Purpose
To map quantified autofluorescence (AF) of human retinal pigment epithelium (RPE) at different stages of AMD relative to age-matched controls with healthy maculas.
Methods
Human chorioretinal tissue (25 donors (69-95 years) were fixed in paraformaldehyde <4 hr after death. Internal fundus examination revealed different AMD stages: incipient and early (n=18), late atrophic (2), and neovascular AMD (5). Incipient and early AMD had qualitatively similar RPE pathology (Ach ARVO2014). RPE flat-mounts preserved the foveal position. RPE cytoskeleton (labeled with AlexaFluor647-phalloidin) and AF (exc. 460-490 nm, em. > 505 nm) were imaged on a spinning disk fluorescence microscope. AF was normalized to a fluorescence reference (Delori, PMID 22016060). Custom FIJI plugins compared normalized AF to 9 age-matched healthy RPE flatmounts (Ach, PMID 25034602). Difference maps plotted mean log pairwise ratios vs. all controls for each eye (i.e., difference of logs).
Results
Normalized AF is variable among AMD donor eyes, as in age-matched normal eyes. AF levels in AMD tissues are below levels measured for controls (Figure for example). Only a few eyes show similar or higher AF levels, presumably independent from AMD stage. In 9 of 20 eyes with incipient, early, or late atrophic AMD, AF was particularly lower on the inner slope of the rod ring (< 4 mm eccentricity from the fovea, Curcio PMID 2324310). In remaining eyes no specific sub-regional predilection was discerned. AF was independent of AMD stage.
Conclusions
This is the first systematic analysis of quantified histological AF in AMD eyes. Normal aging leads to increased AF in the perifovea and near periphery (1.25-2.75 and 2.75-4.25 mm eccentricity per Polyak; Ach, PMID 25034602) reflecting high density of rods. In AMD eyes the perifovea is characterized by noticeably reduced AF, indicating the spatial non-correspondence between areas of high lipofuscin-attributable AF and RPE loss. Unchanged or high AF in some tissues suggests additional factors regulating AF that may be best characterized in longitudinal imaging of genetically defined patient populations.