Ophthalmology Science

Objective: To define CSC genetic architecture and identify implicated ocular tissues, cell types, genes, and circulating proteins. Data Sources: Genome-wide data were assembled from FinnGen, All of Us, Mass General Brigham Biobank, Million Veteran Program, and a Dutch chronic CSC cohort. Serum protein quantitative trait loci, human single-cell ocular atlases, and UK Biobank macular optical coherence tomography (OCT) imaging were used for downstream analyses. Study Selection: Five European-ancestry cohorts with genome-wide data and cohort-specific CSC case-control definitions were included, comprising 2,584 cases and 1,044,455 controls. Variants present in at least 2 cohorts were meta-analyzed. Data Extraction and Synthesis: Cohort-level GWASs were adjusted for age, age squared, sex, genotyping array or batch, and 10 genetic principal components, then combined using fixed-effects inverse-variance meta-analysis. Post-GWAS analyses included gene prioritization, colocalization, Mendelian randomization, single-cell disease-relevance scoring, and testing of a CSC genetic risk score in UK Biobank OCT images. Main Outcome(s) and Measure(s): Genome-wide significant CSC loci, effector genes and proteins, tissue and cell-type enrichment, and CSC-relevant OCT abnormalities. Results: Across 11,068,938 variants, 10 loci reached genome-wide significance (P < 5e-8), including 3 novel loci near TGFB1, LINC00551, and LOC105375630 and 7 replicated loci near CFH, CD46, NOTCH4, PREX1, PTPRB, GATA5, and TNFRSF10A. Integrative analyses prioritized 10 candidate effector genes. Colocalization and Mendelian randomization implicated circulating TNFRSF10A, TGFB1, and CASP10 levels. Single-cell analyses localized genetic risk to sclera (P = 2.0e-4) and vascular endothelial cells (P = 4.0e-4), with fibroblast enrichment. In UK Biobank, OCT abnormalities were more frequent in the top vs bottom 1% of CSC genetic risk (18 of 109 [16.5%] vs 8 of 134 [6.0%]; odds ratio, 4.05; 95% CI, 1.65-10.87; P = .002). Conclusions and Relevance: In this GWAS meta-analysis, CSC susceptibility localized predominantly to scleral and vascular biology rather than primary retinal pigment epithelial dysfunction. These findings support CSC as a sclerovascular disorder and nominate complement regulation, endothelial signaling, and extracellular matrix pathways for future study.

Purpose To investigate the relative efficacy of nine distinct visual field (VF) denoising artificial intelligence (AI) methods and a pathology-aware AI strategy to discourage over-correction of glaucomatous defects. Design Retrospective study. Participants 87,940 paired visual field (VF) and optical coherence tomography (OCT) samples from a tertiary academic center. Methods Denoising models were trained on a separate VF-only dataset and evaluated on an independent structure-function dataset of paired VF-OCT samples. We implemented and evaluated nine distinct VF denoising strategies representing three broad categories: baseline measurements, self-supervised and image restoration models (including Noise2Noise, Noise2Void, and NAFNet), and latent variable compression-based models (autoencoders and variational autoencoders). All models were designed to reconstruct VF sensitivity maps. We then predicted retinal nerve fiber layer thickness (RNFLT) maps from the denoised VFs using a fixed, independently trained VF-to-RNFLT prediction model. Main Outcome Measures Predicted VF and RNFLT maps and resultant evaluation metrics. Results The raw VF baseline achieved a global R2 of 0.5468 and MAE of 16.83 um. Restoration-based models maintained or slightly improved concordance, with the pathology-aware NAFNet achieving the highest global R2 of 0.5485 and a comparable MAE of 16.82 um. In contrast, compression-based models degraded concordance, with CNN-VAE showing a significant reduction (R2 approximately 0.50). In severe glaucoma, concordance decreased across all methods; however, compression architectures exhibited disproportionately greater degradation compared with restoration-based approaches. Conclusions We present a comparative benchmark of AI-based VF denoising strategies paired with structure-function evaluation. While restoration-based models can reduce variability without loss of biological signal, latent compression risks attenuating clinically meaningful defects. Visually smoother fields are not necessarily more biologically accurate.

Purpose: To develop and evaluate a deep learning model that predicts optical coherence tomography (OCT)-equivalent retinal nerve fiber layer thickness (RNFLT) maps directly from color fundus photographs and to assess their diagnostic value for glaucoma detection. Design: Retrospective model development and evaluation study. Participants: 15,031 paired fundus photographs and spectral-domain OCT scans collected at Massachusetts Eye and Ear between 2011 and 2022. Methods: Paired fundus and OCT images were used to train a U-Net-based model to predict pixel-wise RNFLT maps with artifact-corrected supervision. Diagnostic performance was evaluated across single-modality models (fundus photos only, real RNFLT maps, predicted RNFLT maps) and multimodal fusion models (fundus + predicted RNFLT maps). Stratified analyses examined model performance across glaucoma severity and demographic subgroups. Glaucoma was defined based on standard criteria applied to Humphrey 24-2 visual field testing. Main Outcome Measures: Mean absolute error (MAE) and structural similarity index (SSIM) for RNFLT map prediction. Area under the ROC curve (AUC) and accuracy for glaucoma detection. Results: RNFLT map prediction achieved a MAE = 15.4 m and a SSIM = 0.65, measured against artifact-corrected RNFLT maps derived from OCT. For glaucoma detection, the predicted RNFLT-only classifier outperformed the fundus-only classifier (AUC 0.889 vs 0.883, p < 0.005; Accuracy 82.0% vs 78.0%), but performed worse than the real-RNFLT-only classifier (AUC 0.889 vs 0.903, p < 0.005). Multimodal fusion of fundus images with predicted RNFLT maps improved performance, achieving an AUC of 0.909, outperforming all single-modality inputs (p < 0.005 vs fundus-only, predicted-RNFLT-only, and real-RNFLT-only). Performance gains between the fundus-only and the multimodal classifier were greater in early-stage glaucoma compared to severe cases: accuracy increased from 55.3% to 64.0% in mild cases, from 71.5% to 80.4% in moderate cases, and from 90.0% to 94.6% in severe cases. Conclusions: Predicted RNFLT maps derived from fundus photographs provide quantitative, OCT-like structural information and improve glaucoma detection. Unlike prior work that predicted only summary RNFLT values, our model generates full RNFLT maps that better support glaucoma classification than fundus images alone. This approach offers a scalable pathway for early glaucoma screening and expands diagnostic access in resource-limited settings.

PurposeAnti-vascular endothelial growth factor (anti-VEGF) therapy is the standard of care for neovascular age-related macular degeneration (AMD), yet many patients exhibit persistent retinal degeneration, fibrosis, and incomplete therapeutic response. The molecular pathways underlying this incomplete response remain poorly understood. We sought to identify VEGF-independent signaling pathways active in the vitreous of anti-VEGF-treated AMD patients. MethodsWe performed multiplex antibody-based proteomic profiling of 1,000 human proteins in vitreous samples from patients with neovascular AMD receiving anti-VEGF therapy (n=8) and comparative controls (n=6). Differential protein expression was assessed using one-way ANOVA, followed by gene ontology and pathway enrichment analyses. Drug-target relationships were evaluated to identify potential opportunities for therapeutic repositioning. ResultsWe identified 107 differentially expressed proteins (p<0.05), including key regulators of immune signaling, angiogenesis, and metabolism. Notably, multiple components of cytotoxic lymphocyte pathways were dysregulated, including IL-21R, SIGLEC-7, CTLA4, and IL-2-associated signaling. Enrichment analyses revealed significant activation of pathways related to T-cell activation, interleukin signaling, and leukocyte-mediated cytotoxicity. These immune signatures persisted despite suppression of VEGF signaling. Several clinically available immunomodulatory agents--including abatacept, sirolimus, and dupilumab--targeted pathways identified in this dataset. ConclusionsAnti-VEGF-treated neovascular AMD exhibits persistent cytotoxic immune signaling in the vitreous, suggesting that VEGF-independent immune mechanisms may contribute to ongoing retinal damage and incomplete therapeutic response. These findings provide a rationale for combination therapeutic strategies targeting both angiogenic and immune pathways in AMD.

Manufacturer-defined signal-strength indices are frequently employed as quality benchmarks for automated optical coherence tomography analysis, yet their empirical relationship with deep learning segmentation accuracy remains unclear. Because these metrics were originally developed for conventional image-processing pipelines, their ability to predict modern model-based segmentation accuracy has not been empirically validated. To address this gap, we evaluated the Heidelberg Spectralis Q-score against U-Net segmentation performance across 5,047 B-scans from 103 eyes for three anatomical boundaries of the posterior segment of the eye: the Ellipsoid Zone (EZ), Bruch's Membrane (BM), and Choroid Outer Boundary (COB). Alongside standard boundary agreement metrics (MAE, MSE, Dice Similarity Coefficient), we adapted the Earth Mover's Distance (EMD) from optimal transport theory as a boundary evaluation metric. Unlike column-wise averages, EMD quantifies boundary agreement as a 2-D geometric displacement, directly measuring residual spatial displacement between the model segmented boundary and the ground-truth boundary. Our results demonstrate that the Q-score - originally designed to gate image-processing-based automated analysis - is a poor predictor of deep learning boundary segmentation accuracy, with explained variance (R2) failing to exceed 1.4% across all three boundaries. We further observed a monotonically increasing error hierarchy with anatomical depth (EZ < BM < COB), consistent across metrics, which is unexplained by the signal strength. At the COB, correlations were paradoxically positive, explained by a B-scan-level mediation chain: higher Q-scores correspond to greater choroidal thickness (r=0.113, {rho}=0.158), which in turn predicts higher COB segmentation error (r=0.165, {rho}=0.191) - a localization difficulty that global signal strength cannot capture. Collectively, these findings challenge the implicit assumption that signal-strength-based quality thresholds are a reliable proxy for deep learning model performance, and motivate a shift toward task-specific acquisition quality criteria calibrated to model performance rather than signal interpretability.

ObjectiveTo describe the ophthalmic examination protocol within the German National Cohort (NAKO) / NAKO Gesundheitsstudie, to report the baseline profile of participants undergoing ophthalmological assessment, and to illustrate the potential of these data as a population-based open resource for artificial intelligence (AI) research in eye health. DesignBaseline analysis of ophthalmic data within the nationwide, population-based multicenter prospective NAKO study. Participants48,460 adults in the ophthalmological level 2 module of 205,053 adults enrolled in NAKO, aged 19-74 years, with mean age 48.9 {+/-} 12.5 years and 52.7% male. MethodsAll participants underwent standardized assessments of a wide range of biomedical examinations and detailed questionnaire-based data collection, including non-dilated color fundus imaging, visual acuity testing, recording of a brief ocular history. Ocular and systemic health measures were summarized using descriptive statistics. Fundus image quality and morphological features (e.g. cup-to-disc ratio, ateriole-to-venule-ratio) were assessed using open-source deep learning models. Standard deep learning architectures were trained on the fundus images to predict age, sex and blood pressure. Main Outcome MeasuresPercentage of fundus images graded as good quality; mean absolute error for age and blood pressure prediction; accuracy for sex prediction. ResultsThe analysis includes 48,460 participants who successfully completed the level 2 ophthalmological baseline examination across 18 study sites in Germany. Mean visual acuity (logMAR) was 0.01 {+/-} 0.20 (left eye) and 0.03 {+/-} 0.21 (right eye). Self-reported ocular disease prevalence was 4.2% for cataract, 2.0% for glaucoma, and 0.9% for macular degeneration. 68.2% of fundus images were classified as gradable as a consensus of four deep learning-based quality grading models Morphological features such as cup-to-disc ratio and arteriole-to-venule-ratio showed systematic differences across age groups. Standard deep learning architectures showed comparative performance to the state-of-the-art for age, sex and blood pressure prediction (2.96 MAE for age prediction, 0.84 accuracy for sex prediction, 10.78 and 7.01 MAE for systolic and diastolic blood pressure prediction). ConclusionsNAKO provides a large-scale, nationwide population-based resource with visual acuity measurements and systemic health indicators, as well as color fundus images in about 50,000 NAKO participants. The data sets the ground for studying eye health in the general adult population in Germany and can serve as a strong foundation for developing and validating AI tools in eye health research.

After rejection, resubmission of scientific manuscripts often requires substantial journal-specific reformatting. We compared systematic review submission policies across high-impact ophthalmology journals and quantified policy similarity to support resubmission planning. We identified the top 50 ophthalmology journals by SCImago Journal Rank that publish systematic reviews and are not invite-only, extracted policies from author instructions using an a priori data dictionary, and computed pairwise similarity on a 0 to 1 scale using the Gower coefficient across mixed policy variables with available-case denominators for unstated fields. Policies were heterogeneous and frequently unstated. Only 29 of 50 journals (58%) stated a main-text word limit; among journals with numeric limits, the median was 4000 words (interquartile range 3500 to 5500; n = 23). Preferred Reporting Items for Systematic Reviews and Meta-Analyses compliance was explicitly required by 35 of 50 journals (70%), and prospective registration by 6 of 50 journals (12%). Across 1225 journal pairs, similarity was modest, with a median of 0.64 (interquartile range 0.57 to 0.71; range 0.05 to 0.98). Similarity among the top 5 highest-ranking journals ranged from 0.62 to 0.90 (median 0.75). Systematic review submission policies vary widely across high-impact ophthalmology journals, and most journal pairs show only modest similarity. Similarity-based guidance may help identify policy-aligned resubmission targets while anticipating common sources of reformatting burden.

PurposeTo develop and evaluate deep learning models for automated detection of corneal perforation in microbial keratitis using anterior segment optical coherence tomography (ASOCT) images. MethodsWe enrolled 150 patients with microbiologically confirmed keratitis. Contralateral healthy eyes served as controls. Four convolutional neural network models using ResNet architecture were trained and evaluated using ASOCT images to classify the presence or absence of corneal perforation at the eye level. Ground truth labels for perforation were established following consensus grading by two masked ophthalmologist graders. Models differed in inclusion of healthy controls and masking of non-corneal anterior segment anatomy. ResultsThe best-performing model (Model 1), which included healthy controls and randomly applied masking of the inferior image portion during training, achieved an AUC of 0.965 (95% CI, 0.911-0.995), sensitivity of 84.0% (95% CI, 70.0%-97.1%), and specificity of 97.8% (95% CI, 96.1%-99.3%) for detection of corneal perforation. Models including healthy controls outperformed those without, and lens masking improved discrimination. ConclusionsDeep learning models achieved high diagnostic accuracy for detecting corneal perforation on ASOCT imaging in eyes with microbial keratitis. These findings support the potential role of automated ASOCT analysis as a clinical decision support tool for identifying this vision-threatening complication.

Purpose: Psychological distress is highly prevalent in glaucoma and is associated with worse adherence, reduced quality of life, and faster disease progression. However, distress is rarely assessed in ophthalmology settings due to time, workflow, and staffing constraints. We evaluated two artificial intelligence (AI)-based screening strategies, designed to efficiently identify distressed primary open angle glaucoma (POAG) patients during routine care, aiming to achieve effective, resource conscious, low burden clinical screening. Design: Hybrid retrospective cohort and prospective cross-sectional study. Participants: The retrospective cohort included >3,000 POAG patients from the Duke Ophthalmic Registry. Prospective validation was conducted in a separate 300 POAG patient cohort who completed patient-reported distress screening. Methods: Using retrospective data, a neural network model was trained to predict an electronic health record (EHR)-derived computable phenotype of distress ("silver standard"). Prospective validation used the 8-item Patient Health Questionnaire (PHQ-8) as the "gold standard." Three screening strategies were compared against PHQ-8: (1) universal PHQ-2 screening (two-item screener administered to all patients), (2) AI-only screening (fully automated EHR-based screener), and (3) sequential screening, (only patients flagged as high risk by AI screener completed the PHQ-2). Performance metrics included sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), accuracy, and screening burden. Main Outcome Measures: Sensitivity; specificity; PPV; NPV; accuracy; proportion of patients requiring secondary screening (screening burden). Results: Distress prevalence was 17% (PHQ-8 > 6). Universal PHQ-2 screening (> 0) achieved high sensitivity (0.96) but lower specificity (0.73) and PPV (0.41), while requiring screening of all patients. The AI-assisted sequential approach substantially reduced screening burden while maintaining strong diagnostic performance. By administering PHQ-2 to ~25% of patients, sequential screening achieved sensitivity 0.64, specificity 0.93, PPV 0.64, NPV 0.93, and accuracy 0.88, representing a ~50% increase in PPV compared to PHQ-2 alone. AI-only screening reduced burden further but did not achieve comparable sensitivity or predictive performance. Conclusions: AI-assisted sequential screening enables scalable, resource efficient identification of psychological distress in glaucoma care, substantially reducing screening burden while preserving clinically meaningful performance. This framework offers a practical pathway for integrating distress screening into routine ophthalmology workflows and improving the identification and referral of at-risk patients.

Purpose: To predict retinal nerve fiber layer thickness (RNFLT) norms from fundus images. Methods: We selected 18,000 OCT scans and visual fields (VF) from the Massachusetts Eye and Ear Glaucoma Service. A U-Net-based deep learning model was developed to predict RNFLT norms from OCT en face fundus images. A total of 10,000 OCT scans with normal VFs (mean deviation [MD] [&ge;] -1 dB, glaucoma hemifield test within normal limits, and pattern standard deviation probability > 5%) tested within 30 days were used for training, while the remaining 8,000 OCT scans (mean VF MD: 3.3 +/- 4.9 dB), including 2,419 scans with normal VFs, were used for evaluation. Structure-function correlations between RNFLT maps and VFs were assessed using linear regression and VGG-16 across original RNFLT maps, deviation maps, and their combination. Performance was evaluated using correlation coefficients, mean absolute error (MAE), and R-squared. Results: Predicted RNFLT norm maps showed agreement with baseline RNFLT maps in eyes with normal VFs (R-squared = 0.81 +/- 0.13). RNFLT deviation maps correlated more strongly with VF MD than original RNFLT maps (R = 0.42 vs. 0.19, p < 0.01). In deep learning-based VF prediction, combining original and deviation maps achieved the best performance (MAE = 3.31 dB, R-squared = 0.39), outperforming the model (p < 0.05) using original RNFLT maps alone (MAE = 3.36 dB, R-squared = 0.35). Conclusions: Deep learning can estimate individualized RNFLT norms and improve structure-function assessment in glaucoma. Translational Relevance: Personalized RNFLT norm prediction may improve detection of glaucomatous damage.

Automated glaucoma screening from optical coherence tomography (OCT) faces two persistent challenges: scarcity of expert-labeled data and unreliable model predictions on diagnostically ambiguous cases. We present a two-tier diagnostic pipeline that addresses both. In the first tier, an EfficientNetV2-S classifier trained under a semi-supervised pseudo supervisor framework achieves 0.84 AUC on 150 held-out test patients from the Harvard Glaucoma Detection and Progression dataset, using only 350 labeled training samples out of 700. In the second tier, 124 flagged cases are routed to a multi-agent system built on MedGemma 4B, where three specialist agents deliberate over three rounds before rendering a final diagnosis. On these flagged cases, the agent system achieves 100% sensitivity--detecting all 55 glaucoma cases with zero missed diagnoses--and 89.5% overall accuracy (111/124), compared to the classifiers 73.4% (91/124). Uncertainty analysis confirms that the classifiers output probability reliably separates confident predictions (96.3% accuracy, n = 27) from uncertain ones (74.0%, n = 123), producing a 22-percentage-point gap that serves as a triage signal. The agents fix 32 cases the classifier misclassifies while introducing 12 new errors, yielding a net improvement of 20 cases. These results are from a single training run without variance estimates and should be interpreted as preliminary evidence that uncertainty-gated routing to vision-language model agents can meaningfully improve diagnostic accuracy on the cases where automated classifiers are least reliable.

BackgroundNeurotrophic keratopathy (NK) is a rare degenerative corneal disease caused by impaired trigeminal innervation, resulting in reduced corneal sensation, impaired epithelial healing, ulceration, and risk of perforation or vision loss. Corneal innervation is essential for protective reflexes, epithelial maintenance, and ocular surface homeostasis. Conventional medical therapies may promote epithelial healing but do not directly restore corneal innervation. Corneal neurotization (CN) has emerged as a surgical strategy in which healthy donor sensory axons are transferred to denervated corneas to provide innervation. Multiple procedural variations now exist, including differences in donor nerve selection, graft use, and methods of limbal nerve insertion. A broad variety of NK etiologies is also being treated, including congenital, infectious, tumor, or other causes. However, published evidence remains limited by small case series, heterogeneous surgical methods, short follow-up periods, and inconsistent outcome reporting. ObjectiveTo address the need for standardized long-term outcome data in CN, we established the Corneal Neurotization Assessment (CorNeA) Registry, an international multicenter observational registry designed to evaluate patients undergoing CN. MethodsThe CorNeA Registry captures demographic characteristics, disease etiology, surgical technique, and longitudinal ocular outcomes in patients with NK treated with CN. Data are recorded in REDCap and include both retrospective and prospective patient enrollment across participating centers. Patients are followed longitudinally after surgery without a predefined endpoint to permit long-term assessment of corneal sensation recovery, ulcer recurrence, and visual outcomes. At the time of reporting, the registry includes 58 patients from multiple international centers, with active expansion ongoing. ConclusionBecause NK is rare and CN remains an evolving surgical field, long-term comparative data are lacking. The CorNeA Registry provides the first international platform to characterize patient selection, procedural variation, and long-term outcomes after CN, with the goal of informing future surgical decision-making and outcome standardization.

Background and Objective As optical coherence tomography (OCT) has enabled the identification of an expanding set of age related macular degeneration (AMD) risk biomarkers and become central to routine clinical practice, there remains a need for a simplified grading scheme that allows physicians to communicate and synchronize AMD grading directly from standard OCT imaging rather than relying on traditional color fundus imaging. This study aims to establish a standardized OCT based AMD classification that balances diagnostic accuracy with practicality for use across clinical and research settings. Patients and Methods Spectral domain optical coherence tomography scans were independently graded by two retinal specialists following the newly proposed Stanford OCT Based AMD Classification (SOAC). Discrepancies were adjudicated by a third independent retinal specialist. Intergrader agreement was assessed using weighted kappa coefficients. Results Among the 109 eyes from 108 patients, AMD staging based on SOAC was distributed as follows: normal aging in 9 patients (8.3%), early AMD in 16 (14.7%), intermediate AMD in 32 (29.4%), neovascular AMD (nAMD) in 18 (16.5%), geographic atrophy (GA) in 20 (18.3%), and combined nAMD and GA in 14 (12.8%). The overall intergrader agreement demonstrated robust consistency, with a weighted kappa value of 0.95 (95% CI: 0.92 to 0.98), signifying excellent intergrader reliability and reinforcing the validity of SOAC. Conclusion SOAC provides a standardized, OCT based framework for AMD grading that demonstrates high intergrader agreement. By enabling consistent classification from commonly acquired OCT scans, SOAC supports reliable disease staging and facilitates integration across clinical studies and translational research. As imaging and molecular data continue to expand, SOAC can serve as a common OCT based reference for phenotype refinement and longitudinal AMD studies.

Purpose: To evaluate the safety and exploratory outcomes of a single intravitreal injection of OGX110, a peptide agonist of CXCR3, in eyes with persistent fluid secondary to neovascular age-related macular degeneration (nAMD) despite ongoing anti-vascular endothelial growth factor (anti-VEGF) therapy. Methods: This prospective, open-label, sequential dose-escalation phase I study (NCT05904691) enrolled subjects receiving standard-of-care intravitreal anti-VEGF therapy. Subjects received a single intravitreal injection of OGX110 at 0.5 mg, 1.0 mg, or 2.0 mg (n=3 per cohort), 7 to 14 days after the anti-VEGF injection. Results: All nine enrolled subjects completed follow-up through day 56. Two subjects (22%) experienced at least 1 adverse event (AE); all were mild and unrelated to study treatment. Exploratory analyses showed a BCVA change of +1.4 letters following anti-VEGF injection and +4.4 letters from OGX110 baseline to 4 weeks (P < 0.05). Six of 9 subjects gained at least 3 ETDRS letters after OGX110. Anatomic responses were heterogeneous. Four eyes showed a reduction in CRT after anti-VEGF injection that was maintained after OGX110 administration. One additional eye demonstrated a substantial reduction in CRT after OGX110 despite minimal response to anti-VEGF treatment. Conclusions: A single intravitreal injection of OGX110 was well tolerated. Exploratory functional and anatomic findings suggest biologic activity; interpretation is limited by small sample size, open-label design, absence of a concurrent control group, and inter-subject heterogeneity. These results support further study in a controlled trial. Translational Relevance: OGX110 represents a mechanistically distinct investigational approach for nAMD that may warrant further evaluation in eyes with persistent.

ObjectiveTo assess how whole genome sequencing and varying phenotype definitions influence genetic discovery for primary open-angle glaucoma (POAG) in a diverse population. DesignAncestry-stratified genome-wide association studies (GWASs) and cross-ancestry meta-analyses of POAG cases and controls using two phenotype definitions. ParticipantsCases (age>40) and controls (age>65) were identified in the National Institutes of Health All of Us Research Program v8 data release and sub-divided into genetically inferred ancestral groups. Using the relaxed phenotype (ICD codes only), case/control counts were: European (1,846/84,654), African (1,042/15,966), and Latino/Admixed American (305/10,167). Using the stringent phenotype (ICD codes and evidence of glaucoma treatment in the electronic health record), case/control counts were: European (1,528/79,276), African (862/14,076), and Latino/Admixed American (250/9,668). Cross-ancestry meta-analyses included 3,193 cases/110,787 controls for the relaxed phenotype and 2,640 cases/103,020 controls for the stringent phenotype. MethodsGWASs were conducted within European, African, and Latino/Admixed American ancestry groups individually using firth logistic regression with age, sex, and the top 10 genotype principal components included as covariates. The ancestry-stratified GWASs were then meta-analyzed using a fixed-effects, inverse variance-weighted approach. Main Outcome MeasuresIdentification of genome-wide significant loci (P < 5x10-8) for POAG using different phenotype definitions and ancestry groups. ResultsKnown POAG risk loci (e.g., TMCO1, CDKN2B-AS1, and GMDS) reached genome-wide significance in both the European GWASs and cross-ancestry meta-analyses (odds ratio (OR) range: 1.19-1.38). A novel risk locus near CYP2A7 (rs76935404[T], OR = 1.35) was identified in the African ancestry GWAS using the stringent phenotype definition. Effect sizes for known POAG risk loci from prior large-scale meta-analyses strongly correlated with effect sizes in this study (Pearson r = 0.75-0.84, P < 1 x 10- for all). The strength and consistency of these correlations support the robustness of the findings. ConclusionsThis study demonstrates the value of whole genome sequencing, diverse ancestry inclusion, and phenotypic refinement in uncovering novel POAG genetic risk loci. The findings underscore the need to prioritize both genetic diversity and refined case/control definitions to advance understanding of this complex ocular disease. PrecisThis study identifies a novel primary open-angle glaucoma risk locus in individuals of African ancestry using whole genome sequencing and varying phenotype definitions in the diverse All of Us Research Program dataset.

PurposeTo evaluate associations between optical coherence tomography angiography (OCT-A) metrics and diabetic retinopathy (DR) and compare their discrimination against conventional clinical risk factors. MethodsIn this cross-sectional study, 108 adult eyes (right eye if both eligible) with diabetes were recruited from tertiary ophthalmology/optometry clinics. DR was clinically graded using ETDRS categories and dichotomised as no DR vs [&ge;] mild NPDR (primary outcome). Macular 6x6 mm OCT-A (Zeiss AngioPlex) was acquired; scans with signal strength >7 and without major artefact were included. Quantitative metrics from the superficial capillary plexus included vessel density (VD) and perfusion density (PD) (central/inner/outer/full regions); structural OCT measures and FAZ parameters were secondary. Associations with [&ge;] mild NPDR were assessed using multivariable logistic regression adjusted for age, sex, HbA1c, and diabetes duration. Discrimination was evaluated with ROC curves/AUC (95% CI) and DeLong comparisons of AUCs. ResultsDR was present in 63% of eyes. DR was associated with lower VD (central, inner, outer, full) and lower PD (central, inner, full) (all p[&le;]0.04). After adjustment, central VD (OR 0.82, 95% CI 0.68-0.98) and central PD (OR 0.92, 95% CI 0.86-0.99) remained independently associated with DR. The OCT-A model outperformed the clinical model (AUC 0.73 vs 0.60); the combined model yielded AUC 0.76. ConclusionVD and PD from the superficial plexus are independently associated with DR and show superior discrimination versus conventional clinical factors alone, supporting OCT-A as an adjunct for earlier DR detection.

Background: Mitochondrial dysfunction is an emerging metabolic hallmark of age-related diseases, yet tools to directly profile mitochondrial pathways and test metabolic interventions in the living human eye remain limited. Multi-omics ocular liquid biopsy enables real-time proteomic and metabolomic profiling of the intraocular microenvironment, complementing systemic biomarkers and imaging surrogates. Here, we used this approach to define mitochondrial and tricarboxylic acid (TCA) cycle dysregulation in geographic atrophy (GA) and to assess whether oral -ketoglutarate (-KG) supplementation can modulate mitochondrial metabolites within the eye. Methods: Mitochondrial and TCA cycle-related proteins were profiled in aqueous humor (AH) samples from patients with GA using DNA-aptamer-based proteomics. In a phase 0 study, a second cohort undergoing sequential cataract surgery provided paired AH samples collected at first-eye surgery and at second-eye surgery after interim -KG supplementation. These samples underwent targeted metabolomic profiling using hydrophilic interaction liquid chromatography coupled with mass spectrometry. Results: In GA, 64 mitochondrial proteins were differentially expressed, including coordinated TCA-cycle deficiencies marked by reduced expression of enzymes regulating TCA entry and flux, including PDHB and DLST. In the phase 0 cohort, oral -KG supplementation significantly increased intraocular -KG levels and the -KG-to-succinate ratio (P < 0.05), with coordinated shifts across TCA intermediates consistent with enhanced TCA cycle flux. Conclusions: AH proteomics demonstrated mitochondrial pathway depletion in GA, consistent with reduced oxidative bioenergetic capacity. AH metabolomics provided first-in-human in vivo evidence that systemic -KG supplementation can modify intraocular metabolites and may enhance intraocular energy metabolism. These findings support ocular liquid biopsy as a precision-health framework for per-patient biomarker-guided metabolic trials in GA.

Aim: To evaluate the potential of a three-dimensional microscope combining Laser scanning confocal imaging and white-light interferometry for quantitative topographic characterisation of Descemet's membrane (DM) in Fuchs endothelial corneal dystrophy (FECD). Methods: Descemet's membranes were collected from 38 FECD patients undergoing endothelial keratoplasty and 4 healthy donors. After flat-mounting on glass slide and drying, specimens were analysed using the VK-X3000 system (KEYENCE). Entire samples were reconstructed by image stitching at low magnification (x10) in white-light interferometry mode (0.01nm axial resolution). Higher magnifications (x20-x150) in confocal mode (12nm axial resolution) enabled detailed structural analysis. Three-dimensional height maps were generated to calculate standardised surface roughness parameters. Guttae and other DM features were classified according to spatial organisation and elevation profiles. Results: White-light interferometry enabled full-field mapping of whole 8mm diameter DMs with nanometric vertical resolution (~2 hours/sample). Surface roughness (Sa) was higher in FECD than in controls (median{+/-}IQR: 0.571{+/-}0.259 m vs 0.239{+/-}0.161 m ; p = 0.0018). In FECD, three zones were identified: central (guttae buried in the posterior fibrillar layer; Sa 0.442 {+/-} 0.112 m), paracentral (large uncovered guttae; Sa 0.562{+/-}0.170 m ; p = 0.0423), and outer zone (no confluent guttae; Sa 0.261{+/-}0.143 m ; p < 0.0001). Confocal 3D imaging revealed radial striae, embossments and furrows in the DM, confluent central guttae, and fused or buried structures. Conclusions: Combining white-light interferometry and confocal microscopy enables label-free, high-resolution surface characterisation of DM in FECD, providing quantitative metrics to compare histological subtypes and supporting the predominance of radial structural organisation.

PurposeTo compare hypopyon detection using anterior segment optical coherence tomography (ASOCT) versus slit lamp examination (SLE) in microbial keratitis, and to evaluate intra- and inter-grader agreement for ASOCT hypopyon measurement. MethodsTwo masked graders independently evaluated ASOCT images for hypopyon presence or absence in eyes with microbial keratitis, with disagreements resolved by consensus. A subset of hypopyon eyes underwent triplicate height measurement using two methods (endothelial length, vertical height). Cohens kappa, intraclass correlation coefficients (ICC), sensitivity, and specificity were calculated comparing diagnostic performance of ASOCT versus SLE. ResultsInter-grader agreement for hypopyon detection on ASOCT was excellent ({kappa}=0.94; 95% CI 0.84-1.00) and intra-grader agreement was excellent ({kappa}=0.89-1.00). ASOCT detected hypopyon in 67.1% of eyes versus 57.0% by SLE (sensitivity 83.0%, specificity 96.2% using ASOCT as reference). Intra-grader reproducibility was excellent for both endothelial length and vertical height measurements (ICC 0.977-0.996). Inter-grader agreement was good for endothelial length (ICC 0.831) and vertical height (ICC 0.827), though a statistically significant inter-grader bias was identified for vertical height only (Wilcoxon p=0.008). ConclusionsASOCT detected hypopyon with greater sensitivity than SLE and demonstrated excellent intra-grader and good inter-grader measurement reproducibility. Endothelial length showed slightly superior inter-grader concordance to vertical height measurement.

BackgroundDiabetic retinopathy (DR) is the leading cause of preventable blindness among working-age adults worldwide, yet screening coverage remains inadequate, particularly in low-and middle-income countries. Automated deep learning systems offer potential to address the global shortage of expert graders, but most existing models lack lesion-level interpretability and are not aligned with established clinical referral frameworks. We developed and validated DRAGS (Diabetic Retinopathy Automated Grading System), a hybrid deep learning model that grades DR according to the UK Diabetic Eye Screening Programme (DESP) classification and provides lesion-level explainability. MethodsWe trained and validated a DenseNet-201-based convolutional neural network on 20,281 anonymised fundus images from two tertiary eye care institutions in Bangladesh. Images were graded by fellowship-trained retinal specialists using the UK DESP framework, resulting in 10 clinically interpretable classes that combine retinopathy grade (R0-R3) and maculopathy status (M0/M1). A companion dataset of 2,936 pixel-level lesion masks spanning nine pathological categories was used to train a parallel multi-label lesion-detection head. The dataset was partitioned 70:15:15 (patient-stratified). Performance was evaluated using macro-averaged AUROC (DeLong estimator), sensitivity, specificity, F1 score, quadratically weighted Cohens {kappa}, and expected calibration error (ECE), with 95% CIs from 2000 bootstrap resamples. Grad-CAM spatial alignment with ground-truth lesion masks was assessed using Dice and IoU. This study follows the TRIPOD+AI reporting guidelines. FindingsOn the held-out test set (Component I: n = 3,044; Component II: n {approx} 440), DRAGS achieved class-wise precision, recall, and F1 scores ranging from 0{middle dot}88 to 0{middle dot}99 across all ten UK DESP grades, with advanced proliferative stages (R3-M0, R3-M1) consistently exceeding 0{middle dot}95. Overall accuracy was approximately 91{middle dot}1% and quadratically weighted Cohens {kappa} was approximately 0{middle dot}90. For referable versus non-referable DR, sensitivity was 90{middle dot}7% and specificity was 91{middle dot}9%. The companion lesion-detection head achieved macro-averaged sensitivity of 93{middle dot}9%, specificity of 99{middle dot}5%, and AUC of 0{middle dot}997 across nine lesion classes; seven of nine classes achieved AUC = 1{middle dot}00. Grad-CAM activations showed progressive spatial shift from diffuse (normal) to lesion-dense peripheral patterns (proliferative DR), with maximal agreement for microaneurysms and exudates. Mean inference time was 110-160 ms per image. InterpretationDRAGS demonstrates high diagnostic accuracy for nine-class UK DESP-aligned DR grading, with clinically interpretable lesion-level explainability on a large real-world LMIC dataset. External validation and prospective clinical evaluation are warranted before deployment. FundingThe present study received no funding.