Macrophage-like cells are still detectable on the retinal surface after posterior vitreous detachment

The identity of vitreoretinal interface macrophage-like cells (MLCs) remains unknown and potential candidates include retinal microglia, perivascular macrophages, monocyte-derived macrophages, and/or vitreal hyalocytes. Since hyalocytes are detectable on the posterior vitreous surface after vitreous extraction in animals, we imaged patients with and without posterior vitreous detachment (PVD) to determine if hyalocytes are the principal MLC component. We performed repeated foveal-centered 3 × 3 mm OCT-A images from 21 eyes (11 no PVD and 10 PVD eyes). Images were registered, segmented, and averaged. The OCT slab from 0 to 3 microns above the internal limiting membrane was used to detect MLCs. We calculated MLC density and distribution in relation to the superficial vascular plexus for 3 vascular regions—on vessels, perivascular, and non-vascular. MLC density was 1.8-fold greater in the PVD group compared to the no PVD group (P = 0.04). MLCs in eyes with PVD were increased 1.9-fold on-vessel (P = 0.07), 1.9-fold in the perivascular region (P = 0.12), and 2.2-fold in non-vascular areas (P = 0.22). MLC density was not severely reduced after PVD, suggesting that the majority of MLCs are not vitreal hyalocytes. PVD status is an important parameter in future MLC studies.

www.nature.com/scientificreports/ hypothesized that MLCs would be severely reduced after PVD if hyalocytes were the principal MLC component. We used OCT and OCT-angiography (OCT-A) imaging to quantify MLC numbers in patient eyes with and without PVD. We found that MLCs were in fact increased in eyes with PVD, and not severely reduced, suggesting that most MLCs are not vitreal hyalocytes. Furthermore, PVD status is an important potential confounding factor in future MLC studies.

Results
We included 21 eyes from 21 subjects, including 11 eyes without PVD and 10 eyes with PVD. The demographic characteristics of both groups are shown in Table 1. The two groups were significantly different in age and sex. The PVD cohort was significantly older (age 64.5 ± 7.3 vs 46.4 ± 15.3) and consisted of more females (90% vs 45%) than the healthy controls. There was no significant difference in refractive error or the average Q-score between the two groups. Representative images showing MLCs in control ( Fig. 1A-C) and PVD eyes ( Fig. 1D-F) are shown in Fig. 1A-F. We found that the MLC density was 1.8-fold greater in the PVD group (13.21 ± 7.29 cells/mm 2 ) compared to no PVD eyes (7.48 ± 4.43 cells/mm 2 ; P = 0.04, Fig. 1G). MLC density in our no PVD eyes was very similar to our previously published MLC density 4 from healthy eyes (6.4 ± 7.0, p = 0.65), highlighting the reproducibility Table 1. Demographic characteristics. PVD posterior vitreous detachment, SD standard deviation. *Statistically significant (P < 0.05).  www.nature.com/scientificreports/ of our methodology. Since our PVD cohort was older, we performed a Pearson's correlation analysis and found no association between age and MLC density (Fig. 1H). Since MLCs are not severely reduced after PVD, this data suggests that hyalocytes are not the primary MLC cell type. Next, we investigated whether increased MLCs are found in a specific location to understand whether increased MLCs are derived from the vasculature. We found that MLCs were increased by 1.9-fold on-vessel (P = 0.07), 1.9-fold in the perivascular region (P = 0.12), and 2.2-fold in non-vascular areas (P = 0.22, Fig. 2).

Discussion
Since hyalocytes have previously been detected on the vitreous surface in animals after vitreous extraction 12 , we investigated eyes with a PVD to determine if MLCs were severely reduced, which would provide evidence that hyalocytes are the main MLC cell type. We found that MLC density was not decreased in PVD eyes ( Fig. 1), suggesting that most MLCs are less likely vitreal hyalocytes and more likely a retinal macrophage cell type. Furthermore, MLC density increased in PVD eyes, suggesting either increased recruitment of MLCs from the circulation or improved detection of pre-existing MLCs following PVD.
Previous studies have shown that vitreoretinal interface MLCs exhibit characteristics typical of macrophages including a ramified morphology 2 , mobility 1 , and Cx3cr1 + staining in mice 3 . As discussed in the introduction, possible MLC identities include microglia, perivascular macrophages, monocyte-derived macrophages, and/or vitreal hyalocytes. Since MLCs were not reduced or absent after PVD (Fig. 1), our data suggest that many MLCs are not hyalocytes. However, we cannot exclude the possibility that hyalocytes remain at the retinal surface along with remnants of the cortical vitreous after PVD. Histological confirmation in post-mortem eyes would be necessary to examine this question with greater certainty. Based on our data, we hypothesize that MLCs are a mixture of different macrophage sub-types, including microglia and perivascular macrophages at steady state with a minor hyalocyte contribution. This hypothesis is supported by a recent adaptive optics study, which showed that the majority of macrophages on the vitreoretinal interface have a highly ramified morphology and travel short distances, while a minority of cells migrate greater distances with a less ramified morphology 13 . We hypothesize that the less migratory and more ramified vitreoretinal interface macrophages are microglia, while the greater migratory and less ramified cells are hyalocytes.
In addition to greater MLC density in PVD eyes, we found no significant difference in MLC density based upon vascular location (Fig. 2). This finding suggests at least two hypotheses for increased MLC numbers after PVD. First, that MLCs could include monocyte-derived macrophages that extravasate and migrate away from vessels in response to transient injury after PVD. Alternatively, the lack of vitreous could potentially enhance the ability of OCT to detect preexisting microglia and/or perivascular macrophages that were previously masked by the attached vitreous. Further post-mortem and/or animal studies are needed with specific markers for microglia, perivascular macrophages, and monocyte-derived macrophages to better understand the identity and function of MLCs.
A third hypothesis is that PVD-induced MLCs include migrating glial cells. PVD can be an inciting event for epiretinal membrane formation 14 . Muller glia cell migration to the ILM surface and differentiation into myofibroblasts has been proposed as a mechanism for epiretinal membrane pathogenesis 15 . It is therefore possible that increased MLCs after PVD could also include migrating glial cells. Without adaptive optics-OCT confirmation in vivo, which would provide optimal transverse resolution for cell morphology and axial resolution for ILM surface location, or post-mortem histological confirmation, the identity of the increased MLC population after PVD remains uncertain.
Limitations of this study include inherent demographic differences between the cohorts and lack of axial length measurements. The PVD group was significantly older and included more female subjects ( Table 1). The effect of sex on MLCs is unknown and requires further investigation. However, increasing age is correlated with www.nature.com/scientificreports/ reductions, not increases, in MLC density 2 . Our finding of increased MLC density in the older PVD eyes further strengthens our evidence that MLC density is indeed increased by PVD. Furthermore, we found no association between MLC density and age, suggesting that this difference between cohorts did not confound our findings. In addition, most patients included in our study did not have an axial length recorded, which may affect image scaling 16 . However, no difference was found in refractive error, suggesting that axial length is unlikely to confound our data.
In conclusion, our study found that vitreoretinal interface MLC numbers are not significantly reduced after PVD. Our findings suggest that most MLCs are not hyalocytes and are more likely retinal macrophages, including microglia and perivascular macrophages at steady state and potentially monocyte-derived macrophages after injury like during diabetes or retinal vein occlusion. However, we cannot exclude cortical hyalocyte remnants on the retinal surface after PVD or migrating glial cells; further post-mortem and/or animal studies remain necessary to determine the origin, identity, and function of MLCs. Finally, PVD status appears to be an important parameter and potential confounder that needs to be considered in future MLC studies.

Methods
We conducted a prospective, cross-sectional study of healthy subjects with and without PVD seen between July 2020 and January 2021 in the Department of Ophthalmology at Northwestern University in Chicago, Illinois. The study was approved by the Institutional Review Board of Northwestern University (IRB no. STU00200890) and conducted in accordance with the tenets of the Declaration of Helsinki and regulations of the Health Insurance Portability and Accountability Act. Written informed consent was obtained from all subjects before participation.
Inclusion criteria were patients aged 20-75 years with or without a clinical diagnosis of PVD by a boardcertified ophthalmologist and confirmed on OCT imaging. Exclusion criteria were patients with prior eye surgery, retinal disease, vitreoretinal interface pathology like epiretinal membrane or vitreomacular adhesion, or significant media or lens opacity that obscured imaging. In addition, healthy patients with partial PVD detected by OCT were excluded.
Image analysis. Images were registered using the full retinal vascular network OCT-A slab via the Register Virtual Stack Slices Plugin (Feature Extraction Model = Rigid, Registration Model = Elastic) in FIJI, a distribution of the program ImageJ (National Institutes of Health, Bethesda, MD, USA). Next, the saved transformation matrix was applied to the MLC layer slabs from 0 to 3 µm above the internal limiting membrane 1 using the Transform Virtual Stack Slices Plugin. Registered MLC stacks were then averaged. MLCs were identified from averaged images using our previously published 4 semiautomated custom macro in FIJI. Please see our prior publication for a more detailed description, including our high reproducibility of this process (ICC = 0.998, 95% confidence interval = 0.982-0.999, p < 0.05) 4 . We quantified the number of MLCs using the Analyze Particles function, and the MLC density was calculated from the number of MLCs and the total image area (cells/mm 2 ).
The location of MLCs in relation to underlying vessels in the superficial vascular plexus was investigated by sorting cells into three vascular regions-on vessels, perivascular (1-30 microns from a vessel), and non-vascular (> 30 microns from the nearest vessel)-using a custom macro in FIJI as previously described 4 . MLCs in the foveal avascular zone (FAZ) were excluded from this analysis because MLCs in the FAZ would be artificially labeled as non-vascular. We excluded the FAZ by removing the entire central 1-mm circle from the analysis. Cell densities were calculated for each region as the number of cells divided by the area of each respective region.
Statistics. Statistical analysis was performed using GraphPad Prism 9.0.1 (GraphPad Software, San Diego, California USA). We used a two-tailed, unpaired t-test to evaluate the difference between MLC density and demographic data except for age, which was analyzed using Welch's t-test due to unequal variances between groups. Correlation between age and MLC density was tested with Pearson's correlation. We used a two-way analysis of variance followed by Šídák's multiple comparisons test to compare MLC densities among the three vascular regions and between the two patient populations. www.nature.com/scientificreports/