Retinal Vein Occlusion and Traditional Risk Factors for Atherosclerosis

Objective  To determine whether retinal vein occlusion (RVO) is related to systemic hypertension, diabetes mellitus, and hyperlipidemia.

Methods  We systematically retrieved all studies published between January 1985 and July 2007 that compared cases with any form of RVO, including central and branch RVO, with controls. We generated pooled odds ratios (ORs) and estimates of the population-attributable risk percentages for systemic hypertension, diabetes mellitus, and hyperlipidemia.

Results  Of 21 studies, including 2916 cases and 28 646 controls, both hypertension (OR, 3.5; 95% confidence interval [CI], 2.5-5.1) and hyperlipidemia (OR, 2.5; 95% CI, 1.7-3.7) were significantly associated with any form of RVO; the association was less pronounced for diabetes mellitus (OR, 1.5; 95% CI, 1.1-2.0). Similar results were found in cases with central RVO and branch RVO. The percentage of cases with any form of RVO attributed to hypertension was 47.9% (95% CI, 31.2%-63.1%), to diabetes mellitus was 4.9% (95% CI, 0.8%-11.5%), and to hyperlipidemia was 20.1% (95% CI, 5.9%-43.8%).

Conclusions  Hypertension and hyperlipidemia are common risk factors for RVO in adults, and diabetes mellitus is less so. It remains to be determined whether lowering blood pressure and/or serum lipid levels can improve visual acuity or the complications of RVO.

Retinal vein occlusion (RVO) is the second most common form of retinal vascular disease causing vision loss.¹ Branch RVO (BRVO) may result from compression of a branch retinal vein by an adjacent retinal artery, typically at a common crossing point. Central RVO (CRVO) typically arises when the central retinal vein is occluded posterior to or within the lamina cribrosa.^1,2Retinal hemorrhage and edema may complicate either form of RVO, resulting in partial or complete loss of vision. Neovascularization of the iris, anterior chamber angle, and retina in CRVO and BRVO may be a further complication.^1,2

Although thrombophilia defects are often associated with lower limb deep vein thrombosis and pulmonary embolism, they are frequently absent in persons with RVO.³ In a recent meta-analysis by Janssen et al, only hyperhomocysteinemia and the presence of anticardiolipin antibodies were more prevalent in persons with RVO.³ In their conclusions, the authors questioned whether conventional systemic risk factors for atherosclerosis play a dominant role in the pathogenesis of RVO. We undertook a meta-analysis to compare 3 common risk factors for atherosclerosis—hypertension, diabetes mellitus, and hyperlipidemia—in persons with and without RVO.

Two investigators (P.R.A.O. and J.G.R.) independently searched the PubMed and EMBASE databases for articles published from January 1985 to July 2007. The following search expression was used: “(retinal vein) AND (thrombosis OR occlusion OR obstruction) AND (hypertension OR diabetes mellitus OR dyslipidemia OR hyperlipidemia OR lipids OR cholesterol).” All searches were limited to English language and human subject studies.

The abstract of each article was read and determined to be eligible according to the following criteria: (1) the study included at least 10 cases with any form of RVO; (2) a control group without RVO was included; and (3) the number of cases and controls with hypertension, diabetes mellitus, and/or hyperlipidemia was also reported in the study. Full-text articles deemed eligible were reviewed to ensure that they met the inclusion criteria.

Study design and patient characteristics, methods used to define atherosclerosis risk factors, and the prevalence of risk factors in cases and controls were put into standardized tables. Hemiretinal vein occlusion was considered to be a variant of CRVO, in keeping with the studies included herein and the recommendations of others.^2,4 If there was any uncertainty about a study's data, the corresponding author was contacted by e-mail.

Among persons with and without RVO, a DerSimonian-Laird random-effects model was used to derive a pooled prevalence and 95% confidence interval (CI) for each atherosclerosis risk factor.⁵ This was separately performed for any type of RVO, CRVO, and BRVO. This random-effects method weights the relative contribution of each study to a pooled event rate based on the variance both within and among studies.⁵ Statistical heterogeneity for the pooled event rate was determined using the Breslow-Day method, with P < .05 representing significant heterogeneity.⁶

For studies of any type of RVO, CRVO, and BRVO, an odds ratio (OR) and 95% CI were also calculated for each atherosclerosis risk factor, comparing persons with vs those without RVO. Derivation of a pooled OR and 95% CI, along with testing for statistical heterogeneity, were performed with the same meta-analytical methods,^5,6 using Meta-Analyst statistical software.⁷

For any type of RVO, as well as CRVO and BRVO, the population-attributable risk percentage (PAR%; the percentage attributed to each atherosclerosis risk factor [ie, hypertension, diabetes mellitus, and hyperlipidemia]) was calculated for each atherosclerosis risk factor using the following equation:

PAR% = {(Atherosclerosis Risk Factor Rate Among Controls) (Pooled OR − 1) /[(Atherosclerosis Risk Factor Rate Among Controls) (Pooled OR − 1) + 1]} × 100

A 95% CI for the PAR% was also derived with this above equation, using the upper and lower 95% CI for the pooled OR for each risk factor, as well as the upper and lower 95% CI for the pooled rate among unaffected controls.

A total of 295 citations were initially found in PubMed and 215 in EMBASE. Of these, 20 articles comprising 21 studies met the inclusion criteria (Table 1).^8-27 Nine of 21 studies were from the United States, 3 from Italy, 3 from Turkey, and 1 each from Armenia, Australia, Austria, Iran, Israel, and Japan. All but 4 were case-control studies: 2 were cross-sectional studies^10,23 and 2 were cohort studies.²⁷

Of 2916 people with RVO, 1398 (47.9%) had CRVO and 1518 (52.1%) had BRVO (Table 1). Controls typically were recruited from ophthalmology practices at the same center or from the same region as the RVO cases. Most cases and controls were aged 60 to 65 years. Two studies used the same control group.^13,14

All studies assessed cases with RVO for the presence of hypertension, 20 assessed for diabetes mellitus, and 9 for hyperlipidemia (Table 2). The methods used to define the presence of hypertension varied across studies. In most studies, hypertension was defined as a documented elevated blood pressure measurement, although a history of hypertension was also used. In addition, the definitions of diabetes mellitus and hyperlipidemia also varied among studies (Table 2).

Among 2916 persons with any form of RVO, the pooled prevalence of systemic hypertension was 63.6% (95% CI, 58.9%-68.1%) compared with 36.2% (95% CI, 30.7%-42.2%) among 28 646 unaffected controls, which was equivalent to a pooled OR of 3.5 (95% CI, 2.5-5.1) (Table 3 and Figure, A). Although most studies described a positive association between hypertension and RVO (Figure, A), a significant degree of heterogeneity was present for the pooled estimates. Of 21 studies, 1 found a higher rate of hypertension among controls, but the difference was not significant, and the method used to define hypertension was not described.¹² The PAR% for hypertension among persons with any form of RVO was 47.9% (95% CI, 31.2%-63.1%) (Table 3). A significant association was seen between hypertension and both CRVO (pooled OR, 3.8; 95% CI, 1.9-7.4) (Table 4) and BRVO (pooled OR, 3.0; 95% CI, 2.0-4.4) (Table 5).

Diabetes mellitus was slightly more prevalent among 2877 cases with any form of RVO (14.6%; 95% CI, 12.3%-17.3%) than the 13 225 unaffected controls (11.1%, 95% CI, 9.2%-13.4%) (Table 3 and Figure, B). Although the pooled OR was significant (1.5; 95% CI, 1.1-2.0), the risk of RVO was not consistently higher in all studies, with opposing results in 4 studies.^15,22,24,25 The PAR% for diabetes mellitus among persons with any form of RVO was 4.9% (95% CI, 0.8%-11.5%) (Table 3). Although CRVO (pooled OR, 2.0; 95% CI, 1.2-3.4) (Table 4) was significantly associated with the presence of diabetes mellitus, BRVO (pooled OR, 1.1; 95% CI, 0.8-1.5) was not (Table 5).

Hyperlipidemia was not defined in a consistent manner across studies, and in some studies, no definition was provided (Table 2).^16,18,20 For any form of RVO, hyperlipidemia was more than twice as common among the 1393 cases (35.1%; 95% CI, 22.0%-51.0%) than the 2642 controls (16.7%; 95% CI, 9.1%-28.6%), equivalent to a pooled OR of 2.5 (95% CI, 1.7-3.7) (Table 3 and Figure, C). These pooled estimates were statistically heterogeneous, however. Similar significant risk estimates were seen for persons with CRVO (Table 4) and BRVO (Table 5). The PAR% for hyperlipidemia among persons with any form of RVO was 20.1% (95% CI, 5.9%-43.8%) (Table 3).

In this meta-analysis of 21 observational studies including 2915 persons with RVO, we observed a more than 3.5 times higher risk of RVO in cases with systemic hypertension. About a 2.5-times higher risk of RVO was associated with hyperlipidemia, but there was only a modest correlation between RVO and the presence of diabetes mellitus.

The definitions of hypertension, diabetes mellitus, and hyperlipidemia varied both within and among studies (Table 2). This was partly due to the changing diagnostic criteria for each condition during the 2 decades of study and the lack of formal measurements from close to the time of RVO onset. In fact, the time of onset of RVO symptoms was rarely considered in relation to each atherosclerosis risk factor. The time between RVO onset and case enrollment varied from weeks^16,24 to 2 years.¹⁷

Some studies did not define a given atherosclerosis risk factor,^{12,16,18,20,24} and in most studies, the severity of hypertension or hyperlipidemia was not reported. Although many studies matched controls to cases on age and sex, they did not control for other relevant factors, such as body mass index, renal function, smoking, or intraocular pressure. The latter is important because ocular hypertension, which is also associated with systemic hypertension, is a probable risk factor for RVO.^11,14,19 Previously, some studies have distinguished between ischemic and nonischemic RVO,¹⁴ which was not a consideration in most studies included in this meta-analysis. Although combining hemiretinal vein occlusion with CRVO might be questioned, they share similar anatomical features and pathogenesis.^2,4 Finally, we did not attempt to include unpublished or non–English language studies. Taken together, these points may limit the external validity of our risk estimates, especially in the modern era of better-controlled hypertension and dyslipidemia.

Statistical heterogeneity was present for most of the pooled risk estimates. In addition to study differences, including the definition of each risk factor and the interval between RVO onset and its assessment, the method of selecting controls varied largely among studies. Some studies were designed to assess for risk factors other than hypertension, diabetes mellitus, and hyperlipidemia as their main exposure of interest.^{12,15-18,20-22,24-26} This likely manifested as a high degree of variability in the rate of hypertension among controls vs participants with any form of RVO (Table 3), resulting in a high degree of variation in the associated pooled OR (Figure, A). In some studies, the OR may have been further inflated by the biased selection of exceptionally healthy controls, whose reported rates of hypertension and diabetes mellitus were less than 10% and 1%, respectively.^17,19,26 Although a significant degree of statistical heterogeneity was found in this meta-analysis, there was a consistently positive association across studies between RVO and both hypertension and hyperlipidemia (Figure, A and C), suggesting that these findings are robust.

This meta-analysis explored whether RVO may be caused by traditional risk factors for atherosclerosis.³ The pronounced PAR% for hypertension (nearly 50%), hyperlipidemia (20%), and diabetes mellitus (5%) in persons with RVO, if causal, would mean that treatment of these diseases might be important in the primary and secondary prevention of RVO. This would be in addition to the prevention of heart disease and stroke, for which there is high-quality evidence.^28,29Accordingly, we recommend that an assessment of blood pressure and both fasting lipid and glucose levels be routinely performed in adults with any form of RVO.At the same time,evidence is lacking about whether lowering blood pressure, serum lipid levels, and/or blood glucose levels can improve visual acuity or the complications of RVO.Hence, a sensible approach is to gradually normalize these measures in accordance with current guidelines,^28,29 with the main goal being the prevention of cardiovascular and renal disease.

Further observational studies should use standardized measurements and definitions of hypertension, diabetes mellitus, and dyslipidemia, each assessed near the time of RVO diagnosis. As was mentioned elsewhere,² data are needed to determine whether excessive lowering of blood pressure in persons with RVO may have negative effects during the acute phase of the disease, and whether it can reduce the long-term risk of neovascular glaucoma or RVO recurrence.

There is a paucity of level 1 evidence regarding the treatment of CRVO and BRVO, including the use of systemic therapy.^30,31 Moreover, some conventional antithrombotic medications do not appear to be protective against RVO and may be harmful. For example, in a recently published case-control study, Koizumi et al found that use of aspirin (OR, 2.66; 95% CI, 1.52-4.64) and warfarin (OR, 3.34; 95% CI, 1.44-7.80) was significantly associated with a higher risk of CRVO.³²

In an effort to better define the role of lipid reduction in persons with RVO, we have embarked on the Atorvastatin Toronto Retinal Vein Occlusion (ATORVO) study (Trial Registration clinicaltrials.gov identifier: NRA2580025). The ATORVO study is an investigator-initiated randomized, double-masked clinical trial comparing 80 mg of atorvastatin calcium once a day to placebo in 180 persons with RVO. Changes in visual acuity and the presence of secondary complications of RVO will be evaluated 24 weeks after randomization (study protocol available by e-mailing the corresponding author).

Those who treat patients with systemic hypertension, diabetes mellitus, and hyperlipidemia should consider that each poses a risk not only to cardiovascular health but also to ocular health.³³ The current findings suggest that a major proportion of RVO cases are caused by highly prevalent atherosclerosis risk factors that can be measured in a routine clinical setting.

Correspondence: Joel G. Ray, MD, MSc, FRCPC, Department of Medicine, St Michael's Hospital, 30 Bond St, Toronto, ON, M5B 1W8, Canada (rayj@smh.toronto.on.ca).

Submitted for Publication: September 6, 2007; final revision received October 19, 2007; accepted October 30, 2007.

Author Contributions: Mr O’Mahoney and Dr Ray had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Financial Disclosure: None reported.

Funding/Support: This study was supported in part by the Division of Endocrinology and Metabolism, St Michael's Hospital, and a Canadian Institutes for Health Research New Investigator Award (Dr Ray). The ATORVO Study is funded by Pfizer Canada through an investigator-initiated grant.

Role of the Sponsor: The sponsors had no involvement in or control over the design and conduct of the study; the collection, analysis and interpretation of the data; the preparation of the data; or the preparation, review, and approval of the manuscript.