Appropriateness of Ophthalmology Recommendations From an Online Chat-Based Artificial Intelligence Model.

Objective: To determine the appropriateness of ophthalmology recommendations from an online chat-based artificial intelligence model to ophthalmology questions. Patients and Methods: Cross-sectional qualitative study from April 1, 2023, to April 30, 2023. A total of 192 questions were generated spanning all ophthalmic subspecialties. Each question was posed to a large language model (LLM) 3 times. The responses were graded by appropriate subspecialists as appropriate, inappropriate, or unreliable in 2 grading contexts. The first grading context was if the information was presented on a patient information site. The second was an LLM-generated draft response to patient queries sent by the electronic medical record (EMR). Appropriate was defined as accurate and specific enough to serve as a surrogate for physician-approved information. Main outcome measure was percentage of appropriate responses per subspecialty. Results: For patient information site-related questions, the LLM provided an overall average of 79% appropriate responses. Variable rates of average appropriateness were observed across ophthalmic subspecialties for patient information site information ranging from 56% to 100%: cataract or refractive (92%), cornea (56%), glaucoma (72%), neuro-ophthalmology (67%), oculoplastic or orbital surgery (80%), ocular oncology (100%), pediatrics (89%), vitreoretinal diseases (86%), and uveitis (65%). For draft responses to patient questions via EMR, the LLM provided an overall average of 74% appropriate responses and varied by subspecialty: cataract or refractive (85%), cornea (54%), glaucoma (77%), neuro-ophthalmology (63%), oculoplastic or orbital surgery (62%), ocular oncology (90%), pediatrics (94%), vitreoretinal diseases (88%), and uveitis (55%). Stratifying grades across health information categories (disease and condition, risk and prevention, surgery-related, and treatment and management) showed notable but insignificant variations, with disease and condition often rated highest (72% and 69%) for appropriateness and surgery-related (55% and 51%) lowest, in both contexts. Conclusion: This LLM reported mostly appropriate responses across multiple ophthalmology subspecialties in the context of both patient information sites and EMR-related responses to patient questions. Current LLM offerings require optimization and improvement before widespread clinical use.

A Comparative Study of Large Language Models, Human Experts, and Expert-Edited Large Language Models to Neuro-Ophthalmology Questions.

BACKGROUND: While large language models (LLMs) are increasingly used in medicine, their effectiveness compared with human experts remains unclear. This study evaluates the quality and empathy of Expert + AI, human experts, and LLM responses in neuro-ophthalmology. METHODS: This randomized, masked, multicenter cross-sectional study was conducted from June to July 2023. We randomly assigned 21 neuro-ophthalmology questions to 13 experts. Each expert provided an answer and then edited a ChatGPT-4-generated response, timing both tasks. In addition, 5 LLMs (ChatGPT-3.5, ChatGPT-4, Claude 2, Bing, Bard) generated responses. Anonymized and randomized responses from Expert + AI, human experts, and LLMs were evaluated by the remaining 12 experts. The main outcome was the mean score for quality and empathy, rated on a 1-5 scale. RESULTS: Significant differences existed between response types for both quality and empathy (P < 0.0001, P < 0.0001). For quality, Expert + AI (4.16 ± 0.81) performed the best, followed by GPT-4 (4.04 ± 0.92), GPT-3.5 (3.99 ± 0.87), Claude (3.6 ± 1.09), Expert (3.56 ± 1.01), Bard (3.5 ± 1.15), and Bing (3.04 ± 1.12). For empathy, Expert + AI (3.63 ± 0.87) had the highest score, followed by GPT-4 (3.6 ± 0.88), Bard (3.54 ± 0.89), GPT-3.5 (3.5 ± 0.83), Bing (3.27 ± 1.03), Expert (3.26 ± 1.08), and Claude (3.11 ± 0.78). For quality (P < 0.0001) and empathy (P = 0.002), Expert + AI performed better than Expert. Time taken for expert-created and expert-edited LLM responses was similar (P = 0.75). CONCLUSIONS: Expert-edited LLM responses had the highest expert-determined ratings of quality and empathy warranting further exploration of their potential benefits in clinical settings.

Acute Optic Neuropathy in Older Adults: Differentiating Between MOGAD Optic Neuritis and Nonarteritic Anterior Ischemic Optic Neuropathy.

BACKGROUND AND OBJECTIVES: Myelin oligodendrocyte glycoprotein antibody-associated disease optic neuritis (MOGAD-ON) and nonarteritic anterior ischemic optic neuropathy (NAION) can cause acute optic neuropathy in older adults but have different managements. We aimed to determine differentiating factors between MOGAD-ON and NAION and the frequency of serum MOG-IgG false positivity among patients with NAION. METHODS: In this international, multicenter, case-control study at tertiary neuro-ophthalmology centers, patients with MOGAD presenting with unilateral optic neuritis as their first attack at age 45 years or older and age-matched and sex-matched patients with NAION were included. Comorbidities, clinical presentations, acute optic disc findings, optical coherence tomography (OCT) findings, and outcomes were compared between MOGAD-ON and NAION. Multivariate analysis was performed to find statistically significant predictors of MOGAD-ON. A separate review of consecutive NAION patients seen at Mayo Clinic, Rochester, from 2018 to 2022, was conducted to estimate the frequency of false-positive MOG-IgG in this population. RESULTS: Sixty-four patients with unilateral MOGAD-ON were compared with 64 patients with NAION. Among patients with MOGAD-ON, the median age at onset was 56 (interquartile range [IQR] 50-61) years, 70% were female, and 78% were White. Multivariate analysis showed that eye pain was strongly associated with MOGAD-ON (OR 32.905; 95% CI 2.299-473.181), while crowded optic disc (OR 0.033; 95% CI 0.002-0.492) and altitudinal visual field defect (OR 0.028; 95% CI 0.002-0.521) were strongly associated with NAION. On OCT, peripapillary retinal nerve fiber layer (pRNFL) thickness in unilateral MOGAD-ON was lower than in NAION (median 114 vs 201 µm, p < 0.001; median pRNFL thickening 25 vs 102 µm, p < 0.001). MOGAD-ON had more severe vision loss at nadir (median logMAR 1.0 vs 0.3, p < 0.001), but better recovery (median logMAR 0.1 vs 0.3, p = 0.002). In the cohort of consecutive NAION patients, 66/212 (31%) patients with NAION were tested for MOG-IgG and 8% (95% CI 1%-14%) of those had false-positive serum MOG-IgG at low titers. DISCUSSION: Acute unilateral optic neuropathy with optic disc edema in older adults can be caused by either MOGAD-ON or NAION. Detailed history, the degree of pRNFL swelling on OCT, and visual outcomes can help differentiate the entities and prevent indiscriminate serum MOG-IgG testing in all patients with acute optic neuropathy.

Optic Neuropathy and Myelopathy in a Teenager With Biotinidase Deficiency.

ABSTRACT: A 19-year-old man presented with 3 years of gradually progressive, painless vision loss in both eyes. The ophthalmic examination showed bilateral diminished visual acuity, dyschromatopsia, and temporal optic nerve pallor. The neurological examination was consistent with a mild myelopathy with decreased pin-prick sensation starting at T6-T7 and descending through the lower extremities. Hyperreflexia was also present in the lower more than upper extremities. Infectious, inflammatory, and nutritional serum workup and cerebrospinal fluid analysis were both unrevealing. MRI of the brain and spinal cord showed abnormal T2 hyperintensity of the fornix, corpus callosum, optic nerves, and lateral columns of the cervical and thoracic spine, with diffusion restriction in the inferior-posterior corpus callosum and fornix. Biotinidase serum enzyme activity was tested and showed a decreased level of activity. Biotinidase gene testing showed a homozygous pathogenic variant, c.424C>A (p.P142T), confirming the diagnosis of biotinidase deficiency and prompting oral biotin supplementation. Three months after starting treatment, the patient's visual acuity, color vision, visual fields, and MRI spine abnormalities all improved significantly. Biotinidase deficiency is an important diagnostic consideration in patients with unexplained optic neuropathy and/or myelopathy.

Visual Outcomes Following Plasma Exchange for Optic Neuritis: An International Multicenter Retrospective Analysis of 395 Optic Neuritis Attacks.

PURPOSE: To evaluate the effectiveness of plasma exchange (PLEX) for optic neuritis (ON). METHODS: We conducted an international multicenter retrospective study evaluating the outcomes of ON following PLEX. Outcomes were compared to raw data from the Optic Neuritis Treatment Trial (ONTT) using a matched subset. RESULTS: A total of 395 ON attack treated with PLEX from 317 patients were evaluated. The median age was 37 years (range 9-75), and 71% were female. Causes of ON included multiple sclerosis (108), myelin oligodendrocyte glycoprotein antibody-associated disease (MOGAD) (92), aquaporin-4-IgG-positive neuromyelitis optica spectrum disorder (AQP4+NMOSD) (75), seronegative-NMOSD (34), idiopathic (83), and other (3). Median time from onset of vision loss to PLEX was 2.6 weeks (interquartile range [IQR], 1.4-4.0). Median visual acuity (VA) at the time of PLEX was count fingers (IQR, 20/200-hand motion), and median final VA was 20/25 (IQR, 20/20-20/60) with no differences among etiologies except MOGAD-ON, which had better outcomes. In 81 (20.5%) ON attacks, the final VA was 20/200 or worse. Patients with poor outcomes were older (P = .002), had worse VA at the time of PLEX (P < .001), and longer delay to PLEX (P < .001). In comparison with the ONTT subset with severe corticosteroid-unresponsive ON, a final VA of worse than 20/40 occurred in 6 of 50 (12%) PLEX-treated ON vs 7 of 19 (37%) from the ONTT treated with intravenous methylprednisolone without PLEX (P = .04). CONCLUSION: Most ON attacks improved with PLEX, and outcomes were better than attacks with similar severity in the ONTT. The presence of severe vision loss at nadir, older age, and longer delay to PLEX predicted a worse outcome whereas MOGAD-ON had a more favorable prognosis. NOTE: Publication of this article is sponsored by the American Ophthalmological Society.

Stroke Risk before and after Central Retinal Artery Occlusion: A Population-based Analysis.

PURPOSE: To determine the risk of stroke, transient ischemic attack (TIA), and transient monocular vision loss (TMVL) before and after a central retinal artery occlusion (CRAO). DESIGN: Population-based, retrospective case series. PARTICIPANTS: Patients diagnosed with a CRAO in Olmsted County, Minnesota, from 1976 to 2016. METHODS: Patients living in Olmsted County with a diagnosis code of CRAO from 1976 to 2016 were reviewed. New CRAOs were confirmed, and stroke, TIA, and TMVL events in the 15 days before and after CRAO were recorded. MAIN OUTCOME MEASURES: Incidence of stroke, TIA, and TMVL events in the 15 days before and after CRAO. RESULTS: Eighty-nine patients with a CRAO were identified, providing an annual incidence of 2.58/100 000 (95% confidence interval [CI], 2.04-3.11). Median age at the time of CRAO was 76 years (range, 46-100 years); 56.2% were male, and 89.9% of the cohort was White. In the 15 days before and after CRAO, there were 2 ischemic strokes (2.2%), 1 hemorrhagic stroke (1.1%), 2 TIAs (2.2%), and 9 TMVL events (10.1%). Starting in 1999, 15 of 45 patients underwent magnetic resonance imaging within 2 months of CRAO. One patient (6.7%) had evidence of asymptomatic diffusion restriction, and 9 patients (60%) had a remote infarct. CONCLUSIONS: This population-based study demonstrated that the risk of symptomatic ischemic stroke is 2.2% in the 15 days before and after a CRAO, which is slightly lower than most studies from tertiary centers. These data should be considered as practice recommendations are developed regarding the urgency of neurovascular workup in patients with acute CRAO.

Population-Based Incidence of Ocular Neovascularization Following Central Retinal Artery Occlusion in Olmsted County, Minnesota.

PURPOSE: To determine the population-based incidence of ocular neovascularization (NV) following central retinal artery occlusion (CRAO) and systemic risk factors associated with its development. METHODS: Diagnosis of CRAO between January 1, 1976, and September 9, 2016, was identified using the Rochester Epidemiology Project, a medical records linkage system for all medical care provided to residents of Olmsted County, Minnesota. Records were reviewed to confirm the diagnosis of CRAO, and data were collected on ocular NV and associated systemic diseases. RESULTS: There were 89 patients with CRAO. Subsequent ocular neovascularization developed in 14 (16%) patients. Neovascularization of the iris (NVI) was present in 9/14 (64%) of these patients, neovascularization of the angle (NVA) in 10/14 (71%), neovascularization of the disc (NVD) in 2/14 (14%), and neovascularization elsewhere (NVE) in 1/14 (7%). Of these 14 patients with NV, 9 (64%) developed neovascular glaucoma (NVG). The mean time from CRAO diagnosis to NV was 82 days (range 22 to 268 days). Excluding the patients with proliferative diabetic retinopathy or CRAO caused by CRVO, the mean time to NV diagnosis was 80 days and the shortest time to NV diagnosis was 22 days. Diabetes mellitus was present in 64% of those with NV compared to 23% of those without NV (P = 0.003). CONCLUSION: The population-based incidence of ocular neovascularization following CRAO is 16% and developed within 2 months in half the cohort. Patients with diabetes mellitus are at increased risk for NV complications.

Ophthalmic Implications of Chimeric Antigen Receptor T-Cell Therapy.

Chimeric antigen receptor (CAR) T-cell therapy is a revolutionary addition to the burgeoning field of immunotherapy. CAR T-cells are engineered by combining a T-cell receptor with the antigen-binding site of an immunoglobulin that allows the hybrid cell to target antigens of interest. CAR T-cell therapy has been approved to treat various hematologic malignancies, including relapsed or refractory B-cell acute lymphoblastic leukemia and diffuse large B-cell lymphoma. While the treatment efficacy is exciting, challenges remain in understanding the unique spectrum of adverse effects of CAR T-cell therapy, including cytokine release syndrome and neurotoxicity. Innovative research is underway to expand this therapy into solid tumors and fields beyond hematology and oncology. To date, there has been limited research into ophthalmic uses and considerations of CAR T-cell therapy. This review focuses on preclinical investigations into CAR T-cell therapy for retinoblastoma and uveal melanoma, as well as ophthalmic complications of CAR T-cell therapy.

Stroke Risk Before and After Central Retinal Artery Occlusion in a US Cohort.

OBJECTIVE: To determine the risk of ischemic stroke, transient ischemic attack (TIA), and amaurosis fugax around the time of central retinal artery occlusion (CRAO). PATIENTS AND METHODS: Patients at Mayo Clinic sites with a diagnosis code of CRAO from January 1, 2001, through September 9, 2016, were reviewed. New CRAOs were confirmed, and ischemic stroke, TIA, and amaurosis fugax events were tallied within 15 days before and after CRAO. RESULTS: Three hundred patients with CRAO were included in the study cohort. The median age at the time of CRAO was 72 years (range, 19-100 years), and 158 (53%) were male patients. Overall, 16 patients (5.3%) had symptomatic ischemic stroke around the time of CRAO, with 7 strokes (2.3%) occurring 15 days before CRAO, 4 (1.3%) occurring simultaneously with CRAO, and 5 (1.7%) occurring after CRAO. Transient ischemic attack and amaurosis fugax were seen in 5 (1.7%) and 26 (8.7%) patients, respectively. It was found that 7% (9/128) of patients with embolic CRAO had a stroke around the time of CRAO as compared with 1.3% (2/149) of patients with an unknown cause of CRAO. CONCLUSION: Symptomatic stroke, TIA, or amaurosis fugax is common around the time of CRAO. Therefore, CRAOs require urgent embolic work-ups.