Purpose
Anti-VEGF agents pass into the bloodstream after intravitreal injection, and can affect systemic VEGF levels. Although safe in the general population, safety in at-risk populations remains unclear. A large meta-analysis of patients treated with ranibizumab (RBZ) for AMD, RVO and DME pointed to imbalances in the incidence of cerebrovascular accident (CVA), death, and wound healing complications in DME patients, but not AMD or RVO (Avery, et al. ARVO 2013). Two meta-analyses have pointed to an increased risk of CVA from 0.5 mg RBZ over 0.3 mg or sham in AMD. We hypothesized that there could be a potential safety signal in diabetics receiving prolonged high monthly doses of anti-VEGF agents.
Methods
We searched MEDLINE and unpublished randomized trials of diabetics receiving 2 years of monthly anti-VEGF treatment for DME. A meta-analysis was performed using RevMan 5.3 software utilizing Peto odds ratio (OR) with 95% confidence intervals. Outcome measures included arterial thromboembolic events (ATEs), CVAs, and mortality.
Results
Four trials meet search criteria. RISE and RIDE evaluated 0.3 and 0.5 mg of RBZ. We analyzed the 0.5 mg dose alone to evaluate maximum exposure, and secondarily in combination with the 0.3 mg dose. In VIVID and VISTA we analyzed monthly 2 mg aflibercept (AFL). When compared to sham/laser, anti-VEGF treated patients had an increase incidence of mortality, OR=2.98 (1.44-6.14), p=0.003 (fig 1) and CVA, OR= 2.38 (1.01-5.66), p=0.05 (fig 2). A trend not reaching significance was found for increased ATEs, OR= 1.58 (0.95-2.62), p=0.08. When combining the 0.5 and 0.3 mg doses from RISE and RIDE, the mortality rate was similarly increased, OR= 2.57 (1.31-5.05), p=0.006.
Conclusions
An increased incidence of mortality and CVA was observed in diabetics treated with monthly AFL or 0.5 mg RBZ for 2 years. A similar, but not statistically significant trend was seen in ATEs. These results must be interpreted with caution as they are derived from 4 small trials. In addition, the lack of patient-level data impairs interpretation and prevents time to death comparisons, but the imbalance in death developed in the second year of treatment. Further detailed review of the cause deaths will help with interpretation of these findings.