Effect of COX inhibitors on VEGF-induced retinal vascular leakage and experimental corneal and choroidal neovascularization

The primary objective of this study was to evaluate the effect of cyclooxygenase (COX) inhibitors, non-steroidal anti-inflammatory drugs (NSAIDs), in two in vivo models of VEGF-dependent corneal and choroidal angiogenesis and two in vivo models of VEGF-mediated vascular leakage. Non-selective COX inhibitors (the NSAIDs indomethacin and flunixin, p.o. or i.p.), the COX-1 selective inhibitor SC-560 (s.c. or i.p.), and the COX-2 selective inhibitor NS-398 (s.c. or i.p.) were evaluated in four experimental models. Choroidal neovascularization was induced in Brown Norway rats by argon laser photocoagulation and measured after ten days. Corneal neovascularization was induced by alkaline cautery in Sprague-Dawley rats and measured after four days. VEGF protein levels in the cornea were quantified by ELISA. VEGF-induced intradermal extravasation of Evans blue dye (EBD)-albumin was assayed in Hartley guinea pigs. Intravitreal VEGF-induced blood-retinal barrier breakdown was assayed by scanning ocular fluorophotometry in Dutch Belt rabbits. Indomethacin (1 or 3 mg kg(-1) day(-1), p.o.), SC-560 (20 mg kg(-1) day(-1), s.c.), and NS-398 (20 mg kg(-1) day(-1), s.c.) failed to inhibit laser-induced CNV. CNV was inhibited, however, by the corticosteroid dexamethasone (0.5 mg kg(-1) day(-1); p.o. or s.c.; 99% or 90% inhibition; p<0.01 or p<0.001, respectively). In contrast, cautery-induced corneal angiogenesis was inhibited partially by the NSAID indomethacin and the COX-2 selective inhibitor NS-398. Indomethacin, 3.5 or 7 mg kg(-1) day(-1), inhibited corneal neovascularization by 56% (p<0.001) or 68% (p<0.001) respectively. Similar partial inhibition of angiogenesis in the cornea model was observed with NS-398 (10 or 20 mg kg(-1) day(-1), s.c. or i.p.; 54% inhibition, p<0.001), but not with the COX-1 selective SC-560 (10 or 20 mg kg(-1) day(-1), s.c.). In the cornea, VEGF protein is dramatically upregulated 24 and 48 hr after cautery, and both indomethacin and NS-398-but not SC-560-significantly inhibited this VEGF upregulation. In experimental models of VEGF-induced vascular leakage, COX inhibitors had no effect on dermal or retinal vascular responses to VEGF. The NSAIDs indomethacin (7.5 or 20 mg kg(-1), p.o. or i.p.) and flunixin (12.5 mg kg(-1), i.p.) failed to inhibit VEGF-induced dermal extravasation of EBD-albumin in guinea pigs. In contrast, L-NAME (25 or 50 mg kg(-1), p.o.)-an anti-vasodilatory inhibitor of nitric oxide synthase-dose-dependently inhibited up to 64% (p<0.001) of this dermal vascular leakage. VEGF-mediated retinal vascular leakage was not blocked by COX inhibition. Intravitreal VEGF-induced BRB breakdown--which was completely blocked by VEGF neutralizing s-Flt-1/Fc protein (intravitreal co-administration; p<0.001)--was not inhibited by indomethacin (20 mg kg(-1) day(-1), s.c.). Although COX inhibitors were ineffective at blocking experimental CNV, both non-selective and COX-2 selective inhibitors partially blocked severe inflammatory corneal angiogenesis and its concurrent upregulation of VEGF protein. These results suggest that eicosanoids produced by inducible COX-2 are among multiple mediators that modulate VEGF expression as a stimulus in inflammation-associated angiogenesis. The lack of effect with COX inhibitors on either VEGF-mediated dermal extravasation or VEGF-mediated blood-retinal barrier breakdown indicates that COX activity is not required for vascular leakage responses to VEGF.