PURPOSE:To evaluate the 3-year efficacy and safety of phacogoniotomy versus phacotrabeculectomy for advanced primary angle-closure glaucoma (PACG) with cataract. DESIGN:Multicenter, randomized controlled, open-label, non-inferiority trial. METHODS:Patients were randomized 1:1 to undergo either phacogoniotomy (65 eyes) or phacotrabeculectomy (59 eyes). Three years retention was 92.3% (60/65) and 83.1% (49/59) for each group, respectively. Primary outcome was 3-year intraocular pressure (IOP) reduction (noninferiority margin: 4 mmHg). Secondary outcomes included surgical success, complications, hypotensive medications used; additional outcomes were changes in visual acuity (BCVA), visual field (VF), and corneal endothelial cell density (ECD). RESULTS:At 3 years, phacogoniotomy reduced mean IOP from 40.2 (10.3) to 14.1 (2.4) mmHg (-26.1 [10.4] mmHg reduction); phacotrabeculectomy, from 39.7 (9.3) to 14.4 (2.5) mmHg (-25.3 [9.2] mmHg reduction). Adjusted between-group difference in IOP change was -0.37 mmHg (95% CI, -1.32-0.58 mmHg; P = 0.44), meeting noninferiority. Complete (78.3% vs 89.8%; P = 0.13) and qualified (90.0% vs 91.8%; P > 0.999) success rates were comparable. Hypotensive medications declined in both groups (phacogoniotomy: 2.1 [1.2] to 0.2 [0.6]; phacotrabeculectomy: 2.1 [1.3] to 0.0 [0.2]; P = 0.06 for 3-year difference). BCVA improvements (0.1 vs 0.0 logMAR; P = 0.49), VF stability (MD difference:1.42 dB, P = 0.26; PSD difference: 0.22 dB, P = 0.75), and ECD loss (difference: 2%; P = 0.53) were similar. No new complications occurred in extended follow-up. CONCLUSIONS:At 3 years, phacogoniotomy remained non-inferior to phacotrabeculectomy in IOP reduction for advanced PACG with cataract.
PURPOSE:To evaluate the effectiveness of test spectacle lens in slowing myopia progression as compared to single vision (SV) spectacle (SPL). DESIGN:Cross-over trial. SUBJECTS:120 children aged 6-12 yrs, cycloplegic spherical equivalent (SE) of -0.75D to -5.00D. INTERVENTION:Enrolled participants were randomly assigned to wear either SV or test SPL (n = 60 each group). The test SPL incorporates a central clear zone and four annular cylindrical rings in the paracentral zone with a mean surface power of + 4.00D. For first six months (stage I), children wore their assigned lenses and then crossed over to wear the other lens for the second six months (stage II). OUTCOMES:Six- monthly changes in SE and AL from baseline presented as mean and 95% confidence intervals (CIs). Differences between groups were assessed using multiple linear regression, adjusting for confounders. To determine potential rebound effect, progression with SV during stage II was compared to progression with SV in stage I. RESULTS:Myopia progression was significantly slower with test SPL during both stages. During stage I, the difference in progression (Δ) between test SPL and SV were 0.32D (0.21, 0.43) and 0.12 mm (0.09，0.16) (P < 0.001). During stage II, myopia progressed slower by 0.25D (0.15, 0.36) and 0.11 mm (0.07, 0.14) (P < 0.001) with test SPL. Progression with SV during stage I and stage II did not differ significantly (SE: -0.42 ± 0.31D vs -0.51 ± 0.30D, p = 0.119; AL: 0.19 ± 0.11 mm vs 0.22 ± 011 mm, P = 0.182). CONCLUSIONS:Test spectacles significantly slow myopia in majority of eyes in 6 months and no statistically significant rebound signal was detected during the observation period.
Purpose To develop and evaluate a multimodal foundation-model-assisted system for differentiating primary open-angle glaucoma (POAG) from non-glaucoma in highly myopic eyes and assessing whether artificial intelligence (AI) assistance changes ophthalmologist diagnostic performance. Design Retrospective diagnostic model-development study with internal validation and paired sequential multi-reader evaluation. Participants Model development used 603 eye-level cases (502 POAG, 101 non-glaucoma); a fixed 150-case set from 149 patients was used for internal test reporting and the paired reader evaluation. Methods A RETFound-based multimodal classifier used color fundus photographs, optical coherence tomography (OCT) images, and 24 structured OCT/retinal nerve fiber layer (RNFL) variables. Six ophthalmologists reviewed each case without and then with AI assistance after a washout interval. Main Outcome Measures Model area under the receiver operating characteristic curve (AUC); reader sensitivity, specificity, accuracy, F1 score, confidence, reading time, diagnostic switching, and inter-reader agreement. Results The final multimodal model achieved an AUC of 0.911 (95% confidence interval [CI], 0.862-0.950) on the fixed 150-case set. In the sequential paired evaluation, unaided versus AI-assisted mean sensitivity was 68.0% versus 75.3%, accuracy was 74.2% versus 79.6%, F1 score was 0.790 versus 0.844, and mean specificity was 91.3% in both phases; Fleiss kappa was 0.583 versus 0.709. Reader-level diagnostic differences were not significant after multiplicity adjustment and were exploratory. Conclusions In this internal retrospective study, the AI-assisted phase showed numerically higher mean sensitivity, accuracy, confidence, and inter-reader agreement than the unaided phase, with unchanged mean specificity. Reader-level diagnostic differences were not statistically significant after multiplicity adjustment and should be considered exploratory. Prospective multicenter validation is required before clinical deployment.
PURPOSE:To evaluate dispensing trends of publicly funded glaucoma medications in New Zealand from 2012 to 2021, and to assess disparities in prescribing across demographic groups. This study provides population-level insights into real-world glaucoma care and treatment equity in a universal healthcare setting. In New Zealand, although limited information exists on prescribing practices, there has been no comprehensive analysis of national dispensing data. This distinction is critical, as dispensing data more accurately reflect medication access, patient uptake, and treatment adherence than prescribing data alone. METHODS:A retrospective observational study was conducted using de-identified national pharmacy dispensing data from the New Zealand Ministry of Health. Dispensing trends for eleven glaucoma medications were analysed by year, medication, sex, and self-identified ethnicity. Age-adjusted per capita dispensing rates were compared using ANOVA with post-hoc analysis. RESULTS:Over 3 million glaucoma prescriptions were dispensed, representing 27.6% of all ocular medications. The number of treated individuals rose from 39,725 in 2012-50,048 in 2021 (a 25.9% increase), outpacing national population growth. The prevalence of pharmacologically treated glaucoma or ocular hypertension increased from 0.90% in 2012-0.98% in 2021. The annual incidence of newly treated glaucoma was estimated at 125 per 100,000 people per year. Latanoprost was the most frequently dispensed glaucoma medication (40%), followed by timolol (13%) and bimatoprost (11%). Disparities in dispensing patterns were evident. Europeans received 87% of glaucoma prescriptions, Māori and Pasifika peoples, who represent 17.8% and 8.9% of the population, received only 1.9% and 1.4% of glaucoma prescriptions, respectively (p < 0.001), even after adjusting for age. CONCLUSION:This nationwide study provides the most comprehensive analysis to date of glaucoma medication dispensing in Aotearoa New Zealand, capturing real-world treatment patterns across a ten-year period. It offers critical insight into the treated prevalence and incidence of pharmacologically treated glaucoma and ocular hypertension at a population level. Latanoprost has clearly emerged as the dominant first-line therapy, consistent with international clinical guidelines, followed by Timolol. However, the findings also expose significant inequities: Māori-the Indigenous people of New Zealand-and Pasifika populations remain markedly under-represented among those receiving glaucoma treatment, even after adjusting for age. Further research is needed to understand the underlying reasons for these disparities and to ensure equitable access to glaucoma care for all New Zealanders.
BACKGROUND:To identify serum metabolic biomarkers that distinguish corticosteroid and cyclosporin A (CS & CsA) resistant pediatric idiopathic uveitis (PIU) patients from sensitive counterparts. METHODS:Serum samples were collected from 32 CS & CsA-sensitive PIU patients and 24 CS & CsA-resistant PIU patients, respectively. UHPLC-OE-MS was employed for comprehensive metabolic profiling of the serum samples. Bioinformatic analyses were performed to identify differentially expressed metabolites (DEMs) between the two patient groups. A machine learning-based classification model was constructed using the identified DEMs as predictive features. For validation purposes, an independent internal cohort of 16 CS & CsA-sensitive and 10 CS & CsA-resistant patients was recruited to evaluate the model's stability. RESULTS:Compared with the CS & CsA-sensitive PIU patients, serum samples from CS & CsA-resistant PIU patients displayed significant metabolic reprogramming. Among the identified differential metabolites, lipids were the most prominently dysregulated class, accounting for 72.47% of all differential metabolites. A machine learning based multivariate feature selection approach including NNET, LASSO, and XGBoost identified 4 candidate metabolite biomarkers. ROC analysis showed that three of these biomarkers (MG 15:0, PI-Cer 28:0;3 O, and SPB 20:0;2 O) exhibited AUC values of 0.934, 0.953, and 0.904, respectively, and were all upregulated in CS & CsA resistant patients. In contrast, N-acetylaspartic acid showed an AUC of 0.934 and was downregulated in CS & CsA resistant patients. The combined classification model incorporating these 4 metabolites achieved an AUC of 1.0. Validation in an independent internal cohort confirmed the model's excellent performance, with AUC values of 0.971 for NNET, 0.971 for LASSO, and 0.957 for XGBoost. CONCLUSION:We have established a classification model capable of effectively discriminating CS & CsA-resistant from -sensitive PIU patients. The machine learning model leveraging metabolic biomarkers demonstrates exceptional classification accuracy and generalizability, offering potential for clinical subtype classification.
Non-advanced age-related macular degeneration (AMD), encompassing early and intermediate stages, represents a critical therapeutic window before irreversible central vision loss. Traditionally defined by drusen size and pigmentary abnormalities on color fundus photography, disease characterization has evolved substantially with the integration of multimodal imaging biomarkers. Contemporary imaging enables detailed structural phenotyping, including reticular pseudodrusen (also known as subretinal drusenoid deposits), hyperreflective foci, and incomplete retinal pigment epithelium and outer retinal atrophy, which refine risk stratification and provide insight into progression toward advanced disease. Functional assessment has expanded beyond best-corrected visual acuity to include low-luminance visual acuity, contrast sensitivity, dark adaptation, and microperimetry, many of which are more sensitive to early dysfunction. However, standardization and regulatory approval of these endpoints remain ongoing challenges. Non-advanced AMD arises from complex interactions among genetic susceptibility, aging, environmental exposures, and systemic metabolic factors. Smoking remains the strongest modifiable risk factor, while Mediterranean dietary patterns appear protective. Currently, no pharmacologic therapies are approved for non-advanced AMD. The Age-Related Eye Disease Study formulations remain the only interventions proven to reduce progression in high-risk individuals. Emerging approaches, including subthreshold laser therapy and photobiomodulation, show preliminary promise but require validation in robust and adequately powered randomized clinical trials. Recent consensus efforts emphasize biomarker-driven classification and highlight the heterogeneity of AMD phenotypes across populations. Future research priorities include validation of quantitative imaging and functional endpoints, integration of artificial intelligence-based predictive models, and development of targeted therapies to delay progression. These endevours will be essential to advance precision prevention strategies.
PURPOSE:To build a visual question answering (VQA) dataset for fine-tuning and evaluating vision-language models (VLMs) in myopic maculopathy (MM). DESIGN:Cross-sectional study. METHODS:Colour fundus photographs (CFPs) from two publicly available datasets were graded using META-PM classification system. GPT-5 was used to generate clinical captions, true/false [TFQ] and open-ended [OEQ] question-answer pairs, all of which were manually verified. InternVL3-8B was fine-tuned on this dataset and evaluated against Gemini 3 Pro, Claude Sonnet 4.5, Qwen3-VL-30B-A3B-Instruct, and pre-trained InternVL3-8B. OEQ responses was evaluated by GPT-5 using a three-level scoring system (0, completely incorrect; 0.5, partially correct; 1, fully correct) and summarized as weighted accuracy. Overall accuracy was defined as the arithmetic mean of the TFQ and OEQ accuracies. RESULTS:MM-VQA comprises 2591 CFPs and 19,648 question-answer pairs. Fine-tuned InternVL3-8B model achieved an overall accuracy of 0.746, surpassing Claude Sonnet 4.5 (0.596), Qwen3-VL-30B-A3B-Instruct (0.566), and pre-trained InternVL3-8B (0.428) (all P < 0.001), while showing no significant difference compared with Gemini 3 Pro (0.724, P = 0.642). For TFQ, the fine-tuned model reached an accuracy of 0.919, outperforming Gemini 3 Pro (0.881), Qwen3-VL-30B-A3B-Instruct (0.834), Claude Sonnet 4.5 (0.796), and the pre-trained model (0.696) (all P < 0.001). On OEQ, it also ranked highest (0.572), outperforming Gemini 3 Pro (0.567, P = 0.044), Claude Sonnet 4.5 (0.395, P < 0.001), Qwen3-VL-30B-A3B-Instruct (0.297, P < 0.001) and the pre-trained model (0.160, P < 0.001). CONCLUSION:This study provides a valuable VQA dataset for MM, supporting the development of disease-specialised VLMs in ophthalmology.
PURPOSE:To investigate the effect of cataracts on a deep learning (DL) model for cardiovascular disease (CVD) risk prediction. METHODS:This retrospective, dual-cohort study analyzed fundus images at baseline, 1, and 6-months post-cataract surgery from a longitudinal cohort (patients who underwent cataract surgery at Hanyang University Guri Hospital [HUGH]) and a cross-sectional replication cohort (Singapore Epidemiology of Eye Diseases [SEED] study). CVD risk scores were generated using an artificial intelligence (AI) software (Dr.Noon CVD). Longitudinal changes were evaluated using a generalized estimating equation (GEE) model. The association between CVD scores and cataract surgery was also assessed in the SEED study using multivariable linear regression. Subgroup analyses were performed based on diabetic retinopathy (DR) status. RESULTS:In the longitudinal cohort, Dr. Noon CVD scores significantly increased from baseline after cataract surgery at 1 month (β = +2.14; 95% CI, 1.28-3.01; P < 0.001) and 6 months (β = +1.69; 95% CI, 0.82-2.56; P < 0.001). A significant interaction with DR was observed, showing substantially larger score increases at 1 month (interaction β = +3.02; 95% CI, 1.20-4.85; P = 0.001) and 6 months (interaction β = +4.78; 95% CI, 2.99-6.56, P < 0.001). These findings were partially replicated in the SEED cohort, where pseudophakia was significantly associated with a higher CVD score compared to phakic eyes (β = +2.60; 95% CI, 1.71-3.49, P < 0.001); however, the interaction between cataract surgery and DR status was not formally significant in SEED (P = 0.451). CONCLUSIONS:Cataract-induced media opacity is associated with significant attenuation of DL-derived CVD risk scores, particularly in eyes with DR. This score shift following improvement in optical clarity highlights the critical need to account for lens status when interpreting fundus-based AI prediction models.
PURPOSE:To evaluate the visual preservation, safety, and surgical outcomes of phacogoniotomy (phacoemulsification combined with or without goniosynechialysis and goniotomy) in patients with end-stage glaucoma. METHODS:This multicenter retrospective study included 134 eyes of 128 patients (92 primary angle-closure glaucoma [PACG] and 42 primary open-angle glaucoma [POAG]) with end-stage glaucoma (defined as inability to perform perimetry or visual acuity ≤ 20/200). Patients underwent phacogoniotomy and were followed ≥ 12 months. Main outcomes included best-corrected visual acuity (BCVA), intraocular pressure (IOP), medication burden, and complications, with specific attention to the "wipe-out" phenomenon. RESULTS:At a mean follow-up of 19.2 ± 8.0 months, no cases of "wipe-out" (sudden, immediate, irreversible central vision loss) occurred. BCVA improved in 63.1% of eyes and remained stable in 30.3%. The mean IOP decreased significantly from 31.4 ± 9.7 mm Hg to 14.1 ± 3.3 mm Hg (P < 0.001) and mean number of medications dropped from 2.2 ± 1.4-0.6 ± 1.0 (P < 0.001). Complete success rate was 67.9% overall, with 76.1% for PACG and 50.0% for POAG (P = 0.005). Qualified success rate was 97.0% overall, with 96.7% for PACG and 97.6% for POAG (P = 0.695). The most common complications were transient hyphema (11.2%) and IOP spike (3.7%). CONCLUSIONS:In end-stage glaucoma, phacogoniotomy is a safe and effective procedure that significantly lowers IOP while preserving or improving vision in > 90% of cases. The procedure may reduce the risk of "wipe-out" associated with traditional filtration surgery and appears particularly effective in eyes with angle-closure mechanisms.
PURPOSE:To describe asymmetrical normal tension glaucoma (NTG) with obstructive sleep apnoea (OSA; apnoea-hypopnoea-index ≥5) in habitual side-sleepers, and to compare optic nerve head (ONH) blood flow velocity across sleeping positions using laser speckle flowgraphy (LSFG; Softcare, Japan). DESIGN:Cross-sectional study. METHODS:Participants underwent ocular assessments and reported sleeping positions via questionnaires. Intraocular pressure (IOP) and LSFG-derived mean blur rates for vascular (MV), tissue (MT), and entire area (MA) were measured after 5 min in supine and lateral decubitus (LD) positions. RESULTS:We enrolled forty NTG-OSA patients (21 unilateral; 19 bilateral but asymmetrical) and 29 controls (70% male; mean age 62.9 ± 8.7 years). Interocular characteristics were comparable except that worse eyes had thinner central corneas and reduced MV, MT, and MA in LD positions. In worse eyes, MT was less in the lower LD position (9.00 ± 2.07 au) than in the upper LD (9.27 ± 1.97 au; P = 0.037) and supine (9.61 ± 1.95 au; P = 0.011) positions. Among participants who slept with the worse eye lower, MV, MT, and MA were less in the worse eyes in both lower and upper LD positions versus the fellow eye (all P < 0.05), but not in supine. Among those who slept with the worse eye upper, MV and MT were less in the worse eyes only in the lower LD position (both P < 0.05), with no differences in other positions. CONCLUSIONS:We report an association of asymmetrical NTG-OSA who habitually sleep laterally. Interocular differences, particularly reduced ONH blood flow velocity in the lower LD position, may provide insights into glaucoma pathogenesis.
Purpose The process of myopic axial elongation has not been fully uncovered yet. Here we propose a Bruch´s membrane (BM)-related hypothesis und present anatomical and clinical findings supporting it. Methods The hypothesis is that ocular axial elongation beyond the age of 3 years occurs by a retina-triggered annular segmental growth of BM in the posterior fundus midperiphery between approximately 10° to 80° anterior to the posterior pole, with a maximum growth at approximately 25° anterior to the posterior pole. Results The location of the posterior midperiphery is supported by anatomical findings of an axial length-related thinning of retinal layers, sclera, and retinal pigment epithelium in the posterior midperiphery, the location of pathologic changes at the anterior and posterior border of the BM growth zone (patchy atrophies, parapapillary myopia beta zone and gamma zone, optic nerve head canal widening, cobble stones), and results of clinical trials on myopia prevention in adolescents by circular progressive contact lenses or glasses. The notion of BM as compared to sclera as the primary structure elongating the eye is supported by the axial length-related choroidal thinning most marked at the posterior pole, the axial length-related increase in BM volume, the axial length-related shift of BM-opening of the ONHC into the macular direction, the biomechanical properties of BM, and by primarily not involving the retina, RPE and choriocapillaris in the foveal region. Conclusions If BM is the primary effector structure for axial elongation, future research may address the retinal messenger molecule directing the RPE to locally produce BM.
Artificial intelligence (AI) tools are rapidly reshaping ophthalmology by improving screening and diagnosis for diabetic retinopathy, age-related macular degeneration, glaucoma, and increasingly retina-based systemic risk assessment. This narrative review provides a comparative assessment of regulatory pathways governing ophthalmic AI and software as a medical device (SaMD) across the United States, European Union, United Kingdom, Australia, China, Japan, Canada, India, and selected emerging jurisdictions.We used a structured search of public regulator databases, guidance documents, manufacturer disclosures, and peer-reviewed literature to assemble a representative sample of marketed or authorized devices through August 2025; the device inventory is illustrative rather than exhaustive. Key differences persist in device classification, evidence expectations, change management for adaptive algorithms, and post-market oversight. Examples such as LumineticsCore, EyeArt, DrNoon for CVD, CLAiR, and EyeWisdom illustrate how risk-based approaches vary across jurisdictions.These inconsistencies can delay multi-region deployment and complicate implementation, supporting the need for lifecycle-focused and internationally aligned standards for safe, transparent, and equitable use of ophthalmic AI.
PURPOSE:As the prevalence of high myopia increases around the world, the incidence of highly myopic cataract (HMC) would increase and present clinicians with unique management challenges. This modified Delphi consensus study aimed to establish practical recommendations for HMC diagnosis and treatment, addressing key controversies in preoperative evaluation, surgical considerations, and postoperative care. METHODS:An international panel of 30 cataract experts from 14 countries/territories participated in this two-round e-Delphi study. Consensus was defined as ≥ 75% agreement on 5-point Likert-scale statements covering disease characterization, preoperative evaluation, intraoperative precautions, and postoperative care. RESULTS:A formal consensus was reached by a broad majority of the panel (86.67%) to endorse HMC as a separate disease category, typically manifesting 10-20 years earlier than age-related cataracts. Key consensus included: use of combined IOL formula calculations (96.67% agreement), preference for hydrophobic acrylic intraocular lens (96.67%), and mandatory 3-month postoperative retinal exams (96.67%). Controversies persisted regarding immediate sequential bilateral surgery, prophylactic routine implantation of a capsular tension ring, and postoperative steroid regime. CONCLUSIONS:We present commonly agreed recommendations for the clinical management of HMC, which include tailored surgical approaches and vigilant postoperative monitoring to address this growing public health challenge.
