Cynthia Toth, MD: What retinal imaging reveals about the preterm infant brain

A new study published in JAMA Ophthalmology suggests that a routine, low-stress eye scan performed at the NICU bedside may help clinicians identify preterm infants at risk for neurodevelopmental delay years before symptoms become apparent.

The research, part of the NIH-funded BabySTEPS initiative, found that optical coherence tomography (OCT) measurements of retinal nerve fiber layer and choroidal thickness correlated with motor, cognitive, and autism-risk outcomes at 2 years of age—building on earlier cross-sectional findings from 2015 with a prospective, gold-standard cohort of 72 infants.

For retinal specialists and ophthalmologists, these findings reinforce the retina's role as an accessible, non-invasive window into brain development, with potential applications extending beyond prematurity to inherited retinal diseases and unexplained pediatric vision loss.

Cynthia Toth, MD, a professor in the Department of Ophthalmology at Duke University School of Medicine and a co-investigator on the BabySTEPS study, sat down with Modern Retina to discuss the study's key findings, the biological rationale behind the nerve fiber layer signal, and what still needs to happen before bedside OCT screening could become part of standard NICU practice.

Transcript lightly edited for clarity and length.

Can you give an overview of the key takeaways and findings of the study?

Cynthia Toth, MD: This study was a repeat, actually, of a study that we completed back in 2015. Back then, we looked across children that we knew had potential brain injury because they'd had MRI scans done because somebody was worried about something.

These were all preterm infants. And at that time, we had found that, using the Envisu OCT system, in kids with problems with their brains, the OCT imaging linked the nerve fiber layer and macular edema with neurodevelopmental outcomes at 2 years. But that was a cross-sectional study. The kids that were entered weren't prospectively entered into a study, so it had some weaknesses.

So this study was a gold-standard-type study. This is NIH-funded and is part of what's called BabySTEPS. This is a prospective study looking at eye and brain and vision development in preterm infants. So what we just published in JAMA Ophthalmology is that when we looked at 72 of these kids at 2 years out—after they were examined in the nursery, or after they'd been born—we found that, indeed, the neurodevelopment, particularly their motor and their cognitive findings—and you measure it in a test called the Bayley scale scores, which is a test of neurodevelopment in children at age 2.

So those scores were higher for kids who had a thicker nerve fiber layer. They were also higher in kids who had a thicker choroid. We also found that the test of autism risk also correlated with those changes in nerve fiber layer. So everything that had not been done in the earlier study, the OCT couldn't measure that thickness.

But the nerve fiber layer validated our finding from back in 2015. So now, in a whole new cohort—again, we find, in kids who just entered the study because they were preterm, not because they had anything particularly wrong with their brains—we were able to predict their neurodevelopmental findings. And it added to what we would already have guessed based just on their age and birth weight.

Regarding that retinal nerve fiber layer, what's your read on why the retinal nerve fiber layer thickness specifically, rather than other retinal layers, emerged as the strongest signal?

Toth: We think it may be because some of the thoughts about brain development are that inflammation—such as necrotizing enterocolitis, sepsis, and other things that go on and cause inflammation in the infants, either in utero or in the nursery—can have a negative impact on brain development, and that affects the myelination of the tissues in the brain. The ganglion cells in the retina are the only cells that have axons that go and connect in the lateral geniculate within the brain.

So although they're not myelinated within the eye, it's possible that the thickness of this layer in the eye and how it develops in the NICU relates to brain development. After all, the retina, including the ganglion cell layer, is an extension of the diencephalon of the brain. That, and it's the only place that we can look in and, using visible light, see the tissues that are brain tissue.

Would you say it's safe to say, then, that the retina is considered a useful window into brain development? And what makes OCT well suited to studying that connection in very preterm infants?

Toth: Great question. Michael Msall, who's a known child development expert, had initially said the eye really is the window to the brain. OCT is fabulous for this—for infants—for 3 reasons.

One is it's low-stress. So we worry every time we try and peer into a kid or do something that we cause stress that affects their neurodevelopment. So an eye exam with the bright white light and all has stress. A week before, we published a paper showing that, indeed, the testing we're doing using this OCT handheld over babies' eyes is actually lower stress than a regular eye exam of the infants done weekly. So it's low stress.

The second one is it gives us a precise, objective measure of tissue and development that we can follow. So it gives us not just a general idea that we would get from a color photo, but now we can measure these layers and indeed see how measures are changing. And then, third, we've been able to watch that these tissues are growing and changing the way that the infant's brain is.

If you look at the brain of an infant in the third trimester, or in that period that preterm babies are in the NICU, at that same time period, the brain is developing many of its convolutions and much of the change. At the same time, the retinal layers are dramatically changing, and we can actually see this on OCT. So now we can wander in at the bedside, peek into the eye, and get a window into the brain.

And it's something that's low enough stress that it could be repeated on a regular basis.

The study authors described this as a step toward identifying at-risk infants years before symptoms emerge. In practical terms, what would need to happen before bedside OCT screening for neurodevelopmental risk could become part of standard NICU protocol?

So this study actually helped strengthen the use of this as a tool [for] assessing neurodevelopment because of the repetition of a prior study. We'll probably need another prospective study to further validate this. And our challenge is that we actually do not have OCT devices out there for use in kids commercially right now. The device that was developed is no longer carried, since 2023.

So we're waiting for OCT devices to be commercially cleared so that they're available for pediatric use across the US and worldwide. So…that's my conflict of interest…I'm a co-founder of one of the 2 companies that are seeking to develop a handheld OCT device so we can get technology that we know is of use to kids back into regular routine use and not just under our IRB protocols for research.

If a clinician identifies an infant with a thinner retinal nerve fiber layer at 36 weeks, what would you want to be able to do with that information today?

Toth: So the issue is that there are numerous new interventions that are being considered to augment an infant's neurodevelopment.

Much of this is working with the neuroplasticity of a child so that you offer optimal interactions and growth and development to optimize their trajectory for neurodevelopment. But if we look at nutrition and other therapeutics that could be considered, there may be others that could be considered during the NICU stay. So we identified this at 36 weeks, but we're actually looking at kids from early on.

So now we're looking further back in time and saying, what about six weeks earlier—during this really vulnerable period—are there things, are there new drugs that could be tried or other things that could change this trajectory? Can we bump it up? Can we make a difference that's going to affect both [the] eye and brain? We don't have those answers yet.

But this gets us started on a path to identify and be able to use precise measures in studies that are going on now to look for outcomes. If you do a study to improve infant neurodevelopment right now, there's not much that you can monitor for change. You can't put a kid into a big MRI unit and have them hold still and do that repeatedly.

Super expensive, super tough for the kid to hold still. Sometimes they sedate them for that. So that's not routine. Whereas if you could do this—where you can just image at the bedside on a weekly basis and watch, are they growing on a curve, and follow that like normal length or weight curves—then you may have better feedback as to whether drug A or treatment B is better for the child.

This work also sits inside the larger BabySTEPS initiative on ROP and neurodevelopmental findings. How does this neurodevelopmental finding change or expand the goals of that screening program going forward?

Toth: It's really helpful. So this has been NIH-funded research in preterm infants—we thank you for funding it. And this is work on children who—all of them had to reach 2 years [old].

They were enrolled over a 5-year period, so this has been going on for over 10 years. And the children that we followed, we've also been able to find visual acuity outcomes. So we find that the same tissue layer—[the] nerve fiber layer—also predicted poor visual function at age 9 months. So all of this combines together to make a confusing story.

So I wish I had a clear answer—which is, sometimes if you were born and have poorer vision, then potentially you have poorer potential incoming sensory information for neurodevelopment. So we don't know: is the neurodevelopment piece impacted by visual acuity and the vision pathway? Or is the vision pathway, as one quarter of the developing brain, just reflective of an overall neurodevelopmental problem or challenge that could be augmented by, again, whether it's nutraceuticals or health therapies that could improve neurodevelopment?

[Also] the BabySTEPS…study is looking at how do we use these markers to identify direct treatments for retinopathy of prematurity while the kids are in the nursery.

So we've got work going on right now in that area. As far as other areas, for certain there are many children born with inherited retinal diseases or other diseases that affect eye and brain development—whether they're storage diseases, like how materials are stored in cells, or other dysfunctions—that, if they can be picked up early in the retina, then you can step in with your intervention. So you can screen for that disease process by identifying the abnormalities in the retina.

So, for instance, Leber congenital amaurosis is a disease—RPE65—for which there's a gene therapy that can be given to children who have that abnormality. One way to identify the disease is you can look with an OCT in the eye and see abnormalities. So this has use outside of just preterm infants—I think more of infants who have other systemic health problems that could be identified.

Or you could even think about a child who's 3 years old and identified as having amblyopia in 1 eye. Amblyopia is poor vision without a physiologic or physical explanation. But if we don't look with OCT, could we be missing a more subtle disease process that you and I, as adults, would expect someone to evaluate with OCT? So that's been my goal in my research career and in this project: to be able to bring OCT to a population that hasn't had access from birth, whether it's preterm birth up to about age 2 to 4, which is an area where kids just aren't going to sit at a slit lamp and get an eye exam, like you and I would.