Whether anti-VEGF injections lead to systemic adverse events remains unclear, but more studies accounting for potential biases and confounders will help to reach a consensus. Photo: Leonid Skorin, OD. Click image to enlarge.

A number of studies have reported adverse events related to intravitreal anti-VEGF injections for diabetic retinopathy, but many others have found no association. Recently, a large-scale study of more than 1.7 million patients in the US Veterans Health Affairs database reported an association between intravitreal anti-VEGF injections and systemic adverse events.¹ However, experts point out that there are certain difficulties involved in this type of analysis, making it hard to say whether the effects of the injections could be the true cause of such events.

The retrospective, longitudinal population-based study analyzed data from the US Veterans Health Affairs database on adults with type two diabetes seen between 2011 and 2012. A total of 1,731,782 patients (mean age 63.8, 95.7% male) were included, with 27.5% having diabetic retinopathy and 0.8% receiving anti-VEGF injections. The main outcome measure was the proportion of patients with any incident systemic adverse event, acute myocardial infarction, cardiovascular disease or kidney disease at follow-up.

Of those with type two diabetes, 18.6% developed systemic adverse events based on data extracted between 2013 and 2017. The five-year cumulative incidence of any systemic adverse event was significantly higher in the injection group (37%) vs. the non-injection group (18.4%).

After controlling for age, race, sex, ethnicity, tobacco use, diabetic retinopathy severity, mean HbA1c, number of injections, statin use and mortality risk, the researchers found that anti-VEGF injections were independently associated with an increased chance of developing any systemic adverse event.

“Alternative therapies such as intravitreal steroids or laser may be an option for patients with diabetes who are at high risk of systemic adverse events,” the researchers wrote in their paper.

“Adverse effects including hypertension, proteinuria and thromboembolic events have already been well-documented in patients with cancer who were treated with systemic anti-VEGF therapy,” notes Brian L. VanderBeek, MD, in a commentary on the study.² “While relatively tiny compared with the exposures from systemic dosing, all anti-VEGF agents have been shown to enter the bloodstream after intravitreal injection. Whether these small intermittent exposures are enough to trigger systemic adverse effects is still hotly debated with evidence for and against.”

In his editorial, Dr. VanderBeek says it’s incredibly difficult to know which adverse events were directly caused by intravitreal injections and which were the result of the natural disease progression of diabetes. He writes that the field of pharmacoepidemiology, which analyzes drug impact in real-world scenarios, and is thus not limited by clinical trial criteria, can help “move past the collection of case reports documenting adverse events and toward evidence of adverse events.”

“One difficulty in addressing these questions is that ideally when examining a drug or drug class for associated adverse events, a comparable class of medications can be used as an active comparator.” He explains that this method “helps to reduce the potential for indication bias, which occurs when an effect of the disease is falsely ascribed to the drug used to treat the disease instead of the disease itself.”

The challenge with this, however, is finding an active comparator. Dr. VanderBeek explains that since anti-VEGF agents are effective first-line treatments, most studies compare anti-VEGF users with nonusers, rather than with those using a different drug class for the same condition.

Dr. VanderBeek adds that time can also be a confounder for analysis. “It’s well known that the longer a patient has diabetes, the more complications they are likely to accrue,” he writes. “These parallel pathologic processes are why studies can be found to both suggest diabetic retinopathy is a predictor of end-stage kidney disease and vice versa.”

He says that one way to control for concurrent disease complications would be to use the Diabetes Complications Severity Index. Employing this tool to assess diabetes severity “at a more granular level” may help reduce the time confounder.

“Lastly,” he writes, “causal inference mandates that drug exposures occur prior to and within relative proximity to the outcome. For an adverse effect to be associated with an anti-VEGF agent, the adverse effect should occur while the agent is still in the patient’s system.” He adds that such a large time gap, as seen in the present study, is an argument against a causal association and notes that while a dose-response curve isn’t necessary to prove an association, it would be strong evidence of one.

“Zafar and colleagues have put forth an excellent effort, adding an additional data point for consideration,” Dr. VanderBeek concludes in his editorial. “No single study will be able to resolve all the potential biases that can exist in analyzing this question. However, as additional various study designs are used, hopefully we will move toward a consensus on systemic adverse effects of intravitreal anti-VEGF use.”