- Top of page
- MATERIALS AND METHODS
Summary: Purpose: To study the prevalence and features of visual field constrictions (VFCs) associated with vigabatrin (VGB) in children.
Methods: A systematic collection of all children with any history of VGB treatment in fifteen Finnish neuropediatric units was performed, and children were included after being able to cooperate reliably in repeated visual field tests by Goldmann kinetic perimetry. This inclusion criterion yielded 91 children (45 boys; 46 girls) between ages 5.6 and 17.9 years. Visual field extent <70 degrees in the temporal meridian was considered abnormal VFC.
Results: There was a notable variation in visual field extents between successive test sessions and between different individuals. VFCs <70 degrees were found in repeated test sessions in 17 (18.7%) of 91 children. There was no difference in the ages at the study, the ages at the beginning of treatment, the total duration of the treatment, general cognitive performance, or neuroradiologic findings between the patients with normal visual fields and those with VFC, but the patients with VFC had received a higher total dose of VGB. In linear regression analysis, there were statistically significant inverse correlations between the temporal extent of the visual fields and the total dose and the duration of VGB treatment. The shortest duration of VGB treatment associated with VFC was 15 months, and the lowest total dose 914 g.
Conclusions. Because of a wide variation in normal visual-field test results in children, the prevalence figures of VFCs are highly dependent on the definition of normality. Although our results confirm the previous findings that VFC may occur in children treated with VGB, our study points out the need to reevaluate critically any suspected VFC to avoid misdiagnosis. Nevertheless, our study suggests that the prevalence of VFC may be lower in children than in adults, and that the cumulative dose of VGB or length of VGB therapy may add to the personal predisposition for developing VFC.
Vigabatrin (VGB) is an antiepileptic drug (AED) that mediates its antiepileptic effect by irreversible inhibition of brain γ-aminobutyric acid (GABA) transaminase (1). VGB has been widely accepted as the first-line treatment for infantile spasms, especially secondary to tuberous sclerosis, and as an add-on therapy for medically resistant localization-related childhood epilepsies (1–4). Since its introduction into clinical use a decade ago, VGB has been widely appreciated for its relatively rare side effects and lack of metabolic interactions (1). Animal toxicity studies suggested a risk of intramyelinic edema and vacuolization (5,6), which have, however, not been found in humans or other primates treated with VGB (6,7). Conversely, retinal degeneration due to VGB treatment in rats also has been reported (8). Although VGB was recently shown to have a high tendency to accumulate in the retina (9), no studies have been published about histologic changes in human retina after VGB treatment.
Recently, however, an increasing number of VGB-treated patients with concentric visual field constrictions (VFCs) have been reported (10–18). Almost all of the patients with VFC reported so far have been adults. The data gathered suggests that VFC may occur, depending on criteria, in ∼20–50% of VGB-treated patients, and it is currently considered to be irreversible (19). Despite the increasing number of reports about patients with VGB-associated VFC, relatively little is known about the mechanism underlying VFC or about the association between therapeutic paradigms and VFC.
In pediatric patients, the duration of the AED therapy and the drug dose relative to body weight may differ considerably from those in adults, and the maturing nervous system may respond to toxic substances in a very different manner. However, only a handful of pediatric patients with VFCs have been reported, and there are only some small patient series of children published to address the relation between detailed features of VGB treatment and the occurrence of VFC (20–25). A major problem in estimating the prevalence of VFC in pediatric patients is a lack of ability to cooperate reliably in visual field testing due to young age or developmental disability. Although the previous reports suggested that VFC does occur in some VGB-treated children, these studies possessed several practical problems: all study groups have been small, methods of visual field testing have varied even within one study (21,22,24), normality has not been appropriately defined (26) or not defined at all (21–24), and abnormal results have not been reevaluated (21–23). All these aspects hamper the possibility of drawing conclusions that would be clinically relevant in children (27,28).
The purpose of our study was (a) to estimate the prevalence of VFC in children treated with VGB, (b) to identify potential demographic risk factors for VFC, and (c) to reveal the possible relation between VFC and the detailed features of VGB treatment.
MATERIALS AND METHODS
- Top of page
- MATERIALS AND METHODS
The patients were collected retrospectively and prospectively between March 1999 and July 2000 by the Finnish Study Group, which consisted of at least one child neurologist from all 15 Finnish neuropediatric units outside of Helsinki district. Virtually all children and adolescents with epilepsy in Finland are regularly (2–4 times a year) seen by a child neurologist. At the time of the visit, the patient's current and previous AED treatments were reviewed, and all patients ever treated with VGB were considered for inclusion into this study. Some data of seven patients were previously published (20,25). This collection method could not identify the patients (a) whose epilepsy is not active (i.e., no AED and thus no regular follow-up); (b) whose follow-up had been transferred from child to adult neurologists because of the age at the time of the study; and (c) who lived in institutions for persons with mental retardation. We believe, however, that the patients in the last group represent only a minor portion of the VGB-treated pediatric patients who could have otherwise been included (i.e., have adequate cooperation for visual field testing). We also believe that absence of two former patient groups did not confound the results, because they are very unlikely to represent any special group with a suspected higher risk for VFC.
Inclusion criteria were (a) age younger than 18 years, (b) previous or current use of VGB, and (c) sufficient ability to cooperate in visual field testing. One hundred six children met these inclusion criteria; they were examined by their own ophthalmologists, and altogether 177 visual field test sessions by Goldmann perimetry were performed. Fifteen children were later excluded for the following reasons: hemianopia (four because of focal lesions), failure in cooperation (nine), temporal optic atrophy (one), and other brain anomalies (congenital hydrocephalus and diplegia) complicating interpretation of the visual field results (one). Finally, 91 children were included in the study group after they had successfully and reliably undergone at least one Goldmann perimetry (Table 1). Thirty-nine (43%) of these children had abnormal findings in neuroradiology, but they were considered not to influence their visual fields or functioning because they were outside of the brain areas involved in visual processing. The epilepsy (29) was cryptogenic or idiopathic (n = 67), due to known focal lesions (n = 18; tuberous sclerosis, infarction, migration disorders, etc.), or due to pre-/perinatal problems (n = 6). Most (90%) of the children had localization-related epilepsy, with either partial (n = 28) or secondarily generalized (n = 54) seizures. In only one (1%) child, epilepsy had been diagnosed as primarily generalized, and in eight (9%) cases, the type of epilepsy was undetermined whether localization related or generalized.
Table 1. Demographic data of the patients in this study
| ||All patients||Normal visual field||Visual field constriction|
|Total number (% of all)||91 (100%)||74 (81.3%)||17 (18.7%)|
| Mean age (yr) (range)||12.1 (5.6–17.9)||12.0 (5.6–17.9)||12.4 (8.3–16.7)|
|Intelligence, normal/subnormal/slightly retarded||69:19:3||55:17:2||14:2:1|
|Neuroradiologic abnormalities, no/yes||52:39||43:31||9:8|
|Stimulus in the perimetry, I/4:II/4:III/4:IV/4:V/4||5:29:2:40:15||4:26:2:30:12||1:3:0:10:3|
All children were first studied by their own ophthalmologist, and none of them complained about visual symptoms or problems suggesting visual field narrowing. Nine of the patients had other ophthalmologic abnormalities in clinical examination, i.e., strabismus (n = 4), hamartoma retinae (n = 1), esophoria (n = 2), and astigmatism (n = 2). All patients had a normal or near-normal visual acuity, and all but the one with hamartoma retinae had normal funduscopy. These other ophthalmologic findings were not considered to affect patients' performance in visual field testing.
Peripheral visual fields were examined with a kinetic Goldmann perimeter with standard objects (IV/4, n = 40; II/4, n = 29; V/4, n = 15; I/4, n = 5; III/4, n = 2) by a trained and skilled perimetrist in child's own hospital clinic. To minimize interhospital variability, all visual field charts were later reevaluated by the same second ophthalmologist (I.N.). The visual field test was judged reliable if the child had good fixation and concentration during the test session, positive judgment of cooperative capacity (and exclusion of likely confounding psychiatric problems) by the treating neurologist, and reliable-appearing visual field charts as assessed by the second ophthalmologist (I.N.). Abnormal visual fields with suspected VFC were reexamined to exclude false-positive findings. In one case, however, the first abnormal result was considered reliable because of the patient's very good cooperation and a clear constriction in his visual fields (male, 9.2 years, normal intelligence; temporal field, 55 degrees; nasal field, 40 degrees).
We decided to use the manual kinetic perimetry instead of automated perimetry for four reasons: (a) changes in the temporal field are greater and, in children, thus likely easier to detect than those in the nasal field, (b) slight to moderate temporal field losses are not found with an automated perimetry, (c) manual perimetry is often easier for the child (although more difficult for the examiner), and (d) child's cooperation and thus reliability of the test result is easier to assess in the face-to-face setting of the Goldmann test.
Normality and classification of visual field test results
Normality of the visual fields in children has not been unambiguously established, and the determination of the limits for normal visual field extent is challenged by three main factors: (a) the level of attention and/or cooperation varies easily; (b) children often show a learning effect (i.e., visual fields grow in successive test sessions); and (c) for unknown reasons, visual fields have been shown to widen during childhood and adolescence (30,31). The peripheral extent of visual fields also depends somewhat on the stimulus paradigm. We defined normality as being >70 degrees in the temporal meridian. This is supported by the previously published studies (30,31) and our own experience with otherwise healthy children (unpublished observation). In addition, this limit was recently used also in studies about VGB-associated VFC in adult patients (10,11), which enables us, to some extent, to compare results between children and adults. It is noteworthy that possible milder visual field defects are not found when using our criteria. However, the even normally seen marked variation in children's visual fields (30–32) makes it almost implausible to attempt detection of subtle changes in the peripheral visual field. To overcome this unavoidable problem, we evaluated the effect of changing the limits of normality to the prevalence of VFC (Table 3).
Table 3. The proportion of children classified as abnormal when the limits for normality are set at different degrees in temporal meridian
|Alternative limit for normality (degrees)||Number of patients (normal/abnormal)||Prevalence of VFC (%)|
Our classification of the severity of VFC was adopted from the previous studies performed with adult VGB-treated patients, and the visual fields were classified into three categories (10,11): (a) normal, if the fields extended >70 degrees in the temporal meridian; (b) mildly abnormal, if the fields extended from 50 to 70 degrees in the temporal meridian; and (c) severely abnormal, if the fields extended <50 degrees in the temporal meridian.
Student's t test was used to evaluate the significance of the differences of continuous variables, and the χ2 test or Fisher's exact test for categoric variables between the patients with normal fields and those with VFC. Parameters related to the VGB therapy between the groups of patients with normal visual fields and VFC were initially compared with Student's t test. Correlation between the visual field extent and the clinical variables (age, total dose, duration of treatment) were analyzed by linear regression analysis (Analyse-It, Software Ltd, Leeds, U.K.).
This study was approved by the Ethics Committee of the University Hospital of Helsinki. To ensure anonymity, all patient records were blindly coded so that only the doctor who treated the child was able to identify his or her patients.
- Top of page
- MATERIALS AND METHODS
Seventeen (18.7%) of our 91 patients showed VFC (i.e., visual field <70 degrees in the temporal meridian in kinetic perimetry). Only one of these was classified as severe [i.e., <50 degrees (45 degrees; age 17.9 years; male; VGB treatment, 24 months).
There was no significant difference in the age (p = 0.67), general cognitive performance (p = 0.50), or the prevalence of neuroradiologic findings (p = 0.90) between the patients with normal visual fields and those with VFC (Table 1). The proportion of boys in the VFC group (59%) was slightly albeit not statistically significantly higher (p = 0.56) than their proportion in the whole study group (49%), and the relative risk of VFC in boys was not increased (relative risk, 0.92; CI95%, 0.75–1.12). The distribution of the types of stimuli used in Goldmann perimetry did not differ between the patients with normal visual fields and those with VFC (p = 0.58) (Table 1).
Comparison between the patients with normal visual fields and those with VFC showed that the distributions of the duration of VGB treatment or the total dose of VGB were overlapping. The total dose of VGB was significantly higher in the patients with VFC, whereas the age and the total duration of treatment did not reach statistical significance (Table 2).
Table 2. Drug treatment variables of the patients with normal visual fields and those with visual field constrictions
| ||Normal visual field (n = 74)||Visual field constriction (n = 17)|
|Age at beginning of treatment|| || |
| (yr) (95% CI)||9.2||8.4|
| p = 0.19||2.0–15.7||2.6–14.7|
|Total duration of treatment|| || |
| (mo) (95% CI)||33.5||46.1|
| p = 0.050||1.2–90||15–116|
|Total dose (g) (95% CI)||1679||2,552|
| p = 0.018||43–6,770||914–4,800|
Among the patients with VFC, the lowest total dose of VGB was 914 g, and the shortest total duration of VGB treatment was 15 months (boy, 8.5 years; VGB treatment, 29 months). In the group with normal visual fields, 14 children had received VGB for <15 months, and 27 children received a total dose of <914 g. Normal visual fields also were found in patients who received a very high total dose of VGB.
In the whole study group, there was no correlation by linear regression analysis between the age at the last visual field test and the visual field extent (p = 0.58; Fig. 1a). This enabled us to continue the analysis independent of patient's age. The visual field extent showed a slight, statistically significant inverse correlation with both the total dose of VGB (R2 = 0.10; p = 0.002) and the total duration of VGB treatment (R2 = 0.10; p = 0.003; Fig. 2). No correlation was seen between the age at beginning of the treatment and the extent of the visual field (p = 0.28; Fig. 1b).
Figure 1. Correlation between the age of the patient at the last Goldmann test (a) or at the beginning of vigabatrin therapy (b) and the temporal visual field extents by linear regression.
Figure 2. Correlation between the duration of the vigabatrin (VGB) therapy (a) or the total dose of VGB received (b) and the temporal visual field extents by linear regression.
Visual fields of 42 children were tested 2 to 4 times. There was a wide intrasubject variation between successive test sessions, and children often performed better (i.e., learning effect) in later test sessions (Fig. 3a and b). Other practical variables of the test sessions like variation in child's cooperation or attention probably had an effect on the test result. Nevertheless, this follow-up does not support the previously hypothesized progressively worsening of VGB-associated VFC.
Figure 3. Individual variation in successive visual field test session. a: Successive visual fields in 40 patients. Visual field extent in temporal meridian is plotted against the duration of vigabatrin (VGB) therapy at the time of performing the test. b: Three successive visual field charts in a 17-year-old girl. The tests were performed after (1) 6 months/340 g or (2) 39 months/2,420 g of VGB treatment, and (3) 3 months after cessation of VGB treatment (total VGB treatment, 44 months/2,670 g).
Correlation by linear regression analysis between the visual field extent in the temporal and nasal meridians was clearly present (R2 = 0.39; p < 0.0001; Fig. 4). However, visual field extent in nasal meridian did not correlate with the age at beginning of the therapy (p = 0.27), with the current age (p = 0.66), with the total duration of treatment (p = 0.20), or with the total dose of VGB administered (p = 0.23).
Figure 4. Correlation between the visual field extent in temporal and nasal meridians by linear regression analysis.
Download figure to PowerPoint
Finally, we evaluated the effect of changing the limits of normality on the prevalence of the abnormal visual field extent. As might be expected, the proportion of visual fields classified as abnormal increased (or decreased) significantly when the limit was set higher (or lower) than 70 degrees (Table 3).
- Top of page
- MATERIALS AND METHODS
There exists a controversy about whether VFC should be monitored by automated or kinetic perimetry. Automated perimetries are good at detecting a decrease in visual sensitivity in central areas. Because of their limited spatial extent, however, they do not reveal mild to moderate visual field losses in the temporal fields, which are more reliably demonstrated by the kinetic (Goldmann) perimetry. Some authors have emphasized the loss of nasal fields as a good indicator of VGB-associated VFC (33), whereas other studies have shown that temporal fields outside of the limits of automated perimetry are also regularly affected (10–13,17). We found a clear correlation between the temporal and nasal extent of visual fields, which supports the idea that defects develop in both horizontal sides, leading to concentric constriction of the visual fields. However, we could not show any correlation between clinical parameters and the extent of nasal visual field, which is likely due to the relatively slight changes observed in nasal meridian as compared with those found in the temporal meridian. It is theoretically possible that the nasal field defect could have been better detected by automated static perimetry. Practically thinking, however, only temporal field loss is clinically relevant for (especially) outdoor activities (e.g., car driving). The peripheral retina is specialized for detection of moving objects, which is ideally tested by kinetic perimetry, and a recent study reported Goldmann kinetic perimetry to be more sensitive in detection of the peripheral temporal defects (34). Thus we believe that, despite of being tedious and more difficult for the examiner, Goldmann perimetry is more reliable and better reveals a clinically meaningful VFC in children.
In our study a bilateral constriction in the peripheral visual field was found in 17 (18.7%) of our 91 children, and in only one (1.1%) patient was this defect classified as severe. This indicates that children and adolescents also are susceptible to the same type of VFC that has been described in adult patients treated with VGB. The prevalence of severe VFC, which would cause problems in daily living, is likely much lower in children than in adults. Our study, however, cannot exclude the possibility of higher prevalence of still very subtle, slowly progressing visual field defects, which also may have practical consequences in later life in cases of continuing VGB medication.
Although our observations confirm the other recent reports on VFC in children (20–24), the prevalence in our study was much lower than that in any previous study. This might be partly explained by differences between demographics of the study populations. However, it is notable that many of the previous studies have used only automated static perimetry instead of manual kinetic perimetry, which was consistently used in this study. In addition, criteria for normality (21,23,26) or methods of visual field testing (21,22) have not been individually detailed in all of the previous studies. Changing the limit of normality does, obviously, dramatically alter the prevalence of VFC, which highlights the importance of defining reasonable limits for normality. These differences in the study designs essentially hamper possibilities of comparing the present results with those of the previous studies. In addition, as compared with the previously published reports on pediatric patients, our study consists of a many times larger patient group; it presents an essentially unselected population, and it contains a partial follow-up of the patients. Thus we believe that this study is the first to elucidate reliably the prevalence of VGB-associated VFC and its possible associations with demographic and clinical variables in children. Because the age at beginning of the treatment did not correlate with the loss of visual field in our study group (all patients older than 2.0 years), our results likely apply similarly to all pediatric patients older than 2 years.
Our findings are in accordance with the previous studies on adult patients (10,11), in that there were no significant differences in the mean age, intelligence, or neuroradiologic findings between the patients with normal visual fields and those with VFC, whereas boys seemed to be slightly higher represented among the patients with VFC. As to the parameters of VGB treatment, VFC was associated with a higher total dose of VGB, and the visual field extent was inversely correlated with a longer treatment or a higher total dose of VGB. Some previous studies in adults have demonstrated that longer treatments (12,17) or a higher total dose of VGB (35) is associated with a greater loss in visual fields, whereas most of the previous studies have not found such a correlation (10,11). The lack of statistically significant correlations in many previous studies may be due to the relatively small size of the patient series. Nevertheless, our findings suggest that the risk of developing VFC may be cumulative, and it increases as a function of total dose of VGB or duration of VGB treatment. The presence of a number of normal patients with high total dose or very long VGB treatments suggests that the development of VFC also requires some as-yet-unidentified individual predisposing factor(s).
In our study group, none of the patients with VFC had used VGB for a shorter period than 15 months or received a total dose <914 g. If VFC <70 degrees was to occur with a significantly shorter VGB treatment or with a lower total dose, we assume that it would have been observed in our study group, which consisted of a substantial number of patients with short treatments and low total doses. Whereas these observations suggest that development of VFC <70 degrees in temporal meridian may require the given dose and time, we cannot exclude the possibility of very mild, emerging VFC already with lower doses or shorter treatments.
Our follow-up of 40 children revealed a considerable variation between successive visual field test sessions, and many of these children showed a temporary reduction in their visual field extent. Indeed, the visual field test in eight children was classified as showing mild VFC in one test session, but they performed normally in later visual field tests. Because VGB-associated VFC is generally considered irreversible, we assume that the occasional poor performance was due to factors other than VGB-associated retinal dysfunction. These patient cases raise the concern that there may be children who have erroneously been diagnosed as having VFC because of inadequate performance during a visual field test session. Indeed, confirmation of visual field test results has not been systematically performed in previous studies in children, and this may essentially explain the markedly higher prevalence of narrowed visual fields reported in those studies. Although the VFC warrants great caution when using VGB, it also is very important to avoid over- or misdiagnosis. Thus we suggest that all suspected VFC should be carefully retested.
Conversely, the possibility of VFC reversal cannot be solely excluded in those of our patients who had better visual field test results after VGB discontinuation. Recent case reports have, indeed, raised this possibility (25,36,37), and prospective follow-up studies are urgently needed to show whether recovery is a real hope for patients with VFC. Theoretically, it could be possible that the high plasticity of developing nervous system, including the retinal network, might compensate for the retinal toxicity of VGB, and thereby lead to a functional recovery in constricted visual fields. However, our observations of improved visual field test results may as well be explained by alternating performance in successive visual field test sessions.
There are two relative disadvantages in our study.
Our determination of normality in visual field extent was based on the previously published data and on our own unpublished experience instead of collecting a separate control group. We do not believe that any permanent finding was obscured by the lack of such a formal control group. We believe that such a single, selected control group would not provide more relevant control figures than those collected from many previous publications. A formal control group might have solved the problem of intraindividual variation. We suppose, however, that this variation was due to vigilance factors, and we assume that child's best test result represents his or her true field of vision. Finally, the main aim in our study was to find out how the patients with VFC using VGB differ from the patients with normal visual fields, and this question does not need a control group to be answered.
For practical reasons, all children could not be tested by the same ophthalmologist. The result of the manual perimetry is somewhat dependent on the experience of the examiner, and this is especially important in testing children whose cooperation may vary even during a single test session. An absolute comparability between test results from different medical centers is difficult to ensure. However, we had all patient charts reevaluated by the same second ophthalmologist (I.N.), and visual fields were reexamined in all cases with suspected reliability. This critical approach led to the exclusion of 10 patients whose visual fields were preliminarily judged abnormal. They were excluded because either their performance throughout the test sessions was not stabile or clinical observation of their use of vision did not corroborate the perimetry findings. Some patients also had one abnormal field result, which was later “normalized” when cooperation and attention were more carefully monitored. We thus believe that the cases reported as abnormal in our study are truly those with constricted visual fields. Therefore the prevalence figures, despite being much lower than those in the previous studies, better reflect the true order of prevalence of VFC in children and adolescents treated with VGB.
A clinically important group of patients with regard to VGB treatment are those with infantile spasms (West syndrome) (38), in which VGB is widely considered the drug of choice because of its mild side-effect profile and equal clinical efficacy as compared with adrenocorticotropic hormone (ACTH) treatment (39). There is at present no clinical evidence either supporting or speaking against the risk of VFC associated with neonatal VGB treatment. This lack of any evidence strongly emphasizes the urgent need to document those VGB-treated children with infantile spasms who would be capable of undergoing ophthalmologic evaluation. We did not find VFC <70 degrees after short periods or low total doses of VGB in older children. Theoretically, these observations raise the hope that short treatments might not cause severe defects to the visual fields of children with infantile spasms. However, it also is theoretically possible that VFC in our children had developed earlier, and we detected it only later at the time of the first visual field test. This possibility should be studied with prospective follow-up of patients with newly started VGB therapy. The nervous system is at its highest plasticity during the neonatal and infant periods. In addition, the mechanisms of GABAergic system and retinal pharmacokinetics of VGB, both of which have been proposed as potential mechanisms for retinal toxicity of VGB (9), are likely considerably different from those in adults (40–42). Because of these poorly understood factors, it is very difficult to make any reasonable assumptions about the effects of VGB on neonatal brain based on data from adult patients. Indeed, recent experimental evidence has suggested that VGB toxicity may differ between immature and mature nervous systems (5).
In practice, child neurologists have to make therapeutic decisions relying solely on the results obtained by child's own consulting ophthalmologist. Our experience shows that the visual field test results may vary considerably, and care should be taken to avoid artifactual VFC findings. This is very important for clinical decision making and for obtaining relevant information for patients and/or parents. The majority (our rough estimate is ∼80%) of VGB-treated pediatric patients cannot reliably undergo visual field testing because of their mental retardation, attentional problems, or young age. Recently, Harding et al. (43,44) introduced a potential screening test based on a cone flicker response in the electroretinogram (ERG), and a number of abnormalities in electrophysiologic tests have been reported in association with VFC (13–16,45). Abnormal findings in the ERG are not, however, consistently related to visual dysfunction in children (21,33). Many of our own patients also had been examined with various electrophysiologic methods (unpublished data). Our feeling is, however, that at least the pediatric patients with epilepsy often have many other neurologic and ophthalmologic problems that seriously confound studies on the association between VFC and electrophysiologic test results. Although it would be invaluable for many patients, there is as yet no established electrophysiologic method that could reliably reveal VFC in patients with inadequate cooperation for the perimetry test.
In conclusion, our results suggest that VFC occurs in children treated with VGB, whereas the prevalence may be much lower than that reported in adults (10,11). Causality between VGB therapy and the VFC observed in a given patient can be confirmed only after having obtained a normal visual field test before starting VGB therapy. Regular follow-up of the visual fields of patients receiving VGB is mandatory now. Reliable testing of visual fields in children is difficult and often impossible because of the high intraindividual variability in all children and the inadequate cooperation in a large proportion of the VGB-treated patients. Before the introduction of more reliable and cooperation-independent methods (e.g., neurophysiologic testing), we want to emphasize strongly the need to reevaluate critically any suspected abnormality of visual fields to avoid misdiagnosis and unnecessary changes in AED therapy. The overall use of VGB should be based on a balanced risk–benefit estimation in every individual case (28,46), and therefore it poses a major challenge for all physicians using VGB to treat their pediatric patients with epilepsy in the future.