J Neurol (2002) 249 : 1066–1071DOI 10.1007/s00415-002-0789-y ORIGINAL COMMUNICATION

T. SchmidtK. RütherB. JokielS. PfeifferK. Tiel-WilckB. Schmitz

Is visual field constriction in epilepsypatients treated with vigabatrin reversible?

Introduction

Vigabatrin (VGB) is a highly effective antiepileptic drugespecially for patients with otherwise intractableseizures [9]. Its wide clinical use has been jeopardized bythe occurrence of visual field narrowing [1]. Based oncross-sectional studies the prevalence of VGB-associ-ated visual field constriction (VFC) has been estimatedto be approximatly 30–40 % in patients treated with VGBas monotherapy or add-on therapy [4, 17].

As known from animal studies VGB acts by increas-ing the GABA-concentration in the CNS. It even more ef-

fectively raises the GABA level in the retina [14] whichmight lead to its retinal toxicity. Although the patho-physiological background of VGB-associated VFC re-mains unclear at present, there is evidence from electro-physiological findings that VGB might effectsecond-order neural cells that connect to cone photore-ceptors [8, 12]. These functional deficits are well corre-lated with the occurrence of VFC based on perimetricalfindings [8]. Little is known about whether VFC due tovigabatrin are reversible after stopping treatment anddata from a cross sectional study indicates that VFCmight be irreversible [3]. To evaluate the risk of VGBtreatment it is important to know whether VFC in theseJO

N 7

89

■ Abstract Objective To evaluatethe reversibility of vigabatrin asso-ciated visual field constriction.Background Visual field constric-tion (VFC) occurs in approxi-mately 40 % of epilepsy patients

Received: 14 December 2001Received in revised form: 14 February 2002Accepted: 20 February 2002

T. Schmidt (�) · B. Jokiel · S. Pfeiffer · K. Tiel-Wilck, MD · B. Schmitz, MDCharité, Campus-Virchow-KlinikumDepartment of NeurologyAugustenburger Platz 113353 Berlin, GermanyTel.: +49-30/4 50-6 00 38Fax: +49-30/4 50-6 09 01E-Mail: tamara.schmidt@charite.de

K. Rüther, MDDepartment of OphthalmologyUniversitätsklinikum EppendorfHamburg, Germany

under treatment with vigabatrin(VGB). There is still controversyabout whether VGB-associatedVFC is reversible. From a cross-sec-tional study there is evidence thatVFC does not reverse three to sixmonths after stopping VGB treat-ment. So far, there are no long termstudies on this subject. Methods Weperformed a follow-up study on 15epilepsy patients (eight women,seven men, median age 45 (21–58)years) with VGB-associated VFCbut otherwise normal ophthalmo-logical examination. Kinetic andstatic perimetry was performedone and two years after VFC wasdiagnosed (baseline examination).Visual field size at first and at sec-ond year-follow-up was comparedwith the baseline examination. Be-cause discontinuation of VGB-treatment was dependant on clini-cal needs, patients either stopped

VGB treatment before or after VFCwas diagnosed. In a small group ofpatients VGB treatment was contin-ued despite of VFC. Results Therewas no statistically significant dif-ference in visual field size compar-ing baseline values with first yearand second year follow-up exami-nations either in patients whostopped VGB treatment (n = 11) orin patients who continued VGBtreatment on a reduced dosage(n = 4). Conclusion Although ourdata are based on a relatively smallgroup of patients there is evidencethat VFC resulting from VGB treat-ment is not reversible in epilepsypatients after stopping the drug.

■ Key words Visual FieldConstriction · reversibility ·epilepsy · antiepileptic drugs ·vigabatrin · adverse events

1067

patients will reverse.We performed a follow-up study onpatients with VGB-associated VFC with the aim to eval-uate the reversibility of VFC after VGB-treatment wasstopped.

Subjects and methods

We performed a follow-up study on 15 epilepsy patients (eightwomen, seven men, median age 45 [21–58] years) with VGB-associ-ated VFC. 13 of these 15 patients initially were part of a larger groupof 33 patients who were enrolled in a VGB safety study (1992–1995)and had developed concentric VFC by 1998. Three additional patientswith VGB-associated VFC were recruited from our outpatient clinicafter 1998. All patients had been treated with VGB either as mono-therapy or together with other antiepileptic drugs. The dosage andduration of treatment were variable. Because discontinuation of VGBtreatment was dependant on clinical needs, patients either stoppedVGB treatment before or shortly after VFC was diagnosed (n = 11). Ina small group of patients (n = 4) VGB treatment was continued despiteof VFC. All patients underwent complete ophthalmological examina-tion including static and kinetic perimetry when VFC was diagnosed(baseline examination).VFC in these cases could not be attributed toother ophthalmological disorders. Patients were perimetrically re-in-vestigated one year after baseline examination. A second follow-upevaluation was possible in seven out of 15 patients, the remainingeight patients were not re-examined for the following reasons: threepatients refused to participate, one patient was lost to follow-up andin another four patients the interval to the second year follow-up wasless than two years.

Perimetry

Visual field size was determined by performing kineticand static perimetry. Kinetic perimetry was performedusing the Goldmann perimeter (Model 940,Haag-Streit)with three targets of different size and luminance (III(0.43°) 4e (1000 asb), I (0.11°) 4e (1000 asb), I (0.11°) 2e(100 asb). Background illumination was 31.4 asb. Cali-bration and test procedure were performed after theproposals of Goldmann [2] and Schmidt [13]. Targetsare moved centripetally with a velocity of about 2°/s andare presented randomly within 360°. Fixation of the pa-tient’s eye is observed directly with a telescope.With sta-tic perimetry light sensitivity thresholds were estimatedwithin 30° of central visual fields of each eye using theprogram 30–2 of the Humphrey-Field-Analyser (Aller-gan, type 635). Background illumination was 31.4 asb. Atarget of 4 mm2 size was presented randomly on 76 lo-cations for 200 ms. Light sensitivity thresholds were cal-culated and after age adjustment transformed into aglobal sensitivity index, the mean deviation (STATPAK 1analysis,Allergan Humphrey).The results were analysedin terms of the Statpak statistical software package. Thereliability criteria – namely, the number of fixationlosses, false-negative responses, and false-positive re-sponses – were all within normal limits.

■ Analysis of Visual Fields

Qualitative analysis at baseline: visual fields were quali-tatively analysed by an experienced neuroophthalmolo-gist (K. R) who was blinded with respect to clinical data.When comparing visual fields kinetic perimetry wasused as the superior method with static perimetry re-garded as a confirmatory tool. Changes of visual fieldsbetween pre-treatment evaluation and baseline exami-nation were diagnosed using three categories: un-changed, mild VFC, severe VFC.

The analysis was based on the following criteria:Class 1: no significant change

– VFC of less than 10° (III/4e) and 5° (I/2e) (nasal ortemporal axis)

– any change, if the amount of fluctuation was similarto Class 2: mild changes

– VFC between 10° and 25° (III/4e) and/or 5–10° (I/2e)(nasal or temporal axis)Class 3: severe changes

– VFC more than 25° (III/4e) and/or 10° (I/2e) (nasal ortemporal axis)

Quantitative analysis at baseline and at follow-up: Totest the reliability of the visual analysis at baseline, find-ings of the kinetic perimetry were quantified. Weanalysed visual fields by measuring the radius of theouter isopter (III 4e) in the nasal, the nasal inferior (225°or 315°) and the temporal axis. The findings of both eyeswere summed and divided by two. The differencesbetween baseline and first and second follow-up exami-nation were then analysed. Changes of visual field sizewere analysed for patients who had stopped VGB-treat-ment and were continuing medication respectively.

■ Statistics

Data were analysed using the SPSS-program, version8.01. To compare values of VFC within subjects,Wilcoxon signed rank test for paired samples testingwas used.

Results

Patients were not homogeneous with respect to clinicalparameters, in particular to drug regime. The durationof treatment with VGB and other antiepileptic drugs wasdetermined by clinical needs leading to different indi-vidual strategies. Clinical and drug related data of pa-tients who stopped VGB are shown in Table 1.

When comparing visual field size at one and two yearfollow-up examination to baseline values we did not finda significant difference in visual field size in both pa-tients who stopped VGB treatment (n = 11) and in pa-tients who were still exposed to VGB (n = 4). Table 2–3

1068

show results from static and kinetic perimetry. Analysisof individual subjects showed a variability in visual fieldsize according to individual fluctuations (Fig. 1–2) butno consistent increase or decrease in visual field size insingle patients.

Discussion

Our study shows that there is no significant change ofVGB associated VFC after one and two years follow up inpatients who stopped VGB treatment. The only otherstudy to investigate the reversibility of VGB-associatedVFC compared measurements of visual fields in ninepatients while taking VGB and three to six months afterwithdrawal of the drug [3]. No significant recovery ofVFC was found.VFC was initially diagnosed by compar-ing visual field measurements in patients taking VGBwith patients from a control group of epilepsy patientswho did not take VGB. In our study VFC was identifiedby comparing visual field size in VGB users before andafter taking the drug. However, beyond these differencesboth studies show that visual field size did not signifi-cantly change in epilepsy patients after VGB treatmentwas stopped.

Only anecdotal account of reversibility of VFC hasbeen given so far. Krämer [6] reported a case of a 19-year-old woman with severe VGB-associated VFC inwhom an early withdrawal of VGB five weeks after theinitiation of the therapy led to complete recovery ofVFC. There was no comparable patient in our study be-cause patients were exposed to VGB for at least ninemonths. Recovery of VGB-associated VFC has been fur-ther documented in single case reports [7, 16]. It hasbeen speculated [3] that VFC in these cases [7, 16] mighthave been due to methodological limitations as perime-try is a technique that requires considerable patients’ at-

Table 1 Demographic and clinical data of patients

Patients who stoppedVGB treatment beforeor after baselinen = 11

Sex Female 7Male 4

Age at Baseline Median (min-max) 40 (21–56)

Epileptic Syndrome Generalised 1Focal 10GTCS* during sleep 0Unclassifiable 0

Aetiology Posttraumatic 1Infectious 0Vascular 1Tumour 2Idiopathic 1Other 1Cryptogenic 5

Seizure Types Absences 1Simple focal 7Complex focal 7GTCS* 8Other 3Unclassifiable 0

Seizure Frequency at baseline Median (min-max) 4 (0–25)(Seizures/Month)

Duration of Treatment with Median (min-max) 26 (3–37)AED at Baseline (Years)

Duration of VGB Exposure Median (min-max) 24 (9–74)(Months)

VGB Maximum Daily Median (min- max) 4 (2–4)Dosage (g/d)

VGB free period at baseline Median (min-max) 12 (0–43)(Months)

Number of AED in the Median (min-max) 2 (1–4)observation period

* Primary or secondary generalised tonic clonic seizures

AMeridian (mm) Baseline first year follow-up second year follow-up p1*/p2*

N = 11 N = 11 N = 6

Median (min-max) Median (min-max) Median (min-max)Nasal 48.8 (19–58) 52.0 (17.5–59) 52.5 (18.5–65) n. s./n. s.Nasal inferior 46.0 (15.5–58.3) 47.5 (18.5–59.5) 46.3 (19.5–56) n. s./n. s.Temporal 75.5 (35–100.5) 79.0 (36.5–98) 75.3 (37–99.5) n. s./n. s.

BMeridian (mm) Baseline first year follow-up second year follow-up p1*/p2*

N = 4 N = 4 N = 1

Median (min-max) Median (min-max) Median (min-max)Nasal 46.2 (39–53.5) 52.7 (36.5–58) 65.0 n. s./n. s.Nasal inferior 45.5 (36–49.8) 55.0 (37.0–62) 60.0 n. s./n. s.Temporal 61.2 (49–77.0) 68.3 (48.0–78) 74.0 n. s./n. s.

* Wilcoxon signed rank test for paired observationsn. s. not significant; p1* baseline versus first year follow-up; p2* baseline versus second year follow up

Table 2 Kinetic perimetry: comparison of visualfield size (radius of outer isopter: nasal, nasal inferiorand temporal) A 11 patients who stopped VGB treat-ment, B 4 patients who continued VGB treatment

1069

AParameters Baseline first year follow-up second year follow-up p1*/p2*

N = 11 N = 11 N = 5

Median (min-max) Median (min-max) Median (min-max)MD (dB) –4.3 (–17–1.4) –4.4 (–15.3–0.8) –7.3 (–14–(–0.4)) n. s./n. s.PSD (dB) 7.5 (1.7–11.8) 7.3 (2.2–12) 6.7 (2–13) n. s./n s

BParameters Baseline first year follow-up second year follow-up p1*

N = 4 N = 4 N = 1

Median (min-max) Median (min-max) Median (min-max)MD (dB) –5.3 (–19.3–(–2.2) –4.2 (–18.4–(–2.4) –2.2 n. s.PSD (dB) 5.0 (3.3–8.8) 3.6 (2.9–7.5) 2.4 n. s.

* Wilcoxon signed rank test for paired observationsn. s. not significant; p1* baseline versus first year follow-up; p2* baseline versus second year follow up

Table 3 Static perimetry: comparison of visual fieldparameters (MD, PSD) A 11 patients who stoppedVGB treatment, B 4 patients who continued VGBtreatment

Fig. 1 Kinetic perimetry at baseline, one year and second year follow-up: radius (mm) of outerisopter in three meridians: 1 nasal, 2 nasal inferior and 3 temporal. A 11 patients who stopped VGBtreatment, B 4 patients who continued VGB treatment.

Fig. 2 Static perimetry at baseline, one year and second year fol-low-up: MD (dB). A 11 patients who stopped VGB treatment, B 4 pa-tients who continued VGB treatment.

1070

tention and cooperation. This problem is especially rel-evant in epilepsy patients, who might have a delayed re-action or a fluctuating attention due to seizure activityand to drug treatment. Hence, we always confirmed thediagnosis of VFC by a second perimetric examination.Cooperation was additionally measured by reliabilityparameters of static perimetry, which were all withinnormal limits. Although VFC of the study group as awhole did not improve significantly,we nevertheless sawan increase or decrease of VFC in individuals. Thesechanges were mild and were not related to the durationof VGB treatment or the duration of the VGB free periodor any other clinical parameter. It seems unlikely thatminor deteriorations of visual fields are based on re-placement by other antiepileptic drugs (lamotrigine,topiramate, primidon), because none of these drugshave been described as causing VFC. Thus, we believethat these changes rather reflect fluctuations in patients’attention than real and lasting changes due to discon-tinuation of VGB treatment. However, since follow updata are still scarce, one cannot exclude at this point thatVGB-associated VFC recovers in individual patients af-ter VGB is discontinued.

For clinical reasons seven out of eleven patientsstopped VGB treatment 10 to 43 months before baselineexamination. Therefore we cannot exclude that VFC im-proved in these patients before baseline examination.However, this seems unlikely as there was no change ofvisual field size at first and second year follow-up eitherin these patients or in patients who stopped VGB treat-ment shortly after baseline examination.

We additionally looked at a small group of patientswho continued VGB treatment on a reduced dosage anddid not find any significant change of VFC in these pa-

tients. Thus, at least in some patients VFC may remainstable when reducing VGB dosage. This finding is clini-cally relevant and has already been reported in six pa-tients with VFC who had repetitive visual field testingevery three months for up to one year during VGB treat-ment [11]. Since data on the pathophysiological back-ground of VGB-associated VFC are still scarce we do notquite understand why patients who continued VGBtreatment did not do worse than patients who stoppedVGB treatment. Data gathered so far suggest that the cu-mulative incidence of VFC increases rapidly during thefirst two years of treatment and within the first two kg ofVGB intake, stabilising at three years and after a totalVGB dose of three kg [15]. We know from animal stud-ies that the acute administration of VGB can produce athreefold increase in GABA levels of the cortex and a six-fold increase in the retinal GABA concentration. Withprolonged VGB administration retinal GABA levels dropto approximately 60 % [10]. Accumulation of VGB in theretina has been suspected to play a major role in the oc-currence of VFC although there is no conclusive evi-dence that GABA potentiation per se is directly retino-toxic [15]. At present we can only speculate thatreduction in retinal GABA-levels in chronic VGB treat-ment could be an explanation for our results. Some pa-tients clearly improve from VGB treatment and refuse tostop the drug even when VFC is diagnosed. These pa-tients should have regular visual field testing on the low-est dosage of VGB possible.

In conclusion, there seems to be no recovery in thevast majority of patients with VGB associated VFC.However,our findings are based on a relatively small andheterogeneous group of patients. Thus, further studiesare needed to confirm our results.

References

1. Eke T, Talbot JF, Lawden MC (1997) Se-vere persistent visual field constrictionassociated with vigabatrin. BMJ314:180–181

2. Goldmann H (1946) Demonstrationeines neuen Projektionskugelperime-ters samt theoretischen und klinischenBemerkungen über Perimetrie. Oph-thalmologika 111:187–197

3. Johnson MA, Krauss GL, Miller NR, etal. (2000) Visual function loss from vi-gabatrin, effect of stopping the drug.Neurology 55:40–45

4. Kälviäinen R, Nousiainen I, Män-tyjärvi M, et al. (1999) Vigabatrin, agabaergic antiepileptic drug, causesconcentric visual field defects. Neurol-ogy 53:922–926

5. Kälviäinen R, Nousiainen I (2001) Vi-sual field defects with vigabatrin: epi-demiology and therapeutic implica-tions. CNS Drugs 15 (3):217–230

6. Krämer G, Ried S, Landau K, et al.(2000) Vigabatrin: Reversibility of se-vere concentric visual field defects af-ter detection and drug withdrawal: acase report. Epilepsia 41:144

7. Krakow K, Polizzi G, Riordan-Eva P, etal. (2000) Recovery of visual field con-striction following discontinuation ofvigabatrin. Seizure 9:287–290

8. Miller NR, Johnson MA, Paul SR, et al.(1999) Visual dysfunction in patientsreceiving vigabatrin. Clinical and elec-trophysiologic findings. Neurology53:2082–2087

9. Murri L, Judice A (1995) Vigabatrin asfirst Add-on treatment in carba-mazepine-resistant epilepsy patients.Acta Neurol Scand (Suppl.) 162:40–42

10. Neal MJ, Shah MA (1990) Developmentof tolerance to the effects of vigabatrin(gamma-vinyl-GABA) on GABA re-lease from rat cerebral cortex, spinalcord and retina. Br J Pharmacol 100(2):324–328

11. Paul SR, Krauss GL, Miller NR, MeduraMT, Miller TA, Johnson M (2001) Vi-sual function is stable in patients whocontinue long-term vigabatrin ther-apy: implications for clinical decisionmaking. Epilepsia 42 (4):525–530

12. Rüther K, Pung T, Kellner U, Schmitz B,Hartmann (1998) Electrophysiologicevaluation of a patient with peripheralvisual field contraction associated withvigabatrin. Arch Ophthalmol116:817–819

13. Schmidt T (1955) Über die Gesichts-felduntersuchung am Goldmann-perimeter. Klinisches Merkblatt derAugenheilkunde 126:209–217

1071

14. Sills GJ, Patsalos PN, Butler E, et al.(1999) Concentration-related pharma-codynamic studies with vigabatrin andtiagabine in rat brain and eye. Epilep-sia 40:132

15. Sills GJ, Patsalos PN, Butler E, et al.(2001) Visual field constriction: accu-mulation of vigabatrin but nottiagabin in the retina. Neurology57:196–200

16. Versino M, Veggiotti P (1999) Re-versibility of vigabatrin induced vi-sual-field defect. The Lancet 354:486

17. Wild JM, Martinez C, Reinshagen G,Harding GFA (1999) Characteristics ofa unique visual field defect attributedto Vigabatrin. Epilepsia 40:1784–1794

18. Wong IC, Mawer GE, Sander JW (1997)Severe persistent visual field constric-tion associated with vigabatrin. BMJ314:1693–1694