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.

Magnetic Resonance Imaging Findings and Genetic Testing Results in Children With Congenital Corneal Opacities

PURPOSE

This study investigates brain and globe abnormalities identified on magnetic resonance imaging (MRI) in children with congenital corneal opacities (CCO).

DESIGN

Retrospective cohort study.

METHODS

Clinical notes, radiology records, and genetic testing results were reviewed for patients diagnosed with corneal opacification within the first 6 months of life at a tertiary referral academic center between August 2008 and January 2018. Ocular findings, systemic anomalies, neuroimaging, and genetic testing results were summarized.

RESULTS

A total of 135 patients presenting at age 1 day to 12 years (mean age, 1 year) were identified. Children with bilateral CCO were more likely to have systemic disease (P = 0.018). Of the entire cohort, 43 (31.8%) patients received MRI, of whom 27 (62.8%) had abnormal brain findings and 30 (69.7%) had abnormal orbital findings. The most common abnormal brain findings were ventriculomegaly (n = 16, 59.2%) and corpus callosum abnormalities (n = 10, 37.0%) followed by brainstem/pons anomalies (n = 5, 18.5%), and cerebellar anomalies (n = 2, 7.4%). Abnormal brain MRI findings were associated with the presence of neurologic (P = .003) and craniofacial (P = .034) disease. A total of 44 (32.1%) patients underwent genetic testing, of whom 29 (65.9%) had pathogenic results.

CONCLUSIONS

More than 60% of the children with CCO who underwent MRI had abnormal brain and orbit findings that were correlated with significant neurologic disease. Furthermore, almost two-thirds of patients with CCO who underwent genetic testing had pathogenic results. These data demonstrate the value of systemic workup in children with CCO, and highlight the role of ophthalmologists in facilitating the diagnosis of systemic comorbidities associated with CCO.

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.

Ocular Neuromyotonia: Clinical Features, Diagnosis, and Outcomes

PURPOSE

The purpose of this study was to describe the clinical features, management, outcomes, and diagnostic pitfalls in a large series of patients with ocular neuromyotonia.

DESIGN

Retrospective cohort.

METHODS

Patients diagnosed with ocular neuromyotonia from January 1, 2004, through January 1, 2023, seen at one of the 3 Mayo Clinic sites in Rochester, MN, Scottsdale, AZ, and Jacksonville, FL, comprised the study population. We ascertained patients with ocular neuromyotonia through a search using the medical records database. Only patients with an observed episode of ocular neuromyotonia were included and the medical records were reviewed. The main outcome measures were clinical features and outcomes of patients with ocular neuromyotonia.

RESULTS

Forty-two patients who were diagnosed with ocular neuromyotonia were included. The median age was 58 years (range, 16-80 years). A history of cranial radiation therapy was present in 39 patients (93%). The sixth cranial nerve was involved in 31 patients (74%). Bilateral disease was found in 2 patients (5%). The median time from onset of diplopia to diagnosis was 8 months (range, 1 month-25 years), with a high rate of initial misdiagnosis in 52%. Twenty of 42 patients (48%) were treated with oral medication, of whom 95% had significant improvement or resolution of symptoms.

CONCLUSION

Prior cranial irradiation is the most common cause for ocular neuromyotonia, affecting the sixth cranial nerve most often. Although delayed and initial misdiagnosis is common, most patients show improved symptoms on medical treatment.

Transient Bilateral Central Scotoma on Awakening: A Distinct Phenomenon

BACKGROUND

Transient vision loss can be an alarming symptom owing to potentially serious etiologies such as thromboembolism or giant cell arteritis. Transient monocular vision loss (TVML) on awakening has been recently described as a benign phenomenon (Bouffard et al, 2017). Our objective was to describe the unique phenomenon of transient binocular vision loss (TBVL) on awakening.

METHODS

Retrospective observational case series of 5 patients who experienced TBVL on awakening seen in the neuro-ophthalmology clinic at the Mayo Clinic between 2016 and 2020. Patients who described generalized blurred vision and those with monocular transient vision loss were excluded.

RESULTS

The median age was 81.4 years (range, 68-92 years), and all were male. Every patient described a stereotyped transient bilateral central scotoma on awakening with slightly irregular borders and mild asymmetry between the 2 eyes lasting between 15 and 90 minutes. Frequency ranged from 3 to 7 times per week, and there was a median of 319 episodes (range, 126-728 episodes) before evaluation in the neuro-ophthalmology clinic. All patients had normal optic nerves, and no plaques were noted in the retinal vessels. All 5 had macular drusen, which were predominantly extrafoveal and mild. Two patients underwent electrophysiology testing, which were both normal on full-field electroretinogram (ERG), but there was blunted central wave forms on multifocal ERG. Two patients underwent dark adaptation testing, which showed both prolong and diminished dark adaptation. Neuroimaging and thromboembolic workup were unrevealing.

CONCLUSIONS

TBVL is a distinct phenomenon from TMVL on awakening, which has a different demographic and symptomology. The etiology is unclear but seems to be a focal macular process in conjunction with an autoregulatory failure resulting in a supply-demand mismatch during low-light conditions.

Incidence and distribution of ocular disorders in the first year of life

PURPOSE

To describe the incidence and distribution of eye diseases affecting children in the first year of life in Olmsted County, Minnesota.

METHODS

We conducted a population-based, retrospective medical record review of infants (≤1 year of age) residing in Olmsted County diagnosed with an ocular disorder from January 1, 2005, through December 31, 2014.

RESULTS

A total of 4,223 infants were diagnosed with an ocular disorder, yielding an incidence of 20,242/100,000 births per year, or 1 in 4.9 live births (95% CI, 19,632-20,853). The median age at diagnosis was 3 months, and 2,179 (51.5%) were female. The most common diagnoses included conjunctivitis, in 2,175 (51.5%), nasolacrimal duct obstruction, in 1,432 (33.6%), and pseudostrabismus, in 173 (4.1%). Visual acuity was decreased in one or both eyes in 23 (0.5%) infants because of strabismus in 10 (43.5%) and cerebral visual impairment in 3 (13.0%). A majority of the infants (3,674 [86.9%]) were diagnosed and managed by a primary care provider, and 549 (13.0%) were evaluated and/or managed by an eye care provider.

CONCLUSIONS

Although ocular disorders occurred in 1 in 5 infants in this cohort, most conditions were evaluated and managed by primary care providers. Understanding the incidence and distribution of ocular diseases among infants is useful for planning clinical resources.

Accuracy of the International Classification of Diseases, 9th Revision for Identifying Infantile Eye Disease

PURPOSE

To determine the predictive value of International Classification of Diseases, 9th Revision (ICD-9) codes for identifying infantile eye diagnoses.

METHODS

Population-based retrospective cohort study of all residents of Olmsted County, Minnesota diagnosed at ≤1 year of age with an ocular disorder. The medical records of all infants diagnosed with any ocular disorder from January 1, 2005, through December 31, 2014, were identified. To assess ICD-9 code accuracy, the medical records of all diagnoses with ≥20 cases were individually reviewed and compared to their corresponding ICD-9 codes. Main outcome measures included positive predictive value (PPV), negative predictive value (NPV), sensitivity, and specificity of ICD-9 codes.

RESULTS

In a cohort of 5,109 infants with ≥1 eye-related ICD-9 code, 10 ocular diagnoses met study criteria. The most frequent diagnoses were conjunctivitis (N = 1,695) and congenital nasolacrimal duct obstruction (N = 1,250), while the least common was physiologic anisocoria (N = 23). The PPVs ranged from 8.3% to 88.0%, NPVs from 96.3% to 100%, sensitivity from 3.0% to 98.7%, and specificity from 72.6% to 99.9%. ICD-9 codes were most accurate at identifying physiologic anisocoria (PPV: 88.0%) and least accurate at identifying preseptal cellulitis (PPV: 8.3%). In eye specialists versus non-eye specialists, there was a significant difference in PPV of ICD-9 codes for conjunctivitis (26.8% vs. 63.9%, p < .001), pseudostrabismus (85.9% vs. 25.0%, p < .001), and physiologic anisocoria (95.5% vs. 33.3%, p = .002).

CONCLUSION

The predictive value of ICD-9 codes for capturing infantile ocular diagnoses varied widely in this cohort. These findings emphasize the limitations of database research methodologies that solely utilize claims data to identify pediatric eye diseases.Abbreviations/Acronyms PPV: positive predictive value; NPV: negative predictive value; CNLDO: congenital nasolacrimal duct obstruction.

Incidence and Clinical Characteristics of Infantile Conjunctivitis in a Western Population

OBJECTIVE

To describe the incidence and clinical characteristics of conjunctivitis in the first year of life.

DESIGN

Population-based cohort study.

PARTICIPANTS

All infant (≤12 months of age) residents of Olmsted County, Minnesota, diagnosed with conjunctivitis from January 1, 2005, through December 31, 2014.

METHODS

The medical records of all potential cases identified by the Rochester Epidemiology Project database were reviewed.

MAIN OUTCOME MEASURES

Incidence rate and clinical features of infantile conjunctivitis.

RESULTS

A total of 2175 infants were diagnosed during the 10-year period, yielding an incidence of 10,422 per 100,000 children or approximately 1 in 10 infants by 1 year of age. The mean age at diagnosis was 4.9 months (range, 1 day-12 months), and 1001 (46.0%) were female. Both eyes were involved in 1180 (54.3%), the right eye alone in 506 (23.3%), and 489 (22.5%) in the left. Five hundred seventy-six (26.5%) of the 2175 were diagnosed at ≤30 days of life, from which topical cultures were obtained in 111 (19.7%). Only 36 (32.4%) of the cultures showed bacterial agents, with Chlamydia present in 3. Treatment for infantile conjunctivitis, where recorded, included topical antibiotics in 523 (90.8%) and simple observation in 47 (8.2%).

CONCLUSIONS

Conjunctivitis in the first year of life occurred in approximately 10% of infants in this population-based cohort. More than half involved both eyes, one-quarter were identified in the first 30 days of life, and sight-threatening infectious agents were rare.

The 'Fault' Lies in the Choroid: Peripapillary Intrachoroidal Cavitation Presenting with Progressive Vision Loss

A 55-year-old male was referred to the Neuro-ophthalmology clinic due to gradual onset, progressive vision loss. On fundus examination a subtle yellow-orange peripapillary lesion was detected in the left eye. Optical coherence tomography with radial scanning illustrated retinal nerve fibre layer thinning as well as an area of intrachoroidal cavitation that corresponded to the lesion. Visual field testing showed a left inferior arcuate defect. Magnetic resonance imaging of the brain and orbit, and laboratory testing was unremarkable. Clinical examination, imaging, and testing were consistent with peripapillary intrachoroidal cavitation (PICC). Follow-up with serial visual field testing showed mild progression of the field defect. While PICC is not well understood in the literature, studies have reported associated risk factors including pathological myopia, older age, increased ocular axial length, chorioretinal atrophy, and vascular abnormalities. Importantly, glaucoma-like visual field defects as well as structural changes have been noticed in a high proportion of patients with PICC. While these alterations are evident, the pathogenic relationship between them is yet to be uncovered. Treatment with anti-glaucoma medications has been suggested, however, the evidence remains scarce for its true benefits. Care providers must be aware of the presentation of a yellow-orange peripapillary lesion with an associated visual field defect to accurately diagnose and manage this condition.

Anterior segment ischemia after laser for retinopathy of prematurity previously treated with antivascular endothelial growth factor

PURPOSE

To assess the risk of vision-threatening anterior segment ischemia (ASI) among retinopathy of prematurity (ROP) patients treated with anti-VEGF followed by laser photocoagulation.

METHODS

The medical records of all infants treated for threshold ROP with laser photocoagulation with and without prior anti-VEGF injections from January 1, 2002, through December 2018 at Mayo Clinic were retrospectively reviewed for the prevalence of vision-threatening ASI.

RESULTS

A total of 241 eyes of 122 infants were included. Mean gestational age was 25.1 weeks (range, 22.9-28.7); mean birth weight was 687.6 g (range, 360-1310 g). Of the 54 eyes (27 patients) treated with anti-VEGF prior to laser, 4 developed ASI (including corneal edema, cataracts, and choroidal effusion) compared with 2 of the 187 eyes (95 patients) treated with laser therapy alone (P = 0.008). Infants receiving both anti-VEGF and laser had a younger gestational age at birth (24.5 vs 25.3 weeks; P < 0.001) and lower birth weight (591.4 g vs 715.0 g; P < 0.001) than those who received laser alone. In multivariate analysis, early gestational age at birth was associated with development of ASI (P = 0.03); the association with anti-VEGF treatment (P = 0.07) fell short of statistical significance.

CONCLUSIONS

The prevalence of vision-threatening ASI was higher among infants treated with intravitreal anti-VEGF followed by laser compared to those treated with laser alone. Further investigation is warranted to confirm this finding and identify potential factors for decreasing the risk of ASI.

Incidence of Dementia in Patients With Open-angle Glaucoma: A Population-based Study

PRECIS

In this population-based study of 509 open-angle glaucoma (OAG) patients over a 36-year period, we identified a decreased rate of developing dementia compared with the rate in the general population.

PURPOSE

The aim was to determine the incidence of dementia and Alzheimer disease (AD) among patients with OAG.

PATIENTS AND METHODS

Retrospective, population-based cohort study. All residents of Olmsted County, Minnesota (≥40 y) who were diagnosed with OAG between January 1, 1965 and December 31, 2000, were eligible for inclusion in this study. A total of 509 patients were included over the 36-year period. The cumulative probability of developing dementia was calculated and compared with the population risk of dementia.

RESULTS

Of the 509 patients included, 300 (58.9%) were female, the median age was 67.5 years, and 278 patients (54.6%) had primary OAG. Other subgroups were pseudoexfoliation in 15.1%, treated ocular hypertension in 14.1%, normal tension glaucoma in 10.6%, and pigmentary glaucoma in 5.5% of the patients. Respectively, 118 (23.0%) and 99 (19.4%) patients developed dementia and AD. The 10-year cumulative probability of developing dementia and AD was 12.0% and 9.9%, with a 95% confidence interval of 9.3%-15.3% and 7.5%-13%, respectively. The observed 10-year incidence of dementia and AD were significantly lower than the expected population incidence (19.0% and 19.0%; P<0.001). Older age at diagnosis of glaucoma was a strong predictor for the development of dementia by multivariate analysis (hazard ratio: 3.31, 95% confidence interval: 2.61-4.20, P<0.001).

CONCLUSION

The risk of developing dementia or AD was decreased in OAG patients compared with the general population. OAG with onset at a later age may present as a different etiopathogenetic entity compared with onset at a younger age, and represent the optic nerve findings of generalized neurodegenerative processes.

A Population-Based Study of Anterior Ischemic Optic Neuropathy Following Cataract Surgery

PURPOSE

To determine if there is an increased incidence rate of post-cataract surgery (pcs) anterior ischemic optic neuropathy (AION) compared to spontaneous AION (sAION).

DESIGN

Retrospective, population-based cohort.

METHODS

Patients diagnosed with AION from January 1, 1990, through December 31, 2016, while residing in Olmsted County, Minnesota. Patients with cataract surgery preceding AION were included in the pcsAION cohort defined in 2 ways: AION within 2 months and AION within 1 year of cataract surgery. The incidence rates of pcsAION and sAION were compared using Poisson regression models.

RESULTS

During the study period, 102 residents developed AION. The median age was 65 years (range, 40-90 years), 44 (43.1%) were female. Twenty of 102 (19.6%) patients had previous cataract surgery, of which 2 and 9 developed AION within 2 months and 1 year of surgery, respectively. The annual incidence rate of pcsAION within 2 months of surgery (8.6 per 100,000) was not significantly greater than the annual incidence rate of sAION (6.9 per 100,000; P = .78). However, the annual incidence rate of pcsAION within 1 year of surgery (38.9 per 100,000) was significantly higher than the incidence rate of sAION (6.5 per 100,000; P < .001).

CONCLUSION

The incidence of AION is increased in the first year after cataract surgery, but not in the early (i.e., 2 months) postoperative period.

Prenatal or Perinatal Injury? Diagnosing the Cortically Blind Infant

PURPOSE

To document the association of prenatal brain disruption with secondary perinatal distress in children diagnosed as having cortical visual impairment (CVI).

DESIGN

Retrospective case series.

METHODS

Eight children with severe CVI and clinical history of perinatal events were included. Case histories and neuroimaging studies were reviewed. The main outcome measures were perinatal history, visual and neurologic findings, and magnetic resonance (MR) imaging.

RESULTS

In our patient cohort, MR imaging showed signs of cortical dysgenesis leading to congenital brain malformations such as polymicrogyria consistent with a prenatal timing of CNS injury. Although subcortical white matter changes were common, signs of watershed injury to the visual cortex were absent, suggesting that the visual loss was attributable to a prenatal etiology with secondary birth complications.

CONCLUSION

Some children with CVI and a history of perinatal distress have prenatal dysgenesis of the developing brain. Therefore, a clinical history of perinatal hypoxia-ischemia is nonspecific and merits neuroimaging to identify antecedent brain malformations and timing of injury, which can influence patient diagnosis and management.

Test-retest reliability of the revised diplopia questionnaire

PURPOSE

To evaluate misclassification of diplopia "success" when using a standardized diplopia questionnaire (DQ), and to report test-retest reliability of the DQ.

METHODS

We retrospectively identified a cohort of 100 patients with stable strabismus (<5^Δ change in prism and alternate cover test measurements at distance and near), stable visual acuity, no change in treatment, and no clinical evidence of change, with completed DQ at two consecutive office visits (median, 71 days apart; range, 5-350 days). To evaluate the rate of misclassification of "success" and "not success," we compared the second to the first administration of the DQ using two established definitions of success: (1) "rarely" or "never" for straight ahead distance and (2) "rarely" or "never" for straight-ahead distance and reading. For DQ test-retest variability, 95% limits of agreement (LOA) and intraclass correlation coefficients (ICC) were calculated on DQ scores (0-100 scale).

RESULTS

When defining success as rarely or never diplopic for distance, misclassification occurred in 12 (12%) of 100 (95% CI, 6%-20%). When defining success as rarely or never diplopic for distance and reading, misclassification occurred in 14 (14%) of 100 (95% CI, 8%-22%). The 95% LOA for the DQ score were 35.2 points, and ICC was 0.85 (95% CI, 0.79-0.90).

CONCLUSIONS

We have quantified misclassification and test-retest variability when using the DQ dichotomously or as a continuous measure, equipping the clinician to better interpret DQ outcome data in practice and research.

Incidence and Ocular Features of Pediatric Myasthenias

PURPOSE

To report the incidence, demographics, and ocular findings of children with myasthenia.

DESIGN

Retrospective cohort study.

METHODS

The medical records of all children (<19 years) examined at Mayo Clinic with any form of myasthenia from January 1 1966, through December 31, 2015, were retrospectively reviewed.

RESULTS

A total of 364 children were evaluated during the study period, of which 6 children were residents of the Olmsted County at the time of their diagnosis, yielding an annual age- and sex-adjusted incidence of 0.35 per 100 000 <19 years, or 1 in 285 714 <19 years. The incidence of juvenile myasthenia gravis (JMG) and congenital myasthenic syndrome (CMS) was 0.12 and 0.23 per 100 000, respectively. Of the 364 study children, 217 (59.6%) had JMG, 141 (38.7%) had CMS, and 6 (1.7%) had Lambert-Eaton syndrome, diagnosed at a median age of 13.5, 5.1, and 12.6 years, respectively. A majority of the JMG and CMS patients had ocular involvement (90.3% and 85.1%, respectively), including ptosis and ocular movement deficits. Among children with at least 1 year of follow-up (JMG; median, 7.1 years, CMS; median, 7.0 years), improvement was seen in 88.8% of JMG patients (complete remission in 31.3%) and in 58.3% of CMS patients.

CONCLUSION

Although relatively rare, myasthenia gravis in children has 2 predominant forms, CMS and JMG, both of which commonly have ocular involvement. Improvement is more likely in children with the juvenile form.

Fluorescein Angiography in Retinopathy of Prematurity: Comparison of Infants Treated with Bevacizumab to Those with Spontaneous Regression

OBJECTIVE

Medium- and long-term sequelae of intravitreal bevacizumab (IVB) for type 1 retinopathy of prematurity (ROP) are uncertain. Our aim was to describe the fluorescein angiography (FA) findings in patients who received IVB as primary treatment for type 1 ROP and compare them to findings in patients with ROP that spontaneously regressed.

DESIGN

Retrospective cohort.

PARTICIPANTS

Twenty-eight patients with a history of ROP who underwent fluorescein angiography between December 1, 2013, and July 31, 2018. Patients were divided into 2 groups based on whether they had received IVB or had ROP that spontaneously regressed.

METHODS

We reviewed the angiograms in the 2 groups for neovascularization (NV) and other abnormal vascular patterns in both the periphery and the posterior pole.

MAIN OUTCOME MEASURES

Fluorescein angiography findings, including NV, peripheral, and macular vascular abnormalities.

RESULTS

Forty eyes of 20 infants were included in the IVB group and 16 eyes of 8 infants in the untreated group. Median gestational age at birth was similar in the 2 groups (24.5 and 24.7 weeks, respectively; P = 0.44), as was the median birth weight (648.5 and 560.0 g, respectively; P = 0.26). Median postmenstrual age at the time of FA was 65.1 and 83.9 weeks, respectively (P = 0.0002). Review of angiograms demonstrated NV in 30.0% and 37.5% in the IVB and untreated cohorts, respectively (P = 0.75). Abnormal vascular patterns in the periphery were similar in both groups (100.0%), whereas posterior pole findings of vessels encroaching onto the fovea were more prevalent in the IVB cohort (65.0% vs. 25.0%; P = 0.009).

CONCLUSIONS

Fluorescein angiography after bevacizumab for ROP reveals abnormal vascular patterns in all eyes and NV in approximately one-third. Similar abnormal vascular patterns on FA are seen at a similar prevalence after spontaneous regression of ROP. These findings suggest that the abnormal vascular patterns identified by FA in patients with ROP result from the disease process itself rather than as a result of exposure to anti-vascular endothelial growth factor medications.

Nd:YAG capsulotomy for the management of posterior capsular amyloidosis

Purpose

To describe the accumulation of amyloid in Berger's space.

Observations

A 35-year-old man with autosomal-dominant, familial transthyretin-associated amyloidosis and bilateral vitreous opacities had a recurrence of amyloidosis following vitrectomy. The recurrent amyloid was attached to the posterior capsule of the lens. Phacoemulsification followed by neodymium:yttrium-aluminum-garnet (Nd:YAG) capsulotomy was helpful in restoring vision.

Conclusions and importance

Amyloid can reaccumulate in Berger's space, which is difficult to reach in phakic eyes during vitrectomy and can then cause decreased vision. Recognition of this interesting anatomic problem will allow for faster visual rehabilitation of the patient.

The Risk of Primary Open-Angle Glaucoma Following Vitreoretinal Surgery-A Population-based Study

PURPOSE

To determine the risk of primary open-angle glaucoma (POAG) following vitreoretinal surgery.

DESIGN

Retrospective, population-based cohort study.

METHODS

All residents of Olmsted County, Minnesota, undergoing scleral buckle and/or vitrectomy between 2004 and 2015 were included in the operative cohort. The fellow nonoperative eyes were included in the comparison cohort. The study and comparison cohorts consisted of 344 and 277 eyes, respectively. The main outcome measure was the development of POAG. Secondary glaucomas were excluded. The probability of glaucoma in operative eyes and nonoperative fellow eyes was compared. The observed rate of POAG in the operative eyes was also compared to the rate of POAG in the population of Olmsted County.

RESULTS

The mean age was 64.7 years and the median follow-up period was 4.9 years. There were 58, 57, and 229 study eyes in the scleral buckle, scleral buckle with vitrectomy, and vitrectomy only cohorts, respectively. The 10-year cumulative probability of developing glaucoma was significantly greater in the operative group (8.9%, 95% confidence interval [CI] 3.8%-14%) compared to the nonoperative group (1.0%, 95% CI 0-2.4%; P = .02). None of the eyes in the scleral buckle group developed glaucoma. The 10-year probability of POAG was 17.5% (95% CI 0-34.9%) and 10.0% (95% CI 3.0%-17.0%) in the scleral buckle with vitrectomy and vitrectomy alone cohorts, respectively. The rates of POAG in operative eyes undergoing scleral buckle with vitrectomy and vitrectomy alone were significantly greater than the rate of POAG for the Olmsted County general population (1.0%, P < .001).

CONCLUSIONS

The risk of POAG is increased after vitrectomy.