When to retest for inherited retinal dystrophies

Genetic testing for inherited retinal dystrophies (IRDs) often leaves patients without a confirmed diagnosis on the first round of testing, and a substantial share of cases return a variant of uncertain significance rather than a definitive result. As gene-specific and variant-specific therapies move from trials into clinical practice, a negative or inconclusive genetic result can affect a patient’s eligibility for treatment.

In this Q&A conversation with Ophthalmology Times Europe (OTE), João Pedro Marques, MD, PhD, FEBOphth—of the Department of Ophthalmology at ULS de Coimbra; the University Clinic of Ophthalmology, Faculty of Medicine, University of Coimbra; the Association for Innovation and Biomedical Research on Light and Image; and the Unidade de Oftalmologia de Coimbra—discusses findings from a single-centre study at the largest referral centre in Portugal, presented at the American Society of Retina Specialists 44th Annual Scientific Meeting.¹ He shares practical considerations for building a systematic re-evaluation strategy, including the technical blind spots that standard exome sequencing misses, how reanalysis and targeted testing moved his cohort’s solved rate from 59.4% to 70.1%, and what retina specialists who are not genetics experts should understand about the limits of a “negative” genetic test before counselling a patient on gene therapy eligibility.

Note: Transcript edited lightly for clarity and length.

OTE: What are the most common reasons a patient with a clear clinical IRD diagnosis still comes back with an inconclusive or negative genetic result, and how does your re-evaluation strategy address that?

João Pedro Marques, MD, PhD, FEBOphth: This is really the question that motivated the whole study. Even with next-generation sequencing, we know from the literature that yields sit somewhere between 53% and 61%, and in our own cohort, the first round of testing only solved 59.4% of families, so we started with roughly 40% of patients still unexplained. On top of that, about one in five IRD cases came back carrying at least one variant of uncertain significance (VUS) rather than a clean answer.

When we dug into why, the reasons broke down into a few recurring patterns. A lot of it is technical: exome sequencing simply doesn’t cover everything—it does poorly in repetitive or GC-rich regions, the classic example being the ORF15 exon of the RPGR gene, and it largely misses deep intronic variants, like the CEP290 allele we ended up having to test for specifically. Structural variants—deletions and duplications—are another blind spot for standard panels and even some exome data. Since these structural variants explain a significant portion of IRD cases, copy number variant (CNV) screening should be performed routinely in genetic testing for IRDs.

Then there is the interpretive side: a lot of variants get flagged but cannot be classified one way or the other because we do not have segregation data from the family, or we do not know whether two variants are on the same chromosome or opposite ones. And then there is just the pace of discovery—new gene-disease associations are being published constantly, so a variant that looks irrelevant today might be recognized as causal in a year or two. We had a case where a hypomorphic ABCA4 variant had been dismissed as too common to be pathogenic, when in fact, in combination with a second variant, it explained the disease.

Our re-evaluation strategy was built to chip away at each of those specific gaps rather than just repeating the same test. We went back and reanalysed existing exome data, ran updated panels, added targeted assays for the regions we know are historically missed—ORF15, the CEP290 intron, MLPA for deletions—and we did systematic familial segregation studies to resolve phase and inheritance. On top of that, every VUS gets checked against the current literature, ClinVar, gnomAD and run through prediction tools like REVEL and VarSome. That combination is what took us from 59.4% to 70.1% solved—86 more families with a real answer, most of them coming from variant reclassification alone.

OTE: How does getting the genetic diagnosis right or wrong affect a patient’s eligibility for current and emerging treatments, and what is the real-world cost of a missed or incorrect diagnosis?

Marques: This is where precision really matters, and it is not something you can be loose about. Gene therapies are mostly gene specific or even variant specific. The clearest example is RPE65, for which there is already an approved gene therapy on the market. If a patient has biallelic RPE65 variants but one of them sits in a region the initial test did not cover well, or comes back as a VUS instead of being confidently classified as pathogenic or likely pathogenic, that patient is effectively locked out of a treatment they might genuinely qualify for—not because the biology does not fit, but because the paperwork of the genetic diagnosis is not solid enough yet.

We did not set out in this paper to map every current or emerging IRD therapy, but the underlying point holds throughout our data: a molecular diagnosis is not just an academic label, it is what unlocks the next step, whether that is a clinical trial, an approved therapy or even just accurate counselling about prognosis and recurrence risk for other family members. Let’s imagine a patient with HPS6-associated oculocutaneous albinism. If genetic testing was not performed, a diagnosis of Hermansky-Pudlak syndrome would probably be missed. There is no gene therapy for that, but knowing the diagnosis allows the clinical team to anticipate and manage life-threatening systemic complications that would otherwise have gone unrecognized. That is the real-world cost of a missed diagnosis in a nutshell—it is not abstract, it changes what a physician does next, whether that is a treatment decision, a screening protocol, clinical multidisciplinary involvement or what you tell a family about their risk.

OTE: From your experience in Portugal and across Europe, where are the biggest gaps in how molecular diagnostics are being applied in IRD care, and what would standardization look like?

Marques: I’d frame this from what our own data and workflow taught us, since this was a single-centre study at the largest referral centre in Portugal rather than a formal multi-country comparison. What stood out to us is that a lot of centres—understandably, given resource constraints—treat a negative or VUS-heavy genetic test as the end of the road. The patient gets told “we didn’t find anything” or “we found something but we can’t be sure it’s the cause,” and that’s where it stops. What we tried to demonstrate is that treating that result as final is a mistake. In our cohort, retesting patients at a median of about 3 years after their first negative result resolved an additional set of families, and that is exactly in line with the joint clinical practice guidelines from the Portuguese Society of Ophthalmology and the Portuguese Society of Human Genetics,² which recommend case- and variant-level reanalysis approximately every 2 years.

If I had to describe what standardization should look like, it is really three things happening together as routine practice rather than as an occasional extra effort: scheduled reanalysis built into the care pathway, not left to chance; targeted testing for the specific regions we know standard panels and exomes miss—ORF15, deep intronic hotspots, structural variants—before a case is ever called negative; and genuinely multidisciplinary review, where IRD and medical genetics specialists sit down together on unresolved cases rather than genetics working in isolation. That is the model we built into this study, and it is what got us from 59.4% to 70.1% solved. I’d also flag that our data do not capture ethnicity, and our cohort reflects Portugal’s own genetic background and founder effects, so how well any of this generalizes to other populations is something that really needs to be tested centre by centre.

OTE: Is the current diagnostic infrastructure—testing platforms, bioinformatics tools and clinical expertise—keeping pace with the gene therapy pipeline, or is there a risk that patients will be left behind because diagnostics cannot move fast enough?

Marques: Honestly, I’d say we’re not quite there yet, at least not in Portugal. We still do not have access to whole-genome sequencing or long-read sequencing in clinical practice, and these techniques have been reported elsewhere to improve solved rates by up to 25% in cases where exome sequencing came back negative. That is a meaningful gap, and it tells us that some of the patients we still count as unsolved probably do have a findable genetic cause, we just do not yet have the coverage to find it with the tools we used here.

Where I think the field is heading, and what we point to in our paper conclusions, is a combination of things: continued research into gene-disease associations, since new ones are being published constantly, better integration of functional validation—things like in vitro splicing assays—alongside the bioinformatic predictions we already lean on, and broader adoption of whole-genome and especially long-read sequencing. Long-read technology is exciting because it is much better at picking up structural variants and resolving phasing, while still holding up for standard single-nucleotide changes, which are the two areas where short-read panels and exomes consistently fall short. I wouldn’t say patients are being left behind in a dramatic sense, but the diagnostic side needs deliberate, continued investment in reanalysis, better coverage and periodic reassessment to keep up, rather than assuming a single negative test at one point in time is the final word.

OTE: For retina specialists who are not genetics experts, what is the most important thing they should understand about the limitations of molecular diagnostics when counselling a patient on gene therapy eligibility?

Marques: That a negative or inconclusive genetic test is not the same as “this isn’t genetic” or “there’s nothing more to find.” That’s the single biggest misunderstanding I’d want to correct. In our cohort, 40.6% of families were unsolved after the first round of testing, and even after we threw everything we had at the remaining cases—reanalysis, targeted assays, segregation studies and literature review—5.6% still ended up as what we call partially solved, meaning we have a variant that fits the phenotype perfectly but cannot yet be formally classified as causal with the evidence available today. That is not a failure of the patient’s biology; it is a limitation of where the technology and the knowledge base happen to be right now.

What that means practically is that a patient who was told a few years ago that their test did not find anything or came back with an uncertain variant may well become eligible for a gene-specific therapy after reanalysis, especially given how quickly new variant-disease associations and testing methods are emerging—thousands of new variants get reported every year. The most useful thing a retina specialist can do is not treat an old genetic result as permanent, but refer back to a specialized genetics team periodically, particularly before telling a patient they’re not a candidate for a specific therapy. In our experience, retesting at around the 2- to 3-year mark is where a meaningful number of new diagnoses come from.

João Pedro Marques MD, PhD, FEBOphth
E: [email protected]

Marques is affiliated with the Department of Ophthalmology, ULS de Coimbra; the University Clinic of Ophthalmology, Faculty of Medicine, University of Coimbra; the Association for Innovation and Biomedical Research on Light and Image and the Unidade de Oftalmologia de Coimbra.

References

1. Marques JP. Enhancing molecular diagnostic accuracy in inherited retinal dystrophies through a personalized re-evaluation strategy. Paper presented at: American Society of Retina Specialists 44th Annual Scientific Meeting; July 15-18, 2026; Montreal, Canada.

2. Marques JP, Soares CA, Carvalho AL, et al. Portuguese Society of Ophthalmology and Portuguese Society of Human Genetics joint clinical practice guidelines for genetic testing in inherited retinal dystrophies. Clin Genet. 2025;107(6):600-611. doi:10.1111/cge.14691