Value of real-world insights in ophthalmology and GA treatment

Image Credit: AdobeStock/NINENII

Data are powerful. With the ability to gather large data sets more accurately and easily for analysis, researchers can discover trends in the general health of the population and how subpopulations are—or are not—having their needs met. These analyses can empower better treatment and even motivate additional treatment options or strategies to fill the gaps in treatment needs.

In a recent interview, Aracelis Torres, PhD, MPH, senior vice president of data and science at Verana Health in New York, New York, discussed current trends in eye health, geographic atrophy (GA), and ophthalmology. “At the macro level, it’s hard to look anywhere without seeing some mention of artificial intelligence [AI],” she said. “We’ve been seeing a growing trend of figuring out where it makes sense to insert it.”

Torres pointed out that Verana Health unveiled a collaboration with the FDA last year that involved the use of FDA-approved software to deploy AI for screening diabetic retinopathy and understanding the difference in subpopulation engagement and utilization of the technology. “There’s definitely an interest to figure out how we can more effectively, more efficiently, and more objectively identify issues in terms of progression of eye health in the population overall,” she said. “Within the ophthalmology setting, we’ve been seeing an increase in patients presenting with [GA] overall over the past several years, so we are trying to understand where we can accelerate the identification of disease. Because certainly, the earlier you’re able to capture it, the more likely you’re able to impact it in a more meaningful way.”

Torres also explained that there are key trends related to GA in particular that are important for ophthalmic professionals to keep tabs on. She noted that 2023 proved to be an active year in terms of therapeutic agents, with the ophthalmic industry seeing newly approved therapies for GA.

“This was a disease that, until then, lacked any approved treatment,” she said. “It has been helpful to understand the adoption of these new therapies, along with any indication of their effectiveness and safety.”

Although the clinical trials are designed to evaluate a new agent’s efficacy over a certain period, Torres pointed out that they often don’t provide the opportunity to observe long-term safety issues that might emerge years after a study concludes. “We are certainly seeing an increase in the number of patients with [GA],” she said. “These agents are coming at a really important time to try to at least slow down the progression of disease further in the context of this patient population.”

Moreover, these real-world insights can provide additional details to ophthalmologists as they treat their patients. Torres noted that one of the biggest ways that real-world insights provide better information is a pulse on real-time practice patterns.

“The landscape and health care moves so quickly, whether it’s a newly approved drug or a different sequence of treatments that can and should be considered,” she said. “Sometimes it’s hard for any treating clinician to keep up with the latest guidance that is reflective of routine clinical care. Sometimes they rely on more of their immediate network, maybe the practice patterns within the site that they are delivering care in.”

When it comes to real-world insights, Torres pointed out that it can help to zoom out to understand, holistically, what the general use of a particular therapy is, especially with newly approved agents such as the ones available in GA. This, she added, can lead to several questions, including whether there are certain types of populations or subpopulations that the therapy is more likely to be used in. Are we seeing early signs of effectiveness? Are there still areas of unmet needs that are not responding in the way we would want?

These insights also can empower researchers to develop new standards of patient care or therapeutics, which Torres said can end up enabling the development of more targeted therapies, therapies that are going to lend themselves better, because of that increased granularity of the understanding of patient phenotype. “It really gives us an understanding of unmet needs,” she said. “What if our latest and greatest therapeutic agents aren’t resonating or aren’t enabling a good response in those subpopulations? Well, let’s dive deeper into those subpopulations. Do they have common characteristics—whether that’s demographic or clinical attributes—to drive the more specialized development and advance patient care?”

As data have been gathered, there is always the chance that it could uncover something in the research on GA that is unexpected or not at the forefront. “In our work, particularly within electronic health record [EHR] systems, we’ve noticed that information exists in diverse formats and ways,” Torres said. “There are common structured data points of categorizing a patient using [International Classification of Diseases] codes, [Current Procedural Terminology] codes—classification systems that make it very easy, almost out of the box, to use the data.”

Torres said something researchers observed in GA is that there is quite a bit of heterogeneity on where clinicians are documenting their GA diagnoses, which can come either in the clinical notes, in free text format, or in clinical images to extract diagnoses.

“This may shift and change over time now that we have 2 newly approved drugs, because sometimes the difference in documentation could be driven by an incentive to document in a certain way, whether that’s billing requirements or other requirements that helped to drive a specific type of documentation workflow,” Torres said. “Overall, the more we understand how clinicians document, the more that it enables us when we’re organizing the data to be able to really flesh out a comprehensive picture of the underlying prevalence of [GA].”

Torres added that if there is no difference in documentation, it can be a case of having been undercounted for a period of time using some of the more traditional data sources. “At Verana, we’ve been able to extract the full picture from the EHR,” she said. “Because we don’t solely rely on structured data, we also pull from some of those clinical texts, which often have really meaningful clinically rich data and details within it. We’ve also been able to augment this and corroborate it with imaging data, ensuring there is alignment and corroboration across those different data sources and types, to be able to understand how many patients truly have this disease, because that informs the broader impact that a newly approved drug can have.”

Torres also pointed out that whenever a new therapy comes to market, there often can be a period of time before there is a unique code that’s assigned to that therapy. “Often in the initial [several] months, there could be more of an unspecified drug code that’s assigned to it,” she said. “What that means is it can limit understanding to a level of granularity the types of patients that are receiving the drug.”

Torres further explained that the value of the company’s platform is that it is able to extract from clinical notes and other sources of information. “We’re able to apply that specificity early on before the rest of the coding system picks up on the assignment of a formal code,” she said. “When we think about these early approvals and the early users of that, the more we can home in on the types of patients sooner rather than later; it unlocks the conversation of real-time, real-world insights. The longer you wait to understand what that patient population looks like, the more stale the data become and the less useful the insights become. We want to make sure that we put robust real-time insights at the fingertips of the clinicians [as much as we can].”