Nifty new approaches to genetic variants on display at ARVO 2025 are accelerating understanding of underlying disease mechanisms in eye disease.
Researchers at The Association for Research in Vision and Ophthalmology 2025 Annual Meeting (ARVO 2025) believe its time to move beyond association in genetic research’s quest to understand ophthalmic disease.
At a time when genetic discoveries in ophthalmology are outpacing our ability to understand their biological impact, presenters showcased their innovative approaches to bridge this gap and accelerate the pace of understanding how genetic variants map to functional changes in eye diseases like Fuchs’ dystrophy, glaucoma and more.
Presenters showed how they are leveraging modern technology and methods to elucidate the oft-elusive functional consequences of genetic variation—and why this represents the next waypoint in human ophthalmic genetics and the hunt for novel therapeutic approaches.
READ MORE: The Promise of Gene Therapy in Retinal Disorders
Exploring the non-coding genome
Dr. Timothy Cherry (United States) opened the session with his work on non-coding variants in retinal disorders. His presentation revolved around his team’s development of machine learning approaches to predict how variants in regulatory elements—enhancers and promoters that control gene expression—affect function.
“We were inspired by first-in class-CRISPR Cas9 based gene therapy used to treat sickle cell disease and beta thalassemia,” Dr. Cherry said. “This inspiration was that not only can we think of the non-coding genome as harboring disease causal variants, but we can also think of functional elements within the non-coding genome as therapeutic targets to modulate disease.”
Dr. Cherry’s team identified a non-coding regulatory element associated with age-related macular degeneration that controls a microRNA involved in glial cell function. They developed a computational model that can predict the impact of variants on regulatory element function, validated through massively parallel reporter assays.
Most recently, they created a CRISPR-based method to test multiple regulatory elements simultaneously, significantly accelerating functional validation.
The key takeaway? “If you only take one thing away from this talk, I hope you take away an appreciation for the fact that there are disease associated variants within the non-coding genome—and that it’s well worth our while not only to identify those variants and characterize them, but also to understand more about what these elements are doing during normal retinal development,” he said.
READ MORE: One Day, Three IRD Gene Therapy Triumphs
High-throughput splicing spice
Dr. Kinga Bujakowska (United States) took the topic from non-coding variants to variants of unknown significance (VUS) as she presented her lab’s high-throughput splicing assay that helps resolve these vexing variants—a persistent challenge in genetic testing.
“About 40 percent of cases where we perform genetic diagnostic testing come with variants of nonspecific significance,” Dr. Bujakowska noted, highlighting the urgency of the problem. Her team’s proposed solution was to developed a mini-gene approach that can evaluate thousands of variants simultaneously for their effect on RNA splicing.
After testing 4,500 variants, they found approximately 10% caused aberrant splicing. Interestingly, prediction tools like SpliceAI missed many variants that their assay detected, highlighting the value of experimental validation. Dr. Bujakowska emphasized in her conclusion, however, that simply detecting a splicing effect isn’t enough—and there is more that can be done. “I feel like there needs to be like a gene-by-gene guideline to interpret which of those variants could be truly pathogenic.”
Evolution of prediction tools
Dr. Carlo Rivolta (Switzerland) traced the evolution of computational tools for predicting variant pathogenicity, from simple substitution matrices to modern AI systems. Unlike straightforward genetic changes, missense variants present unique challenges.
“In contrast to loss of function, which are very clear cut, there’s no real automatic way of telling whether a missense variant is pathogenic or not,” Dr. Rivolta explained. “In the ideal world, we would have to validate every single one of them by experimental tests.”
His comparative analysis revealed that no single prediction tool works optimally for all genes. “Are all genes equal? No, some genes can be very sensitive to loss of function mechanisms and DNA functions. Therefore, not all genes should be treated with the same approach.”
Dr. Rivolta also warned about circular reasoning: researchers sometimes classify variants as pathogenic based solely on prediction tools, then use those classifications to train new predictors. This practice undermines model reliability and highlights the need for gene-specific approaches and unsupervised learning models that don’t rely on potentially flawed training data.
READ MORE: A New Dawn for Retinal Care
Animal models in glaucoma
Dr. Benjamin Thomson’s talk focused on glaucoma—and specifically using model organisms to validate glaucoma-associated variants, particularly protective variants that reduce disease risk. He highlighted work on angiopoietin-like 7 (ANGPTL7) and angiopoietin-2 (ANGPT2), where animal models confirmed mechanisms by which these variants decrease intraocular pressure (IOP).
Dr. Thomson’s main point was one rooted in the practical realities of this sort of research, including a call to action for faster approaches. “Geneticists are putting out these studies all the time… they’re averaging 3 and a half years for each one of these genes that we look at, right? So we have 150 years of work here ahead of us,” he said. “I think our patients would like us to do a little bit better.”
His lab is adopting innovative approaches to bridge this gap, including AAV-delivered CRISPR/Cas9 for faster variant testing and nanocarriers for tissue-specific gene targeting in the eye’s drainage structures.
His conclusion called for a mixture between traditional approaches and searching for new ways to expedite impractically long research timelines. “Animal models are still the gold standard for validating these genes and variants,” Dr. Thomson stated, “but the traditional systems are just far too time and labor intensive.”
Lessons from Fuchs’ dystrophy
Dr. Alice Davidson (United Kingdom) closed the session with insights into corneal-specific repeat instability mechanisms in Fuchs’ dystrophy. Approximately 79% of Fuchs’ dystrophy cases have CTG repeat expansions in the TCF4 gene, making it surprisingly prevalent in the landscape of genetic disorders.
“Fuchs is actually considered the most common repeat-mediated disease within this much larger family. It’s also a bit of a black sheep, because unlike the vast majority of these conditions, it’s not a neurological or neuromuscular disease,” Dr. Davidson said.
Using optical genome mapping and long-read sequencing technologies, her team discovered that these repeats are dramatically more unstable in corneal endothelial cells than in blood cells from the same individuals. This tissue-specific instability appears central to disease pathogenesis.
“We now know that the repeat is highly vulnerable and unstable within these post-mitotic endothelial cells,” Dr. Davidson summarized. “Our updated hypothesis is that the repeat first needs to become really big over an individual’s lifetime in order for these other downstream cascades to be elicited.”
What it all means
With the research in this session, it appears that the trend of genetics in ophthalmology is moving beyond simply finding genetic associations to developing a mechanistic understanding of variant effects. One main point is that complementary approaches—computational prediction, high-throughput functional assays, and animal models—are needed to fully understand genetic contributions to eye diseases.
These innovations seek to accelerate the pace at which genetic discoveries can be translated into clinical insights, potentially leading to new therapeutic targets for previously untreatable eye conditions. As tools become more efficient and tissue-specific, the gap between genetic association and functional understanding continues to narrow, bringing the promise of precision medicine in ophthalmology closer to reality.
READ MORE: Want more updates from ARVO 2025? Follow our daily coverage here!
Editor’s Note: Reporting for this story took place during the annual meeting of The Association for Research in Vision and Ophthalmology (ARVO 2025) being held from 4-8 May in Salt Lake City, Utah, United States.