We read with great interest the editorial by Peracha and Rosenfeld,1 “Anti-vascular endothelial growth factor therapy in pregnancy: what we know, what we don't know, and what we don't know we don't know.” Retina 2016;36:1413–1417. Within the editorial, the authors discuss the systemic effects of anti-VEGF therapy for the treatment of ROP, as a proxy for assessing the potential effects on the developing fetus. The review of the literature highlights the difficulties and limitations of studying systemic effects in the ROP patient population, because of confounding comorbidities.
Since the preparation of the editorial, two studies were published in the interim that implicated neurodevelopmental sequelae as a possible complication of anti-VEGF treatment in infants with ROP.2,3 This has caused a stir in the ophthalmic and pediatric communities, because the use of anti-VEGF treatment has been increasing. However, we would like to highlight that these studies have many limitations, similar to those outlined by Peracha and Rosenfeld. As a disclosure, our groups at the time of writing usually use laser as first-line treatment, and have no biases in favor of anti-VEGF therapy for ROP.
Numerous studies have now shown that anti-VEGF agents injected intravitreally enter the systemic circulation and suppress systemic VEGF levels.4 Organogenesis is still occurring in these premature infants, and VEGF is a critical cytokine that modulates neurogenesis, although no studies to date have shown that the transient VEGF suppression by intravitreal injections have any clinical implications.5 Lien et al recently reported in a small retrospective study that infants treated with a combination of bevacizumab and laser (n = 16) had higher odds of having psychomotor development index score <70 at 24 months, compared with infants treated with laser alone (n = 33), or bevacizumab alone (n = 12), while controlling for sex, gestational age, and birth weight.2 There were no differences at other time points, or with the mental development index. The issue was that the combination group had the highest percentage of eyes with Zone I disease who received treatment 1.5 weeks earlier, indicating worse ocular disease and systemic issues that are difficult to control for in such a small study.
Several months later, the Canadian Neonatal Network published their retrospective data of infants treated with bevacizumab (n = 27) or laser (n = 125).3 The authors reported that infants treated with bevacizumab had higher odds of “severe neurodevelopment disability” (one of the poorly defined secondary end points, as discussed below) at 18 months. This study has garnered attention, but it should be interpreted with caution because of methodology concerns: 1) The authors do not explain the criteria for which infants were treated with bevacizumab or laser. 2) The bevacizumab group had more Zone I disease, thus having smaller and sicker infants at baseline. This difference is not accounted for in their regression analysis, in addition to birth weight, use of mechanical ventilation, and hospitalization length. 3) The laser group contained 11 infants who were treated for Type 2 ROP. These infants are usually healthier, which disadvantages the bevacizumab group. 4) The bevacizumab group, however, contained the infants who were treated with a combination of laser + bevacizumab. Eyes with the most aggressive disease tend to require combination therapy, and hence usually have more systemic comorbidities also. 5) If the children were too developmentally delayed, neurodevelopmental scoring was not performed. There were nine such cases in the laser group and one in the bevacizumab group. This excluded nine very developmentally delayed infants from the laser group, further biasing the results against bevacizumab. 6) Some of the data, such as estimates of visual acuity, were obtained from parental interviews.