The media loves a good story and when a medication that has been given to children may have a link to “brain damage” you better believe it will gain some traction.
Ever since the first reports began emerging suggesting laser treatment for ROP may have an alternative in Avastin (bevicizumab) skeptics have abounded. After the link was established between high vascular endothelial growth factor (VEGF) and ROP it was not long before the thought of using Bevicizumab to reduce abnormal vessel proliferation of the developing retina. This medication is an anti-VEGF medication that had been used for treatment of colorectal cancer and with this link to ROP a new use had been found for an existing drug. It was big news and created a great stir in many units around the world.
What followed was solid research demonstrating efficacy with the BEAT-ROP being the most influential trial. The full text of this study can be found here. The results of this study of 150 infants (300 eyes) clearly demonstrated that for those with 3+ disease in zone 1, outcomes were better than if laser was used for treatment. For Zone 2 disease there really was no benefit to Bevicizumab but due to the ablative nature of laser, infants in the long run (particularly for Zone 1 disease) can be left with visual compromise later in life including myopia, strabismus as well as amblyopia.
It is for these reasons that I have to confess I have been quite keen on Bevicizumab although at least in our own centre laser in general continues to be presented as standard of care with the option for injection. For Zone 1 disease though Bevcizumab is becoming increasingly recommended due to the aforementioned visual sparing effects in the long run.
What is the concern with Avastin?
The main issue as with any new treatment is the question of whether there are any long term consequences to using a medication that could distribute systemically. We know that many ocular medications can enter the rest of the body and given that VEGF is found in virtually every organ system in the body it is appropriate to express some caution before laser is replaced as standard of care. Anti-VEGF activity could for example cause abnormal vascular development in the lungs of these fragile infants who are trying to recover from BPD and make a disease that already has a vascular component worse. A more direct concern could be the effect of altering vascular development so close to the brain with the potential fear of causing developmental impairment as a result. It may seem surprising that there is so little out there on the topic until you consider that the BEAT-ROP study was published in 2011. This is the time five years later that we should expect to see some outcome results and now we have.
This was the title this week of an AAP update after the following article was published Neurodevelopmental Outcomes Following Bevacizumab Injections for Retinopathy of Prematurity by Morin J using data from the Canadian Neonatal Network. The trial is important as data is needed on the outcome of these infants but the message that is coming out of this publication is worrisome to me in that conclusions are being drawn from a study that has significant concerns. The main findings that have raised many an eyebrow were the following comparing injection to laser in that order for the percentages.
Neurodevelopmental Impairment: 78% vs 56%
Severe Neuro disability: 52% vs 29%
The usual concerns are here in that the study is retrospective and therefore prone to the usual types of error in having groups unbalanced. One of these imbalances is in the severity of illness at admission as measured by the SNAP-II score. The babies treated with injections were sicker and significantly so. The scores in this case were medians of 24 in the Bevicizumab group vs 19 in the laser group. This is quite significant as another group using the CNN data previously published their findings in 2013 that among other things the SNAP-II score was predictive of adverse outcome in a group of infants from 23-30 weeks (same group that is screened for ROP) if the score was above 20. Indeed in the current study the median score for the group treated with Bevicizumab was above this threshold so is the treatment the issue or the fact that the babies were sicker to begin with?
The second significant concern I have with this study is the nature of the babies who were treated with injection vs laser. Based on the BEAT-ROP study most centres reserve injections for the most severe Zone 1 disease and that was no different here. The babies receiving injections had aggressive Zone 1 disease 19% of the time compared to the laser group at 7%. These babies had worse eye disease but how does that tie in to long term neurodevelopment? Several papers have linked adverse developmental outcome by itself to the severity of ROP that the infants had. One such paper by Msall looked at the outcome from the CRYOROP study and found exactly this in a follow up of 1199 survivors so a very large study to draw conclusions from.
So we have the current study suggesting that adverse neurodevelopmental outcome is associated with receiving injections for ROP. We also have two very clear confounders that one cannot ignore. The babies who received injections were both sicker at admission and had worse eye disease overall which both are predictive of adverse outcome. Add to this that another retrospective study that is not much smaller than this found no difference in outcome and I think the conclusions of the current CNN paper are suspect. Interestingly I don’t recall seeing any media attention for this negative study but then that tends to be the way these types of results are handled. Kind of boring so not worth reporting?
Where do we go from here?
What we really need now is a double blind randomized controlled trial to sort this out. It needs to be large enough such that the SNAP-II scores are not significantly different and injections and laser need to truly go head to head. This could be accomplished by following up the BEAT-ROP group or by repeating another trial with the primary outcome being neurodevelopment. In the meantime I would urge everyone to take a breath and not panic. While I confess I have a bias towards Bevicizumab for its vision sparing advantages there could be an impact on development that we cannot ignore. We need to study this well though and that will take some time. Until we do both the scientific community and the lay press need to carefully weigh the merits of releasing studies with significant sources of confounding risks as a little fear can be a strong motivator and potentially deprive infants in the NICU of an important option for treatment.