Objective To evaluate the off-label use of subconjunctival bevacizumab for corneal neovascularisation (CoNV).

Methods 30 patients with recent-onset CoNV from various causes were randomly assigned into a double-masked, placebo-controlled trial. Each received three 0.1 ml injections containing either 2.5 mg bevacizumab or 0.9% saline at monthly intervals. Dexamethasone 0.1% drops were used four times a day for the first month, when the dose was modified if clinically indicated. The primary outcome was change in area of corneal involvement by CoNV from baseline to 3 months measured using specialised imaging technology.

Results The mean area of CoNV reduced by −36% (range −92% to +40%) in the 15 eyes that received bevacizumab compared with an increase of 90% (range −58% to +1394%) in eyes that received saline placebo (analysis of covariance (ANCOVA); p=0.007). One outlier in the placebo arm developed corneal graft rejection with aggressive neovascularisation (+1384%), but even when this patient was excluded the mean reduction in CoNV in the placebo group (−3%, range −58% to +40%) was still significantly different from the treatment arm (ANCOVA; p=0.016). Changes in best-corrected visual acuity, central corneal thickness, intraocular pressure and endothelial cell counts were similar between groups. The intervention was well tolerated with no major safety concerns.

Conclusions Three subconjunctival injections of 2.5 mg bevacizumab are more effective than placebo at inducing the regression of recent-onset CoNV. Further studies are needed to confirm this effect and our data suggest that a sample size of 40 patients per treatment group is required.