Local intra-arterial fibrinolysis for acute occlusion of the central retinal artery: a meta-analysis of the published data

Abstract

Background/aim: Central retinal artery occlusion (CRAO) is typically associated with a poor visual outcome. Several favourable reports of local intra-arterial fibrinolysis (LIF), which involves the superselective administration of a thrombolytic agent directly into the ophthalmic artery, have appeared in the recent literature. The aim of this study was to critically appraise these studies in a collective fashion.

Methods: A meta-analysis was performed of all the published literature germane to LIF in cases of CRAO.

Results: Of the 16 studies identified, all were retrospective and non-randomised. After correction for data duplication, the results of LIF in 100 patients can be reported. A final acuity of 6/6 or better was seen in 14% of patients following LIF, and a visual result of 6/12 or better was seen in 27% of subjects. A poor final acuity of 3/60 or worse was seen in 60.6% of eyes treated with local intra-arterial fibrinolysis. These results compare favourably with conventional forms of therapy. Potentially serious complications were seen in four patients, but no patient suffered a permanent neurological deficit.

Conclusion: The results of this study suggest that there may be a marginal visual benefit associated with LIF compared with conventional management of CRAO. However, the methodology of the cited studies was often unsatisfactory, and a randomised controlled trial of LIF in cases of CRAO is justified. Outside of a randomised clinical trial, the use of superselective fibrinolytic therapy for CRAO cannot be recommended on the basis of current evidence.

Figure 1

Scattergram illustrating the relation between visual outcome and presenting acuity following local intra-arterial fibrinolysis for central retinal artery occlusion. In order to generate a scattergram, categories of Snellen acuity were ascribed an integer score ranging from 1 (NPL) to 13 (6/6 or better). Random noise of not more than plus or minus 0.5 was then superimposed onto the results to ensure that a maximum number of data points were visually represented. NPL = no perception of light, PL = perception of light, HM = hand movements, CF = counting fingers.