Since the 50/10 OIR model consistently develops IVNV and naturally undergoes regression of IVNV followed by vascularization of previously avascular retina, we were able to test the effects of neutralizing VEGF with antibody on these quantifiable outcomes by analyzing two different time points. We found that neutralizing VEGF with an intravitreous injection of 50 ng of VEGFab administered at p12 caused a significant reduction in IVNV at p18. Furthermore, this dose sustained the effect at p25 and did not interfere with ongoing retinal vascularization. However, the lower dose of antibody did not sustain the inhibitory effect on IVNV.
Antibodies generally are cleared from that eye over several days–on average 5.6 days in monkey vitreous [42
]. Since vitreous VEGF concentration at p18 was approximately 1/10 of the retinal VEGF (), we propose that an adequate dose of VEGFab may neutralize sufficient vitreous VEGF without severely interfering with intraretinal signaling necessary for ongoing retinal vascular development [23
]. If the concentration of antibody binds both retinal and vitreous VEGF sufficiently to cause vitreous VEGF bioactivity to be effectively zero, then it is possible a compensatory increase in intraretinal VEGF protein, as we found, may promote intraretinal vascularization. If the level of antibody is too low, then both IVNV and intraretinal vascularization would be reduced, but chemoattractive forces in both vitreous and retina may be insufficient to inhibit, but only slow retinal vascularization. Indeed, we found a slight, although insignificant, increase in avascular retina compared to IgG control at p25. Still the avascular retinal area at p25 was less than that at p18.
There is increasing evidence that the role of VEGF in retinal vascular development and oxygen stresses is complex. It has been reported that, a front of migrating cells, e.g., astrocytes in cat [24
] or angioblasts in dog [45
], sense physiologic hypoxia and express VEGF. The ensuing endothelial cells are attracted to VEGF and migrate to create blood vessels [24
]. More recently, it has been found that VEGF concentration may regulate endothelial cell division rate [46
], but the presentation of VEGF, as in a gradient, regulates endothelial tip cells at the migrating front and direct the growth of endothelial cells [48
]. Furthermore, there is evidence that a gain in VEGF signaling through VEGFR2 can cause disoriented endothelial daughter cell divisions rather than orderly angiogenesis [49
]. In immunohistochemical qualitative assessment of immunohistochemical sections taken from eyes one day after injections, we found that, compared to control or noninjected eyes, 50 ng of VEGFab appeared to reduce but not entirely inhibit signaling of VEGFR2, the receptor believed most associated with angiogenic processes [50
]. Although we were unable to confirm the reduction in VEGFR2 phosphorylation by western blot analysis, we suspect that analyzing whole retinas may dilute effects seen in a small percentage of cells. VEGFR1 phosphorylation did not appear to be affected. This may have been because VEGFR1 has greater affinity for VEGF [50
] and would have bound free VEGF not associated with VEGFab. Effects from VEGFab on intraretinal VEGFR2 signaling appeared to have resolved by p18. We also found that free VEGF in the retina measured one day after an injection of 50 ng VEGFab was increased compared to control. Possibly, this represents a compensatory effect and may partly explain the insignificant but increased number of clock hours of IVNV at p25 in eyes injected with 25 ng of VEGFab when compared to Ig G control.
The VEGFab used in this study is greater than 78kD molecular weight, which is the limit above which diffusion beyond the inner plexiform layer becomes extremely slow [43
]. From our immunohistochemical sections, much of the reduction in VEGFR2 phosphorylation appeared to be in the region of the ganglion cell/nerve fiber layers. However, there is experimental and clinical evidence that antibodies of higher molecular weight, including the humanized mouse monoclonal antibody to VEGF (bevacizumab), penetrate into the retina and can affect signaling in deeper retinal layers [52
Several papers have shown immunolabeling of VEGF receptors but not phosphorylation, in developing retinal vessels and in IVNV-like vessels or endothelial buds after oxygen stress [12
]. VEGFR2 immunoreactivity was also reported to be strong within IVNV in a beagle model of OIR and weak within newly forming intraretinal vessels during normal development [12
]. In addition, VEGF and its receptors were found within neural retina [56
], mainly the ganglion cells, astrocytes, and Mueller cells. In development, VEGF receptor inhibitors led to a reduction in the thickness of the retina and in the ganglion cell layer [54
]. We report that activation of VEGFR2 signaling was associated with intraretinal and intravitreous blood vessels and within the ganglion cell and nerve fiber layers of retinas from pups exposed to the 50/10 OIR model. VEGFR1 phosphorylation was associated mainly with the ganglion cell and outer plexiform layers. We propose that inhibition of VEGFR2 signaling by VEGFab occurred in developing vessels, and within Mueller cell processes, astrocytes, and ganglion cells at p13, and was resolved by p18. Definitive confirmation of what cells are affected will require colabeling of retinal sections in future studies. Although reduced VEGFR2 phosphorylation was associated with decreased IVNV in our experiments, further study is needed to determine if there are significant effects on neuronal and glial cells, particularly if higher doses of VEGFab or prolonged and repeated doses are considered. However, ganglion cells treated with the VEGF antibody, bevacizumab, were not reported to have reduced viability in vitro [59
Although we found a modest decrease in IVNV compared to control, it would represent a clinically significant outcome in reducing the risk of poor vision in preterm infants with ROP [19
]. Many studies have shown that blocking other signaling pathways can inhibit IVNV more completely [60
]. However, in ROP, aggressive angiogenic inhibition is not desired, because ongoing retinal vascular development may both lead to improved visual function and reduce the hypoxic stimulus for pathologic IVNV.
The effects of inhibiting VEGF on pathologic IVNV and its presumed stimulus, the avascular retina, are relevant questions when considering anti-VEGF strategies for ROP in which it is undesired to inhibit retinal vascular development but necessary to prevent or treat IVNV. Our data suggest that an intravitreous antibody to neutralize VEGF may be effective and safe, but dose appears important. Determining an effective dose in individual infant eyes is difficult because vitreous VEGF protein produced by the hypoxic retina may vary in separate eyes depending on the zone of ROP--i.e., the extent of avascular retina. Currently, vitreous VEGF measurements cannot be obtained safely in human infants with stage 3 ROP. In our study, measuring vitreous VEGF was not always possible in the animal model, because the concentration of VEGF in the vitreous was at the lower limits of detection by ELISA. Very high concentration or slow release formulations of an anti-VEGF antibody may carry a risk to retinal neurons and also theoretically inhibit ongoing retinal vascular development [22
]. In addition, too low a dose of anti-VEGF antibody may not be effective or require repeated injections, which increase risk. Furthermore, unlike in the 50/10 OIR model, human infants with ROP do not always undergo disease regression, in part, because of the effect of other factors [1
When administering an intravitreous injection of any drug into an infant eye, the high vitreous/blood volume of the premature infant to the adult must be considered. The resultant drug concentration from absorption into the systemic circulation is greater in the infant than that in adult and may cause systemic effects [37
]. Systemic anti-VEGF agents used in metastatic colon or renal cell carcinoma have been reported associated with several serious effects, including hypertension and vascular events [63
]. In this study, compared to control injected eyes, we found no adverse effect on weight gain or on outcomes in the fellow eyes, suggesting little effect from systemic absorption at the time points we analyzed. Nor did we find an adverse effect on capillary densities of newly vascularized retina within the treated eyes. Although questions remain, in certain severe forms of ROP with poor outcomes [66
], use of anti-VEGF agents should be studied and considered.