigation, eight had received CBZ as a concomitant drug at some point in their history. It is likely that CBZ, in combination with VGB, had an effect on color discrimination, and this interaction may explain why the color-discrimination error scores did not correlate with VGB cumulative dosage and treatment duration. A tentative exploration of the effect of CBZ cumulative dosage and duration on color discrimination showed no significant correlations, suggesting that the relation between color discrimination and AED drug therapy is a complex one. Previous studies have found VGB-related color defects to predominate in the tritanoptic axis, whereas CBZ-related color deficiencies are more random (18). A possible combined effect of CBZ, VGB, and PHT on color discrimination may explain why the zones of confusion were so disparate and unresolved.
Our results showed that the FM 100-hue color discrimination error score correlated with SWAP central visual field mean deviation, a finding that was independent of drug dosage and duration. This is similar to previous reports in which FM 100-hue error score correlated with the extent of peripheral white–white visual field loss (18).
Short-wavelength automated perimetry (SWAP)
This study was the first to carry out SWAP 10-2 visual field testing on VGB-treated epilepsy patients. In total, 14 eyes of the eight patients tested yielded an abnormal result. In this study, an investigation of right and left eyes separately did not reveal a significant correlation between SWAP mean deviation and VGB cumulative dosage; however, when the eyes were considered together, a trend was apparent (Fig. 4). Although some previous reports have identified a dose-dependent relation between VGB and visual-field defects (29,30), this has not been a universal finding (2,4). No correlation was apparent between SWAP pattern standard deviation and VGB cumulative dose or duration, suggesting, as expected, that a diffuse change in the visual field is more readily related to diffuse drug toxicity than is localised damage.
The FASTPAC algorithm (Humphrey Instruments) was used as the threshold algorithm in this study because it yields greater staircase efficiency than the standard 4-2dB strategy in SWAP (31), and it is our experience that patients with epilepsy fatigue more readily than do other patient groups. The between-subject variability is greater for SWAP than for standard perimetry (31); consequently, the confidence intervals delineating normality in the visual field are wider than those for white–white perimetry, potentially hindering statistical interpretation. Perimetric research has identified that central visual-field locations demonstrate less between-subject variability than do peripheral locations (19,31); thus SWAP is likely to be statistically more sensitive when measured in the central 10 degrees compared with more peripheral locations.
A recent study by Daneshvar et al. (2), identified SWAP 30-2 defects in eight of the nine VGB-treated epilepsy patients tested. Moreover, in both glaucoma and diabetic retinopathy, SWAP has enabled the detection of subtle defects before white–white perimetry (32,33). Defects of the central 10-degree field have not previously been reported from 30-2 visual field testing in VGB-treated epilepsy patients. In other diseases, SWAP is known to be more sensitive to early visual defects, and although white–white visual fields were not included in this study, our findings of SWAP defects in the central 10 degrees imply a preliminary finding that this technique may be more sensitive for the characterisation of central defects in epilepsy, which have not been evident on the central white–white field.
SWAP differs from conventional perimetry by preferentially stimulating the parvocellular SWS pathway. A number of theories attempt to explain why SWAP allows the detection of defects before white–white perimetry. One theory proposed, known as the fragile receptor hypothesis, states that blue cones are in some way more susceptible to damage by light, chemicals, or retinal diseases (34). Quigley et al. (35) showed that in glaucoma, larger-diameter nerve fibres are selectively damaged, and the SWS pathway is mediated by parvocellular fibres that are larger in diameter than those serving the medium-wavelength sensitive (MWS) and long-wavelength sensitive (LWS) pathways. Nousiainen et al. (18) reported that tritanoptic color deficiencies occur in patients treated with VGB monotherapy, which is suggestive of a selective loss of SWS cones. It is known that GABA has a modulatory role in photoreceptor signal transduction to ganglion cells, and Daneshvar et al. (2) suggested that VGB-associated visual-field loss may be attributable to enhanced GABA-transaminase inhibition of the rod bipolar cells. Anatomically SWS cones are more similar to rods than they are to MWS and LWS cones (36), and this may offer an explanation of why the SWS cones are more susceptible to VGB damage over MWS and LWS cones. A further hypothesis called the reduced redundancy, or undersampling, theory assumes that all chromatic pathways are equally damaged in a disease process and that because of the relative paucity of blue cones, their functional response is impaired earlier than other pathways (37). This theory could also be consistent with our findings.
There is some controversy regarding the retinal cell types that are damaged in VGB drug toxicity. Through the inactivation of GABA-transaminase, VGB causes an increase in cerebral and retinal GABA levels, and there is evidence to suggest a greater inactivation in the retina (38). GABA is an inhibitory neurotransmitter that has been identified in numerous subpopulations of amacrine, horizontal, and retinal ganglion cells (39–42). Abnormal ERG b-wave responses have been attributed to Müller cell dysfunction (2), and indeed, the Müller cell has been repeatedly implicated as a primary site for VGB damage. Daneshvar et al. (2) proposed the built-in redundancy theory for the central retina, which suggests that the higher density of Müller cells located in the central retina provides a protective effect when compared with the less populated peripheral retina (2). Although the main functional effects of GABA are concentrated in the inner retina, this neurotransmitter also has been identified in cone synaptogenesis in a rabbit model (43), and a number of electrodiagnostic studies point toward a dysfunction of the cone system through abnormal ERG 30-Hz flicker responses (5,13–15). The results from this study, together with those from previous studies, suggest that cones are affected in VGB toxicity in the central retina, and it is conceivable that the reduced redundancy of the SWS pathway is responsible for the ability of SWAP to detect this dysfunction. Moreover, the redundancy theories suggest a multifactorial hypothesis in which the combined redundancy effect resulting from the general paucity of the SWS cones in the central retina, and of Müller cells in the peripheral retina, may be responsible for inducing achromatic visual loss beyond 10 degrees and SWAP defects within the central 10-degree field.
LogMAR visual acuity
A normal high-contrast logMAR visual acuity consistent with 20/20 vision was recorded for the group. The finding of normal high-contrast visual acuity, while central visual deficits exist in color perception, SWAP and low contrast sensitivity, is not unusual and has been demonstrated before in a number of diseases. In human immunodeficiency virus–positive patients without retinopathy, even when visual acuity is 20/20 or better, central function anomalies have been identified in up to one third of cases when tested with SWAP, FM 100-Hue color vision, or contrast sensitivity (44). Similarly, in early diabetic patients with good visual acuity, low-contrast sensitivity has been shown to be seriously reduced (23). The data from this study, in conjunction with these reports, confirms that conventional high-contrast visual acuity assessment alone is insufficient and that a series of visual tests is necessary to assess central visual function.
The results from this study confirm that central visual function deficits exist in epilepsy patients who have been treated with VGB, and are suggestive of a diffuse retinal drug toxicity in which visual function is affected centrally as well as peripherally. The majority of patients investigated were receiving concomitant AEDs, and it is likely that the combined effect of several AEDs, particularly VGB and CBZ, is responsible for some of these findings. FM 100-hue testing yielded unresolved abnormal color-discrimination defects, which were most likely the result of complex drug interactions. The lengthy test time, combined with the disparity of the defects, suggests that this test is of limited use for monitoring the effect of VGB in patients treated with several AEDs. Measures of LogMAR high-contrast visual acuity were normal overall, similarly suggesting that this test in isolation is insufficient as a measure of central visual function. SWAP 10-2 provided an indication of visual defects in the macular region of some patients, and spatial contrast sensitivity yielded a rapid technique for the measurement of foveal function.
Although the examination of peripheral vision is important, central vision is arguably more relevant in the day-to-day life of the patient and must not be ignored. The current ophthalmic testing regimen used for VGB-treated patients uses techniques that simultaneously assess all visual pathways. If subtle damage has occurred either specifically, or more readily, in a given visual pathway, it is possible that the change may be masked by the responses of the other pathways. Testing strategies that are tuned selectively to evaluate particular visual pathways are likely to be rendered more sensitive, especially for the central retina, where defects are more difficult to detect. The authors recommend that central vision in VGB-treated patients should be investigated by using SWAP 10-2 and spatial-contrast sensitivity.