Abstract
Recent attention has been given to immune cells near the inner limiting membrane (ILM) of the retina due to their potential use as a biomarker or therapeutic target. Non-invasive phase contrast imaging via optical coherence tomography or non-confocal adaptive optics scanning light ophthalmoscopy has allowed investigations into cellular dynamics and morphology. The cells have been variously referred to as microglia, hyalocytes, or ILM macrophages, with no consensus yet reached as to their identity. Here, we employed AOSLO with an improved non-confocal detection scheme to investigate motile immune cells and their relation, if any, to nearby retinal vasculature. Image series were systematically acquired to observe cells over time (30-60 minutes) at different depth planes (9.2 µm separation), in 3 healthy human subjects, within 2°x2° fields between 8° and 14° temporally and 0° and 8° superior from fixation. A total of 77 cell somas and their process tips were labelled, with a further 20 cells tracked with higher frequency to explore moment-to-moment variations. Cell somas were separated from the nearest blood vessel by 39.1 ± 17.3 µm (mean ± standard deviation). The cells occupied an effective monolayer, averaging 25.0 ± 15.7 µm anterior to the nearest vessel and with no cell observed posterior to a vessel. Compared to cell distributions known from histology, our observations suggest that only a subset of resident immune cells is amenable to phase contrast imaging. Monte Carlo simulations were carried out to determine whether immune cells are located differently from random within their monolayer; we found that cells did not co-localize with the nearby vasculature (p > 0.05), but did have a strong tendency to avoid each other (p ∼ 10(-12)). We note, however, that cell processes (often rapidly extending and contracting) are often long enough to reach the closest vessels, averaging 34.8 ± 11.5 µm in length with a maximum observed extension of 101.9 µm. Finally, soma speeds averaged 0.52 ± 0.49 µm/min when followed every 5 minutes, consistent with previous reports. A sub-analysis with data collected every 12-15 seconds demonstrated an almost eight-fold increase in measured speed (averaging 4.04 ± 1.47 µm/min), reflecting the stop-start nature of cellular motility.