Search

Diane N Sayah,1– 3 Itika Garg,1,2 Raviv Katz,1,2 Ying Zhu,1,4 Ying Cui,1,5 Rebecca Zeng,1,2 Rachel Tandias,1,2 Jade Y Moon,1,2 Filippos Vingopoulos,1,2 Hannah E Wescott,1,2 Grace Baldwin,1,2 Kira Wang,1,2 Tobias Elze,1,2 Cassie Ann Ludwig,1,2 Demetrios G Vavvas,1,2 Joan W Miller,1,2 Deeba Husain,1,2 Leo A Kim,1,2 Nimesh A Patel,1,2 John B Miller1,2 1Harvard Retinal Imaging Lab, Boston, MA, USA; 2Retina Service, Department of Ophthalmology, Massachusetts Eye and Ear, Harvard Medical School, Boston, MA, USA; 3College of Optometry, University of Houston, Houston, TX, USA; 4Eye Center of Xiangya Hospital, Central South University, Changsha, Hunan, People’s Republic of China; 5Guangdong Eye Institute, Department of Ophthalmology, Guangdong Provincial People’s Hospital, Guangdong Academy of Medical Sciences, Guangzhou, People’s Republic of ChinaCorrespondence: John B Miller, Retina Service, Massachusetts Eye and Ear, Department of Ophthalmology, Harvard Medical School, 243 Charles St, Boston, MA, 02114, USA, Tel +1 617 573-3750, Fax + 1 617 573-3698, Email John_Miller@meei.harvard.eduPurpose: Visual prognosis and treatment burden for macular neovascularization (MNV) can differ between myopic macular degeneration (MMD) and age-related macular degeneration (AMD). We describe and compare MNV associated with MMD and AMD using swept-source (SS)-OCTA.Patients and Methods: Adult patients with documented MNV associated with MMD or AMD were consecutively recruited. Qualitative and quantitative features were assessed from 6x6mm angiograms, including the MNV area and vessel density (VD). Descriptive statistics and linear regression analyses were carried out.Results: Out of 75 enrolled eyes with diagnosed MNV (30 MMD-MNV and 45 AMD-MNV; mean age 55± 19 and 75± 8 years, respectively), 44 eyes had discernible MNV (11 MMD-MNV and 33 AMD-MNV) on SS-OCTA at the time of the study and were included in the analysis. The MMD-MNV group exhibited a three-fold smaller sized MNV (p=0.001), lower greatest linear dimension (p=0.009) and greatest vascular caliber (p< 0.001) compared to AMD-MNVs, and had a higher prevalence of tree-in-bud pattern. Eyes with AMD showed a higher prevalence of type 1 MNVs with medusa pattern. There was no difference in the location of the MNV, shape’s regularity, margins, presence of core vessel, capillary fringe, peripheral loops, or perilesional dark halo (p> 0.05) between both conditions. After adjustment, decreased MNV area and increased VD were associated with the tree-in-bud pattern, whereas the diagnosis did not significantly influence those parameters.Conclusion: While larger studies are warranted, this study is the first to describe and compare MMD-MNV and AMD-MNV using SS-OCTA, providing relevant clinical insight on MNV secondary to MMD and AMD. These findings also further validate OCTA as a powerful tool to detect and characterize MNV non-invasively.Keywords: macular neovascularization, age-related macular degeneration, high myopia, swept source, OCT angiography

Nerve growth factor treatment significantly improved visual acuity and non-visual outcomes and reversed the trend of visual decline in 1,476 patients with neurotrophic keratopathy within the IRIS Registry., (Copyright © 2024. Published by Elsevier Inc.)

Background and Objective: This study aimed to investigate the rate of rhegmatogenous retinal detachment (RRD) in the fellow eye of patients after developing an RRD in one eye., Patients and Methods: This is a retrospective cohort study of patients with a new RRD diagnosis in the American Academy of Ophthalmology IRIS Registry (Intelligent Research in Sight) from 2016 to 2020. The association between risk factors of RRD and fellow-eye RRD was evaluated using Cox regression., Results: Among the 201,622 patients who developed a new RRD in at least one eye, 6,740 (3.34%) developed an RRD in the fellow eye at a mean time of 406.02 days (SD = 373.74) from the first eye diagnosis. Of those 6,740 eyes, 1,889 (28.03%) had lattice degeneration, 3,938 (58.43%) had PVD, 588 (8.72%) had cataract surgery, and 38 (0.56%) had complex cataract surgery. Adjusted regression modeling showed that male sex (HR 1.73, [95% confidence interval (CI) 1.64 to 1.83]; P < 0.001), younger age at first eye RRD diagnosis (HR 1.02, [95% CI 1.01 to 1.02]; P < 0.001), active smoking (HR 1.13, [95% CI 1.05 to 1.22]; P = 0.002), and cataract surgery (HR 1.33, [95% CI 1.23 to 1.45]; P < 0.001) were associated with increased risk of a fellow eye RRD. On the other hand, fellow eye PVD was associated with a lower risk of RRD (HR 0.80, [95% CI 0.76 to 0.84]; P < 0.001)., Conclusion: Among patients with an RRD in one eye, male sex, younger age of first eye RRD diagnosis, active smoking, and cataract surgery were associated with a higher risk of developing a fellow eye RRD, whereas fellow eye PVD was associated with a lower risk of fellow eye RRD. [ Ophthalmic Surg Lasers Imaging Retina 2024;55:XX-XX.] .

Purpose: To assess the diagnostic performance of two chatbots, ChatGPT and Glass, in uveitis diagnosis compared to renowned uveitis specialists, and evaluate clinicians' perception about utilizing artificial intelligence (AI) in ophthalmology practice., Methods: Six cases were presented to uveitis experts, ChatGPT (version 3.5 and 4.0) and Glass 1.0, and diagnostic accuracy was analyzed. Additionally, a survey about the emotions, confidence in utilizing AI-based tools, and the likelihood of incorporating such tools in clinical practice was done., Results: Uveitis experts accurately diagnosed all cases (100%), while ChatGPT achieved a diagnostic success rate of 66% and Glass 1.0 achieved 33%. Most attendees felt excited or optimistic about utilizing AI in ophthalmology practice. Older age and high level of education were positively correlated with increased inclination to adopt AI-based tools., Conclusions: ChatGPT demonstrated promising diagnostic capabilities in uveitis cases and ophthalmologist showed enthusiasm for the integration of AI into clinical practice.

Introduction: Accurate diagnosis and timely management are vital for favorable uveitis outcomes. Artificial Intelligence (AI) holds promise in medical decision-making, particularly in ophthalmology. Yet, the diagnostic precision and management advice from AI-based uveitis chatbots lack assessment., Methods: We appraised diagnostic accuracy and management suggestions of an AI-based chatbot, ChatGPT, versus five uveitis-trained ophthalmologists, using 25 standard cases aligned with new Uveitis Nomenclature guidelines. Participants predicted likely diagnoses, two differentials, and next management steps. Comparative success rates were computed., Results: Ophthalmologists excelled (60-92%) in likely diagnosis, exceeding AI (60%). Considering fully and partially accurate diagnoses, ophthalmologists achieved 76-100% success; AI attained 72%. Despite an 8% AI improvement, its overall performance lagged. Ophthalmologists and AI agreed on diagnosis in 48% cases, with 91.6% exhibiting concurrence in management plans., Conclusions: The study underscores AI chatbots' potential in uveitis diagnosis and management, indicating their value in reducing diagnostic errors. Further research is essential to enhance AI chatbot precision in diagnosis and recommendations.

Purpose: To determine if intravitreal injection of antibiotics alone versus early pars plana vitrectomy (PPV) plus injection of intravitreal antibiotics predicted better or worse visual outcomes for patients with endophthalmitis after anti-VEGF injections., Design: Retrospective cohort study., Participants: Patients developing endophthalmitis after receiving an intravitreal anti-VEGF injection from the American Academy of Ophthalmology IRIS® (Intelligent Research in Sight) Registry between 2016 and 2020., Methods: Inclusion criteria were endophthalmitis diagnosis within 1 to 28 days after anti-VEGF injection and a recorded visual acuity (VA) at baseline, on the day of diagnosis, and posttreatment. Patients in the Injection Only group underwent intravitreal injection of antibiotics alone and in the Early Vitrectomy group received PPV with intravitreal antibiotics or intravitreal injection followed by PPV within 2 days of diagnosis. Patients were excluded if they had cataract surgery during the study, intravitreal steroids before endophthalmitis, or intermediate/posterior uveitis or cystoid macular edema. The study created a 1:1 matched cohort using Mahalanobis distance matching, accounting for the differences in VA at baseline and diagnosis., Main Outcome Measures: Posttreatment logarithm of the minimum angle of resolution (logMAR) VA., Results: A total of 1044 patients diagnosed with postinjection endophthalmitis met the inclusion and exclusion criteria. In the unmatched cohort, there were 935 patients in the Injection Only and 109 in the Early Vitrectomy group. In 1:1 matched cohort, 218 patients (109 in each group) were included; the median logMAR VAs were 0.32 (20/40-20/50) at baseline, 0.88 (∼20/150) at diagnosis, and 0.57 (20/70-20/80) posttreatment. There were no statistically significant differences in the visual outcomes between the 2 matched treatment groups (b = 0.05; P = 0.23); including the subgroup of patients with VA worse than 1.0 logMAR (b = 0.05; P = 0.452)., Conclusions: There was no significant difference in final VA outcomes between patients receiving Injection Only and those treated with Early Vitrectomy for postinjection endophthalmitis. The findings support the use of either treatment strategy., Financial Disclosure(s): Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article., (Copyright © 2024. Published by Elsevier Inc.)

Purpose: To predict 10-2 Humphrey visual fields (VFs) from 24-2 VFs and associated non-total deviation features using deep learning., Methods: We included 5189 reliable 24-2 and 10-2 VF pairs from 2236 patients, and 28,409 reliable pairs of macular OCT scans and 24-2 VF from 19,527 eyes of 11,560 patients. We developed a transformer-based deep learning model using 52 total deviation values and nine VF test features to predict 68 10-2 total deviation values. The mean absolute error, root mean square error, and the R2 were evaluation metrics. We further evaluated whether the predicted 10-2 VFs can improve the structure-function relationship between macular thinning and paracentral VF loss in glaucoma., Results: The average mean absolute error and R2 for 68 10-2 VF test points were 3.30 ± 0.52 dB and 0.70 ± 0.11, respectively. The accuracy was lower in the inferior temporal region. The model placed greater emphasis on 24-2 VF points near the central fixation point when predicting the 10-2 VFs. The inclusion of nine VF test features improved the mean absolute error and R2 up to 0.17 ± 0.06 dB and 0.01 ± 0.01, respectively. Age was the most important 24-2 VF test parameter for 10-2 VF prediction. The predicted 10-2 VFs achieved an improved structure-function relationship between macular thinning and paracentral VF loss, with the R2 at the central 4, 12, and 16 locations of 24-2 VFs increased by 0.04, 0.05 and 0.05, respectively (P < 0.001)., Conclusions: The 10-2 VFs may be predicted from 24-2 data., Translational Relevance: The predicted 10-2 VF has the potential to improve glaucoma diagnosis.

Purpose: To evaluate racial disparities in vision outcomes and eye care utilization among glaucoma patients., Design: Retrospective cohort study., Methods: In this population-based IRIS Registry (Intelligent Research in Sight) study, we included patients with minimum one diagnosis code for glaucoma at least 6 months prior to January 1, 2015 and at least one eye exam, visual field (VF), optical coherence tomography (OCT), or eye-related inpatient or emergency department (ED) code in 2015. Multivariable logistic and negative binomial regression models were used to assess vision and utilization outcomes, respectively, across race and ethnicity from January 1, 2015 to January 1, 2020. Vision outcomes included cup-to-disc ratio (CDR) progression > 0.80, poor vision (visual acuity 20/200 or worse), low vision codes, and need for glaucoma filtering surgery. Utilization outcomes included outpatient eye exams, OCTs, VFs, inpatient/ED encounters, and lasers/surgeries., Results: Among 996,297 patients, 73% were non-Hispanic White, 15% non-Hispanic Black, 9% Hispanic, 3% Asian/Pacific Islander, and 0.3% Native American/Alaska Native. Compared to White eyes, Black and Hispanic eyes had higher adjusted odds of CDR progression (odds ratio [OR] = 1.12, 95% confidence interval [CI] = 1.08-1.17; OR = 1.28, 95% CI = 1.22-1.34), poor vision (OR = 1.26, 95% CI = 1.22-1.29; OR = 1.26, 95% CI = 1.22-1.31), glaucoma filtering surgery (rate ratio (RR) = 1.47, 95% CI = 1.42-1.51; RR = 1.13, 95% CI = 1.09-1.18). Hispanic eyes also had increased odds of low vision diagnoses (Hispanic OR = 1.18, 95% CI = 1.07-1.30). Black and Hispanic patients were less likely to have eye exams (RR = 0.94, 95% CI = 0.94-0.95; RR = 0.99, 95% CI = 0.99-0.99) and OCTs (RR = 0.86, 95% CI = 0.85-0.86; RR = 0.97, 95% CI = 0.96-0.98), yet Black patients had higher odds of inpatient/ED encounters (RR = 1.64, 95% CI = 1.37-1.96) compared to White patients. Native American patients were more likely to have poor vision (OR = 1.17, 95% CI = 1.01-1.36) and less likely to have outpatient visits (RR = 0.89, 95% CI = 0.86-0.91), OCTs (RR = 0.85, 95% CI = 0.82-0.89), visual fields (RR = 0.91, 95% CI = 0.88-0.94) or lasers/surgeries (RR = 0.87, 95% CI = 0.79-0.96) compared to White patients., Conclusions: We found that significant disparities in US eye care exist with Black, Hispanic, and Native American patients having worse vision outcomes and less disease monitoring. Glaucoma may be undertreated in these racial and ethnic minority groups, increasing risk for glaucoma-related vision loss., (Copyright © 2024 Elsevier Inc. All rights reserved.)

Purpose: To elucidate the impact of demographics, including gender, race, ethnicity, and preferred language, on regional visual field (VF) loss and progression in glaucoma., Methods: Multivariable linear mixed regressions were performed to determine the impact of race, ethnicity, and preferred language on regional VF loss with adjustment for age and gender. Regional VF loss was defined by pointwise total deviation values and VF loss patterns quantified by an unsupervised machine learning method termed archetypal analysis. All cross-sectional and longitudinal analyses were performed both without and with adjustment for VF mean deviation, which represented overall VF loss severity. P values were corrected for multiple comparisons., Results: All results mentioned had corrected P values less than 0.05. Asian and Black patients showed worse pointwise VF loss than White patients with superior hemifield more affected. Patients with a preferred language other than English demonstrated worse pointwise VF loss than patients with English as their preferred language. Longitudinal analyses revealed Black patients showed worse VF loss/year compared to White patients. Patients with a preferred language other than English demonstrated worse VF loss/year compared to patients preferring English., Conclusions: Blacks and non-English speakers have more severe VF loss, with superior hemifield being more affected and faster VF worsening., Translational Relevance: This study furthered our understanding of racial, ethnic, and socioeconomic disparities in glaucoma outcomes. Understanding the VF loss burden in different racial, ethnic, and socioeconomic groups may guide more effective glaucoma screening and community outreach efforts. This research could help reduce vision loss and improve quality of life in disproportionately affected populations by guiding public health efforts to promote glaucoma awareness and access to care.

Purpose: This study assesses the case frequencies, underlying causes, and vision outcomes in patients with a diagnosis of corneal opacity in the United States., Design: Retrospective cohort study., Participants: Patients in the IRIS® Registry (Intelligent Research in Sight) who received a diagnosis of corneal opacity between January 1, 2013, and November 30, 2020., Methods: The IRIS Registry contains demographic and clinical data of 79 887 324 patients who sought treatment at eye clinics during the study period. We identified patients with corneal opacity using International Classification of Diseases (ICD), Ninth and Tenth Revisions, codes of 371 (corneal scar) and H17 (corneal opacity), respectively. The analyzed data comprised demographic parameters including age, sex, race, ethnicity, and geographical location. We evaluated clinical data including laterality, cause, disease descriptors, and best-corrected visual acuity (VA) up to 1 year before the onset (± 30 days), at the time of diagnosis, and at 1 year after diagnosis (± 30 days)., Main Outcome Measures: Case frequencies, causes, and vision outcomes in patients with a diagnosis of corneal opacity., Results: We identified 5 220 382 patients who received a diagnosis of corneal opacity and scars using H17 (ICD, Tenth Revision) and 371.0 (ICD, Ninth Revision) codes over 7 years. The case frequency of corneal opacity during the study period was 6535 cases per 100 000 patients (6.5%). The mean age of the patients was 63.36 ± 18.14 years, and most were female (57.6%). In the cohort, 38.39% and 30.00% of patients had bilateral and unilateral corneal opacity, respectively. Most of the patients affected by corneal opacity were White (69.13%), followed by Black or African American (6.84%). Corneal dystrophies (64.66%) were the most common cause of corneal opacity in the study cohort. Visual acuity of the patients worsened significantly because of corneal opacity (0.46 ± 0.74 logarithm of the minimum angle of resolution [logMAR]) and did not improve to the baseline (0.37 ± 0.68 logMAR) after management (0.43 ± 0.77 logMAR). The multiple linear regression analysis showed worse vision outcomes in female patients (compared with male patients), and Asian, Black or African American, and American Indian or Alaska Native (compared with White) patients. Additionally, worse vision outcomes were observed in patients with opacity associated with corneal malformation, degenerative disorders, edema, injury, and ulcer compared with those with hereditary corneal dystrophy., Conclusions: Our study showed that corneal opacity was diagnosed in 6.5% of patients in the IRIS Registry and primarily associated with corneal dystrophies. The final vision outcomes in patients with corneal opacity were significantly worse compared with baseline., Financial Disclosure(s): Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article., (Copyright © 2024 American Academy of Ophthalmology. Published by Elsevier Inc. All rights reserved.)

Purpose: To compare the incidence of strabismus after upper and lower blepharoplasty in the United States., Methods: Retrospective cohort study of adults (age ≥18 years) in the IRIS Registry (Intelligent Research in Sight) who underwent blepharoplasty between January 1, 2013 and December 31, 2020. The primary outcome was the Kaplan-Meier estimated cumulative incidence of strabismus diagnosis and surgery within 3 years of blepharoplasty. Multivariable Cox regression was used to determine the association of blepharoplasty type with strabismus diagnosis and surgery, adjusting for patient age, sex, and geographic region., Results: Blepharoplasty was performed in 368,623 patients (median [interquartile range] age, 69 [63-75] years, and 69% female). Compared with those undergoing upper eyelid blepharoplasty, patients treated with lower eyelid blepharoplasty were slightly younger (median age, 66 vs. 69 years; p < 0.001) and more likely to be female (71% vs. 69%; p < 0.001). There was a greater 3-year incidence of strabismus diagnosis (2.0% vs. 1.5%; p < 0.001) and a greater 3-year incidence of strabismus surgery (0.15% vs. 0.06%; p = 0.003) for individuals undergoing lower vs. upper blepharoplasty. After adjusting for age, sex, and geographic region, lower blepharoplasty was associated with a higher 3-year risk of strabismus diagnosis (HR, 1.49; 95% CI, 1.23-1.81; p < 0.001) and surgery (HR, 2.53; 95% CI, 1.27-5.03; p = 0.008)., Conclusions: This registry-based analysis found that individuals undergoing lower eyelid blepharoplasty were at higher risk of strabismus compared with those undergoing upper eyelid blepharoplasty. Using large databases to understand the incidence of complications of frequently performed procedures may improve ophthalmologists' ability to provide data-driven counseling on surgical risks prior to intervention., (Copyright © 2024 The American Society of Ophthalmic Plastic and Reconstructive Surgery, Inc.)

Fairness (also known as equity interchangeably) in machine learning is important for societal well-being, but limited public datasets hinder its progress. Currently, no dedicated public medical datasets with imaging data for fairness learning are available, though underrepresented groups suffer from more health issues. To address this gap, we introduce Harvard Glaucoma Fairness (Harvard-GF), a retinal nerve disease dataset including 3,300 subjects with both 2D and 3D imaging data and balanced racial groups for glaucoma detection. Glaucoma is the leading cause of irreversible blindness globally with Blacks having doubled glaucoma prevalence than other races. We also propose a fair identity normalization (FIN) approach to equalize the feature importance between different identity groups. Our FIN approach is compared with various state-of-the-art fairness learning methods with superior performance in the racial, gender, and ethnicity fairness tasks with 2D and 3D imaging data, demonstrating the utilities of our dataset Harvard-GF for fairness learning. To facilitate fairness comparisons between different models, we propose an equity-scaled performance measure, which can be flexibly used to compare all kinds of performance metrics in the context of fairness. The dataset and code are publicly accessible via https://ophai.hms.harvard.edu/datasets/harvard-gf3300/.

Objective: Angle-based minimally invasive glaucoma surgery (ab-MIGS) has grown substantially, although long-term efficacy is poorly understood. We analyze ab-MIGS effectiveness with and without preceding laser trabeculoplasty (LTP)., Design: Retrospective cohort study., Subjects: Eyes undergoing ab-MIGS (Canaloplasty, Goniotomy, Trabectome, and iStent) with and without prior laser trabeculoplasty (< 2 years preceding MIGS) were identified in the IRIS® Registry (Intelligent Research in Sight) 2013 to 2018., Methods: Propensity score matching (PSM) was undertaken to define the following 4 cohorts: (1) standalone ab-MIGS, no prior LTP vs. (2) standalone ab-MIGS, with prior LTP; and (3) ab-MIGS + phacoemulsification, no prior LTP vs. (4) ab-MIGS + phacoemulsification, with prior LTP., Main Outcome Measures: Failure was defined as subsequent glaucoma reoperation after ab-MIGS (either MIGS or traditional glaucoma surgery). Time-to-event outcome and incidence rates were calculated using survival analysis, and adjusted hazard ratios (aHRs) were generated using multivariate Cox proportional hazards models. Medication data were not available for analysis., Results: A total of 164 965 unique MIGS procedures were performed, from 2013 to 2018. After PSM, we identified 954 eyes undergoing standalone ab-MIGS and 7522 undergoing ab-MIGS + phacoemulsification. For eyes undergoing standalone ab-MIGS, those with prior LTP (n = 477) were more likely to undergo reoperation vs. those without LTP (n = 477) at 6 and 12 months. In multivariable models, those with prior LTP were more likely to undergo reoperation over the 36-month period vs. those without prior LTP (aHR, 1.53; CI, 1.15-2.04; P = 0.004). For eyes undergoing ab-MIGS + phacoemulsification, those with prior LTP (n = 3761) were more likely to undergo reoperation vs. those without LTP (n = 3761) at 12, 24, and 36 months. In multivariable models, those with prior LTP were more likely to undergo reoperation over the 36-month period vs. those without prior LTP (aHR, 1.53 CI, 1.15-2.04; P = 0.004)., Conclusions: Prior LTP may be associated with a higher chance of subsequent glaucoma surgery following ab-MIGS, either with or without concurrent phacoemulsification. These findings have important implications for understanding who may benefit most from ab-MIGS, and for guiding patient and surgeon treatment expectations., Financial Disclosures: Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article., (Copyright © 2024 American Academy of Ophthalmology. Published by Elsevier Inc. All rights reserved.)

Background: Machine learning (ML) can differentiate papilloedema from normal optic discs using fundus photos. Currently, papilloedema severity is assessed using the descriptive, ordinal Frisén scale. We hypothesise that ML can quantify papilloedema and detect a treatment effect on papilloedema due to idiopathic intracranial hypertension., Methods: We trained a convolutional neural network to assign a Frisén grade to fundus photos taken from the Idiopathic Intracranial Hypertension Treatment Trial (IIHTT). We applied modified subject-based fivefold cross-validation to grade 2979 longitudinal images from 158 participants' study eyes (ie, the eye with the worst mean deviation) in the IIHTT. Compared with the human expert-determined grades, we hypothesise that ML-estimated grades can also demonstrate differential changes over time in the IIHTT study eyes between the treatment (acetazolamide (ACZ) plus diet) and placebo (diet only) groups., Findings: The average ML-determined grade correlated strongly with the reference standard (r=0.76, p<0.001; mean absolute error=0.54). At the presentation, treatment groups had similar expert-determined and ML-determined Frisén grades. The average ML-determined grade for the ACZ group (1.7, 95% CI 1.5 to 1.8) was significantly lower (p=0.0003) than for the placebo group (2.3, 95% CI 2.0 to 2.5) at the 6-month trial outcome., Interpretation: Supervised ML of fundus photos quantified the degree of papilloedema and changes over time reflecting the effects of ACZ. Given the increasing availability of fundus photography, neurologists will be able to use ML to quantify papilloedema on a continuous scale that incorporates the features of the Frisén grade to monitor interventions., Competing Interests: Competing interests: None declared., (© Author(s) (or their employer(s)) 2024. Re-use permitted under CC BY-NC. No commercial re-use. See rights and permissions. Published by BMJ.)

Optical coherence tomography imaging provides a crucial clinical measurement for diagnosing and monitoring glaucoma through the two-dimensional retinal nerve fiber layer (RNFL) thickness (RNFLT) map. Researchers have been increasingly using neural models to extract meaningful features from the RNFLT map, aiming to identify biomarkers for glaucoma and its progression. However, accurately representing the RNFLT map features relevant to glaucoma is challenging due to significant variations in retinal anatomy among individuals, which confound the pathological thinning of the RNFL. Moreover, the presence of artifacts in the RNFLT map, caused by segmentation errors in the context of degraded image quality and defective imaging procedures, further complicates the task. In this paper, we propose a general framework called RNFLT2Vec for unsupervised learning of vectorized feature representations from RNFLT maps. Our method includes an artifact correction component that learns to rectify RNFLT values at artifact locations, producing a representation reflecting the RNFLT map without artifacts. Additionally, we incorporate two regularization techniques to encourage discriminative representation learning. Firstly, we introduce a contrastive learning-based regularization to capture the similarities and dissimilarities between RNFLT maps. Secondly, we employ a consistency learning-based regularization to align pairwise distances of RNFLT maps with their corresponding thickness distributions. Through extensive experiments on a large-scale real-world dataset, we demonstrate the superiority of RNFLT2Vec in three different clinical tasks: RNFLT pattern discovery, glaucoma detection, and visual field prediction. Our results validate the effectiveness of our framework and its potential to contribute to a better understanding and diagnosis of glaucoma., Competing Interests: Declaration of competing interest The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Mengyu Wang reports financial support was provided by National Institute of Health. Mengyu Wang reports a relationship with Schepens Eye Research Institute of Massachusetts Eye and Ear that includes: employment., (Copyright © 2024. Published by Elsevier B.V.)

Aims/hypothesis: As the prevalence of insulin resistance and glucose intolerance is increasing throughout the world, diabetes-induced eye diseases are a global health burden. We aim to identify distinct optical bands which are closely related to insulin and glucose metabolism, using non-invasive, high-resolution spectral domain optical coherence tomography (SD-OCT) in a large, population-based dataset., Methods: The LIFE-Adult-Study randomly selected 10,000 participants from the population registry of Leipzig, Germany. Cross-sectional, standardised phenotyping included the assessment of various metabolic risk markers and ocular imaging, such as SD-OCT-derived thicknesses of ten optical bands of the retina. Global and Early Treatment Diabetic Retinopathy Study (ETDRS) subfield-specific optical retinal layer thicknesses were investigated in 7384 healthy eyes of 7384 participants from the LIFE-Adult-Study stratified by normal glucose tolerance, prediabetes (impaired fasting glucose and/or impaired glucose tolerance and/or HbA 1c 5.7-6.4% [39-47 mmol/mol]) and diabetes. The association of optical retinal band characteristics with different indices of glucose tolerance (e.g. fasting glucose, area under the glucose curve), insulin resistance (e.g. HOMA2-IR, triglyceride glucose index), or insulin sensitivity (e.g. estimated glucose disposal rate [eGDR], Stumvoll metabolic clearance rate) was determined using multivariable linear regression analyses for the individual markers adjusted for age, sex and refraction. Various sensitivity analyses were performed to validate the observed findings., Results: In the study cohort, nine out of ten optical bands of the retina showed significant sex- and glucose tolerance-dependent differences in band thicknesses. Multivariable linear regression analyses revealed a significant, independent, and inverse association between markers of glucose intolerance and insulin resistance (e.g. HOMA2-IR) with the thickness of the optical bands representing the anatomical retinal outer nuclear layer (ONL, standardised β=-0.096; p<0.001 for HOMA2-IR) and myoid zone (MZ; β=-0.096; p<0.001 for HOMA2-IR) of the photoreceptors. Conversely, markers of insulin sensitivity (e.g. eGDR) positively and independently associated with ONL (β=0.090; p<0.001 for eGDR) and MZ (β=0.133; p<0.001 for eGDR) band thicknesses. These global associations were confirmed in ETDRS subfield-specific analyses. Sensitivity analyses further validated our findings when physical activity, neuroanatomical cell/tissue types and ETDRS subfield categories were investigated after stratifying the cohort by glucose homeostasis., Conclusions/interpretation: An impaired glucose homeostasis associates with a thinning of the optical bands of retinal ONL and photoreceptor MZ. Changes in ONL and MZ thicknesses might predict early metabolic retinal alterations in diabetes., (© 2024. The Author(s).)

Purpose: To describe the prevalence of missing sociodemographic data in the IRIS® (Intelligent Research in Sight) Registry and to identify practice-level characteristics associated with missing sociodemographic data., Design: Cross-sectional study., Participants: All patients with clinical encounters at practices participating in the IRIS Registry prior to December 31, 2020., Methods: We describe geographic and temporal trends in the prevalence of missing data for each sociodemographic variable (age, sex, race, ethnicity, geographic location, insurance type, and smoking status). Each practice contributing data to the registry was categorized based on the number of patients, number of physicians, geographic location, patient visit frequency, and patient population demographics., Main Outcome Measures: Multivariable linear regression was used to describe the association of practice-level characteristics with missing patient-level sociodemographic data., Results: This study included the electronic health records of 66 477 365 patients receiving care at 3306 practices participating in the IRIS Registry. The median number of patients per practice was 11 415 (interquartile range: 5849-24 148) and the median number of physicians per practice was 3 (interquartile range: 1-7). The prevalence of missing patient sociodemographic data were 0.1% for birth year, 0.4% for sex, 24.8% for race, 30.2% for ethnicity, 2.3% for 3-digit zip code, 14.8% for state, 5.5% for smoking status, and 17.0% for insurance type. The prevalence of missing data increased over time and varied at the state-level. Missing race data were associated with practices that had fewer visits per patient ( P  < 0.001), cared for a larger nonprivately insured patient population ( P  = 0.001), and were located in urban areas ( P  < 0.001). Frequent patient visits were associated with a lower prevalence of missing race ( P  < 0.001), ethnicity ( P  < 0.001), and insurance ( P  < 0.001), but a higher prevalence of missing smoking status ( P  < 0.001)., Conclusions: There are geographic and temporal trends in missing race, ethnicity, and insurance type data in the IRIS Registry. Several practice-level characteristics, including practice size, geographic location, and patient population, are associated with missing sociodemographic data. While the prevalence and patterns of missing data may change in future versions of the IRIS registry, there will remain a need to develop standardized approaches for minimizing potential sources of bias and ensure reproducibility across research studies., Financial Disclosures: Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article., (© 2024 by the American Academy of Ophthalmology.)

Purpose: To identify sociodemographic factors associated with the visual outcomes of retinoblastoma survivors., Methods: Retrospective cohort study using a US-based clinical data registry. All individuals < 18 years of age with a history of retinoblastoma in the Intelligent Research in Sight (IRIS®) Registry (1/1/2013-12/31/2020). The primary outcome was visual acuity below the threshold for legal blindness (20/200 or worse) in at least one eye. Multivariable logistic regression was used to evaluate the association between visual outcomes and age, sex, laterality, race, ethnicity, type of insurance, and geographic location., Results: This analysis included 1545 children with a history of retinoblastoma. The median length of follow-up was 4.1 years (IQR, 2.2-5.9 years) and the median age at most recent clinical visit was 12 years (IQR, 8-16 years). Retinoblastoma was unilateral in 54% of cases. Poor vision in at least one eye was identified in 78% of all children and poor vision in both eyes in 17% of those with bilateral disease. Poor visual outcomes were associated with unilateral diagnosis (OR, 1.55; 95% CI,1.13-2.12; p  = .007), Black race (OR, 2.03; 95% CI, 1.19-3.47; p  = .010), Hispanic ethnicity (OR, 1.65; 95% CI, 1.16-2.37; p  = .006), and non-private insurance (OR, 1.47; 95% CI, 1.02-2.10; p  = .037)., Conclusions: Poor visual outcomes appear to be more common among Black, Hispanic, and publicly insured children with a history of retinoblastoma, raising concerns regarding healthcare inequities. Primary care physicians should ensure that young children receive red reflex testing during routine visits and consider retinoblastoma in the differential diagnosis of abnormal eye exams.

Background and Objective: This study aimed to examine conversion rates from non-exudative to exudative age-related macular degeneration (AMD) in the fellow eye of patients with unilateral exudative AMD using the Academy IRIS ® Registry., Patients and Methods: This study was a retrospective, cohort analysis from 2016 to 2019. Patient and disease characteristics including initial AMD stage were collected. Cox proportional-hazard (PH) and logistic regression modeling were performed., Results: The risk of conversion was lower for men relative to women and for Asians and Blacks relative to Whites. Compared to never-smokers, active smokers were at increased risk of conversion, and compared to initially early non-exudative AMD eyes, intermediate and advanced non-exudative AMD eyes had higher rates of conversion. Compared to active choroidal neovascularization eyes, eyes with inactive choroidal neovascularization and inactive scars had lower rates of fellow eye conversion., Conclusions: In this cohort analysis of unilateral exudative AMD patients, women, Whites, and active smokers had higher rates of non-exudative to exudative AMD conversion in the fellow eye. [ Ophthalmic Surg Lasers Imaging Retina 2024;55:220-226.] .

Purpose: Dry-eye disease (DED) is a chronic progressive ocular surface disorder with limited studies in the pediatric population. The Academy of Ophthalmology's IRIS® Registry was leveraged to investigate the prevalence of DED in the pediatric population (PDED, patients <18 years old) and the demographic differences of DED between pediatric and adult patients (ADED)., Methods: Retrospective cohort study. Patients with DED between January 1st, 2013 and December 31st, 2019 (N = 4,795,979) were included. Descriptive statistics, Pearson's chi-squared tests and two-sample proportions tests were conducted to compare key demographic distributions between the ADED and PDED cohorts., Results: The average age at onset for ADED patients was 61.06 (±14.75) years and for PDED patients was 12.51 (±3.86). The overall tests for independence and the individual tests of proportions of each category were statistically significant for all demographic characteristics (p < 0.001). Characteristics with the largest discrepancies between patients of PDED and the IRIS Registry pediatric patient pool (PIRIS) included female sex (58.08 % vs. 50.60 %), male sex (41.58 % vs. 48.78 %) and Asian race (6.02 % vs. 3.11 %) respectively. Within the PDED cohort, females were at higher risk of PDED (58 % vs. 42 %). PDED was more prevalent in children with refractive errors (76 %) and eyelid/conjunctival disorders (41 %). Characteristics with the largest discrepancies between PDED and ADED patients included female sex (58.08 % vs. 68.12 %), male sex (41.58 % vs. 31.55 %) and Caucasian race (50.24 % vs. 67.06 %) respectively., Conclusions: Significant differences in the PDED cohort are demonstrated in this study. PDED was more prevalent in the female sex and Caucasian race compared to PIRIS and was more commonly associated with refractive errors and eyelid/conjunctival disorders., (Copyright © 2024 Elsevier Inc. All rights reserved.)

Prcis: Using a large data set, we showed structural and functional differences between primary angle closure glaucoma (PACG) and primary open angle glaucoma (POAG). Primary angle closure glaucoma has relative structural preservation and worse functional loss inferiorly., Purpose: To identify structural and functional differences in PACG and POAG., Materials and Methods: In this large cross-sectional study, differences in structural and functional damage were assessed among patients with POAG and PACG with optical coherence tomography and reliable visual field testing., Results: In all, 283 patients with PACG and 4110 patients with POAG were included. Despite similar mean deviation on visual fields (mean [SD] -7.73 [7.92] vs. -7.53 [6.90] dB, P =0.72), patients with PACG had thicker global retinal nerve fiber layer (RNFL), smaller cup volume, smaller cup-to-disc ratio, and larger rim area than POAG (77 [20] vs. 71 [14] µm, 0.32 [0.28] vs. 0.40 [0.29] mm 3 , 0.6 [0.2] vs. 0.7 [0.1], 1.07 [0.40] vs. 0.89 [0.30] mm 2 , P <0.001 for all), while patients with POAG had more pronounced inferior RNFL thinning (82 [24] vs. 95 [35] µm, P <0.001). In a multivariable analysis, hyperopia [odds ratio (OR): 1.24, confidence interval (CI): 1.13-1.37], smaller cup-to-disc ratio (OR: 0.69, CI: 0.61-0.78), thicker inferior RNFL (OR: 1.15, CI: 1.06-1.26) and worse mean deviation (OR: 0.95, CI: 0.92-0.98) were associated with PACG. Functionally, POAG was associated with superior paracentral loss and PACG with inferior field loss. After adjusting for average RNFL thickness, PACG was associated with more diffuse loss than POAG (total deviation differences 1.26-3.2 dB)., Conclusions: Patients with PACG had less structural damage than patients with POAG despite similar degrees of functional loss. Regional differences in patterns of functional and structural loss between POAG and PACG may improve disease monitoring for these glaucoma subtypes., Competing Interests: Disclosure: L.R.P.: Consultant for Twenty-Twenty and Character Bio. L.Q.S.: Consultant for FireCyte Therapeutics and AbbVie. T.E.: Financial support from Genentech. M.W.: Financial support from Genentech. D.S.F.: Consultant for Thea Pharmaceutical and Life Biosciences. M.V.B.: Consultant for Carl Zeiss Meditec, Topcon Healthcare, Janssen, Allergan. The remaining authors declare no conflicts of interest., (Copyright © 2023 Wolters Kluwer Health, Inc. All rights reserved.)

Importance: Primary open-angle glaucoma (POAG) is a highly heritable disease, with 127 identified risk loci to date. Polygenic risk score (PRS) may provide a clinically useful measure of aggregate genetic burden and improve patient risk stratification., Objective: To assess whether a PRS improves prediction of POAG onset in patients with ocular hypertension., Design, Setting, and Participants: This was a post hoc analysis of the Ocular Hypertension Treatment Study. Data were collected from 22 US sites with a mean (SD) follow-up of 14.0 (6.9) years. A total of 1636 participants were followed up from February 1994 to December 2008; 1077 participants were enrolled in an ancillary genetics study, of which 1009 met criteria for this analysis. PRS was calculated using summary statistics from the largest cross-ancestry POAG meta-analysis, with weights trained using 8 813 496 variants from 449 186 cross-ancestry participants in the UK Biobank. Data were analyzed from July 2022 to December 2023., Exposures: From February 1994 to June 2002, participants were randomized to either topical intraocular pressure-lowering medication or close observation. After June 2002, both groups received medication., Main Outcomes and Measures: Outcome measures were hazard ratios for POAG onset. Concordance index and time-dependent areas under the receiver operating characteristic curve were used to compare the predictive performance of multivariable Cox proportional hazards models., Results: Of 1009 included participants, 562 (55.7%) were female, and the mean (SD) age was 55.9 (9.3) years. The mean (SD) PRS was significantly higher for 350 POAG converters (0.24 [0.95]) compared with 659 nonconverters (-0.12 [1.00]) (P < .001). POAG risk increased 1.36% (95% CI, 1.08-1.64) with each higher PRS decile, with conversion ranging from 9.52% (95% CI, 7.09-11.95) in the lowest PRS decile to 21.81% (95% CI, 19.37-24.25) in the highest decile. Comparison of low-risk and high-risk PRS tertiles showed a 2.0-fold increase in 20-year POAG risk for participants of European and African ancestries. In the subgroup randomized to delayed treatment, each increase in PRS decile was associated with a 0.52-year (95% CI, 0.01-1.03) decrease in age at diagnosis (P = .047). No significant linear association between PRS and age at POAG diagnosis was present in the early treatment group. Prediction models significantly improved with the addition of PRS as a covariate (C index = 0.77) compared with the Ocular Hypertension Treatment Study baseline model (C index = 0.75) (P < .001). Each 1-SD higher PRS conferred a mean hazard ratio of 1.25 (95% CI, 1.13-1.44) for POAG onset., Conclusions and Relevance: Higher PRS was associated with increased risk for POAG in patients with ocular hypertension. The inclusion of a PRS improved the prediction of POAG onset., Trial Registration: ClinicalTrials.gov Identifier: NCT00000125.

Purpose: To evaluate the potential of retinal optical coherence tomography (OCT) measurements and polygenic risk scores (PRS) to identify people at risk of cognitive impairment., Methods: Using OCT images from 50 342 UK Biobank participants, we examined associations between retinal layer thickness and genetic risk for neurodegenerative disease and combined these metrics with PRS to predict baseline cognitive function and future cognitive deterioration. Multivariate Cox proportional hazard models were used to predict cognitive performance. P values for retinal thickness analyses are false-discovery-rate-adjusted., Results: Higher Alzheimer's disease PRS was associated with a thicker inner nuclear layer (INL), chorio-scleral interface (CSI) and inner plexiform layer (IPL) (all p<0.05). Higher Parkinson's disease PRS was associated with thinner outer plexiform layer (p<0.001). Worse baseline cognitive performance was associated with thinner retinal nerve fibre layer (RNFL) (aOR=1.038, 95% CI (1.029 to 1.047), p<0.001) and photoreceptor (PR) segment (aOR=1.035, 95% CI (1.019 to 1.051), p<0.001), ganglion cell complex (aOR=1.007, 95% CI (1.002 to 1.013), p=0.004) and thicker ganglion cell layer (aOR=0.981, 95% CI (0.967 to 0.995), p=0.009), IPL (aOR=0.976, 95% CI (0.961 to 0.992), p=0.003), INL (aOR=0.923, 95% CI (0.905 to 0.941), p<0.001) and CSI (aOR=0.998, 95% CI (0.997 to 0.999), p<0.001). Worse future cognitive performance was associated with thicker IPL (aOR=0.945, 95% CI (0.915 to 0.999), p=0.045) and CSI (aOR=0.996, 95% CI (0.993 to 0.999) 95% CI, p=0.014). Prediction of cognitive decline was significantly improved with the addition of PRS and retinal measurements., Conclusions and Relevance: Retinal OCT measurements are significantly associated with genetic risk of neurodegenerative disease and may serve as biomarkers predictive of future cognitive impairment., Competing Interests: Competing interests: None declared., (© Author(s) (or their employer(s)) 2024. No commercial re-use. See rights and permissions. Published by BMJ.)

Purpose: To compare demographic and clinical factors associated with glaucoma following cataract surgery (GFCS) and glaucoma surgery rates between infants, toddlers, and older children using a large, ophthalmic registry., Design: Retrospective cohort study., Participants: Patients in the IRIS® Registry (Intelligent Research in Sight) who underwent cataract surgery at ≤ 17 years old and between January 1, 2013 and December 31, 2020., Methods: Glaucoma diagnosis and procedural codes were extracted from the electronic health records of practices participating in the IRIS Registry. Children with glaucoma diagnosis or surgery before cataract removal were excluded. The Kaplan-Meier estimator was used to determine the cumulative probability of GFCS diagnosis and glaucoma surgery after cataract surgery. Multivariable Cox regression was used to identify factors associated with GFCS and glaucoma surgery., Main Outcome Measures: Cumulative probability of glaucoma diagnosis and surgical intervention within 5 years after cataract surgery., Results: The study included 6658 children (median age, 10.0 years; 46.2% female). The 5-year cumulative probability of GFCS was 7.1% (95% confidence interval [CI], 6.1%-8.1%) and glaucoma surgery was 2.6% (95% CI, 1.9%-3.2%). The 5-year cumulative probability of GFCS for children aged < 1 year was 22.3% (95% CI, 15.7%-28.4%). Risk factors for GFCS included aphakia (hazard ratio [HR], 2.63; 95% CI, 1.96-3.57), unilateral cataract (HR, 1.48; 95% CI, 1.12-1.96), and Black race (HR, 1.61; 95% CI, 1.12-2.32). The most common surgery was glaucoma drainage device insertion (32.6%), followed by angle surgery (23.3%), cyclophotocoagulation (15.1%), and trabeculectomy (5.8%)., Conclusions: Glaucoma following cataract surgery diagnosis in children in the IRIS Registry was associated with young age, aphakia, unilateral cataract, and Black race. Glaucoma drainage device surgery was the preferred surgical treatment, consistent with the World Glaucoma Association 2013 consensus recommendations for GFCS management., Financial Disclosure(s): Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article., (Copyright © 2023 American Academy of Ophthalmology. Published by Elsevier Inc. All rights reserved.)

Purpose: The Ocular Hypertension Treatment Study (OHTS) identified risk factors for primary open-angle glaucoma (POAG) in patients with ocular hypertension, including pattern standard deviation (PSD). Archetypal analysis, an unsupervised machine learning method, may offer a more interpretable approach to risk stratification by identifying patterns in baseline visual fields (VFs)., Methods: There were 3272 eyes available in the OHTS. Archetypal analysis was applied using 24-2 baseline VFs, and model selection was performed with cross-validation. Decomposition coefficients for archetypes (ATs) were calculated. A penalized Cox proportional hazards model was implemented to select discriminative ATs. The AT model was compared to the OHTS model. Associations were identified between ATs with both POAG onset and VF progression, defined by mean deviation change per year., Results: We selected 8494 baseline VFs. Optimal AT count was 19. The highest prevalence ATs were AT9, AT11, and AT7. The AT-based prediction model had a C-index of 0.75 for POAG onset. Multivariable models demonstrated that a one-interquartile range increase in the AT5 (hazard ratio [HR] = 1.14; 95% confidence interval [CI], 1.04-1.25), AT8 (HR = 1.22; 95% CI, 1.09-1.37), AT15 (HR = 1.26; 95% CI, 1.12-1.41), and AT17 (HR = 1.17; 95% CI, 1.03-1.31) coefficients conferred increased risk of POAG onset. AT5, AT10, and AT14 were significantly associated with rapid VF progression. In a subgroup analysis by high-risk ATs (>95th percentile or <75th percentile coefficients), PSD lost significance as a predictor of POAG in the low-risk group., Conclusions: Baseline VFs, prior to detectable glaucomatous damage, contain occult patterns representing early changes that may increase the risk of POAG onset and VF progression in patients with ocular hypertension. The relationship between PSD and POAG is modified by the presence of high-risk patterns at baseline. An AT-based prediction model for POAG may provide more interpretable glaucoma-specific information in a clinical setting.

Purpose: To investigate the demographic and clinical characteristics of patients with sympathetic ophthalmia (SO) and define the risk factors for its incidence following trauma and ophthalmic procedures., Design: Retrospective cohort study., Participants: Patients in the American Academy of Ophthalmology's (Academy) IRIS Registry (Intelligent Research in Sight) who were (n=1523) or were not diagnosed with SO following a documented procedure or trauma between January 1, 2013, and December 31, 2019., Methods: Multiple demographic and clinical factors were collected, descriptive statistics and prevalence were calculated, and multivariate linear regression models were fit to the data., Main Outcome Measures: Prevalence of SO, demographic and clinical characteristics, and beta coefficient (β) estimates of demographic and clinical characteristics impacting time to SO onset after procedure (Procedure Only cohort) or trauma (Trauma cohort)., Results: Of 65,348,409 distinct IRIS Registry patients, 1523 (0.0023%) were diagnosed with SO between 2013 and 2019, and also had a documented preceding trauma or procedure. Of these, 927 (60.87%) were female, 1336 (87.72%) belonged to the Procedure Only cohort, and 187 (12.28%) belonged to the Trauma cohort. The prevalence of SO after trauma was 0.0207%, whereas after procedures it was 0.0124%. The highest risk of procedure-related SO was seen in patients with history of "other anterior segment" (0.122%) followed by glaucoma (0.066%) procedures, whereas the lowest prevalence was noted with cataract surgeries (0.011%). The average time to onset of SO across both cohorts combined was 527.44 (±715.60) days, with statistically significant differences between the 2 cohorts (P < .001). On average, the time to onset from inciting event to SO was shorter with increasing age, by 9.02 (95% CI: -11.96, -6.08) days for every 1-year increase., Conclusions: SO following trauma and ophthalmic procedure is potentially rarer than previously reported, as measured in this large ophthalmic medical record database. Female sex may be a risk factor for SO. Older age may be a risk factor for quicker onset. These findings can guide clinical decision-making and management., (Copyright © 2023 Elsevier Inc. All rights reserved.)

The human retina is a multilayered tissue that offers a unique window into systemic health. Optical coherence tomography (OCT) is widely used in eye care and allows the noninvasive, rapid capture of retinal anatomy in exquisite detail. We conducted genotypic and phenotypic analyses of retinal layer thicknesses using macular OCT images from 44,823 UK Biobank participants. We performed OCT layer cross-phenotype association analyses (OCT-XWAS), associating retinal thicknesses with 1866 incident conditions (median 10-year follow-up) and 88 quantitative traits and blood biomarkers. We performed genome-wide association studies (GWASs), identifying inherited genetic markers that influence retinal layer thicknesses and replicated our associations among the LIFE-Adult Study ( N = 6313). Last, we performed a comparative analysis of phenome- and genome-wide associations to identify putative causal links between retinal layer thicknesses and both ocular and systemic conditions. Independent associations with incident mortality were detected for thinner photoreceptor segments (PSs) and, separately, ganglion cell complex layers. Phenotypic associations were detected between thinner retinal layers and ocular, neuropsychiatric, cardiometabolic, and pulmonary conditions. A GWAS of retinal layer thicknesses yielded 259 unique loci. Consistency between epidemiologic and genetic associations suggested links between a thinner retinal nerve fiber layer with glaucoma, thinner PS with age-related macular degeneration, and poor cardiometabolic and pulmonary function with a thinner PS. In conclusion, we identified multiple inherited genetic loci and acquired systemic cardio-metabolic-pulmonary conditions associated with thinner retinal layers and identify retinal layers wherein thinning is predictive of future ocular and systemic conditions.

Importance: There is no consensus on the optimal surgical treatment for children with intermittent exotropia (IXT)., Objective: To compare the 5-year reoperation rates for children with IXT treated with horizontal muscle strabismus surgery using bilateral lateral rectus recession (BLR) vs unilateral lateral rectus recession with medial rectus resection (RR)., Design, Setting, and Participants: This cohort study examined data obtained from the Intelligent Research in Sight (IRIS) Registry on 7482 children (age, <18 years) with IXT who underwent horizontal eye muscle strabismus surgery between January 1, 2013, and December 31, 2017. Children undergoing initial surgeries involving 3 or more horizontal muscles, vertical muscles, or reoperations were excluded., Main Outcomes and Measures: The primary outcome was the adjusted cumulative incidence of repeat horizontal muscle surgery within 5 years after the initial surgery. Reoperation risk was analyzed using adjusted hazard ratios (AHRs) derived from multivariable Cox regression models, adjusting for individual demographic and surgical factors (age, sex, race and ethnicity, US Census region, and surgeon subspecialty). Data were analyzed between January 16 and September 20, 2023., Results: The study included 7482 children (median [IQR] age at initial surgery, 6 [4-9] years; 3945 females [53%]) with IXT treated with horizontal muscle strabismus surgery. Bilateral lateral rectus recession was performed more frequently than RR (85.3% vs 14.7%, P < .001), especially in younger children (rates of BLR vs RR by age: age 0 to ≤4 years, 88.4% vs 11.6%; age 5 to ≤11 years, 84.7% vs 15.3%; age 12 to ≤17 years, 78.1% vs 21.9%; P < 0.001). After data adjustment, the 5-year cumulative incidence of reoperation was 21.3% (95% CI, 20.1%-22.5%). The adjusted 5-year cumulative incidence of reoperation was higher for BLR than for RR (22.2% vs 17.2%; difference, 4.9%; 95% CI, 1.9%-8.0%). Unilateral lateral rectus recession with medial rectus resection was associated with a lower 5-year reoperation risk compared with BLR (AHR, 0.77; 95% CI, 0.64-0.93). Younger age at time of initial surgery was associated with a higher reoperation risk (AHR per 1-year decrease, 1.09; 95% CI, 1.07-1.11) after adjusting for all other covariates., Conclusions and Relevance: In this nationwide registry, approximately 1 in 5 children with IXT underwent reoperation within 5 years after the initial surgery. Children treated with RR were less likely to require a reoperation within 5 years compared with those treated with BLR. Further efforts to identify modifiable risk factors for reoperation are needed to reduce the surgical burden and improve outcomes for children with IXT.

Prcis: Participant surveys taken after using tablet-based and smart visual function analyzer (SVFA) perimetry tests suggest that patients may prefer novel perimetry tests over traditional visual field machines., Purpose: Compare patient experience using the IMOvifa SVFA and the tablet-based Melbourne Rapid Fields visual field (VF) tests to the Humphrey Field 24-2 Swedish Interactive Threshold Algorithm Standard., Patients and Methods: Prospective observational cohort study on adult participants with diagnoses of glaucoma suspect, ocular hypertension, or glaucoma. Participants attended 2 study visits ~3 months apart. During the first visit, participants were trained to use the 2 novel perimeters, took 1 test on both new devices and the Humphrey Field Analyzer, then were surveyed. Participants received tablets and performed weekly tablet VF tests at home between study visits. At the final study visit, participants re-took the VF tests and completed the same surveys., Results: Eighty-one participants were surveyed twice. At the baseline survey, participants preferred the SVFA (71.7%) and tablet tests (69.2%) over the Humphrey VF. Most were willing to perform weekly monitoring at home on the SVFA (69.1%) and tablet (75.4%). Participants generally had a "very good" overall experience when testing on the SVFA (71.6%) and tablet (90.1%). At the final visit, fewer participants were willing to test on the tablet daily (23.5% to 9.9%; P = 0.02 for change) and more were willing to test monthly (18.5% to 33.3%; P = 0.03 for change)., Conclusion: Users reported a preference for novel VF devices. Overall participant experience using these devices was positive, supporting the feasibility of home monitoring of VFs from an experience perspective., Competing Interests: Disclosure: D.S.F. receives research support from Genentech Inc. and has received speaking honoraria and travel expenses from Thea Pharmaceuticals. D.S.C. is employed by Genentech Inc. T.E. receives research support from Genentech Inc. The remaining authors declare no conflict of interest., (Copyright © 2023 Wolters Kluwer Health, Inc. All rights reserved.)

Purpose: To evaluate the relationship between contrast sensitivity (CS) and widefield swept-source optical coherence tomography angiography (WF SS-OCTA) vascular metrics in diabetic macular edema (DME) was the purpose., Methods: This prospectively enrolled cross-sectional observational study included 61 eyes of 48 patients that were tested with the quantitative CS function (qCSF) test on the same day as imaging with WF SS-OCTA (PLEX® Elite 9000, Carl Zeiss Meditec) 3 × 3, 6 × 6, and 12 × 12 mm scans. Outcomes included visual acuity (VA) and multiple qCSF metrics. Vascular metrics included vessel density (VD) and vessel skeletonized density (VSD) in the superficial (SCP) and deep capillary plexus (DCP) and whole retina (WR) and foveal avascular zone (FAZ) parameters. Mixed effects multivariable linear regression models controlling for age, lens status, and diabetic retinopathy stage were performed. Standardized beta coefficients were calculated by refitting the standardized data., Results: SS-OCTA metrics had a significant association with CS and VA. The effect size of OCTA metrics was larger on CS compared to VA. For example, the standardized beta coefficients for VSD and CS at 3 cpd (β SCP  = 0.76, β DCP  = 0.71, β WR  = 0.72, p < 0.001) were larger than those for VA (β SCP  =  - 0.55, p < 0.001; β DCP  =  - 0.43, p = 0.004; β WR  =  - 0.50, p < 0.001). On 6 × 6 mm images, AULCSF, CS at 3 cpd, and CS at 6 cpd were significantly associated with VD and VSD in all three slab types (SCP, DCP, and WR), while VA was not., Conclusion: Structure-function associations in patients with DME leveraging the qCSF device suggest microvascular changes on WF SS-OCTA are associated with larger changes in contrast sensitivity than VA., (© 2023. The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature.)

Purpose: To develop an automated deep learning system for detecting the presence and location of disc hemorrhages in optic disc photographs., Design: Development and testing of a deep learning algorithm., Methods: Optic disc photos (597 images with at least 1 disc hemorrhage and 1075 images without any disc hemorrhage from 1562 eyes) from 5 institutions were classified by expert graders based on the presence or absence of disc hemorrhage. The images were split into training (n = 1340), validation (n = 167), and test (n = 165) datasets. Two state-of-the-art deep learning algorithms based on either object-level detection or image-level classification were trained on the dataset. These models were compared to one another and against 2 independent glaucoma specialists. We evaluated model performance by the area under the receiver operating characteristic curve (AUC). AUCs were compared with the Hanley-McNeil method., Results: The object detection model achieved an AUC of 0.936 (95% CI = 0.857-0.964) across all held-out images (n = 165 photographs), which was significantly superior to the image classification model (AUC = 0.845, 95% CI = 0.740-0.912; P = .006). At an operating point selected for high specificity, the model achieved a specificity of 94.3% and a sensitivity of 70.0%, which was statistically indistinguishable from an expert clinician (P = .7). At an operating point selected for high sensitivity, the model achieves a sensitivity of 96.7% and a specificity of 73.3%., Conclusions: An autonomous object detection model is superior to an image classification model for detecting disc hemorrhages, and performed comparably to 2 clinicians., (Copyright © 2023 The Authors. Published by Elsevier Inc. All rights reserved.)

Purpose: Correcting retinal nerve fiber layer thickness (RNFLT) artifacts in glaucoma with deep learning and evaluate its clinical usefulness., Methods: We included 24,257 patients with optical coherence tomography and reliable visual field (VF) measurements within 30 days and 3,233 patients with reliable VF series of at least five measurements over ≥4 years. The artifacts are defined as RNFLT less than the known floor value of 50 µm. We selected 27,319 high-quality RNFLT maps with an artifact ratio (AR) of <2% as the ground truth. We created pseudo-artifacts from 21,722 low-quality RNFLT maps with AR of >5% and superimposed them on high-quality RNFLT maps to predict the artifact-free ground truth. We evaluated the impact of artifact correction on the structure-function relationship and progression forecasting., Results: The mean absolute error and Pearson correlation of the artifact correction were 9.89 µm and 0.90 (P < 0.001), respectively. Artifact correction improved R2 for VF prediction in RNFLT maps with AR of >10% and AR of >20% up to 0.03 and 0.04 (P < 0.001), respectively. Artifact correction improved (P < 0.05) the AUC for progression prediction in RNFLT maps with AR of ≤10%, >10%, and >20%: (1) total deviation pointwise progression: 0.68 to 0.69, 0.62 to 0.63, and 0.62 to 0.64; and (2) mean deviation fast progression: 0.67 to 0.68, 0.54 to 0.60, and 0.45 to 0.56., Conclusions: Artifact correction for RNFLTs improves VF and progression prediction in glaucoma., Translational Relevance: Our model improves clinical usability of RNFLT maps with artifacts.

Purpose: Circumpapillary retinal nerve fiber layer thickness (RNFLT) measurement aids in the clinical diagnosis of glaucoma. Spectral domain optical coherence tomography (SD-OCT) machines measure RNFLT and provide normative color-coded plots. In this retrospective study, we investigate whether normative percentiles of RNFLT (pRNFLT) from Spectralis SD-OCT improve prediction of glaucomatous visual field loss over raw RNFLT., Methods: A longitudinal database containing OCT scans and visual fields from Massachusetts Eye & Ear glaucoma clinic patients was generated. Reliable OCT-visual field pairs were selected. Spectralis OCT normative distributions were extracted from machine printouts. Supervised machine learning models compared predictive performance between pRNFLT and raw RNFLT inputs. Regional structure-function associations were assessed with univariate regression to predict mean deviation (MD). Multivariable classification predicted MD, pattern standard deviation, MD change per year, and glaucoma hemifield test., Results: There were 3016 OCT-visual field pairs that met the reliability criteria. Spectralis norms were found to be independent of age, sex, and ocular magnification. Regional analysis showed significant decrease in R2 from pRNFLT models compared to raw RNFLT models in inferotemporal sectors, across multiple regressors. In multivariable classification, there were no significant improvements in area under the curve of receiver operating characteristic curve (ROC-AUC) score with pRNFLT models compared to raw RNFLT models., Conclusions: Our results challenge the assumption that normative percentiles from OCT machines improve prediction of glaucomatous visual field loss. Raw RNFLT alone shows strong prediction, with no models presenting improvement by the manufacturer norms. This may result from insufficient patient stratification in tested norms., Translational Relevance: Understanding correlation of normative databases to visual function may improve clinical interpretation of OCT data.

Ophthalmic images, along with their derivatives like retinal nerve fiber layer (RNFL) thickness maps, play a crucial role in detecting and monitoring eye diseases such as glaucoma. For computer-aided diagnosis of eye diseases, the key technique is to automatically extract meaningful features from ophthalmic images that can reveal the biomarkers (e.g., RNFL thinning patterns) associated with functional vision loss. However, representation learning from ophthalmic images that links structural retinal damage with human vision loss is non-trivial mostly due to large anatomical variations between patients. This challenge is further amplified by the presence of image artifacts, commonly resulting from image acquisition and automated segmentation issues. In this paper, we present an artifact-tolerant unsupervised learning framework called EyeLearn for learning ophthalmic image representations in glaucoma cases. EyeLearn includes an artifact correction module to learn representations that optimally predict artifact-free images. In addition, EyeLearn adopts a clustering-guided contrastive learning strategy to explicitly capture the affinities within and between images. During training, images are dynamically organized into clusters to form contrastive samples, which encourage learning similar or dissimilar representations for images in the same or different clusters, respectively. To evaluate EyeLearn, we use the learned representations for visual field prediction and glaucoma detection with a real-world dataset of glaucoma patient ophthalmic images. Extensive experiments and comparisons with state-of-the-art methods confirm the effectiveness of EyeLearn in learning optimal feature representations from ophthalmic images.

Importance: Better understanding of primary open-angle glaucoma (POAG) genetics could enable timely screening and promote individualized disease risk prognostication., Objective: To evaluate phenotypic features across genetic burden for POAG., Design, Setting, and Participants: This was a cross-sectional, population-based study conducted from 2006 to 2010. Included participants were individuals from the UK Biobank aged 40 to 69 years. Individuals with non-POAG forms of glaucoma were excluded from the analysis. Data were statistically analyzed from October 2022 to January 2023., Main Outcomes and Measures: POAG prevalence based on structural coding, self-reports, and glaucoma-related traits., Results: Among 407 667 participants (mean [SD] age, 56.3 [8.1] years; 219 183 majority sex [53.8%]) were 14 171 POAG cases. Area under receiver operating characteristic curve for POAG detection was 0.748 in a model including polygenic risk score (PRS), age, sex, and ancestry. POAG prevalence in the highest decile of PRS was 7.4% (3005 of 40 644) vs 1.3% (544 of 40 795) in lowest decile (P < .001). A 1-SD increase in PRS was associated with 1.74 times higher odds of POAG (95% CI, 1.71-1.77), a 0.61-mm Hg increase in corneal-compensated intraocular pressure (IOP; 95% CI, 0.59-0.64), a -0.09-mm Hg decrease in corneal hysteresis (95% CI, -0.10 to -0.08), a 0.08-mm Hg increase in corneal resistance factor (95% CI, 0.06-0.09), and a -0.08-diopter decrease in spherical equivalent (95% CI, -0.11 to -0.07; P < .001 for all). A 1-SD increase in PRS was associated with a thinning of the macula-region retinal nerve fiber layer (mRNFL) of 0.14 μm and macular ganglion cell complex (GCC) of 0.26 μm (P < .001 for both). In the subset of individuals with fundus photographs, a 1-SD increase in PRS was associated with 1.42 times higher odds of suspicious optic disc features (95% CI, 1.19-1.69) and a 0.013 increase in cup-disc ratio (CDR; 95% CI, 0.012-0.014; P < .001 for both). A total of 22 of 5193 fundus photographs (0.4%) in decile 10 had disc hemorrhages, and 27 of 5257 (0.5%) had suspicious optic disc features compared with 9 of 5158 (0.2%) and 10 of 5219 (0.2%), respectively, in decile 1 (P < .001 for both). CDR in decile 10 was 0.46 compared with 0.41 in decile 1 (P < .001)., Conclusion and Relevance: Results suggest that PRS identified a group of individuals at substantially higher risk for POAG. Higher genetic risk was associated with more advanced disease, namely higher CDR and corneal-compensated IOP, thinner mRNFL, and thinner GCC. Associations with POAG PRS and corneal hysteresis and greater prevalence of disc hemorrhages were identified. These results suggest that genetic risk is an increasingly important parameter for risk stratification to consider in clinical practice.

Purpose: The tablet-based Melbourne Rapid Fields (MRF) visual field (VF) test and the IMOvifa Smart Visual Function Analyzer (SVFA) are portable perimeters that may allow for at-home monitoring and more frequent testing. We compared tablet and SVFA results with outputs from the Humphrey Field Analyzer (HFA) 24-2 Swedish Interactive Threshold Algorithm Standard program., Design: Observational cross-sectional study., Subjects: Adult participants with a diagnosis of glaucoma, suspected glaucoma, or ocular hypertension seen in the Massachusetts Eye and Ear glaucoma clinic were enrolled. All participants were reliable and experienced HFA testers., Methods: Participants were tested with the SVFA and HFA. The study staff also trained participants on the MRF tablet with instructions to take weekly tests at home for 3 months. Visual field results from the 3 devices were compared., Main Outcome Measures: Mean deviation (MD), pattern standard deviation (PSD), reliability parameters, and point sensitivity., Results: Overall, 79 participants (133 eyes) with a mean age of 61 ± 13 years (range, 26-79 years) were included; 59% of the participants were female, and the mean HFA MD was -2.7 ± 3.9 dB. The global indices of MD and PSD did not significantly vary between HFA and the 2 novel devices, except that the tablet VF reported a 0.6 dB higher PSD compared with HFA. However, tablet and SVFA sensitivities significantly differed from those of the HFA at 36 and 39 locations, respectively, out of 52 locations. Relative to HFA, the tablet overestimated light sensitivity in the nasal field while underestimating the temporal field. The SVFA generally underestimated light sensitivity, but its results were more similar to HFA results compared with the tablet., Conclusions: Although average MD values from the 2 novel devices suggest that they provide similar results to the HFA, point-by-point comparisons highlight notable deviations. Differences in specific point sensitivity values were significant, especially between the tablet and the other 2 devices. These differences may in part be explained by differences in the devices' normative databases as well as how MD is calculated. However, the tablet had substantial differences based on location, indicating that the tablet design itself may be responsible for differences in local sensitivities., Financial Disclosure(s): Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article., (Copyright © 2023 American Academy of Ophthalmology. Published by Elsevier Inc. All rights reserved.)

Purpose: To develop an objective glaucoma damage severity classification system based on OCT-derived retinal nerve fiber layer (RNFL) thickness measurements., Design: Algorithm development for RNFL damage severity classification based on multicenter OCT data., Subjects and Participants: A total of 6561 circumpapillary RNFL profiles from 2269 eyes of 1171 subjects to develop models, and 2505 RNFL profiles from 1099 eyes of 900 subjects to validate models., Methods: We developed an unsupervised k -means model to identify clusters of eyes with similar RNFL thickness profiles. We annotated the clusters based on their respective global RNFL thickness. We computed the optimal global RNFL thickness thresholds that discriminated different severity levels based on Bayes' minimum error principle. We validated the proposed pipeline based on an independent validation dataset with 2505 RNFL profiles from 1099 eyes of 900 subjects., Main Outcome Measures: Accuracy, area under the receiver operating characteristic curve, and confusion matrix., Results: The k -means clustering discovered 4 clusters with 1382, 1613, 1727, and 1839 samples with mean (standard deviation) global RNFL thickness of 58.3 (8.9) μm, 78.9 (6.7) μm, 87.7 (8.2) μm, and 101.5 (7.9) μm. The Bayes' minimum error classifier identified optimal global RNFL values of > 95 μ m , 86 to 95 μ m , 70 to 85 μ m , and < 70 μ m for discriminating normal eyes and eyes at the early, moderate, and advanced stages of RNFL thickness loss, respectively. About 4% of normal eyes and 98% of eyes with advanced RNFL loss had either global, or ≥ 1 quadrant, RNFL thickness outside of normal limits provided by the OCT instrument., Conclusions: Unsupervised machine learning discovered that the optimal RNFL thresholds for separating normal eyes and eyes with early, moderate, and advanced RNFL loss were 95 μ m , 85 μm, and 70 μ m , respectively. This RNFL loss classification system is unbiased as there was no preassumption or human expert intervention in the development process. Additionally, it is objective, easy to use, and consistent, which may augment glaucoma research and day-to-day clinical practice., Financial Disclosures: Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article., (© 2023 by the American Academy of Ophthalmology.)

Purpose: The purpose of this study was to assess metabolites associated with intraocular pressure (IOP) and inner retina structure., Methods: We cross-sectionally assessed 168 non-fasting plasma metabolites measured by nuclear magnetic resonance (NMR) spectroscopy with IOP (n = 28,195), macular retinal nerve fiber layer thickness (mRNFL; n = 10,584), and macular ganglion cell inner plexiform layer thickness (mGCIPL; n = 10,554) in the UK Biobank. We used multiple linear regression models adjusting for various covariates with probit-transformed metabolite levels as predictors for each outcome. Each estimate represents the difference in outcome variable per standard deviation increase in the probit-transformed metabolite values. We used the number of effective (NEF) tests and false discovery rate (FDR) to adjust for multiple comparisons for metabolites and metabolite classes, respectively., Results: In individual metabolite analysis, multiple amino acids, especially branched-chain amino acids, were associated with lower IOP (-0.12 mm Hg; 95% confidence interval = -0.16 to -0.07; NEF = 2.7E-05). Albumin, 3 hydroxybutyrate, lactate, and several lipids were associated with higher IOP (range = 0.07 to 0.18 mm Hg, NEF = ≤ 0.039). In IOP-adjusted analyses, five HDL-related metabolites were associated with thinner mRNFL (-0.15 microns for all metabolites, NEF = ≤ 0.027), whereas five LDL-related metabolites were associated with thicker mGCIPL (range = 0.17 to 0.20 microns; NEF = ≤ 0.044). In metabolite class analysis, the lipid components of lipoproteins (cholesterol, triglycerides, etc.) were not associated with our outcomes (FDR > 0.2 for all); yet multiple lipoproteins were significantly (FDR < 0.05) associated with all outcomes., Conclusions: Branched-chain amino acids were associated with lower IOP, HDL metabolites were associated with thinner mRNFL, and LDL metabolites were associated with thicker mGCIPL.

Aims: To assess whether associations of cardiopulmonary conditions and markers with glaucoma differ by background genetic risk for primary open angle glaucoma (POAG)., Methods: We constructed a POAG polygenic risk score (PRS) using genome-wide association study summary statistics from a large cross-ancestry meta-analysis. History of glaucoma (including self-report and codes for POAG, 'other glaucoma' or unspecified glaucoma), history of common cardiopulmonary conditions and cardiopulmonary measures were assessed in the UK Biobank. Stratifying by PRS decile 1 (lowest risk) versus decile 10 (highest risk), separate multivariable models were estimated to assess the associations of cardiopulmonary diseases or factors with glaucoma, adjusting for age, sex, smoking and medication use. A Bonferroni correction was used to adjust p values for multiple comparisons., Results: Individuals in POAG PRS decile 1 (417 cases, 44 458 controls; mean age 56.8 years) and decile 10 (2135 cases, 42 413 controls; mean age 56.7 years) were included. Within decile 1, glaucoma cases had significantly higher glycated haemoglobin (38.5 vs 35.9 mmol/mol) and higher prevalence of diabetes (17.5% vs 6.5%), dyslipidaemia (31.2% vs 18.3%) and chronic kidney disease (CKD) (6.7% vs 2.0%) than controls (adjusted p<0.0013 for each). Within decile 10, glaucoma was associated with higher prevalence of dyslipidaemia (27.7% vs 17.3%, p=6.9E-05). The magnitude of association between glaucoma and diabetes, CKD and glycated haemoglobin differed between deciles 1 and 10 (contrast test p value for difference <0.05)., Conclusion: The relations between systemic conditions and glaucoma vary by underlying genetic predisposition to POAG, with larger associations among those who developed glaucoma despite low genetic risk., Competing Interests: Competing interests: LRP is a consultant to Twenty Twenty, Skye Bioscience and Eyenovia. These consultancies do not relate to the topic of this manuscript. No conflicting relationship exists for the other authors., (© Author(s) (or their employer(s)) 2023. No commercial re-use. See rights and permissions. Published by BMJ.)