Alzheimer's Disease, a global health challenge, may have a new ally in the fight for early detection: the retina.
A recent study employed machine learning to analyze the 3D morphology of microglial cells in retinal tissue, uncovering critical differences between Alzheimer's Disease (AD) patients and healthy controls. The findings documented disease mechanisms and could become the basis for non-invasive diagnostic opportunities, the authors noted.
Microglia, the immune cells of the central nervous system (CNS), play a crucial role in maintaining neural homeostasis by clearing harmful substances such as amyloid-beta (Aβ). However, in AD, prolonged exposure to Aβ leads to microglial dysfunction, reduced clearance, and ultimately neurotoxicity. The researchers said that while microglial changes in the brain are well-documented, their role in the retina—a CNS extension—remains underexplored.
The study, published in Acta Neuropathologica Communications, leveraged machine learning techniques to analyze the morphology and activity of retinal microglia, offering a window into AD pathology and its potential retinal biomarkers.
Researchers analyzed retinal tissues from age-matched AD patients and healthy controls using confocal microscopy. Microglia were labeled with markers IBA-1 and CD68, highlighting immune activity and lysosomal presence. By employing Ilastik, an interactive machine-learning image analysis software, a custom Python pipeline, the team achieved high-resolution segmentation of 3D microglial structures. This approach was chosen to minimize observer bias, reduce analysis time, and enhance accuracy compared to manual methods.
Manual and machine learning analyses showed a decline in microglial density in AD retinas compared to controls. While manual data showed statistical significance, machine learning data revealed a similar trend, suggesting that microglial loss could impair Aβ clearance, the researchers noted.
Additionally, AD-associated microglia exhibited significantly larger sizes (volume and area) compared to controls, with the changes confined to CD68-positive microglia—those actively engaged in phagocytosis. This suggested heightened activation in response to Aβ accumulation. Surprisingly, measures of microglial shape complexity, such as solidity and circularity, showed no significant differences between AD and control groups.
CD68 labeling indicated elevated lysosomal activity in AD microglia, likely due to increased Aβ uptake. This aligned with the hypothesis that microglial enlargement reflects their response to heightened pathological burdens, the researchers wrote.
These findings highlight the potential of retinal imaging as a non-invasive tool for early AD detection. Technologies including Optical Coherence Tomography could, in the future, enable real-time monitoring of microglial changes, aiding in early diagnosis and tracking therapeutic efficacy.
The researchers noted that the study underscored the importance of preserving microglial health. Promising therapies such as Lecanemab target Aβ clearance, and maintaining functional microglia may enhance treatment outcomes and slow disease progression.
The study’s small sample size and variability in retinal regions analyzed highlight the need for larger, standardized datasets. Future research could explore how microglial populations change across AD stages and investigate the interplay between microglial activity and tau pathology in the retina, according to the researchers.
The authors declared no competing interests with this study.