Intraocular injections of anti–vascular endothelial growth factor (VEGF) agents have yielded dramatic improvements in the management of a wide variety of neovascular ocular diseases including age-related macular degeneration (AMD), diabetic retinopathy (DR), retinal vein occlusion, and now even retinopathy of prematurity. These agents have thus become one of the most commonly employed treatments, with 824 525 anti-VEGF intraocular injections given in 2008 in the United States. There are currently 2 widely used agents: ranibizumab (Lucentis), which is FDA-approved, and bevacizumab (Avastin), given off-label. Despite its off-label status, a majority of the intravitreal anti-VEGF injections given in the United States and around the world are for bevacizumab, accounting for 58% of all injections in a US review of Medicare fee-for-service Part B claims for neovascular AMD during 2008.

Two new studies add to the growing debate on the relative merits and concerns on the use of ranibizumab vs bevacizumab for the treatment of AMD. The 1-year results of the Comparison of Age-Related Macular Degeneration Treatments Trial (CATT) have for the first time demonstrated in a randomized controlled trial that bevacizumab and ranibizumab have somewhat equivalent efficacy, with few differences in visual outcomes using either a monthly injection schedule or treatment on an as-needed basis. In terms of safety, the frequencies of arteriothrombotic and venothrombotic events and mortality did not differ significantly between the 2 compounds. However, the proportion of patients with serious systemic adverse events was 24.1% for bevacizumab and 19.0% for ranibizumab ( P = .04). After adjustment for demographic features and coexisting illnesses at baseline, the risk of serious systemic adverse events for bevacizumab was higher by 29% ( P = .04), although no specific organ system consistently accounted for the difference in adverse events, with differences in rates greatest for infections and gastrointestinal disorders. It should be re-emphasized, however, that the CATT was not primarily designed or adequately powered to conclusively detect differences in adverse event rates between these 2 agents.

Another recent study addressed systemic safety. In a retrospective analysis of 146 942 Medicare case records, Curtis and associates reported higher risks of stroke and all-cause mortality with intravitreal injections of bevacizumab as compared to ranibizumab for the treatment of AMD. Further analysis of the Medicare claims database presented at the 2011 Association for Research in Vision and Ophthalmology (ARVO) annual meeting (Gower EW, et al. ARVO 2011 E-Abstract 6644) indicated an 11% higher risk in all-cause mortality and 57% higher risk of hemorrhagic stroke with bevacizumab, with no statistically significant differences in the risk of either myocardial infarction or ischemic stroke.

The systemic safety of intraocular anti-VEGF agents has never been definitively established. Much remains unknown about the pharmacokinetics of these agents in the eye. In theory, bevacizumab appears to confer higher systemic risk for both cardiovascular and cerebrovascular events. The larger molecular weight of bevacizumab, a full-length antibody, has been suggested to have a longer half-life in the eye than ranibizumab, an antibody fragment. Furthermore, the Fc component of bevacizumab, which ranibizumab lacks, has also been postulated to increase risk of systemic adverse events by facilitating systemic absorption, and the systemic half-life of bevacizumab may also exceed that of ranibizumab by up to 3 weeks.

Until recently, clinical data on the systemic safety of both bevacizumab and ranibizumab have been sparse, with studies to date of insufficient quality and size to adequately address safety concerns. The landmark ranibizumab trials individually did not show a significant increase in stroke risk. However, other studies suggest that patients with a history of previous stroke may be at higher risk, as the SUSTAIN study found that 10% of patients who had had an earlier stroke suffered another stroke in the first 12 months. A meta-analysis of all trials of anti-VEGF therapy for AMD up until 2008 also found that ranibizumab was linked to a significant increase in nonocular hemorrhage. The study, however, was inconclusive for bevacizumab due to insufficient data. Only 11 prospective and 6 retrospective case series evaluating bevacizumab were identified, each with a relatively small median enrollment of 48 subjects.

The main obstacle to obtaining good safety data for intraocular bevacizumab is further highlighted in a recent randomized clinical trial of bevacizumab vs laser for retinopathy of prematurity in 150 preterm infants. The Bevacizumab Eliminates the Angiogenic Threat of Retinopathy of Prematurity (BEAT-ROP) study on 150 preterm infants suggested that intraocular injection of bevacizumab may be useful to treat severe retinopathy of prematurity. This study, however, reported a 2-fold higher mortality rate with bevacizumab than laser (6.6% vs 2.6%), with 4 of 5 deaths from respiratory failure. The authors of the study suggested that a trial of 2800 infants would be needed to establish systemic safety, a number that is clearly difficult to achieve.

In the absence of more convincing clinical trial data, the Medicare claims data provide the best evidence to date in evaluating the safety of anti-VEGF agents. In addition, there are 2 indirect but increasingly compelling lines of evidence that may provide further insights.

First, there is clear evidence of greater treatment-related mortality with intravenous bevacizumab in cancer patients. A new meta-analysis of treatment-related mortality with intravenous bevacizumab in cancer patients by Ranpura and associates reported increased mortality with intravenous bevacizumab. Compared with chemotherapy alone, the addition of bevacizumab was associated with an increased risk of fatal adverse events, with a relative risk of 1.46. The most common causes of mortality were hemorrhage, neutropenia, and gastrointestinal tract perforation (7.1%). Another meta-analysis of serious hemorrhage with bevacizumab in cancer patients likewise found that bevacizumab significantly increased the risk of bleeding overall, relative risk 2.48, and the risk of high-grade bleeding, relative risk 1.91.

Second, it is becoming apparent that intraocular injections of bevacizumab can significantly suppress systemic VEGF levels. Although the dose and volume of delivery with intraocular injections (1.25 mg in 0.05 mL typically) are much lower than the typical dose of 2.5 to 5 mg/kg/week administered intravenously, there is accumulating evidence that intraocular injections of bevacizumab can still suppress systemic VEGF levels, in some cases by as much as intravenous injections. Clinical studies have also found evidence of a fellow eye effect following unilateral bevacizumab injections. For example, regression of retinal and iris neovascularization secondary to diabetes mellitus in the injected eye was associated with a subtle decrease in leakage of retinal and iris neovascularization in the fellow uninjected eye in 1 study. Another study reported an acute retinal pigment epithelial rip in the untreated fellow eye following repeated bevacizumab injections. Direct assays of plasma levels of VEGF after a single intravitreal injection of bevacizumab have also shown suppression by almost 117-fold 1 day after injection. This suppression persisted with 4-fold reductions at 1 month after injection. Similarly, Qian found that plasma levels of VEGF were suppressed after intravitreal bevacizumab prior to vitrectomy for proliferative DR. Consistent with these observations, Ranpura reported that the risk of mortality with intravenous use did not differ between high- and low-dose intravenous regimens.

A satisfactory conclusion to the controversy surrounding bevacizumab remains elusive despite the completion of the 1-year CATT results. It also seems unlikely that any of the other comparison randomized clinical trials in progress will provide a more definitive answer in relation to systemic safety, because of their relatively small individual sample size. In the meantime, emerging data emphasize the need for heightened surveillance for systemic adverse events with intraocular anti-VEGF injections for different eye diseases. It is worth emphasizing again that most elderly patients with AMD who might benefit from anti-VEGF therapy are already at increased risk of cardiovascular and cerebrovascular diseases.