Background: Early age-related macular degeneration (AMD) is common among the elderly. While only a small number progress to sight-threatening stages of AMD, identifying prognostic functional markers remains paramount. Here, we objectively evaluate retinal function in patients with large drusen by multifocal pupillographic objective perimetry (mfPOP). Different temporal presentation rates and luminances were compared to optimize parameters for high signal to noise ratios (SNR) and diagnosticity for early AMD.
Methods: Pupil responses were recorded from 19 early AMD patients (30 eyes) and 29 age-matched control subjects. We compared a luminance-balanced stimulus ensemble and two unbalanced stimulus variants, each consisting of 44 independent stimulus regions per eye extending from fixation to 15˚ eccentricity. Video cameras recorded pupil responses for each eye under infrared illumination. The amplitudes and delays of the peak responses were analysed by multivariate linear models. The diagnostic accuracy of the stimulus variants was compared using areas under the curve (AUC) of receiver operator characteristic (ROC) plots.
Results: Early AMD eyes differed significantly from normal in their mean constriction amplitudes (-2.22 ± 0.15 dB, t = -14.8) and delays (17.92 ± 1.2 ms, t = 14.9). The brightest stimulus ensembles produced the highest median SNRs of 3.45 z-score units; however, the balanced method was found to be the most diagnostic. AUC values of 0.95 ± 0.03 (mean ± SE) for early AMD were obtained when the asymmetry of response amplitudes between eyes was considered.
Conclusions: The mfPOP responses of early AMD eyes showed significant abnormality in response amplitudes and peak time. The ROC AUCs of 95 % suggest that mfPOP is a sensitive tool for detecting early abnormalities in AMD and longitudinal studies measuring progression of retinal dysfunction are warranted.