Abstract
PURPOSE: Mouse models of Alzheimer's disease (AD) are designed to replicate a multitude of pathologies. However, a direct comparison between models is virtually impossible due to vastly differing data acquisition and image processing protocols. Based on the hypothesis that a well-defined experimental framework is key to determining subtle differences in their retinal phenotype, we use a standardized anesthesia protocol, the same optical coherence tomography (OCT) system and image processing pipeline to compare commercially available AD mouse models. METHODS: Two models of amyloidosis (5xFAD and APP/PS1) and one of tau pathology (PS19) were investigated (age = 42.5 ± 6.5 weeks). A high-resolution OCT prototype was used to image the retina of animals under isoflurane anesthesia. Retinal OCT parameters were mapped and compared. RESULTS: Total retinal thickness was comparable for the amyloidosis models (transgenic [tg] and non-transgenic [ntg] APP/PS1 = 209.3 µm and 210.7 µm, and tg and ntg 5xFAD = 212.1 µm and 211.2 µm), but, on average, approximately 17% thicker for the tg and ntg PS19 mice (256.4 µm and 236.8 µm). APP/PS1 mice had approximately 10 µm thicker outer retinal layers (ORL) in tg and ntg models and 10 µm thinner inner retinal layers (IRL) in comparison to 5xFAD mice. PS19 mice had between 10 and 20 µm thicker IRL and ORL than the amyloidosis models. The vascular density in superficial vascular, intermediate, and deep capillary plexuses differed significantly for 5xFAD mice. CONCLUSIONS: Despite standardized conditions, the retinal parameters were not constant across different mouse models of AD, indicating fundamental phenotypical differences.