Background: The retina is an accessible extension of the central nervous system, yet the protein-coding architecture linking retinal structure, visual function, and major blinding diseases remains poorly defined. Methods: We analyzed whole-exome sequencing data from 356,982 participants and performed exome-wide gene-based tests of rare coding variants and single-variant analyses of common coding variants. Rare-variant findings were further evaluated in an independent All of Us cohort (N = 245,388). We then assessed cross-phenotype pleiotropy, functional validation and biological characterization, clinical relevance, and exploratory analyses of brain-related and systemic traits. Findings: We identified 22 significant rare-variant gene-based associations involving 16 genes, including 12 novel genes, 10 of which were independently supported in the All of Us cohort. Single-variant analyses identified 243 independent common coding variants in 126 genes, including 24 novel genes. CFI, C3, and RIOX1 showed associations across retinal structure, visual function, and disease phenotypes, supporting cross-domain pleiotropy. Among the disease-associated genes, the novel gene FYB2 was selected for experimental validation because it showed expression support in retinal pigment epithelium (RPE)-related contexts and clinical relevance in Cox analyses. FYB2 knockdown aggravated barrier dysfunction in human induced RPE (iRPE) cells, supporting a potential role in diabetic retinopathy. Interpretation: These findings define the protein-coding architecture of retinal phenotypes and support shared genetic links across retinal structure, visual function, and disease. The identified genes provide targets for mechanistic investigation in blinding retinal disorders.

Purpose:To describe the visual and anatomic outcomes after being lost to follow-up (LTFU) among patients with diabetic macular edema (DME) that treated by intravitreal anti-VEGF therapies, and to determine the risk factors of persistent vision loss. Design:A retrospective cohort study using national registry data. Participants:A total of 5660 patients who were LTFU >6 months after anti-VEGF treatments from April 2020 through July 2024. Methods:Visual and anatomic outcomes at the visit before being LTFU, the return visit, 3 months, 6 months, and 12 months after return, and the final visit were collected. Multivariable logistic regression models were used to estimate odds ratios (ORs) and 95% confidence intervals (CIs) of risk factors for persistent vision loss. Main Outcome Measures:Visual acuity (VA) and central foveal thickness (CFT). Results:After a mean LTFU period of 11.3 months, 2712 (47.9%) patients had vision decline at return visit. Identified risk factors included increasing age (45-64 years vs. ≤44 years: OR, 1.30; 95% CI, 1.07-1.56; P = 0.007; ≥65 years vs. ≤44 years: OR, 1.41; 95% CI, 1.15-1.72; P = 0.001), being treated in less developed regions (OR, 1.20; 95% CI, 1.08-1.34; P = 0.001), better baseline VA (20/50-20/200 vs. worse than 20/200: OR, 3.20; 95% CI, 2.67-3.83; P < 0.001; 20/40 or better vs. worse than 20/200: OR, 5.89; 95% CI, 4.86-7.12; P < 0.001), and being LTFU for >12 months (OR, 1.22; 95% CI, 1.09-1.37; P = 0.001). A total of 792 patients returned with worsened vision and underwent ≥12 months of follow-up after resuming therapies. After a mean follow-up time of 17.3 months, 522 (65.9%) patients had persistent vision loss. Central foveal thickness presented paralleled changes with VA. Odds of persistent vision loss were greater among patients with VA of 20/40 or better before being LTFU (OR, 2.54; 95% CI, 1.25-5.15; P = 0.010) or having CFT of 350μm or more at return (OR, 1.57; 95% CI, 1.15-2.16; P = 0.005). In retreatment, odds of persistent vision loss were lower for patients who switched anti-VEGF agents (OR, 0.61; 95% CI, 0.45-0.84; P = 0.002) after return. Conclusions:Nearly half of the patients had vision decline after being LTFU, and after resuming therapies, vision loss still persisted in 2 of 3 patients. Clinical adherence should be strengthened for patients with baseline VA of 20/40 or better or having CFT of 350 μm or more at return. Multiple agents used were associated with vision recovery. Financial Disclosures:Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

The retina is an accessible extension of the central nervous system, yet the protein-coding architecture linking retinal structure, visual function, and major blinding diseases remains poorly defined. Here, using large-scale whole-exome sequencing data from 356,982 UK Biobank participants, exome-wide gene-based tests of rare coding variants and single-variant analyses of common coding variants are performed. A total of 22 significant rare-variant gene-based associations involving 16 genes are identified, including 12 novel genes, 10 of which are independently supported in the All of Us cohort (N = 245,388). Single-variant analyses identify 243 independent common coding variants in 126 genes, including 24 novel genes. CFI, C3, and RIOX1 show associations across retinal structure, visual function, and disease phenotypes, supporting cross-domain pleiotropy. Among the disease-associated genes, the novel gene FYB2 is prioritized for experimental validation, supported by retinal pigment epithelium (RPE)-related expression evidence and clinical relevance in Cox analyses. FYB2 knockdown aggravates barrier dysfunction in human induced RPE (iRPE) cells, supporting a potential role in diabetic retinopathy. These findings define the protein-coding architecture of retinal phenotypes, and support shared genetic links across retinal structure, visual function, and disease. The identified genes provide candidate targets for mechanistic investigation in blinding retinal disorders.

BACKGROUND/AIMS:To identify sociodemographic and clinical factors associated with best-corrected visual acuity (BCVA) restoration at the visit after lost to follow-up (LTFU) in patients with myopic choroidal neovascularization (mCNV) treated with antivascular endothelial growth factor intravitreal injections (IVI). METHODS:A multicentre, retrospective cohort study was conducted in China of mCNV patients receiving injections who were LTFU >6 months. Data were collected from baseline visit, pre-LTFU, initial post-LTFU visit and 6 months post-LTFU, and the final visit. Logistic regression analyses assessed associations between sociodemographic/clinical factors and BCVA restoration. RESULTS:The study included 1155 LTFU patients with mCNV (mean (SD) age: 54 (16) years; 61.6% female), with 66.6% achieving BCVA restoration at the initial post-LTFU visit after LTFU (mean(SD) duration: 315 (154) days. Patients with BCVA restoration were younger (p=0.004), more often from eastern China (p=0.022), and had shorter LTFU duration (p=0.013). Stratification by pre-LTFU BCVA showed that patients with BCVA ≥20/40 had lower restoration odds (OR 0.65, 95% CI 0.44 to 0.96; p=0.032) compared with those with BCVA <20/200. BCVA decline was associated with increased central subfield thickness(CST) (OR 1.52, 95% CI 1.15 to 2.01; p=0.004). Among 223 patients with ≥6 months post-LTFU, those with initial post-LTFU CST >280 µm had higher BCVA restoration odds (OR 1.98, 95% CI 1.02 to 3.92; p=0.046). CONCLUSIONS:Better pre-LTFU BCVA (≥20/40) increased post-LTFU deterioration risk (associated with progressive CST thickening), while elevated post-LTFU CST (>280 µm) predicted higher BCVA restoration odds in patients with sustained follow-up, supporting CST-guided monitoring and early intervention for high-risk groups.

PURPOSE:To identify peripheral retinal vascular phenotypes in retinitis pigmentosa (RP) using ultra-widefield OCT angiography (UWF-OCTA), validate a quantitative vascular biomarker, and develop a clinically applicable severity grading system. DESIGN:Development and validation of a severity grading system. METHODS:We analyzed 544 eyes from 272 patients with primary RP from three tertiary centers. Peripheral vascular phenotypes were defined using UWF-OCTA. Structure-function correlations between the CNZ area and multimodal visual functions-including kinetic perimetry (KP), full-field stimulus threshold (FST), and static perimetry (SP)-were evaluated using linear mixed-effects models (LMMs) to account for inter-eye correlation. A four-stage severity grading system was developed using principal component analysis (PCA) to construct a composite functional score and regression tree analysis with cluster bootstrapping (1,000 resamples) to determine optimal CNZ cut-points; the system was then validated in internal and external datasets. Age-related severity distribution was evaluated through a Kaplan-Meier age-at-observation simulation. RESULTS:Two distinct peripheral vascular phenotypes were identified: Type 1 (Terminal Remodeling; 306 eyes, 56.2%) and Type 2 (Simple Atrophy; 238 eyes, 43.8%). Type 1 eyes showed greater disease severity than Type 2 eyes at comparable ages. The CNZ area demonstrated strong structure-function correlations, explaining substantial variance in KP (Marginal R2 up to 0.77) and FST (Marginal R2 up to 0.64). A CNZ-based 4-stage severity grading (Stage I, ≥ 260 mm²; Stage II, 200-260 mm²; Stage III, 150-200 mm²; Stage IV, < 150 mm²) effectively stratified disease severity, demonstrating high explanatory power for visual function (KP η² up to 0.75; FST η² up to 0.66) and maintaining discriminatory performance for best-corrected visual acuity (BCVA) in both internal (η² = 0.17) and external (η² = 0.28) validation cohorts. CONCLUSIONS:Peripheral retinal vascular alterations captured by UWF-OCTA define distinct RP phenotypes and identify the CNZ as a robust quantitative biomarker of disease severity. The CNZ-based severity grading system provides an objective framework for clinical severity assessment, disease monitoring, and patient stratification in future interventional trials.

PURPOSE:To evaluate 1-year efficacy and safety of aflibercept 8 mg in Asian patients with neovascular age-related macular degeneration (nAMD) in the PULSAR (NCT04423718) trial. DESIGN:Subgroup analysis of the Phase 3 PULSAR trial. METHODS:Patients aged ≥ 50 years with nAMD were randomized to receive intravitreal aflibercept 8 mg every 12 (8q12) or 16 (8q16) weeks, or aflibercept 2 mg every 8 weeks (2q8), following 3 initial monthly injections. Outcomes included change from baseline in best-corrected visual acuity (BCVA) and central subfield retinal thickness (CRT), durability, and safety at week (W) 48. FINDINGS:Overall, 234 Asian patients participated and received study treatment (8q12 [n = 74], 8q16 [n = 77], and 2q8 [n = 83]). At W48, the least squares (LS) mean (95 % CI) BCVA change from baseline was + 9.9 (6.9, 12.9), + 9.2 (7.1, 11.2), and + 7.6 (4.7, 10.5) letters in the 8q12, 8q16, and 2q8 groups, respectively. The LS mean (95 % CI) change in CRT from baseline to W48 for the 8q12, 8q16, and 2q8 groups was -141 (-152, -129), -156 (-167, -146), and -142 (-155, -129) µm, respectively. Most patients receiving 8q12 (85 %) or 8q16 (87 %), maintained their randomized dosing interval, through W48. The safety profile of aflibercept 8 mg was comparable to aflibercept 2 mg. CONCLUSIONS:Aflibercept 8 mg was effective and well tolerated in Asian patients with nAMD, with demonstrated improvements in functional and anatomic outcomes comparable to aflibercept 2 mg at W48. Aflibercept 8 mg outcomes were achieved with fewer injections than with aflibercept 2 mg, consistent with overall PULSAR results.

ABSTRACT Artificial intelligence holds promise for addressing the uneven distribution of medical resources across different regions. Compared with convolutional neural network models, foundation models offer significant advantages: on one hand, their pre‐training does not rely on massive labeled data; on the other hand, they possess the capability to fuse multimodal images with clinical metadata. This review systematically searched PubMed and IEEE databases to select five representative foundation models for ophthalmology. We focus on analyzing their core architectural design strategies and diagnostic performance, while summarizing technical approaches for enhancing generalization capabilities and explainability across different models. Foundation models typically employ masked autoencoders techniques for self‐supervised learning. This paradigm expands diagnostic coverage from common ophthalmic diseases to rare conditions and even systemic diseases. However, challenges persist: homogeneity in training data can lead to unstable performance, and more prospective trials are required to validate their efficacy in real clinical settings.

Objective To evaluate the safety, tolerability and preliminary efficacy of IBI324, a vascular endothelial growth factor-A/angiopoietin-2 bispecific antibody, in participants with diabetic macular oedema (DME).Methods and analysis This multicentre, open-label, phase 1 dose-escalation clinical trial consisted of a single ascending dose (SAD) stage and a multiple ascending dose (MAD) stage. 24 participants with fovea-involving DME were enrolled. SAD participants received a single intravitreal injection (IVT) of 0.5 mg, 2 mg or 4 mg IBI324 and were followed up until Day 42 post injection. In the MAD stage, six participants each received 3 monthly IVTs of 2 mg or 4 mg IBI324 and were followed up until determined as ‘treatment needed’ per prespecified criteria or until 24 weeks after first dose. The primary endpoints were incidence of dose-limiting toxicities (DLTs), adverse events (AEs) and changes in vital signs and laboratory test findings.Results No DLT, treatment-related AE, AE of special interest or ocular serious AE was reported. Treatment-emergent adverse events (TEAEs), all mild or moderate in severity, were observed in 4 (33.3%) SAD participants and 11 (91.7%) MAD participants. TEAEs of the study eye included intraocular pressure increased, conjunctival haemorrhage, allergic conjunctivitis, posterior capsular opacification and visual acuity decreased. Mean best-corrected visual acuity increase and mean central subfield thickness decrease from baseline in the study eye were observed in all dose groups, accompanied by intraretinal fluid/subretinal fluid improvements. 16 weeks after the last dose, 7 (58.3%) MAD 2 mg and 6 (50.0%) MAD 4 mg participants remained free of ‘treatment needed’.Conclusion IBI324 was well tolerated with evidence of functional and anatomical improvement in patients with DME.Trial registration number NCT05489718.

Background To evaluate the central-field stimulus threshold (CST) values in patients with age-related macular degeneration (AMD) and investigate the relationship between CST values and conventional visual function assessments, as well as retinal anatomical parameters, thereby exploring the role of CST in assessing anti-vascular endothelial growth factor (VEGF) therapy. Methods Forty patients (80 eyes) with AMD were included in the analysis. All patients underwent complete ophthalmic examination, including slit-lamp examination, fundus examination, intraocular pressure measurement, Early Treatment Diabetic Retinopathy Study best-corrected visual acuity (BCVA), fundus photography, optical coherence tomography (OCT), OCT angiography, microperimetry (MP), and CST testing. Linear mixed-effects models and generalized linear mixed models were employed to analyze the correlations between CST and BCVA, MP, OCT features, and choroidal neovascularization (CNV) area, as well as changes in CST following anti-VEGF therapy. Results Compared to normal eyes, nAMD eyes exhibited higher CST values for white (-47.02 ± 9.02 vs. -51.09 ± 3.61 dB), blue (-52.50 ± 14.46 vs. -60.28 ± 8.60 dB), and red (-30.22 ± 8.21 vs. -36.24 ± 6.41 dB) stimuli. Approximately 39% of nAMD eyes exhibited cone-mediated vision under scotopic conditions. In nAMD eyes, BCVA was negatively correlated with blue and white CST, indicating that better BCVA was associated with higher central retinal light sensitivity. Red and blue CST were positively correlated with center retinal thickness (CRT) and CNV area (All P < 0.05). In the thicker CRT group, the blue-red CST difference decreased from -18.03 ± 10.01 dB to -22.85 ± 2.22 dB after 1 month of anti-VEGF treatment (P = 0.047), 9 of 14 eyes were cone-mediated at baseline, which decreased to 3 eyes after three months of treatment. Conclusions CST assessment revealed that both cone and rod were affected in nAMD eyes. CST was correlated with CRT and CNV area. In some nAMD patients, severe rod dysfunction leads to a shift to cone-mediated vision under scotopic conditions, but anti-VEGF therapy may ameliorate this phenomenon. These findings suggest that CST assesses the visual function of AMD patients from a more microscopic, cellular functional standpoint, which may be beneficial for monitoring AMD progression and evaluating the treatment efficacy.

Purpose:The purpose of this study was to evaluate the clinical characteristics of the foveal sparing phenotype among patients with inherited retinal diseases (IRDs) and to identify the predictive imaging markers for visual prognosis. Methods:Consecutive patients with definitive clinical and genetic diagnoses of IRD who first visited our clinic from November 2021 to December 2022 and had high-quality spectral-domain optical coherence tomography (SD-OCT) images were included in this retrospective cohort. The foveal sparing phenotype was defined by foveal preservation on autofluorescence (FAF) and OCT images. Best-corrected visual acuity (BCVA) and dimensions of residual structures of outer retinal layers on OCT and FAF images were measured and analyzed. Spearman correlation analysis was performed to evaluate the correlation between retinal imaging features and BCVA. Receiver operating characteristic (ROC) curve was performed to assess the predictive performance of residual ellipsoid zone (EZ) area on OCT. Results:A total of 601 eyes of 308 Chinese patients with IRD were enrolled. Foveal sparing phenotype was observed in 195 (32.4%) eyes of 105 (34.1%) patients. There were 46.7% of cases with USH2A-related retinopathy and 76.9% EYS-related retinopathy presented foveal sparing phenotype, whereas their structural integrity showed no significant difference. Spearman correlation analyses revealed significant association between residual EZ area (P < 0.01) and BCVA. ROC curve analysis demonstrated that the residual EZ area at initial diagnosis could predict the degree of BCVA deterioration within 2 years (area under the curve [AUC] = 0.70). Conclusions:Our findings indicate that the foveal sparing phenotype is associated with better visual prognosis and is more frequently observed in IRDs associated with EYS and USH2A mutations. The residual EZ area on OCT can serve as a predictor of the timeframe for central vision deterioration to blindness in patients with IRD. Translational Relevance:Residual EZ area on OCT can serve as a predictive biomarker to support clinical decision making in monitoring and managing the progression of IRD.

Diabetic retinopathy (DR) is driven by chronic oxidative stress and mitochondrial dysfunction, yet effective, non-invasive strategies for mitochondrial quality control (MQC) that can traverse the blood-retinal barrier (BRB) remain limited. Here, we systematically develop and investigate a bioengineered nanoplatform-endothelial mitochondria-derived vesicles (EMDVs) from retinal microvascular cells. Through optimized functional extraction and membrane potential-preserving bioengineering, EMDVs retain critical mitochondrial membrane potential and adenosine triphosphate synthesis capabilities. Upon Coenzyme Q10 (CoQ10) modification, EMDVs exhibit potent antioxidant activity. Kinetic and phenotypic recovery assays demonstrate that EMDVs non-invasively penetrate the retina and efficiently deliver CoQ10 across the BRB to all retinal layers, restoring mitochondrial homeostasis and remodeling antioxidant defenses. Transcriptomic and mechanistic analyses further reveal that topical administration of EMDVs and CoQ10-engineered vesicles dynamically modulates the disease-dependent eIF2 alpha-ATF4-CHOP-integrated stress response axis, facilitating repair and remodeling of the retinal barrier, particularly the microvasculature, in both early- and late-stage DR. Long-term in vivo safety evaluation confirms no systemic toxicity or local inflammation. This study introduces a mitochondria-derived vesicle nanoplatform with high BRB permeability and efficient MQC function, highlighting its translational potential as a dynamic, targeted therapeutic strategy for mitochondrial dysfunction-related degenerative diseases and versatile drug delivery across biological barriers.

BACKGROUND:This study aims to investigate the role of inflammation in proliferative vitreoretinopathy (PVR) by analysing inflammation-related proteins in vitreous samples and identifying potential biomarkers for PVR severity. METHODS:Vitreous samples were collected from patients with different stages of PVR (control group, n = 15; PVR A/B, n = 15; PVR C/D, n = 14). Using an Onlink proteomics panel, the levels of 92 inflammation-related proteins were measured. Differential expression analysis was conducted to identify proteins significantly altered between the PVR and control groups. AlphaFold3 modelling was employed to predict common binding peptide segments among differentially expressed proteins. Additionally, the impact of PVR vitreous on inflammatory cytokine production in retinal pigment epithelium (RPE) cells was assessed to explore the functional consequences of the observed protein changes. RESULTS:The proteomic analysis revealed differential expression of 63 inflammation-related proteins between the PVR and control groups. Among these, MCP-1 exhibited significant variations across PVR stages and demonstrated high diagnostic value for PVR presence (AUC = 0.949) and severity (PVR C/D, AUC = 0.719). AlphaFold3 modelling identified a common binding peptide segment, INAPVTCCYN, shared by MCP-1, CCL3, CCL4 and CXCL9, suggesting their potential collaborative role in PVR progression. Furthermore, vitreous samples from PVR patients significantly stimulated inflammatory cytokine production in RPE cells, indicating a functional link between vitreous inflammation and retinal pathology. CONCLUSIONS:PVR is associated with elevated levels of vitreous inflammation and MCP-1 is a promising biomarker for predicting and diagnosing PVR severity. The identification of a common binding peptide segment and the stimulation of cytokine production in RPE provide mechanistic insights into PVR progression.

INTRODUCTION:γ-Aminobutyric acid (GABA), a classical neurotransmitter, also regulates skeletal muscle-an endocrine organ secreting irisin. This FNDC5-derived myokine induces white adipose tissue browning, augmenting thermogenesis and metabolic function. OBJECTIVES:This study investigated the effects of GABA on irisin secretion and its molecular mechanisms. METHODS:In L6 myotubes, GABA activated GABAAR, causing membrane depolarization and calcium influx, which upregulated CREB phosphorylation and increased PGC-1α and FNDC5 expression, significantly elevating irisin secretion. In vivo, GABA treatment increased PGC-1α/FNDC5 expression in mouse skeletal muscle and raised serum irisin levels. Additionally, GABA-induced irisin promoted the browning of white adipose tissue by upregulating thermogenic genes and remodeling fat metabolism. CONCLUSION:These findings reveal a novel role for GABA in regulating myokine secretion and suggest its potential as a therapeutic target for metabolic diseases such as obesity and type 2 diabetes.

Purpose:RHO mutations are the primary cause of autosomal dominant retinitis pigmentosa (adRP), with Class 1 mutations typically exhibiting more severe phenotypes than Class 2. This study aims to clarify the mechanistic basis for this clinical disparity by systematically comparing protein degradation pathways, mitochondrial stress, and neuroinflammation. Methods:Humanized mouse lines carrying Class 1 (P347L) or Class 2 (L125R) RHO mutations were generated via CRISPR/Cas9-mediated knock-in. Retinal function, ultrastructure, and transcriptomic profiles were characterized through electroretinography (ERG), transmission electron microscopy (TEM), and RNA-sequencing (RNA-seq). To further elucidate molecular mechanisms, protein trafficking and degradation pathways were analyzed in transfected HEK293T cells using HiBiT extracellular quantification, pharmacological inhibition of lysosomal and proteasomal pathways, and BRET2 visual arrestin recruitment assay. Results:The P347L mutant failed to undergo efficient outer-segment-directed trafficking and was predominantly degraded via the lysosomal pathway, consistent with its enhanced visual arrestin recruitment and endocytosis. In contrast, the L125R mutant showed protein misfolding and was degraded by both proteasomal and lysosomal pathways. In vivo, P347L mice exhibited more pronounced mitochondrial dysfunction than L125R mice, accompanied by elevated cGMP levels and lysosomal overload. Neuroinflammation was similarly present in both mutants, indicating a shared pathological mechanism rather than a differential contributor. Conclusions:We propose a pathogenic model in which elevated endocytosis and mitochondrial dysfunction contribute to the accelerated photoreceptor degeneration in RHO P347L-associated adRP.

Anti-AAV neutralizing antibody (NAb) levels vary widely across regions and serotypes, yet current data on their seroprevalence in the Chinese population remain limited. This multi-provincial observational study recruited 286 adult participants across six Chinese provinces to assess NAbs against six clinically relevant serotypes using transduction inhibition assays. Results showed widespread NAb prevalence in the Chinese population, with anti-AAV2 having the highest median titer (1:1675) and anti-AAV5 the lowest (1:15). Prevalence patterns aligned with capsid sequence conservation, with higher similarity serotypes showing comparable prevalence. Geographically, southern provinces showed significantly greater seroprevalence of anti-AAV6 and anti-AAV8 NAbs than northern provinces. Healthcare workers performing invasive procedures had higher titers for anti-AAV5 NAb than staff performing non-invasive procedures. Tree-based regression revealed serotype-specific correlates: total cholesterol and HBsAb showed broad statistically positive associations with NAb titers, while ApoA1 was selectively positive and ApoB negative. This work provides real-world reference data for NAb prevalence in China, demonstrating widespread prevalence with particularly high AAV2 titers influenced by geographical and demographic factors. Considering the hepatotropic nature of AAV and the importance of hepatic lipid metabolism, the correlation between NAb titers and lipid profiles (cholesterol and apolipoproteins) points to potential biological mechanisms that merit further study.

The exposome, defined as the totality of external exposures individuals experience throughout the lifespan and the internal biological responses they trigger, is increasingly recognized as a major, modifiable determinant of health. The eye is uniquely vulnerable to exposome risk factors because it comprises an external interface directly exposed to the environment, a transparent lens highly sensitive to cumulative oxidative injury, and metabolically active retina prone to inflammation and vascular dysregulation. Research linking ocular diseases to air pollution, climate and weather factors, heavy metals, persistent organic pollutants, and emerging threats such as micro-/nanoplastics has expanded rapidly, yet the evidence remains fragmented and lacks systematic integration. In this review, we propose an anatomy-informed ocular exposome atlas along the visual axis to comprehensively evaluate evidence from clinical studies, pathological mechanisms, and intervention strategies across major disease groups-including ocular surface diseases, cataract, refractive errors, glaucoma, and retinal diseases. We further integrate cross-disease mechanistic pathways (oxidative stress, cellular responses, inflammation, immune reactions) and discuss causality. Finally, we identify priority research directions and provide actionable recommendations for prevention and policy.

Importance:In the Study of the Effects of High-Dose Aflibercept Injected Into the Eye of Patients With an Age-Related Disorder That Causes Loss of Vision Due to Growth of Abnormal Blood Vessels at the Back of the Eye (PULSAR) phase 3 randomized clinical trial, treatment with aflibercept, 8 mg, demonstrated noninferior (4-letter margin) best-corrected visual acuity (BCVA) gains vs aflibercept, 2 mg, in participants with neovascular age-related macular degeneration (nAMD). This post hoc subgroup analysis evaluated clinical outcomes in participants with polypoidal choroidal vasculopathy (PCV). Objective:To compare the efficacy and safety of aflibercept, 8 mg vs 2 mg, monotherapy among participants with PCV in the PULSAR trial. Design, Setting, and Participants:This was a post hoc subgroup analysis of the PULSAR randomized clinical trial. The setting included hospitals and clinics in 12 countries where indocyanine green angiography (ICGA) was performed to identify PCV. Included were a subgroup of adults with nAMD enrolled in the PULSAR trial with ICGA-confirmed PCV. Study data were analyzed from August 2020 to July 2022. Interventions:Participants were randomly assigned 1:1:1 to aflibercept, 8 mg, every 12 weeks or 16 weeks, or aflibercept, 2 mg, every 8 weeks, each after 3 initial monthly doses. From week 16, dosing intervals in the treatment arms receiving 8 mg every 12 weeks and every 16 weeks were shortened if predefined disease activity criteria were met at prespecified visits. Main Outcomes and Measures:Least-squares (LS) mean change in BCVA from baseline at week 48. Results:A total of 139 participants were included in this analysis. ICGA-confirmed PCV was present in 44 participants in the treatment group receiving aflibercept, 8 mg, every 12 weeks (mean [SD] age, 72.2 [8.1] years; 50% male), 41 participants receiving 8 mg every 16 weeks (mean [SD] age, 73.2 [8.7] years; 63% male), and 54 participants receiving 2 mg every 8 weeks (mean [SD] age, 72.6 [8.2] years; 69% male). Mean baseline BCVA letter score (approximate Snellen) was 56.3 (20/80), 60.1 (20/63), and 57.6 (20/80), respectively, with 41, 37, and 51 participants completing week 48 and receiving a mean (SD) of 6.1 (0.4), 5.1 (0.5), and 7.0 (0.2) injections, including 68 of 78 (87%) treated with aflibercept, 8 mg, who maintained dosing intervals of 12 weeks or longer. In the treatment arms receiving aflibercept, 8 mg, every 12 and 16 weeks and aflibercept, 2 mg, every 8 weeks, LS mean BCVA change from baseline at week 48 was +9.5, +8.4, and +9.1 letters, respectively (estimated difference, 0.40; 95% CI, -4.4 to 5.2 letters for 8 mg every 12 weeks vs 2 mg every 8 weeks; -0.7; 95% CI, -4.6 to 3.2 letters for 8 mg every 16 weeks vs 2 mg every 8 weeks), and polypoidal lesions were absent in 37%, 47%, and 38% of participants, respectively, who completed week 48. Conclusions and Relevance:Results of this post hoc analysis of the PULSAR randomized clinical trial in participants with PCV demonstrated similar visual and anatomic outcomes with aflibercept, 8 mg vs 2 mg, as administered in this trial, supporting the use of aflibercept, 8 mg, as an alternative monotherapy for PCV. Trial Registration:ClinicalTrials.gov Identifier: NCT04423718.

PULSAR (NCT04423718) was a global, phase 3, randomized, double-masked, non-inferiority study of adults with neovascular age-related macular degeneration (nAMD). Patients were randomized 1:1:1 to receive aflibercept 8 mg every 12 (8q12), or 16 weeks (8q16), or aflibercept 2 mg every 8 weeks (2q8), following 3 initial monthly doses. This subgroup analysis investigated the efficacy and safety of aflibercept 8 mg vs. aflibercept 2 mg in patients from China with nAMD from the PULSAR trial. This exploratory analysis evaluated the change from baseline in best-corrected visual acuity (BCVA), central retinal thickness (CRT), durability, and safety outcomes through week 48 in patients from China. All results were descriptive in nature. Least squares (LS) mean (95

The longitudinal management of blinding fundus diseases constitutes a Partially Observable Markov Decision Process (POMDP) necessitating a critical precision-risk trade-off between intervention and over-treatment, as true pathology is often obscured in static observations. However, existing paradigms fail to address this complexity. Traditional vision models remain uninterpretable and memoryless, and while Vision-Language Models (VLMs) excel in semantic understanding, they rely on unsafe open-loop text reasoning lacking the anatomical grounding essential for clinical safety. Furthermore, robust learning is hindered by the scarcity of process supervision in sparse clinical records. To bridge this gap, we introduce the {Logic-Constrained Abductive Data Engine}. Operating on a ``Propose-and-Verify'' paradigm, it validates MLLM-Proposed biomarkers against clinical and temporal logic to reconstruct dense pathological states from sparse outcomes. Building on this foundation, we propose ORBIT, the first ophthalmic Prognostic World Model. Uniquely, ORBIT employs counterfactual visual foresight to imagine anatomical futures under different treatments, anchoring decisions in Closed-Loop Anatomical Verification rather than linguistic probabilities. Experiments demonstrate that ORBIT effectively captures disease evolution and establishes a new paradigm for autonomous diagnosis and reliable decision-making in complex ophthalmic environments.

In the Study of the Effects of High-Dose Aflibercept Injected Into the Eye of Patients With an Age-Related Disorder That Causes Loss of Vision Due to Growth of Abnormal Blood Vessels at the Back of the Eye (PULSAR) phase 3 randomized clinical trial, treatment with aflibercept, 8 mg, demonstrated noninferior (4-letter margin) best-corrected visual acuity (BCVA) gains vs aflibercept, 2 mg, in participants with neovascular age-related macular degeneration (nAMD). This post hoc subgroup analysis evaluated clinical outcomes in participants with polypoidal choroidal vasculopathy (PCV). To compare the efficacy and safety of aflibercept, 8 mg vs 2 mg, monotherapy among participants with PCV in the PULSAR trial. This was a post hoc subgroup analysis of the PULSAR randomized clinical trial. The setting included hospitals and clinics in 12 countries where indocyanine green angiography (ICGA) was performed to identify PCV. Included were a subgroup of adults with nAMD enrolled in the PULSAR trial with ICGA-confirmed PCV. Study data were analyzed from August 2020 to July 2022. Participants were randomly assigned 1:1:1 to aflibercept, 8 mg, every 12 weeks or 16 weeks, or aflibercept, 2 mg, every 8 weeks, each after 3 initial monthly doses. From week 16, dosing intervals in the treatment arms receiving 8 mg every 12 weeks and every 16 weeks were shortened if predefined disease activity criteria were met at prespecified visits. Least-squares (LS) mean change in BCVA from baseline at week 48. A total of 139 participants were included in this analysis. ICGA-confirmed PCV was present in 44 participants in the treatment group receiving aflibercept, 8 mg, every 12 weeks (mean [SD] age, 72.2 [8.1] years; 50% male), 41 participants receiving 8 mg every 16 weeks (mean [SD] age, 73.2 [8.7] years; 63% male), and 54 participants receiving 2 mg every 8 weeks (mean [SD] age, 72.6 [8.2] years; 69% male). Mean baseline BCVA letter score (approximate Snellen) was 56.3 (20/80), 60.1 (20/63), and 57.6 (20/80), respectively, with 41, 37, and 51 participants completing week 48 and receiving a mean (SD) of 6.1 (0.4), 5.1 (0.5), and 7.0 (0.2) injections, including 68 of 78 (87%) treated with aflibercept, 8 mg, who maintained dosing intervals of 12 weeks or longer. In the treatment arms receiving aflibercept, 8 mg, every 12 and 16 weeks and aflibercept, 2 mg, every 8 weeks, LS mean BCVA change from baseline at week 48 was +9.5, +8.4, and +9.1 letters, respectively (estimated difference, 0.40; 95% CI, −4.4 to 5.2 letters for 8 mg every 12 weeks vs 2 mg every 8 weeks; −0.7; 95% CI, −4.6 to 3.2 letters for 8 mg every 16 weeks vs 2 mg every 8 weeks), and polypoidal lesions were absent in 37%, 47%, and 38% of participants, respectively, who completed week 48. Results of this post hoc analysis of the PULSAR randomized clinical trial in participants with PCV demonstrated similar visual and anatomic outcomes with aflibercept, 8 mg vs 2 mg, as administered in this trial, supporting the use of aflibercept, 8 mg, as an alternative monotherapy for PCV. ClinicalTrials.gov Identifier: NCT04423718