Purpose: A reliable classification of AMD subgroups is necessary prior to any genetic or clinical investigation. For this purpose, we evaluated a selected list of items for AMD based on color photographs of the fundus but also on fluorescein and ICG angiographies and OCT. The first step (SPA-1) is to evaluate the interobserver agreement for AMD precursors.
Methods: The 30 items for AMD precursors, choroidal neovascularization and atrophy, were chosen from the literature. We selected 20 eyes from 20 consecutive patients affected by various forms of these precursors from the prospective NAT-2 study. Grading was based on color photographs of fundus, red-free frames, and fluorescein angiography. For each eye, five grids were filled in by four independent, experienced readers from our reading center and one expert. Within-subject variance was studied.
Results: Within subject variance showed the greatest agreement between graders for drusenoid pigment epithelial detachment (variance=0.000). The worst value for within subject variance (0.255) was recorded for the item "intermediate drusen between 5 and 20."
Conclusion: This evaluation of our first grid for AMD precursors shows high agreement between graders for most of the items. The item with the lowest agreement should be modified according to the comments of the physicians. This grid could be implemented for phenotypic classification of ARM and AMD in future clinical or genetic studies.