In their article “Risk of Myocardial Infarction and Stroke With Single or Repeated Doses of Intravitreal Bevacizumab in Age-Related Macular Degeneration,” Etminan and associates perform a retrospective study and a nested case-control study in order to examine the risk of myocardial infarction (MI) and stroke associated with bevacizumab intravitreal injections for neovascular age-related macular degeneration (AMD).

In the retrospective study, the authors compare 5644 neovascular AMD patients who were administered bevacizumab to 2564 neovascular AMD patients not taking inhibitors of vascular endothelial growth factor (VEGF). The unexposed group consists of subjects diagnosed with neovascular AMD between 2000 and 2007, according to ICD-9 codes, while the treated group received bevacizumab between 2009 and 2013.

Both cohorts were followed to the first MI event or the last date of follow-up. The rate of MI was 11/1000 person-years in bevacizumab-treated patients and 14.9/1000 person-years in patients not exposed to bevacizumab. The adjusted relative risk (RR) for MI was 0.70 (95% CI = 0.46–1.2).

The authors conclude that bevacizumab injections do not increase the risk of MI and stroke in patients with neovascular AMD. Their results would even indicate that receiving bevacizumab decreases the risk of having an MI, which would contradict most prospective studies comparing bevacizumab to other anti-VEGF agents.

We think that several methodological issues challenge those conclusions.

First, the incidence of stroke and MI in the studied population is not constant between the 2 analyzed periods.

Then, anti-VEGF therapy was available prior to 2007 in Canada (though reimbursed by the British Columbia Provincial Health Service Authority since June 2009). Let’s consider the subset of patients that could have been treated by anti-VEGF but were not treated. It is likely that physicians preferred not to experiment with novel agents with possible systemic adverse effects in their more severely diseased patients. Those patients would be included in the control group but not in the bevacizumab-treated group. Moreover, the control group may include patients in such poor general condition that they could not receive anti-VEGF therapy after this option was proposed to them (at least 1 ophthalmology visit is enough to be included in this group). Hence, the baseline systemic health of the control arm is possibly worse than the bevacizumab-treated cohort. Regression models cannot correct for all those morbidities, particularly those affecting mobility.

Regarding the nested case-control study, the authors find a median number of 3 intravitreal injections in the year prior to MI and calculate an adjusted RR for MI among those receiving 3 or more bevacizumab injections of 0.71 (95% CI = 0.41–1.22) compared to those receiving fewer than 3 injections. Once more, a cardiovascular protective effect of bevacizumab injections would be disturbing. We must undoubtedly consider with scrutiny patients started with anti-VEGF but not achieving induction. Those patients, whatever may be the causes of noncompliance, should certainly not be seen as a reference group for the assessment of dose-related risk.

For all those reasons, we cannot share the reassuring views of the authors on bevacizumab systemic side effects.