Central retinal vein occlusion (CRVO) remains a common cause of unilateral visual loss. Although it was originally described over 150 years ago by Richard Liebreich,1 our understanding of its pathogenesis, our ability to modify the final visual outcome and the availability of treatments to effectively intervene in the progression of the disorder are all relatively limited. Recently, a number of randomized controlled clinical trials have shown treatment options that provide improved visual acuity (VA) outcomes over the natural history.
Prevalence rates reported from population-based studies range from 0.1% to 0.5% of the middle-aged to older age groups.2,3 After diabetic retinopathy, retinal vein occlusions (RVOs), both branch and central are the second most common cause of visual loss from retinal vascular disorders. Pooled data from individual population-based studies have indicated that the prevalence for CRVO is 0.8 per 1000 (confidence interval 0.61–0.99).4 Prevalence increases with age, and for CRVO, this does not differ by gender or race/ethnicity. Branch RVO was found to be 5.6 times more common than CRVO. It was estimated using these prevalence rates and the 2008 world population that 2.5 million adults were affected by CRVO worldwide.4
CRVO is characterized by superficial and deep intraretinal haemorrhages in all four quadrants of the retina associated with variable degrees of retinal venous engorgement and tortuosity, optic disc swelling, cotton wool spots and cystoid macular oedema (CME; Fig. 1). Baseline VA reflects the severity of the venous occlusion and is influenced by the degree of macular intraretinal haemorrhage, CME and retinal ischaemia. In most reported studies, presenting VA is less than 6/12 and decreases to less than 6/60 for many with the ischaemic-type CRVO.5
Common ocular investigations for CRVO include fluorescein angiography (FA) and optical coherence tomography (OCT). FA will show normal choroidal filling, but there is usually a variable delay in retinal vascular filling because of the obstruction to venous outflow in a vascular system that is normally a closed loop-type circulation. The later phases of the angiogram will show variable staining of the optic nerve head and retinal veins together with variable degrees of vascular leakage in the macular and capillary nonperfusion (CNP). OCT will often show cystic fluid-filled spaces in the macular with retinal thickening and occasionally submacular fluid.
Variable terminology has been applied to CRVO. Liebreich in 1855 first described CRVO as retinal apoplexy.1 Hayreh used the terms venous stasis retinopathy for the milder degrees of CRVO and haemorrhagic retinopathy for the more severe forms.6 More recently, the central vein occlusion study (CVOS) divided CRVO into non-ischaemic and ischaemic forms based on the degree of CNP as defined by FA.7 CRVOs were considered ischaemic when there were more than 10 disc areas (DAs) of CNP seen on the FA. In the CVOS study at presentation, approximately 2/3 were classified as non-ischaemic, and 1/3 were classified as ischaemic.7 In about 7%, the perfusion status of the CRVO could not be determined because of the degree of intraretinal haemorrhage, and these were initially classified as indeterminate, although most were later reclassified to ischaemic.7 The differentiation into the two categories in the CVOS was based purely on FA results. Other investigators have found that functional testing is of significant value in making this distinction especially in the early phases when FA results can be difficult to interpret. VA < 6/60, relative afferent pupillary defects of ≥1.2 log units, significant central visual field defects, extensive haemorrhagic retinopathy with cotton wool spots and reduction in the b-wave amplitude on electroretinography have all been found to be associated with increased retinal ischaemia. While none of these tests has 100% sensitivity and specificity, when combined, they are indicative that an ischaemic CRVO is present, and the risk of neovascular consequences is significant.8
In the acute phase of CRVO, visual reduction occurs due to macular hypoxia, CME and intraretinal haemorrhage. While in some cases, normalization of the retinal venous circulation may occur with minimal sequelae from a visual point of view, there is in most cases a variable degree of long-term visual impairment. Chronic CME may result in retinal pigment epithelial dispersion and clumping underneath the fovea with photoreceptor loss. Other long-term causes of reduced VA include epiretinal membrane formation and macular ischaemia. Anterior segment neovascularization with the development of neovascular glaucoma may also occur depending on the degree of retinal ischaemia. The risk of both anterior segment neovascularization and more rarely posterior segment neovascularization increases with the degree of retinal ischaemia.
In the CVOS study, 34% of initially non-ischaemic eyes converted to the ischaemic category by the conclusion of the 3-year study. The development of ischaemia was most rapid in the first 4 months and progressed continuously throughout the entire 3-year duration of the study.9 Anterior segment neovascularization developed in 16% of patients enrolled in the CVOS study and was seen in 10% of patients classified as initially non-ischaemic. In the central vein bypass study (CVBS), the rate of conversion of eyes initially non-ischaemic in the control group to the ischaemic category over the 18-month follow-up was 20.8%.10 Other authors have found higher rates of ischaemic conversion. Glacet-Bernard et al.11 reported that 54% of eyes presenting initially as non-ischaemic had converted to the ischaemic category at 1 year with 12% developing iris neovascularization and 9% neovascular glaucoma. In the CVOS study for eyes that were initially classified as ischaemic, there was a 35% risk of developing anterior segment neovascularization.9
Most studies of CRVO report deterioration in VA over time with the decrease being most pronounced in the ischaemic group.5 A meta-analysis of studies where the change in VA could be calculated using logMAR units indicated a pooled mean decrease in VA of 10 letters from baseline to 6 months and three letters from baseline to ≥12 months for non-ischaemic CRVO. For ischaemic CRVO, the pooled mean decrease was 15 letters from baseline to 6 months and 35 letters from baseline to ≥12 months.5 The CVOS involved 725 patients with a CRVO of less than 12 months duration and follow-up of 3 years.7,9 This study found that final VA outcomes were highly influenced by VA at presentation and the degree of retinal ischaemia. Sixty-five per cent of patients presenting with initially good VA (6/12) or better maintained this level of VA. For those in the intermediate group of presentation VA (6/15–6/60), VA improved to greater than 6/15 for 19%, 44% stayed in the same range and 37% deteriorated to less than 6/60, For those with initial poor VA (less than 6/60), there was an 80% chance of having a VA of less than 6/60 at the conclusion of the study. The results in the control group of the CVBS study, which had a VA entry criteria similar to the CVOS intermediate group, showed a very similar final VA result at the end of the 18-month follow-up.10 Other studies have found a worse visual prognosis for CRVO, even in eyes that remain non-ischaemic.11 Prognostic factors for a better VA outcome include good baseline VA, smaller extent of retinal ischaemia, younger age, female sex and absence of hypertension.3,11
The pathogenesis of CRVO remains incompletely understood and controversial. The obstruction to central retinal venous outflow is thought to occur in the region of and just posterior to the lamina cribrosa. A combination of factors including local anatomical susceptibility, vessel wall changes, haemorrheologic and thrombotic tendencies may all contribute. The hypothesis that a CRVO is due to a thrombus located in the central retinal vein (CRV), in the region of the lamina cribrosa, is based on anatomic features of the lamina cribrosa and also on an autopsy study of eyes with longstanding CRVO where evidence of thrombosis was found in the region of the lamina cribrosa.12
The CRV in the region of the lamina cribrosa may be prone to obstruction because of distinct anatomical arrangements. The central retinal artery shares a common adventitial sheath with the CRV, and compression of the vein by the more rigid arterial wall, especially in those with arteriosclerotic vascular disease, may occur. Post-mortem studies have demonstrated a natural constriction in the CRVO in this region, in healthy eyes, and more pronounced changes are seen with increasing age.13 High blood velocities in the CRV at this site have been demonstrated, also suggesting the presence of a constriction.14 These changes could then lead to greater shear stresses causing endothelial cell damage that in turn could lead to endothelial cell proliferation and possible thrombus formation. Retinal venous endothelial cell morphology is normally homogenous and distinct from arterial cells; however, in the region of the posterior lamina cribrosa, the venous endothelial cells show arterial-like characteristics supporting the theory that this is an area of haemodynamic stress.15
The earlier hypothesis assumes that the degree of obstruction to venous outflow in the region of the lamina cribrosa is variable, and there is either a complete or incomplete occlusion of the CRV producing the variation in the clinical picture. Other investigators believe the variation in the clinical picture stems from the location of the blockage in the CRV.16,17 Their hypothesis is that an occlusion of the CRVO in the lamina cribrosa region will produce a more complete obstruction to CRV outflow and hence a more severe or ischaemic-type clinical picture, whereas an occlusion further posterior to the lamina cribrosa will produce a less severe or non-ischaemic clinical condition. This is due to the fact that an occlusion posterior to the lamina cribrosa will allow some venous outflow via venous tributaries that occur in the CRV in the retro cribrosal region that can form anastomoses with surrounding veins.
Major risk factors for CRVO are increasing age, hypertension, diabetes, arteriosclerotic vascular risk factors and glaucoma.18–23 Risk of CRVO is reduced with increased levels of physical activity, increasing alcohol consumption and, in postmenopausal women, the use of oestrogens.19,21
The use of antithrombotic drugs is common in patients in the older age group. One study21 found an increased risk of CRVO in patients taking aspirin and warfarin, even when controlling for other risk factors. Whether this reflects the greater cardiovascular risk profile of patients on these agents or is due to other factors is unclear. In any case, the use of anticoagulants in patients at risk of CRVO does not appear to offer any significant protective benefit as other groups have also reported CRVOs developing in patients already on warfarin.24,25
Haemorrheologic alterations in blood components may also play a role in RVO. The retinal venous circulation has a high resistance to flow that may be more pronounced in the region of the lamina cribrosa because of its unique anatomical characteristics.26 This resistance not only prevents the retinal veins from collapsing with elevated intraocular pressure but also limits its ability to adapt to altered blood viscosity. The possible role of blood viscosity in RVO has been raised for over 30 years.27 Blood viscosity is determined by four factors: haematocrit level, plasma viscosity, erythrocyte aggregability and erythrocyte deformability.26 A number of studies have examined the role of altered haemorrheologic characteristics of the blood in RVO with often-conflicting results.27–31 The variation in conclusions may be due to small numbers in many of the studies, heterogeneous patient populations and variations in methods of measuring the rheologic factors. A more recent case-controlled study32 has found a significant association between whole blood viscosity and the erythrocyte deformability index with RVO indicating that haemorrheologic factors may play some role in the pathogenesis.
Dehydration has also been reported as a risk factor in younger patients with CRVO further supporting the role of whole blood viscosity as a contributing factor in the pathogenesis of CRVO in certain situations.33 The role of systemic thrombolytic factors in the pathogenesis of CRVO similarly remains controversial because of conflicting results in the literature. Associations with factor V Leiden, hyperhomocysteinemia, protein C and S deficiency, antiphospholipid antibody syndrome and factor XII deficiency have all been reported with CRVO.34–46 Most of these studies, however, are retrospective, have included all forms of RVO, have small sample sizes and have, for many of the factors investigated, shown conflicting results.
The role of hyperhomocysteinemia in RVO remains controversial as some studies have shown plasma homocysteine levels to be significantly higher in patients with RVO34-38 whereas other studies have failed to show any association.39,40 Elevated plasma homocysteine is a known risk factor for arterial and venous thrombosis. As serum homocysteine levels can vary with age, renal function, the presence of the 5,10-methylenetetrahydrofolate reductase genotype as well as plasma vitamin B12 and folate levels, further studies may prove useful.
Antiphospholipid antibodies are associated with arterial and venous occlusion, and several studies have linked them to RVO, often in those with relatively younger age.41–43 Other studies, however, have not confirmed this association. A recent case-controlled study found no association between RVO patients and thrombophilic risk factors including antiphospholipid antibodies and homocysteine.44 A meta-analysis of thrombophilic risk factors in RVO found associations only with hyperhomocysteinemia and anticardiolipin antibodies, factors that are known risk factors for venous as well as arterial vascular disease.45
The role of thrombophilia in the pathogenesis of RVO remains unclear, and the value of widespread screening of patients with CRVO for these coagulation disorders is yet to be made.46 The exception may be, however, in those with evidence of a family history of venous thromboembolism, venous thromboses at multiple sites or perhaps those in the younger age group.
Treatment options for CRVO up until recently have been limited to largely laser for the neovascular complications of retinal ischaemia.47 Recently, a number of randomized-controlled multicentre clinical trials (RCTs) have examined new treatment options. Therapeutic interventions can be broadly classified into two major categories. The first is those aimed at the causal pathology that is the obstruction to venous outflow. The second category is those aimed at the sequelae of the venous occlusion that includes both macular oedema and the consequences of retinal ischaemia.
Treatments aimed at causal pathology
Abnormalities of platelet aggregation have been demonstrated in RVO, and these appear experimentally to be more completely reversed by antiplatelet drugs such as ticlopidine rather than aspirin.48–50 There is some experimental evidence that this agent may potentially have an effect on RVO, but this needs to be supported by clinical trials. Warfarin usage does not appear to offer any benefit in either a prophylactic role or in the management of CRVO.21,24,25
Thrombolytic agents have been delivered both systemically and locally into the eye in an attempt to lyse the presumed thrombus causing the obstruction to venous outflow. In one RCT, systemic thrombolytic agents were shown to be associated with an improvement in VA; however, these agents delivered systemically are associated with significant haemorrhagic side-effects and are now no longer used.51–53 The risks with systemic thrombolytics have led to these agents being tested locally within the eye. This has included intravitreal tissue plasminogen activator54,55 and direct injection of tissue plasminogen activator into a retinal vein.56,57 The hypothesis by which this approach may confer a beneficial effect is controversial as the presumed thrombus causing the obstruction to venous outflow would be the penultimate event in the process that caused the obstruction, and these changes would remain and presumably cause the thrombus to reoccur. The results from these small-uncontrolled trials are variable with some risk of complications.57
Optic nerve sheath decompression
Sectioning of the posterial scleral ring via an orbital approach was originally proposed by Vasco-Posada.58 Despite one small series showing some improvement in VA parameters in patients with CRVO, others have shown that it is associated with a significant risk, and it is now no longer practised.59,60
Radial optic neurotomy
This surgical procedure involves a vitrectomy following which a radial incision into the nasal scleral ring of the optic nerve is made. The original hypothesis was that a CRVO was similar to a compartment syndrome resulting from increased pressure within the confined space of the scleral outlet affecting the CRV and that this incision would release this. The initial non-randomized studies did show some visual improvement in up to 73%;61–63 however, other studies have not confirmed a similar degree of improvement.64,65 The theoretical basis of radial optic neurotomy (RON) remains contentious. Blood flow in the CRV in the region of the lamina cribrosa has not been found to alter following this procedure.14,66 It has also been proposed that as the lamina cribrosa is a relatively rigid band of collagen and not elastic, the procedure would be unlikely to offer any significant benefits as an incision would have no decompressive effect.67 Chorioretinal anastomoses have been found to occur in up to 47% of eyes treated with a RON, and this may represent the mechanism of any beneficial effect from the procedure.68,69 The creation of a RON involves a vitrectomy, and some have suggested that the beneficial effects are due to this alone.70
Complications of the procedure include vitreous haemorrhage, haemorrhagic retinal detachments, retinal detachment, visual field defects, central retinal artery occlusion and neovascularization.64,65,71–75 Current evidence does not support the use of RON in CRVO, and RCTs will be required to define its role.
Laser chorioretinal anastomosis (L-CRA)
This technique seeks to create a bypass of the occluded CRV by creating an anastomosis between a retinal vein and the choroidal venous circulation (Fig. 2). The procedure involves the use of a high-power argon laser, sometimes combined with an Nd:YAG laser to rupture both Bruch's membrane and the overlying retinal vein to allow an anastomosis to develop, thereby providing an alternative route for the obstructed venous blood to exit the retinal circulation.76,77 The CVBS was the first RCT of this procedure and involved a follow-up of 18 months.10 A successful L-CRA was created in 76.4% of patients in the treatment group. Over the 18-month follow-up, treated eyes that developed an L-CRA achieved an 11.7 letter-mean improvement from baseline over the control group (P = 0.004). The creation of an L-CRA also conferred a protective benefit on progressive retinal ischaemia. In this study, all eyes were initially non-ischaemic, and conversion to the ischaemic category occurred in 20.8% of the control group, whereas, in the group that developed an L-CRA, where the retinal ischaemia was due to progression of the CRVO, only 4.9% progressed to the ischaemic form10 (P = 0.03).
The procedure does have some complications with 18% of eyes developing neovascularization at the L-CRA site. This was often small and easily controlled with sectorial laser. While laser was used in this study, currently more effective agents to control this neovascularization are available with modern anti-vascular endothelial growth factor (VEGF) agents.78 Vitrectomy surgery was required in 9% because of either non-resolving vitreous haemorrhage or macular traction.
Treatments aimed at sequelae
Panretinal laser photocoagulation (PRP)
The CVOS indicated that prophylactic PRP in eyes with ischaemic CRVO decreased but did not eliminate the risk of anterior segment neovascularization. In this RCT, 20% of those receiving prophylactic PRP developed anterior segment neovascularization compared with 35% in the observation group (P = 0.17).47 They did find greater resolution of anterior segment neovascularization in those that were previously untreated with PRP (56%) compared with those that had had previous prophylactic PRP (22%). The recommendation of the CVOS study group was that eyes with CRVO identified at high risk of neovascularization be followed at least monthly for at least 6–8 months when the risk of anterior segment neovascularization is highest. High-risk characteristics include those with large areas of CNP, recent onset, that is, <1 month and those with VA < 6/60. If signs of neovascularization develop, PRP should be performed promptly.47 The problem with the CVOS guidelines is that ischaemia was defined at baseline as 10 DAs or more of CNP in the posterior retina, and as FAs were only performed yearly in the study, the risk of neovascularization was not quantified against the area of CNP. The CVOS did find the rates of anterior segment neovascularization increased from 16% for eyes with 10–29 DAs of CNP at baseline to 52% for eyes with 75 or more DAs. In the future studies with ultrawide field, FA may provide more definitive estimations of risk of anterior segment neovascularization.79 Until then, the decision about prophylactic PRP will need to be made individually based on the risk factors for anterior segment neovascularization and the patient's ability to attend for regular close follow-up.
Macular grid laser
The CVOS also investigated the effectiveness of macular grid laser in those with perfused macular oedema and VA ≥ 6/15. They found that while there was some angiographic reduction in macular leakage, there was no VA benefit, and therefore macular grid laser for perfused macular oedema in CRVO is not recommended.80
Heamodilution decreases blood viscosity and acts on other rheologic characteristics of the blood such as red blood cell deformability and aggregation. The rationale is that by lowering blood viscosity by reducing the haematocrit, there will be improved retinal microcirculation and oxygenation. Haemodilution has shown to improve VA in some studies,81,82 but not in others.83 A more recent RCT of haemodilution using automated erythrocytapheresis did show a VA improvement of 1.7 lines in the haemodilution group versus a loss of 2.3 lines in the control group at 12 months.84
Troxerutin is a flavonoid derivative of rutin, which is able to inhibit platelet and red blood cell aggregation and to increase red blood cell deformability. The hypothesis for its use in RVO is that it would be able to reduce blood viscosity and to improve retinal microcirculation. A small RCT did show an improvement in VA at the end of the 4-month follow-up (P = 0.03).85 The small sample size and short follow-up make conclusions from this study difficult to interpret.
Intravitreal triamcinolone (IVTA)
Intraocular steroids can help stabilize the blood-retinal barrier86 and can down regulate the expression of VEGF.87,88 The Standard Care vs. Corticosteroid for Retinal Vein Occlusion (SCORE) study compared the efficacy and safety of 1 mg and 4 mg doses of a preservative-free, nondispersive formulation of triamcinolone intravitreally with observation for perfused macular oedema in 271 eyes with CRVO89. At 1 year, 27% (1 mg) and 26% (4 mg) of eyes treated with IVTA gained ≥15 letters, compared with 7% of observation eyes (P = 0.01). There was no difference in the effectiveness between the two doses of IVTA. At month 24, however, there was some attenuation of the benefit with IVTA. A loss in VA letter scores of ≥15 letters was noted in 48% of the observation group compared with approximately 30% of those in the IVTA group (P = 0.06). Imaging with OCT showed a negative correlation with VA improvements, with no difference in centre point thickness at all study visits apart from the 4-month visit.
There was also a higher incidence of intraocular pressure (IOP) elevation in the 4 mg (39%) versus the 1 mg group (20%) at 12 months (P = 0.02).
The results from the SCORE study show a significant improvement in VA with IVTA at 12 months. The discordance between VA and retinal thickness is a concern, and it implies that the IVTA has little effect on macular oedema. This, combined with the attenuation of the VA benefit in the second year, indicates that longer term follow-up is required before the full effect of IVTA in CRVO is known.
Corticosteroid intravitreal implants
Sustained release devices have been trialled to provide a more prolonged duration of effect.
In a small prospective clinical trial of 14 patients with CME, secondary to CRVO refractory to other treatments, insertion of a sustained release fluocinolone acetonide intravitreal implant (Retisert; Bausch and Lomb, Rochester, NY, USA) was performed. While there was a significant improvement in VA and OCT thickness from baseline at 1 year, all phakic patients developed a cataract, and 92% of patients required intervention for glaucoma.90
The OZURDEX (Allergan, Inc., Irvine, CA, USA) trial examined in an RCT the effect of a sustained release dexamethasone implant on CME in eyes with RVO.91 This was a 6-month study with two (0.35 and 0.7 mg) implant concentrations. In the CRVO subgroup, there was a significant improvement in VA at days 30, 60 and 90 but not at day 180. At day 90, the 0.35 and 0.7 mg implant groups had a 9.5 and 10 letter, respectively, mean change in VA from baseline compared with −0.5 letter for the sham group (P < 0.001). At day 180, however, the improvement in VA for the two groups had decreased to a mean of 2 and 0 letters, respectively, compared with −2 letters for the sham group (P = 0.3). There was no statistical difference at any stage between the two dexamethasone concentrations. A lower rate of cataract and glaucoma was reported at 6 months compared with other intraocular steroids with and IOP rise seen in less than 16% of eyes.
Intravitreal anti-VEGF agents
VEGF is known to be upregulated in CRVO.92 VEGF antibodies have been investigated as possible treatments for CME associated with CRVO.
- 1Intravitreal bevacizumab (Avastin, Genentech, San Francisco, CA, USA) has shown in a number of small non-randomized trials to have a short-term beneficial effect on CME with an improvement in vision.93–95 The effect lasts in most cases 4–6 weeks, and repeated injections are required to maintain the visual improvement. Longer term data are required to fully evaluate the role of this agent as most studies have only followed patients for up to 1 year with evidence that the CME is still recurring.
- 2Pegaptanib sodium (Macugen, Pfizer Pharmaceuticals, New York, USA) was investigated in a phase two trial with a non-significant result seen at 30 weeks.96
- 3Intravitreal ranibizumab (Lucentis, Genentech) was investigated in the phase three RCT, the CRUISE trial. This 6-month trial investigated monthly intravitreal Lucentis injections (either 0.3 mg or 0.5 mg; Genentech, Inc., South San Francisco, CA, USA) compared with a sham group in patients with centre-involved macular oedema secondary to CRVO.97 At 6 months, there was a mean gain of 12.7 and 14.9 letters, respectively, in the two treatment arms compared with 0.8 letters in the sham group (P < 0.0001). The mean change in OCT central foveal thickness at 6 months was −434 microns and −452 microns for the control groups, respectively, compared with −168 microns for the sham group (P < 0.0001). There was no difference at any time point seen between the two Lucentis doses. At 6 months, the gains in BCVA were maintained on average with a prn treatment schedule extending out to 12 months. Patients in the sham group at the 6-month time point began treatment with monthly 0.5 mg Lucentis on a prn basis, however, despite this gained fewer letters at 12 months than either of the two treatment groups (+7.3 letters vs.+13.9 letters) (P < 0.001). During the second 6 months of the study, recurrent or persistent macular oedema was common necessitating an injection of ranibizumab approximately two thirds of the time in each of the groups.98 While these are impressive results, longer term data are still needed as normalization of retinal venous flow after a CRVO can take a considerable time.