Abstract
Purpose:
To estimate the test-retest variability of photopic ERG amplitudes as obtained during examination under anesthesia in normal fellow eyes of patients under treatment for retinoblastoma to aid in the interpretation of changes in ERG amplitudes associated with therapy.
Methods:
Ganzfeld ERGs were obtained during regularly scheduled examinations under anesthesia from patients undergoing evaluation or treatment for retinoblastoma at Memorial Sloan Kettering Cancer Center. We report repeated-measure analysis of variance of the amplitude of the response to 30-Hz flicker stimulation, which we have found to be representative of the entire ISCEV standard ERG protocol in these patients. ERGs were recorded with ERG-jet contact lens electrodes using the Diagnosys Espion electrodiagnostic system with a hand held ColorBurst ganzfeld stimulator. We estimate the variability of ERG amplitudes under these conditions, based on recordings from the normal fellow eyes of patients with monocular retinoblastoma (mostly sporadic cases), so as to prevent confusion of inherent variability in the ERG with treatment effects. As we have previously identified signicant effects of scleral depression on the ERG of both the manipulated and untouched fellow eye, we include here only the results of ERG recordings obtained immediately following induction of general inhalation anesthesia, before manipulation or examination of either eye. Repeated-measure ANOVA was carried out using NCSS statistical software.
Results:
Photopic 30-Hz flicker amplitude data were available from 1129 ERG recordings of normal fellow eyes in 101 patients, after outliers due to factors such as inadequate pupillary dilation were removed from the database. The sum-of-squares between subjects was 4758, whereas the sum-of-squares within subjects (that is, associated with the several recording sessions for individual patients) was 376, corresponding to a standard deviation of approximately 19.4 μV.
Conclusions:
The test-retest variability of 30-Hz flicker amplitudes in normal eyes under our recording conditions yields a standard deviation of approximately 20 μV. This estimate serves as the basis for our standard separation of ERG responses in our clinic into broad classes: Extinguished (no ERG flicker response distinguishable from background noise); Poor (between 1 and 25 μV); Fair (26 - 50 μV); Good (51 - 75 μV); Very Good (76 - 100 μV); and Excellent (greater than 100 μV).