RESUMO
PURPOSE: To measure the predictive values of relative afferent pupillary defect (RAPD) assessed semi-quantitatively, and visual acuity (VA) at onset of central retinal vein occlusion (CRVO), for neovascularization. METHODS: Retrospective analysis of the TROXHEMO trial that included patients with CRVO within 30 days after the onset. Inclusion criteria were as follows: semi-quantitative RAPD assessment at diagnosis and/or at one month. RAPD was 'severe' if ≥ 0.9 log. Exclusion criteria were as follows: prophylactic panretinal photocoagulation (PRP) before neovascularization. RESULTS: Among the 119 patients enrolled in the main centre, 101 were analysed. 26 had a neovascular complication during the twelve months of follow-up: rubeosis (19), glaucoma (7) and posterior neovascularization (15). The mean time to onset of a neovascular complication was 4.7 months (1 to 12, median 3 months). All the patients who had a neovascular complication had RAPD at first examination or at one month (negative predictive value (NPV) = 100%) but the positive predictive value (PPV) was low (31%, 95% CI [21%; 42%]). The association 'severe RAPD or VA < 35 letters (ETDRS) at inclusion or at one month' was the best compromise between PPV (53%, [39%; 68%]) and NPV (96%, [92%; 100%]). CONCLUSION: To predict neovascularization, RAPD should be routinely evaluated with filters: the risk of neovascular complication is (a) almost nil if there is no RAPD, (b) very low if there is no severe RAPD and if VA is higher than 35 letters, and (c) higher than 50% if RAPD is ≥ 0.9 log or if VA is less than 35 letters.