Systemic Safety of Intravitreal Anti-vascular Endothelial Growth Factor Agents in Age-Related Macular Degeneration

Vaidehi S. Dedania; Sophie J. Bakri


Curr Opin Ophthalmol. 2016;27(3):224-243. 

In This Article

Text of Review

Of the studies examining systemic adverse events associated with intravitreal anti-VEGF agent use, fourteen examined ranibizumab;[11,12,13,25–36] five examined bevacizumab;[37–41] two examined pegaptanib;[42–45] three examined aflibercept;[46–49] seven examined ranibizumab and bevacizumab;[6,15–17,50–53,54] three examined bevacizumab and photodynamic therapy (PDT);[55–57] two examined ranibizumab and PDT;[9,10,58,59] one examined ranibizumab and aflibercept;[5,14] one examined bevacizumab, pegaptanib and PDT;[60] and one examined ranibizumab, bevacizumab, pegaptanib, and PDT[61] (Table 1).

Study duration varied, with 25 long-term studies (≥ 52 weeks)[5,6,9–17,25–35,42–44,50,53,54,57–61] and 11 short-term studies (< 52 weeks).[36–41,45,48,49,55,56] One study reported on short-term follow-up in patients treated with ranibizumab and long-term follow-up in patients treated with bevacizumab.[51] Another study reported short and long-term results in patients treated with aflibercept.[46,47]

Efficacy Measures of Major Trials Evaluating Intravitreal Antivascular Endothelial Growth Factor Agents

Intravitreal anti-VEGF therapy has the potential to stabilize vision in patients CNV lesions. Bevacizumab, the most widely used intravitreal anti-VEGF agent, is not currently United States Food and Drug Administration (FDA)-approved for the treatment of neovascular AMD. CATT was one of the first trials to provide level 1 support of bevacizumab use for neovascular AMD, reporting noninferiority of: ranibizumab versus bevacizumab monthly; ranibizumab monthly versus pro re nata (PRN); ranibizumab versus bevacizumab PRN and ranibizumab PRN versus bevacizumab monthly.[6,17] IVAN[15,16] also showed similar efficacy between ranibizumab and bevacizumab.

At 2 years, ANCHOR[9,10] found a greater clinical benefit with ranibizumab than PDT and MARINA[11] reported prevention of vision loss and improved visual acuity with ranibizumab use. The FOCUS trial compared ranibizumab plus PDT with PDT and found combination therapy to be superior.[58,59] HORIZON was an extension of patients that completed 24 months in ANCHOR, MARINA, or FOCUS and showed that less frequent follow-up and treatment was associated with decreased visual acuity gains.[25] Similar findings were also reported in SAILOR,[28] PIER,[26,27] and EXCITE (Efficacy and Safety of Ranibizumab in Patients with Subfoveal Choroidal Neovascularization Secondary to Age-Related Macular Degeneration)[30] and for aflibercept in VIEW 1/VIEW 2.[5,14]

Definition of Adverse Events and Serious Adverse events

The definition of adverse events and serious adverse events (SAE) was vaguely specified in many studies. Fourteen studies[5,9–11,13,14,25–28,42,58,60] reported events as SAE if they met the Antiplatelet Trialists' Collaboration criteria: nonfatal myocardial infarction, ischemic or hemorrhagic stroke or death owing to vascular or unknown causes. In CATT,[6,17] FDA-specified criteria defined SAE, including events that: resulted in death, threatened death, required or prolonged inpatient hospitalization, resulted in persistent or significant disability or resulted in a congenital anomaly or birth defects. In studies that did not specify criteria defining SAE or report events as SAE the authors of this review categorized the adverse events per the Antiplatelet Trialists' Collaboration and/or FDA criteria.

Nonocular Serious Adverse Events

Incidence of at least one nonocular SAE was reported in 25 studies,[5,6,9,13–17,25,29–35,39,40,46–50,54,55,57–60] ranging from 0[17,25,32,39,40,48,55] to 30.4%,[59] 31.7,[17,25] 39.3[17] and 25.3%[14] in controls, ranibizumab, bevacizumab, and aflibercept groups, respectively (Table 2 and Table 3).

In IVAN,[15,16] there was no difference between ranibizumab and bevacizumab in the proportion of patients experiencing a SAE at years 1 and 2 (P = 0.25 and P = 0.82, respectively), whereas in CATT[6,17] nonocular SAE were higher in bevacizumab-treated patients at years 1 and 2 (P = 0.04 and P = 0.004, respectively). This data from CATT are difficult to interpret though, as participants receiving fewer doses of bevacizumab had a greater risk of adverse events. A meta-analysis of ANCHOR, MARINA, PIER, and FOCUS showed a higher rate of adverse events in ranibizumab-treated patients at year 2.[62]

Vascular Adverse Events

Vascular adverse events reported include arterial thromboembolic events (ATE), venous thromboembolic events (VTE), myocardial infarction, cerebrovascular accident (CVA), transient ischemic attack (TIA), HTN, cardiac failure, and nonocular hemorrhage. There was variation among studies in the definition of vascular adverse events. Some studies reported CVA without breakdown of included events,[26,27,38,58,59] whereas others reported stroke and/or cerebral infarction separately.[9–11,25,29,60] Similarly, some studies included myocardial infarction, CVA, and/or death when reporting overall ATE[6,13,14,15,29,43] and others reported these events separately.[9–11,26,27,38,44,54,59]

Arterial Thromboembolic Events

Incidence of ATE was reported in 26 studies,[5,6,9–17,25–31,34,36,37,40–43,50,51,53,54,56,58–60] ranging from 0[26,37,41,56,60] to 7.1%,[59] 10,[36] 5.0,[17] and 3.0%,[42] and 1.0[5] to 3.6%[14] in controls, ranibizumab, bevacizumab, pegaptanib and aflibercept groups, respectively.

In IVAN,[15,16] more patients reported an ATE in the ranibizumab-treated group than in the bevacizumab-treated group at year 1 (P = 0.03). In year 1 of CATT[6] there was no difference in ATE or VTE between ranibizumab and bevacizumab, although there was a higher rate of adverse events with bevacizumab. In year 2 of CATT[17] there were no differences between ranibizumab and bevacizumab in the rate of death or ATE (P = 0.62 and P = 0.89, respectively). A meta-analysis of ANCHOR, MARINA, PIER, FOCUS, and SAILOR showed that the rates of thromboembolic events were similar in all treatment groups at 1 year.[62] Additionally, although they were retrospective studies, Carneiro et al.[51] showed that bevacizumab was associated with a higher risk (approximately 10 times) of ATE than ranibizumab and Curtis et al.[61] reported an increased rate of stroke with bevacizumab.

Venous Thromboembolic Events

Incidence of VTE was reported in 16 studies[5,6,13,15,16,17,31,34,37,38,40–43,50,52,54,56,58,59] and seven major studies[5,6,12,15–17,42,43,58,59] reported no significant risk of VTE with anti-VEGF agent use.

Myocardial Infarction

The incidence of myocardial infarction was reported in 30 studies,[5,6,9–11,13–17,25–32,34,35,37,38,41,44,45,50,52–54,56–61] ranging from 0[37,60] to 5.4%,[59] 0[13,27,30,34,35,53,59] to 3.6%,[10] 0[37,38,41,56] to 2.0%,[60] 0[60] to 4.8%,[45] and 0.2[5] to 1.3%[14] in controls, ranibizumab, bevacizumab, pegaptanib, and aflibercept groups, respectively.

Two studies reported a significant difference in the incidence of myocardial infarction. In one, there was a higher rate in patients receiving PDT compared to ranibizumab,[61] and in the other there was a higher rate in patients treated with ranibizumab compared to bevacizumab (P = 0.014).[54] The former study was retrospective[61] and both had only 1 year of follow-up. Meta-analysis of ANCHOR, MARINA, and FOCUS reported no association between ranibizumab use and myocardial infarction (P = 0.19).[63]


Stroke incidence was reported in 29 studies,[5,6,9–11,13–17,25,32,34,35,37,38,41,44,47,50–52,54,56–61] ranging from 0[10,15,26,27,30,32,37,41,47,50,56,59,60] to 2.6%,[27] 4.8,[59] 2.7,[51] and 1.1%[14] in controls, ranibizumab, bevacizumab, and aflibercept groups, respectively. No strokes were reported in patients treated with pegaptanib.[44,60]

Reports on the association of stroke and intravitreal anti-VEGF agent use varied greatly. A meta-analysis of ANCHOR, MARINA, and FOCUS found a significant association between ranibizumab use and incidence of CVA (P = 0.045).[63] On the other hand, there were similar rates of stroke across all treatment groups in VIEW 1/VIEW 2.[5,14] Campbell et al.[52] reported no significant association between ischemic stroke and ranibizumab or bevacizumab exposure. Curtis et al.[61] found that bevacizumab was associated with a greater risk of stroke than ranibizumab, although when controlled for socioeconomic status, there was no difference. Findings from these last two studies may not be as applicable as they were retrospective with shorter, varying periods of follow-up.

One risk factor for stroke during intravitreal anti-VEGF therapy may be history of stroke. In SUSTAIN,[29] approximately 10% of patients with a history of stroke, suffered another stroke. Similarly, SAILOR[28] reported a higher risk of stroke in patients with a history of stroke receiving ranibizumab 0.5 mg versus 0.3 mg. The significance of these risk factors for stroke is unclear as there was no control group in these studies.

Atrial Fibrillation

The incidence of atrial fibrillation was reported in six studies.[5,6,14,29,31,58–60] None reported significant rates of atrial fibrillation in treatment groups.

Transient Ischemic Attacks

The incidence of TIA was reported in 15 studies,[5,6,14,15,16,29,34,36,38,41,50,51,53,54,56–59] ranging from 0[34,50,51] to 10%,[36] 0[6,15,41,50,56] to 1.0%,[6] and 0.2[5] to 1.2%[14] in ranibizumab, bevacizumab, and aflibercept groups, respectively. No studies found significant rates of TIA in treatment groups.


Hypertension incidence was reported in 23 studies,[6,9–13,17,25–34,38–40,42–44,48,50,54,59,60] ranging from 0[6,12,13,28,32–34,39,40,44,59,60] to 16.1%,[10] 33.3,[32] 0.9,[54] and 10%[42] in controls, ranibizumab, bevacizumab, and pegaptanib groups, respectively. The incidence of HTN may differ greatly among studies secondary to study design. Some study groups were very small (<=3 patients), reporting high rates of HTN[32] On the other hand, Fung et al.[38] reported a low rate of HTN in a large group, although this may be biased secondary to self-reporting. Additionally, the definition of HTN was not clearly stated in many studies. Some studies reported 'investigator-defined hypertension',[11] whereas others reported 'treatment-emergent hypertension'.[9]

Hypertension is a well documented side-effect of systemic anti-VEGF therapy. In an analysis of ANCHOR, MARINA, and PIER[64] no increased risk of treatment-emergent HTN in ranibizumab-treated patients over controls was found. No association between intravitreal anti-VEGF agents and HTN was reported in any study.

Cardiac Failure

Incidence of cardiac failure, including congestive heart failure, was reported in nine studies.[5,6,13–16,29,46,49,52,59] Only one study, IVAN,[15] reported a significantly higher incidence in the ranibizumab-treated group (P = 0.03).

Nonocular Hemorrhage

Incidence of nonocular hemorrhage was reported in sixteen studies.[5,9–11,13,16,25–30,31,34,36,42–44,60,61] In MARINA,[11] nonocular hemorrhage was defined by Medical Dictionary for Regulatory Activities preferred terms, including: epistaxis, hematoma, ecchymosis, rectal hemorrhage, gastrointestinal hemorrhage, hemorrhagic diarrhea, hematochezia, vaginal hemorrhage, hemoptysis, and cerebral hemorrhage. The rate of nonocular hemorrhage in ANCHOR[9,10] and MARINA[11] at 2 years was consistently higher in the ranibizumab groups. Although both of these were large randomized-controlled trials, in consideration of all data, the association between ranibizumab and nonocular hemorrhage remains ambiguous.

Gastrointestinal Perforation

The incidence of gastrointestinal perforation was reported in two studies,[5,11] and no association with anti-VEGF use was found.


Mortality rates were reported in 26 studies,[5,6,9–12–17,25–33,35,38,42–44,46,47,50,54,57–59,61] ranging from 0[5,12,26,27,30,32,33,35,37,43,58,60] to 11.1%.[25] Curtis et al.[61] reported significantly increased mortality in patients treated with pegaptanib versus PDT, ranibizumab, and bevacizumab at year 1 (P = 0.001).

Vascular Deaths

Incidence of death because of vascular causes was reported in seventeen studies,[5,6,9–17,25,28,30,31,35,37,46,47,54,57,60] ranging from 0[12,15,25,30,35,37,47,60] to 2.1%,[10] 3.7,[25] 2.4,[17] and 1.8%[14] in controls, ranibizumab, bevacizumab, and aflibercept groups, respectively. None of the studies reported an association with treatment with anti-VEGF agents.

Nonocular Adverse Events

Rates of nonocular adverse events are high as these events are less severe than the SAE and include common conditions, such as, nasopharyngitis, urinary tract infection and constipation.[27]

Incidence of nonocular adverse events was reported in 12 studies,[5,6,25,27,29–33,35,36,49] ranging from 71.4[25] to 77.4%,[27] 33.3[32] to 86.9%,[27] 81.1 to 83.7%,[6] and 0[48] to 74%[5] in controls, ranibizumab, bevacizumab, and aflibercept groups, respectively (Table 4).


Hypertension was reported in 15 studies,[5,6,11,25–33,35,42,48,58] ranging from 0[59] to 4.0%,[43] 0[32] to 17.2%,[11] 5.9 to 6.3%,[6] 4.0[43] to 10%,[42] and 0[48] to 9.7%[5] in controls, ranibizumab, bevacizumab, pegaptanib, and aflibercept groups, respectively.

Thirteen studies reported two categories of HTN, SAE, and adverse events, although most did not define the threshold blood pressure (BP).[5,6,11,25,28–33,42,43,48,58] None of these studies reported significant BP elevation in anti-VEGF treatment groups or a dose-dependent relationship.


The incidence of proteinuria was reported in six studies, ranging from 0[11,25,28,31,48] to 1.6%,[25] 2.3[5] and 2.3%[5] in controls, ranibizumab, and aflibercept groups, respectively. VIEW 1/VIEW 2 results at year 1 showed the highest rate of proteinuria, 2.3 to 3.3% in treatment groups, although there was no untreated control group for comparison.[5]

Cardiac Disorders

Incidence of cardiac disorders was reported in six studies.[5,6,15–17,42–44,50,54] Two-year results from CATT[17] showed a higher rate of cardiac disorders in patients treated with bevacizumab (P = 0.11), whereas Kodjikian, et al.[50] reported a higher rate in patients receiving ranibizumab (P = 0.28). Neither was significant.

Delayed Wound Healing

Delayed wound healing incidence was only reported in two studies.[5,29] No association between anti-VEGF use and delayed wound healing was found.


The incidence of gastrointestinal disorders was reported in six studies[5,6,15–17,42,44,50,54] and a significant difference in incidence of gastrointestinal disorders was only reported in one study (P = 0.005).[17] No association between anti-VEGF use and incidence of infection, benign or malignant neoplasms, or nervous system disorders was found.

Confounding Factors

Results may be confounded by control of risk factors. Nine studies[12,37,39,45–47,49,51,55,60] excluded patients with a history of certain cardiovascular risk factors. HARBOR[13] excluded patients with a history of stroke, atrial fibrillation, and uncontrolled BP. Carniero et al.[51] excluded patients with a history of myocardial infarction or stroke in the previous 6 months at the discretion of the treating physician. Selective exclusion may account for the more favorable profile of pegaptanib, although other factors, discussed below, may also contribute. Five studies[9–12,28,52] specified that patients were not excluded because of a history of cardiovascular disease and one[28] included patients if their cardiovascular disease was controlled. ANCHOR[9,10] and MARINA[11] did not exclude patients with preexisting cardiovascular, cerebrovascular, or peripheral vascular conditions. In one study, patients with a history of aspirin use 14 days prior to enrollment were excluded.[36]

Other factors confounding results include reporting criteria, definition of adverse events, and enrollment bias. Some studies only reported events thought to be associated with anti-VEGF therapy. In ANCHOR,[9,10] MARINA,[11] and FOCUS[58,59] no reference was made to whether the reported incidences represented total occurrences or only those attributed to treatment. Finally, study enrollment may be inherently biased because of self-reporting or selective enrollment, as patients with comorbidities are often less likely to participate. This is important to consider when evaluating retrospective studies and surveys as the reported incidence and prevalence of adverse events and risk factors may not accurately reflect the population of patients with AMD treated with intravitreal anti-VEGF injections.

Theories on Mechanism of Increased Risk of Adverse Events

Although the exact mechanisms of increased risk of nonocular adverse events in patients treated with intravitreal anti-VEGF agents are unclear, the presence of the medications in the serum might contribute. Studies have shown a significant decrease in plasma VEGF levels[65,66] and the presence of anti-VEGF agents in the systemic circulation[28] after intravitreal injection. With this in consideration, there may be a dose-dependent relationship with adverse events, although this was not supported by the results of SAILOR.[28]

Thromboembolic Risk

Data on systemic bevacizumab use in patients with metastatic carcinoma indicate an increased risk of ATE. One possible mechanism addresses the role of endothelial cell damage and the initiation of the coagulation cascade. As VEGF is necessary for normal functioning of endothelium, promoting vascular integrity and endothelial cell survival, drug-related toxicity may be secondary to endothelial cell dysfunction.[19] Another mechanism relies on specific drug properties. Pegaptanib selectively targets the VEGF-165 isoform, whereas bevacizumab and ranibizumab target all VEGF-A isoforms.[67] As some isoforms of VEGF may exert positive effects, VEGF-121 and VEGF-189, their inhibition might affect normal cardiovascular function.[67]


One theory of the mechanism of elevated BP in patients receiving anti-VEGF therapy focuses on the role of VEGF in nitric oxide synthase upregulation.[68] Inhibition of VEGF leads to decreased endogenous nitric oxide, resulting in vasoconstriction.[68] Another mechanism addresses the increase in systemic vascular resistance from VEGF inhibition. Such an increase may lead to a decrease in vessel number[69] and affect the renin–angiotensin system.[70,71]


As VEGF is involved in the regulation of the glomerular and tubular capillary network, dysfunction of endothelial cells of the glomerular capillary network may result in disruption of the glomerular filtration barrier, resulting in vascular leak and proteinuria.[72–74]

Nonocular Hemorrhage

VEGF inhibition can cause defects in the endothelium, exposing procoagulant phospholipids and diminishing the regenerative capacity of endothelial cells, ultimately leading to thrombosis or hemorrhage.[75]