Current Issue

---

In this week’s issue

• Implantable devices such as EYEMATE-SC show promise in safely and accurately measuring IOP without the assistance of a clinician.
• There is promise in using machine learning algorithms in glaucoma risk assessment as an objective tool for measuring RNFL thickness from optic disc images.
• Hormonal contraceptive use confers a higher relative risk of retinal vein occlusion, though the relative risk of RVO is low. 
• Treatment discontinuation with anti-VEGF injections in the setting of neovascular AMD is multifactorial. For young patients, discontinuation is most likely due to patient choice. 
• A potential interaction between Intense Pulsed Light (IPL) therapy and glaucoma medications may explain the limited sustained effectiveness of IPL for MGD in glaucoma patients. 

Too busy to read the lens? Listen to our weekly summary on your daily:  https://open.spotify.com/episode/3wsxYChP4OzmI6pVYkMAQz?si=b695818b5273460d

Ophthalmology
Safety & performance of a suprachoroidal pressure sensor system

Safe longitudinal surveillance of IOP via suprachoroidal implant: fact or fiction? Presently, intraocular pressure (IOP) remains the only modifiable risk factor for glaucoma. Implantable devices that can safely and accurately measure IOP without the assistance of a clinician may allow IOP modulating therapies to be better tailored to the individual. This multinational follow-up study analyzed the long-term safety of the EYEMATE-SC suprachoroidal sensor system (Implandata), which includes an implantable IOP sensor and a handheld reading device. 22 eyes from 22 patients who received EYEMATE-SC implants in conjunction with non-penetrating glaucoma surgery (canaloplasty or deep sclerectomy) were included at a mean follow-up time of 2.7 ± 0.6 years after implantation. The main outcome measured was level of concordance between IOP readings obtained from the EYEMATE-SC system and those obtained via Goldmann applanation tonometry (GAT). Adverse event (AE) rates (increased IOP, reduced visual acuity, cataract formation, etc.) were also tabulated. The mean IOP difference between EYEMATE-SC and GAT readings was 2.4 mmHg, with 94% of paired readings measuring within 5 mmHg of each other. Though no serious AEs were recorded, raised IOP was noted in 5 patients, reduced visual acuity in 3 patients, and cataract in 3 patients. Overall, this study suggests the EYEMATE-SC system is capable of monitoring IOP longitudinally with reasonable safety and accuracy, though future studies with greater sample sizes are warranted before widespread adoption of this technology.

JAMA Ophthalmology

Predicting glaucoma risk from optic disc photographs using machine learning

Unveiling objective insights using deep learning. Current glaucoma risk assessment in patients with ocular hypertension relies on evaluating the cup-to-disc ratio, a subjective measure prone to interobserver variability. Machine learning provides a more objective alternative by directly measuring RNFL thickness from optic disc photographs. The machine-to-machine (M2M) algorithm is a deep-learning convolutional neural network trained on fundus photographs and their associated OCT scans. The M2M algorithm was applied to the Ocular Hypertension Treatment Study (OHTS) to retrospectively predict RNFL thickness in primary open-angle glaucoma (POAG) suspects over their natural progression. Among 2864 eyes included in the study, 347 (12.1%) converted to POAG. The mean baseline M2M-predicted RNFL thickness was lower in converted eyes (94.1 μm) compared to unconverted eyes (97.1 μm; 95% CI, 2.23-3.81). For each 10-μm decrease in predicted RNFL thickness, the risk of conversion to POAG nearly doubled (HR 1.97, 95% CI, 1.60-2.42). Furthermore, a faster rate of RNFL decrease (1 μm/year) was associated with an increased risk of POAG conversion (HR, 6.01; 95% CI, 3.33-10.64). These results demonstrate the potential of M2M predictions as robust measures for early glaucoma risk assessment in patients with ocular hypertension. 

American Journal of Ophthalmology (AJO)

Hormonal contraceptives and retinal vascular occlusion risk

The pill protects against pregnancy, but is there a cost to vision? Retinal vascular occlusions (RVOs) are serious eye conditions that can lead to vision loss. While hormonal contraceptive use is a known risk factor for thromboembolism, its role in the development of RVOs remains unclear, especially with modern formulations. This nationwide, prospective Danish cohort study followed over 2 million women aged 15-49 for a median of 10 years, using data from national prescription and hospital registries to assess the association between hormonal contraceptive use and the risk of both retinal vein and artery occlusion. Compared to non-use, current hormonal contraception use was associated with a 40% increased rate of retinal vein occlusion (adjusted incidence rate ratio [AIRR] 1.4; 95% CI: 1.1–1.8), corresponding to 4.6 additional cases per 1,000,000 person-years. This higher rate was most notable for combined oral contraceptives containing 30-40 µg of estrogen. No increased rate was observed for retinal artery occlusion. Although there was a statistically significant increase in the risk of retinal vein occlusions, the absolute risk remained very low. These findings contribute to the overall safety profile of hormonal contraceptives and support their continued use for most women of reproductive age, while highlighting a small but important risk.

British Journal of Ophthalmology

Why do patients with neovascular AMD stop anti-VEGF therapy?

Not all anti-VEGF discontinuations are a failure of follow-up. Intravitreal anti-VEGF injections have significantly improved outcomes for patients with neovascular age-related macular degeneration (AMD). However, real-world results often fall short of those seen in clinical trials, potentially due to treatment discontinuation. This retrospective, multicenter study analyzed data from 2,302 patients with treatment-naive neovascular exudative AMD who initiated anti-VEGF therapy between 2019 and 2021. The study investigated the frequency and reasons for discontinuation categorized into five groups: patient’s decision, clinician’s decision, continuation of therapy at another clinic, systemic diseases preventing attendance, or death. Of the group, 28.5% of patients discontinued treatment. The most common reason for discontinuation was clinician-driven decisions, primarily due to absence of exudation or poor response or non-response. Younger patients were more likely to discontinue therapy by choice, while greater distance from the clinic increased the likelihood of continuing care elsewhere. Anti-VEGF discontinuation in neovascular AMD is common and multifactorial. Understanding the main drivers behind treatment interruption may help improve adherence and outcomes.  

Eye

When dry eye meets glaucoma: How drops may dampen IPL efficacy

Intense pulsed light (IPL) has become a popular therapy for stubborn meibomian gland dysfunction. Do chronic glaucoma medications interfere with its long-term benefits? Meibomian gland dysfunction (MGD) is a common contributor to chronic dry eye. IPL therapy has gained traction as a treatment for cases unresponsive to traditional therapies. However, many patients with glaucoma, who often suffer from MGD as well, use long-term topical hypotensive agents, raising the question of whether these medications affect IPL’s efficacy. This retrospective study compared 21 eyes of 21 glaucoma patients with MGD to 21 matched eyes of 21 non-glaucoma MGD patients, all of whom underwent IPL therapy combined with meibomian gland expression (MGX). Outcomes such as SPEED scores, tear film breakup time (TBUT), lid telangiectasia, and meibum quality were tracked over a 12-month period. Both groups saw significant improvement in dry eye symptoms and tear film stability at 9 months. However, only the non-glaucoma group maintained those improvements at 12 months and demonstrated continued improvement in lid telangiectasia and meibum expressibility. These findings suggest that IPL therapy is beneficial for treating refractory MGD in both glaucoma and non-glaucoma patients. However, the diminished long-term response in those using hypotensive eye drops points to a potential interaction that may reduce IPL’s sustained effectiveness in these patients.

Ophthalmology Retina

Can interleukin-6 serve as a biomarker in diabetic retinopathy?

Diabetic retinopathy (DR) lacks a predictive marker of disease progression– interleukin-6 (IL-6) may be the solution. DR is one of the leading causes of vision loss globally, yet disease markers such as HbA1c levels are a poor marker of disease progression. Given mounting evidence that DR has an inflammatory component, this study, a part of the prospective Inflammatory Mediators in the PathophysIology of Diabetic Retinopathy (INSPIRE) study, sought to investigate aqueous IL-6 in 328 eyes from 164 patients with type 2 diabetes mellitus. Median IL-6 levels significantly increased with worsening DR: 5.4 pg/ml in no DR, 9.25 pg/ml in moderate nonproliferative diabetic retinopathy (NPDR), and 15.71 pg/ml in proliferative diabetic retinopathy (PDR). While median HbA1c significantly differed across all groups, there was no correlation with IL-6. IL-6 was also positively correlated with central subfield thickness and macular volume, and was significantly associated with the presence of diabetic macular edema (DME). The findings strengthen the possible role of IL-6 as a biomarker of DR disease activity and as a potential therapeutic target. Unlike HbA1c, IL-6 has a short half-life and can reflect real-time intraocular inflammation, thus potentially proving to be a better predictor of DR risk and treatment response.

NPJ Digital Medicine

Multi model AI spots glaucoma using fundus images 

Lightweight AI spots glaucoma before vision fades, even on your phone. Glaucoma is a silent thief of sight worldwide, often detected too late for effective treatment. Screening remains underutilized, especially in primary care and telehealth settings. This study introduced AI-based Glaucoma Screening (AI-GS), a hybrid network of six lightweight deep learning models (total ~110 MB) designed to analyze macula-centered fundus photos for glaucomatous changes, such as optic disc cupping, hemorrhages, and nerve fiber layer defects. AI-GS was trained on diverse datasets totaling over 8,000 fundus images, with segmentation models matching expert accuracy for cup/disc delineation. It achieved 93.5 % sensitivity at 95 % specificity in test datasets and, in real-world screening, maintained a high sensitivity of 80.5 % at 91.1 % specificity which has substantially outperforming standalone classifiers. Moreover, the tool reports detailed numeric features (e.g., cup-to-disc ratio, disc indices), enhancing interpretability and clinical trust. With its compact size and fast processing (<1 s), AI-GS is primed for deployment on smartphones and tele-ophthalmology platforms. This is an exciting development as it brings specialist-level glaucoma screening to under-resourced areas and community environments, marking a major leap toward accessible, equitable eye care.

International Optic Nerve Trauma Study (IONT) - 1999

Maybe nothing is usually better than something in neuro-ophthalmology. Prior to the 1999 IONT study, there were a variety of management options used for patients with traumatic optic neuropathy, but there was no clear formula for success. The 1999 IONT treated patients with traumatic optic neuropathy with corticosteroids (n = 85), surgery (n = 33), or no treatment (n=9) with the primary outcome of visual acuity.

Key Points

• After adjusting for baseline visual acuity, there were no significant differences in outcomes between the treatment groups and the observation group
• The dosage or timing of corticosteroid therapy did not affect visual acuity outcome

The IONT study is a landmark study because it determined that the method of treatment did not affect visual outcome in traumatic optic neuropathy. As a result, and due to potential adverse effects of treatment, neither corticosteroid therapy nor surgery should be considered the standard of care. Instead, observation as the standard and further treatment based on the individual patient are most appropriate.