Clinical and histological aspects of CNV formation: studies in an animal model.

The purpose of the present thesis was to develop an animal model of CNV in order to study the early formation of CNV and to test the effects of an anti-angiogenic treatment. Porcine eyes were chosen as a substrate for CNV induction, since they are similar to human eyes in terms of both macroscopic and microscopic morphology. However, a major difference is that pigs lack a fovea; instead they have a visual streak, with a relatively stable and high concentration of cones. By surgical perforation of Bruch's membrane we were able to induce formation of CNV membranes. The morphology and cellular composition of these membranes varied with the surgical technique employed. When RPE cells were locally removed at the time of perforation, the resulting CNV was thinner, contained fewer blood vessels and was less prone to leak on fluorescein angiography than when RPE cells were left intact at induction. The neuroretina overlying the perforation site was not damaged by any of the surgical techniques, thus allowing the subsequent retinal damage to be ascribed to the actual process of CNV formation. Using this animal model allowed us to directly map histological findings onto fluorescein angiograms and thereby perform meaningful correlations between histopathologic and photographic features. Such correlations have been hampered in human subjects, since human eyes are not enucleated as a consequence of CNV and are therefore only available for post-mortem studies. In such studies there often is a considerable time-gap between the death of the patient and the latest available fluorescein angiogram, thereby allowing macular pathology to evolve in the interim. Histological examination of the porcine membranes demonstrated that they were composed of RPE cells, glial cells, macrophages, endothelial cells, collagen and smooth muscle fibres, which are the same cellular and fibrillar elements that dominate human CNV membranes. The porcine model was applied to test the effects, in a randomized and masked fashion, of intravitreally injected bevacizumab. Bevacizumab, a pan VEGF A antibody, was found to reduce both the proliferation of endothelial cells in CNV membranes and the propensity to leak in fluorescein angiograms. Immunohistochemically, bevacizumab was detected in the inner limiting membrane, in retinal blood vessels and binding uniformly to the entire CNV membrane without any cellular predisposition. Based on the above findings we believe that the porcine CNV model shows a bearing to human disease and therefore might be used as a tool to obtain improved treatments for this debilitating condition.