Evaluation of VEP perimetry in normal subjects and glaucoma patients

Purpose: To estimate sensitivity to glaucomatous visual field loss using multifocal visual evoked potential (VEP) perimetry, to compare these findings to those of conventional achromatic perimetry and to determine specificity of VEP perimetry in normal subjects. Methods: A total of 33 glaucoma patients with known visual field defects in at least one eye on standard computerized perimetry and 33 healthy subjects were tested with VEP perimetry. The glaucoma patients were also tested with standard computerized perimetry using the 30-2 SITA Fast program of the Humphrey Field Analyzer (HFA). Visual evoked potential perimetry classification and VEP probability maps were used to determine the sensitivity and specificity of the technique. Results: Visual evoked potential perimetry classified 68% of all eyes in the glaucoma group (45/66) as pathological; sensitivity increased to 81% (38/47) when considering only those eyes with HFA field defects. It also identified more test locations with significant loss at the p < 5% level in both groups (48% and 37%, respectively) than did HFA, while HFA identified more loss at the higher significance levels p < 2%, and p < 1%. Visual evoked potential perimetry showed more significant loss in eyes with almost normal or slightly damaged standard fields, while HFA identified more significant field loss in eyes with severe conventional field damage. The mean VEP amplitude of the 66 glaucoma eyes was 1.461e(-7) V; it was 1.676e(-7) V for the 66 control eyes. This difference was significant (p = 0.0033), but the overlap between groups was large. Visual evoked potential perimetry classified 42% of the control eyes as 'outside normal limits', and VEP probability maps showed 30.0% of test segments as significantly depressed at the p < 5% level, 10.8% of sites at p < 2%, and 4.6% at the p < 1% level. Conclusion: Mean VEP amplitude differed significantly between normal and glaucoma eyes, but the overlap was considerable. Visual evoked potential perimetry falsely classified a large number of normal eyes as pathological and showed many more significantly depressed test locations than expected. Agreement between VEP and standard perimetry was relatively poor for the glaucoma group. Further refinements are needed before VEP perimetry can be regarded as a reliable clinical method of mapping glaucomatous visual fields.