This Week in Ophthalmology — Sep 5, 2026

Welcome to This Week in Ophthalmology. This week we're covering 10 notable papers spanning new drug therapies for retinal and refractive disease, systemic risk and drug safety questions that land in our clinics, and a set of studies on how we detect, image, and treat disease across glaucoma, cornea, and ocular oncology. Let's dive in.

We start with pharmacotherapy, where two trials tested drugs for conditions that have had nothing to offer. In JAMA Ophthalmology, Kay and colleagues report TEASE-1, a two-year, multicentre, double-masked, placebo-controlled trial of oral gildeuretinol acetate in Stargardt disease, the most common macular dystrophy and one with no approved treatment [1]. Fifty participants were randomised across two doses and placebo, supplemented by fifty-four natural history cases selected before the analysis plan was finalised. Atrophic lesion growth on fundus autofluorescence was slower on drug, a mean difference of about 0.05 millimetres per year, which the authors frame as roughly a twenty percent relative reduction. In the more conservative prespecified sensitivity analysis restricted to randomised participants only, the reduction fell to about fifteen percent. Safety looked reassuring, with mostly mild or moderate adverse events, no treatment-related serious events, and importantly no reported night blindness or dark adaptation difficulty, which matters for a drug acting on the visual cycle. The honest caveat, which the authors state themselves, is that the clinical relevance of a fifty-micron-per-year difference in lesion growth is not yet established, and the reliance on external natural history controls to reach the headline effect deserves scrutiny. The second drug story comes from the American Journal of Ophthalmology, where Pepose and colleagues report LYNX-1, a phase 3 trial of 0.75 percent phentolamine ophthalmic solution for dim light vision disturbances [9]. One hundred forty-five subjects with mesopic pupils of at least five millimetres and poor mesopic low-contrast acuity were randomised to a nightly drop or placebo for fourteen days. By day fifteen, about a fifth of treated subjects gained the acuity endpoint compared with three percent on placebo, and patients reported significantly less glare, halo, and starburst. Adverse events were mild and transient, with no excess conjunctival hyperemia. For the post-refractive patient with intractable night vision complaints, this is the first pharmacologic option with randomised phase 3 support, though the trial is small and only two weeks long.

Our second theme is systemic risk, and here two papers should change how you counsel patients. In Retina, Wang and colleagues used the TriNetX network with propensity score matching to compare systemic vascular outcomes after retinal artery versus retinal vein occlusion [5]. Arterial occlusions carried substantially higher risk across the board: roughly double the risk of ischaemic stroke compared with vein occlusions, along with meaningfully higher risk of ischaemic heart disease, myocardial infarction, atrial fibrillation, and death. Central retinal artery occlusion was the worst actor, with more than twice the stroke risk of central retinal vein occlusion. The practical message is one many of us already say but do not always enforce: a retinal artery occlusion is a stroke equivalent and warrants urgent, same-day systemic workup, not a routine outpatient referral. Vein occlusions still need cardiovascular risk factor management, but they do not carry the same acute thromboembolic urgency. The drug safety question of the year gets a careful answer in Ophthalmology Retina, where Lee and colleagues emulated a target trial using United States claims data from 2012 to 2024 to test whether GLP-1 receptor agonists raise the risk of non-arteritic anterior ischaemic optic neuropathy [2]. Among nearly twenty thousand patients with type 2 diabetes, using SGLT-2 inhibitors and DPP-4 inhibitors as active comparators, overall GLP-1 use was not significantly associated with higher NAION risk. That is the headline, and it should be reported as a negative primary finding. There were only twenty-nine NAION events in total, so the study is underpowered for a rare outcome, and the authors did see signals in subgroups, specifically among liraglutide users at twelve to eighteen months, and among men and older adults. So this neither exonerates nor convicts the class. For now, the reasonable stance is to mention the uncertainty to patients with crowded discs or prior NAION rather than to withhold a drug with substantial cardiometabolic benefit.

Our third theme is detection and imaging. In JAMA Ophthalmology, Hashemi and colleagues compared event-based progression detection across OCT, OCT angiography, and visual fields in 180 eyes from the Diagnostic Innovations in Glaucoma Study, followed a mean of five years [8]. About sixty percent of eyes progressed by at least one modality, and when eyes did progress, OCT angiography vessel density was the earliest detector most often, in a little over a third of cases, with a mean lead time of about two point three years ahead of visual field progression, compared with about one and a half years for circumpapillary nerve fibre layer thickness. The catch is specificity: among stable eyes, OCT angiography was correct about sixty-nine percent of the time versus about seventy-six percent for OCT. So earlier detection comes with more false alarms, and the authors are explicit that the cost-benefit of adding another test remains unproven. Treat it as complementary, not a replacement. On the uveitis side, Ophthalmology Retina publishes the Multimodal Imaging in Uveitis taskforce consensus on ocular toxoplasmosis from Trinco and colleagues [6]. The experts reaffirm that classic toxoplasmosis remains a clinical diagnosis, with colour fundus photography sufficient for typical focal necrotising retinochoroiditis. OCT earns its place for lesion depth and longitudinal monitoring, autofluorescence for staging lesion evolution, and OCT angiography for detecting choroidal neovascularisation. Fluorescein and indocyanine green angiography are explicitly not recommended routinely. In an era of reflexive multimodal imaging, that is a useful de-escalation.

Finally, surgical and treatment outcomes. In the American Journal of Ophthalmology, Baudere and colleagues compared postcataract endophthalmitis across 23 years at two French tertiary centres, 126 eyes from the mid-2000s versus 108 eyes from 2016 onward [4]. Presentations are milder now, with more eyes seeing 20/400 or better and fewer needing vitrectomy, but the microbiology has shifted uncomfortably. Coagulase-negative staphylococci fell from roughly thirty-eight to twenty-five percent of isolates while Enterococcus faecalis rose from two to ten percent, and Gram-negatives more than doubled. Cefuroxime resistance rose from about a quarter of tested isolates to about two thirds, methicillin resistance among staphylococci rose from about thirty percent to three quarters, and multidrug-resistant phenotypes went from three percent to nearly forty percent. That is a direct challenge to empiric regimens and to intracameral cefuroxime prophylaxis strategies. In the British Journal of Ophthalmology, Yao and colleagues report 1,566 treatment-naive unilateral retinoblastoma eyes from eleven Chinese centres [3]. Intra-arterial chemotherapy achieved higher globe salvage than intravenous chemotherapy in cT3b and cT3c eyes, roughly three to four times the salvage hazard, with about four in five salvaged cT3b or cT3c eyes retaining at least light perception. Critically, overall and metastasis-free survival did not differ from primary enucleation, and high-risk pathological features were not more common. For cT3d eyes, though, salvage rates did not differ and the authors advise prompt enucleation. And in Cornea, Karaca and colleagues compared the sub400 protocol against contact lens-assisted cross-linking in 88 eyes with corneas thinner than 400 microns [10]. At twelve months tomographic stability was around ninety percent in both groups with no endothelial decompensation, and although the contact lens technique produced a significantly deeper demarcation line, that did not translate into better visual or topographic outcomes.

If you only have time for one paper this week, make it the retinal occlusion analysis in Retina [5]. It is the finding most likely to change what happens in your clinic tomorrow, because it converts a familiar intuition about artery occlusions into hard comparative risk data that justifies same-day systemic triage.

Here are the key takeaways from this week in Ophthalmology. Retinal artery occlusion carries roughly twice the stroke risk of vein occlusion and should be managed as an acute stroke equivalent. GLP-1 receptor agonists were not significantly linked to NAION overall in a target trial emulation, but the event count was small and a liraglutide subgroup signal persists, so counsel with nuance rather than withdrawing therapy. Oral gildeuretinol slowed atrophic lesion growth in Stargardt disease by a modest margin whose clinical meaning is still unclear. Postcataract endophthalmitis is presenting more mildly but with markedly more cefuroxime and methicillin resistance, so revisit your empiric coverage. And OCT angiography may flag glaucoma progression about two years before the visual field, at the cost of some specificity.

That's your roundup for This Week in Ophthalmology. The full transcript and references are available on the episode page in your AudioScholar library. This is an AI-curated summary — for clinical decisions, always consult primary sources and current guidelines. See you next week.

If this weekly briefing is useful, follow the show in your podcast app so new episodes arrive automatically. And if it saves you time, a quick rating in your podcast app helps other physicians find the show.

This is an automated summary generated by artificial intelligence, which can make mistakes. Always review the original source materials.