Central retinal artery occlusion (CRAO) is an ophthalmic emergency, associated with catastrophic visual loss. Since von Graefe described it in 1859,1 many treatments have been proposed to improve visual outcome, including ocular massage, reduction in intraocular pressure, antiplatelet therapy, heparin therapy, vasodilators, isovolumic haemodilution, hyperbaric oxygen, and embolectomy. None, however, has stood the test of time. Intra-arterial thrombolysis is currently the most widely advocated therapy,2^–12 and success has been enthusiastically claimed for it. The role of intra-arterial thrombolysis in CRAO has recently been discussed in reviews,13 14 one of which is published in this issue of this Journal (see page 588).15

The objective of the present discussion is to consider the question of intra-arterial thrombolysis in CRAO in the light of available scientific facts. A brief discussion is first required of those various basic crucial issues, which determine the effectiveness or ineffectiveness of any therapy. These include the following:

1. Scientific basis: The first essential, does a therapy have a scientific rationale? Without it, any treatment must eventually prove useless or even harmful. One classic recent example of this was optic nerve sheath decompression for non-arteritic anterior ischaemic optic neuropathy.16 That study was first published on an expedited basis and became popular worldwide but was proved by a multicentre randomised clinical trial17 to be harmful: 24% of those with decompression suffered further visual loss, compared with only 12% who were left alone, because the whole procedure had no scientific rationale.18

2. Natural history of disease: The gold standard is to compare the outcome of a treatment with the natural history of the disease. It is not unusual to find natural history being attributed as the beneficial effect of a treatment. For example, Richard et al4 claimed visual acuity improvement in 66% of CRAO eyes treated by local intra-arterial thrombolysis, irrespective of the time when the therapy was given after the onset of CRAO. However, it transpired that the final visual acuity in that study was no different from that of untreated patients.19

3. Factors specific to CRAO: In addition, there are the other important considerations when evaluating any treatment for CRAO.

4. • Retinal tolerance time to acute retinal ischaemia: For management of CRAO, this is the most crucial information, because the chance of recovery of vision only exists as long as the retina has reversible ischaemic damage. An experimental study of CRAO in elderly, atherosclerotic, hypertensive rhesus monkeys (similar to most patients with CRAO) showed that the retina suffers no detectable damage with CRAO of up to 97 min, but after that, the longer the CRAO, the more extensive the irreversible ischaemic retinal damage.20 21 CRAO lasting for about 4 h results in massive and irreversible ischaemic retinal damage.20 21 A review of the various reported studies on thrombolysis in CRAO shows that the time interval between the onset of CRAO and start of thrombolytic therapy has always been longer than 4 h, varying between 6 and 18 h or even longer; by that time, the retina is already dead. Claims of visual improvement in such cases have no scientific rationale.

   • Nature of the embolus causing CRAO: The study by Arruga and Sanders22 showed that retinal emboli are made of cholesterol in 74%, calcific material in 10.5% and platelet-fibrin in only 15.5%. Fibrinolytic agents cannot dissolve cholesterol or calcified material. Therefore, there is no scientific rationale for the use of fibrinolytic agents in at least 85% of CRAO cases. Moreover, there is a little-realised important fact that when the central retinal artery (CRA) is completely occluded, there is no blood flow in the artery from its site of origin from the ophthalmic artery to the site of occlusion; consequently, there is little chance of fibrinolytic agent getting to the site of the thrombus in adequate concentration to dissolve the thrombus.

   • Proof of improvement of retinal circulation following thrombolysis: In CRAO, the key demonstration that thrombolysis is indeed restoring the retinal circulation to restore retinal function would be fluorescein angiography before and immediately after the treatment, showing improvement in retinal blood flow. That information is universally missing from the studies claiming benefit with thrombolysis.

   • Proof of genuine visual improvement: Visual acuity is a function of the fovea only, and not of the rest of the retina; since CRAO involves the entire retina, visual acuity does not provide all the required information for evaluating visual outcome. By contrast, visual fields plotted with a Goldmann perimeter provide information about the function of the entire retina. An effective therapy must result in improvement in both visual acuity and visual fields. Central scotoma is invariably present in CRAO.13 Visual acuity improvement without a corresponding improvement in central visual fields implies that the patient has learnt to fixate eccentrically, rather than that a genuine improvement has occurred.13 23 24

5. Prevalent misconceptions on CRAO: The subject of CRAO is plagued with misconceptions. These are responsible for much confusion and controversy on management. These are discussed in detail elsewhere.25 The following are a few of the major misconceptions:

6. • That the site of occlusion is always at the level of lamina cribrosa: Anatomical study of the CRA has shown that the lumen of the CRA is narrowest where it pierces the dural sheath of the optic nerve and NOT at the lamina cribrosa.26 27 Since embolism is the most common cause of CRAO, the most common site of impaction of the embolus is where the lumen is narrowest.

   • That in CRAO the central retinal artery is usually only partially and not completely occluded. This is perhaps the most prevalent misconception25 and has invariably been used as an argument for the use of thrombolysis therapy many hours or even days after the occlusion. Unfortunately, this concept has been shown to be invalid; in experimental CRAO studies in rhesus monkeys,20 28 29 in eyes with cherry red spot and marked retinal infarction, fluorescein fundus angiography done immediately after experimental clamping or cutting the CRA showed the presence of a variable amount of slow retinal circulation in the vast majority. The mechanism25 29 of that residual retinal circulation and its lack of any impact on retinal function20 are discussed elsewhere.

   • That a treatment which is effective in myocardial infarction and stroke is equally effective in CRAO. This assumption has invariably been put forward for the use of thrombolysis in CRAO, implying that the pathogenesis and management of CRAO are similar to those in those disorders. However, the morphology, physiology and responses to acute ischaemia of the retina are very different from those in the brain and heart. Moreover, CRAO is usually embolic in nature, whereas myocardial infarction is thrombotic—which is why thrombolytic agents work in myocardial infarction. The response of the brain to acute ischaemia and its tolerance to ischaemia are very different from those of the retina.20 Thus, it is an error of judgment to equate CRAO, myocardial infarction and stroke.

   • That CRAO is always embolic or thrombotic in nature. Recent studies have shown that a fall of perfusion pressure in the CRA below the level crucial for maintaining circulation in the artery can result in CRAO.13 Thrombolysis has no role in this case.

   • That eyes with CRAO are at risk of developing neovascular glaucoma. Studies have shown no valid scientific proof of this.25 30 When neovascular glaucoma develops in eyes with CRAO, it is primarily a complication of associated ocular ischaemia.30 31

   • That CRAO is one clinical entity. Recent studies have shown that CRAO actually consists of four distinct clinical entities, with different visual outcome and fundus findings.13 32 These are (i) non-arteritic CRAO, (ii) non-arteritic CRAO with cilioretinal artery sparing, (iii) transient non-arteritic CRAO, and (iv) arteritic CRAO.

7. Natural history of visual outcome in CRAO: There is an almost universal belief that an eye with CRAO has no chance of spontaneous visual improvement. However, a recent prospective study of 260 consecutive eyes with CRAO showed the opposite.13 In that study, in eyes with visual acuity of counting fingers or worse, it improved in 82% of transient non-arteritic CRAO, 67% of non-arteritic CRAO with cilioretinal artery sparing and 22% of non-arteritic CRAO. Visual acuity improvement occurred primarily within the first 7 days (p<0.0001). Central visual field improved in 39% of transient non-arteritic CRAO, 25% of non-arteritic CRAO with cilioretinal artery sparing and 21% of non-arteritic CRAO. The peripheral visual field soon after the onset of CRAO was normal in 63% of eyes with transient non-arteritic CRAO and 22% in those with non-arteritic CRAO. Peripheral fields improved in non-arteritic CRAO (39%) and in transient non-arteritic CRAO (39%). The presence of normal peripheral visual fields is essential for navigation, even in the presence of a central scotoma. Arteritic CRAO, by contrast, does not show any visual improvement, because it is invariably associated with arteritic anterior ischaemic optic neuropathy.33

Thrombolysis is currently the most popular therapy, and success has been enthusiastically claimed for it.2^–12 A critical review of the published studies showed several fundamental problems, including the following. (a) Almost all studies are retrospective. (b) There is almost no randomised, controlled, masked study. (c) The fundamental flaw in most studies claiming visual improvement is the one stated above: they contain no fluorescein fundus angiographic evidence to document improved blood flow immediately after the therapy compared with that beforehand. (d) Practically all the studies lack comparison with a satisfactory natural history control. (e) Almost all studies have lumped CRAO into one category and not classified it into various types to determine the visual outcome; as shown above, a recent study13 has shown that visual outcome varies greatly among the different types. (f) This therapy lacks a scientific rationale in the vast majority, because, as mentioned above, only 15% of the emboli are platelet-fibrin in nature, which fibrinolytic therapy is likely to dissolve—in the remaining 85%, the emboli are made of cholesterol or calcified material, which the fibrinolytic agents cannot dissolve. (g) Most importantly, thrombolytic therapy has invariably been administered 6–18 h or even longer after the onset of CRAO, by which time the retina is already dead.20 21 Other serious problems have also been pointed out.19

Beatty and Au Eong34 performed a meta-analysis of all the published literature germane to local intra-arterial fibrinolysis in CRAO. They concluded that all studies were retrospective and non-randomised, their methodology was often unsatisfactory, and outside a randomised clinical trial, the use of superselective fibrinolytic therapy for CRAO cannot be recommended based on current evidence. Moreover, Framme et al,35 on comparison of visual recovery after intra-arterial fibrinolysis with conventional treatment, found no statistical difference between patients treated with the two types of therapies with regard to improvement of visual acuity, and noted that, additionally, thrombolytic therapy carries an increased risk of a stroke. Fraser and Siriwardena36 searched the Cochrane Controlled Trials Register, which included only randomised controlled trials in which one treatment for CRAO was compared with another treatment, but found no such trials on CRAO that met their inclusion criteria. They found that treatments reported in the literature are of unproven efficacy, and concluded that there is currently not enough evidence to decide which, if any, interventions for non-arteritic CRAO have any beneficial effect. Recently, the “Multicenter study of the European Assessment Group for Lysis in the Eye (EAGLE) for the treatment of CRA occlusion” has been launched to evaluate the role of local intra-arterial fibrinolysis in CRAO.37 The most important consideration in any study is its study design, which determines its outcome, irrespective of the fact that a study is prospective in nature. Unfortunately, several fundamental flaws have been pointed out in this EAGLE study design.38 For example: (a) it is not masked; (b) there is no control group; (c) the eyes with CRAO of up to 20 h are included; (d) its protocol is sketchy and vague, and would be interpreted differently by 16 centres participating in the study resulting in unreliable data; (e) primary end-point criterion is change in visual acuity without visual field information; and (f) there is no classification of CRAO into its various categories. A study with such basic flaws of design is bound to lead to flawed conclusions.

Thus, in conclusion, CRAO is a classic case of “a disease without any treatment has many treatments.” None of the claimed treatments for CRAO has so far stood the test of time, in spite of enthusiastic claims. As regards intra-arterial thrombolytic therapy, a beneficial effect is possible only if this treatment is started aggressively in much less than 4 h from the onset of CRAO, and the occlusion is caused by a fibrin embolus and not cholesterol or calcific embolus.