Lesion-Centered Functional Trajectories Reveal Early Retinal Sensitivity Decline in Intermediate AMD: PINNACLE Study Report 15.

Abstract

To characterize lesion-centered functional decline in intermediate age-related macular degeneration (iAMD), including measurement variability, spatial vulnerability, pre-detection functional changes, and lesion-specific sensitivity trajectories. Longitudinal analysis of microperimetry from the prospective, multicenter PINNACLE study (NCT06682455). Overall, 264 participants contributing 422 eyes were included. Focal event analyses included 338 OCT-defined events from 131 eyes of 113 participants. Participants underwent mesopic microperimetry using a standard 24-point grid. OCT-defined events triggered lesion-centered 5-point testing. Repeatability was assessed using linear mixed-effects models. Lesion-centered sensitivity trajectories were estimated using spatiotemporal generalized additive mixed models incorporating measured post-detection focal-grid data and spatially interpolated pre-detection standard-grid data. A secondary piecewise linear mixed-effects model quantified changes before and after event detection. Point-wise microperimetry repeatability, sensitivity loss of ≥7 dB, relative and absolute scotoma, and model-estimated lesion-centered sensitivity trajectories. The point-wise coefficient of repeatability was 7.61 dB (95% confidence interval [CI], 7.58-7.64) for standard-grid testing and 13.00 dB (95% CI, 12.76-13.21) for lesion-centered testing. The 338 events comprised 240 ellipsoid zone/interdigitation zone loss events, 85 drusen-collapse events, and 13 subretinal-fluid events. Within 2 years, the risk of sensitivity loss of ≥7 dB was lower at parafoveal (hazard ratio, 0.66; 95% CI, 0.51-0.84) and perifoveal locations (hazard ratio, 0.68; 95% CI, 0.53-0.87) than at foveal locations. Model-estimated sensitivity at ellipsoid zone/interdigitation zone loss sites declined before detection by 0.57 dB/year (95% CI, 0.25-0.89 dB/year) and continued after detection, reaching 0.94 dB/year (95% CI, 0.55-1.34 dB/year) during the late post-detection period. Drusen-collapse sites showed no clear longitudinal decline. Subretinal-fluid events showed an immediate loss of 2.18 dB (95% CI, 0.84-3.53), followed by early decline and possible later improvement. Matched control locations remained stable. Localized functional decline in iAMD is characterized by measurement variability, central spatial vulnerability, and lesion-specific trajectories. Model-estimated sensitivity decline was most evident at sites of ellipsoid zone/interdigitation zone loss. These findings support structurally and spatially informed microperimetry endpoints while highlighting substantial measurement variability.