OBJECTIVES

To investigate the topographic relationship between glial tissue and active neovessels in epiretinal membranes (ERMs) in proliferative diabetic retinopathy (PDR).

MATERIALS AND METHODS

Phase-contrast and immunofluorescence microscopy were performed on 17 surgically removed ERMs from 17 eyes of 17 PDR patients. Clusters of active neovessels and the surrounding posterior hyaloid membrane were excised en bloc. ERMs were immunolabeled with anti-glial fibrillary acidic protein (GFAP) antibodies to identify glia, and with anti-collagen IV or anti-von Willebrand factor (VWF) antibodies to identify neovessels. All ERMs were analyzed as whole-mounted preparations, each including the area of leading neovessels.

RESULTS

GFAP-immunopositive glial cells (GCs) were identified in 11 of 17 specimens (65%). These cells also co-expressed type IV collagen. Fibrils immunopositive for type IV collagen (GFAP-negative) were detected in all cases. The topography, structure, and GFAP immunoreactivity distinguished GCs from GFAP-negative hyalocytes. GCs had bipolar shape, small cell bodies, very long, sparsely branching, bidirectional processes, and showed a tendency to form clumps. The structure of GCs was more consistent with that of Müller cells. In all ERMs, the majority of GCs were localized around the epicenter of neovascular clusters (where neovessels branched from the maternal vessel), which also corresponded to the highest density of collagen fibrils. In four cases (23.5%), GCs were also identified in the area of the leading capillaries; however, no signs of direct interaction between GCs and developing neovessels was observed in these cases.

CONCLUSION

Our study found no evidence of direct interaction between GCs and leading neovessels in PDR, opposite to what was shown in embryonic retinal angiogenesis. The findings may suggest that the presence of GCs near the neovascular cluster epicenter and around leading capillaries reflects different phases of the proliferative process in PDR. In the first case, GFAP+ cells appear to be involved in the involution of neovessels, which occurs during vascular remodeling or regression. In the second case, when GCs were located around the leading neovessels, their proliferation was not directly related to blood vessel formation; in our opinion, these processes may represent independent events that might have common triggers.