Predicting Choroidal Nevus Transformation to Melanoma Using Machine Learning.

Purpose: To develop and validate machine learning (ML) models to predict choroidal nevus transformation to melanoma based on multimodal imaging at initial presentation. Design: Retrospective multicenter study. Participants: Patients diagnosed with choroidal nevus on the Ocular Oncology Service at Wills Eye Hospital (2007-2017) or Mayo Clinic Rochester (2015-2023). Methods: Multimodal imaging was obtained, including fundus photography, fundus autofluorescence, spectral domain OCT, and B-scan ultrasonography. Machine learning models were created (XGBoost, LGBM, Random Forest, Extra Tree) and optimized for area under receiver operating characteristic curve (AUROC). The Wills Eye Hospital cohort was used for training and testing (80% training-20% testing) with fivefold cross validation. The Mayo Clinic cohort provided external validation. Model performance was characterized by AUROC and area under precision-recall curve (AUPRC). Models were interrogated using SHapley Additive exPlanations (SHAP) to identify the features most predictive of conversion from nevus to melanoma. Differences in AUROC and AUPRC between models were tested using 10 000 bootstrap samples with replacement and results. Main Outcome Measures: Area under receiver operating curve and AUPRC for each ML model. Results: There were 2870 nevi included in the study, with conversion to melanoma confirmed in 128 cases. Simple AI Nevus Transformation System (SAINTS; XGBoost) was the top-performing model in the test cohort [pooled AUROC 0.864 (95% confidence interval (CI): 0.864-0.865), pooled AUPRC 0.244 (95% CI: 0.243-0.246)] and in the external validation cohort [pooled AUROC 0.931 (95% CI: 0.930-0.931), pooled AUPRC 0.533 (95% CI: 0.531-0.535)]. Other models also had good discriminative performance: LGBM (test set pooled AUROC 0.831, validation set pooled AUROC 0.815), Random Forest (test set pooled AUROC 0.812, validation set pooled AUROC 0.866), and Extra Tree (test set pooled AUROC 0.826, validation set pooled AUROC 0.915). A model including only nevi with at least 5 years of follow-up demonstrated the best performance in AUPRC (test: pooled 0.592 (95% CI: 0.590-0.594); validation: pooled 0.656 [95% CI: 0.655-0.657]). The top 5 features in SAINTS by SHAP values were: tumor thickness, largest tumor basal diameter, tumor shape, distance to optic nerve, and subretinal fluid extent. Conclusions: We demonstrate accuracy and generalizability of a ML model for predicting choroidal nevus transformation to melanoma based on multimodal imaging. Financial Disclosures: Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

Evaluating Large Language Models on their Accuracy and Completeness: Immune Checkpoint Inhibitors and their Ocular Toxicities.

PURPOSE: To analyze the accuracy and thoroughness of 3 large language models (LLMs) to produce information for providers about immune checkpoint inhibitor (ICI) ocular toxicities. METHODS: Eight questions were created about the general definition of checkpoint inhibitors, their mechanism of action, ocular toxicities, and toxicity management. All were inputted into ChatGPT 4.0, Bard, and LLaMA programs. Utilizing the 6-point Likert scale for accuracy and completeness, four ophthalmologists who routinely treat ocular toxicities of immunotherapy agents rated the LLMs answers. ANOVA testing was used to assess significant differences among the three LLMs and a post-hoc pairwise t-test. Fleiss kappa values were calculated to account for interrater variability. RESULTS: ChatGPT responses were rated with an average of 4.59 for accuracy and 4.09 for completeness; Bard answers were rated 4.59 and 4.19; LLaMA results were rated 4.38 and 4.03. The three LLMs did not significantly differ in accuracy (p=0.47) nor completeness (p=0.86). Fleiss kappa values were found to be poor for both accuracy (-0.03) and completeness (0.01). CONCLUSIONS: All three LLMs provided highly accurate and complete responses to questions centered on ICI inhibitor ocular toxicities and management. Further studies are needed to assess specific ICI agents and the accuracy and completeness of updated versions of LLMs.

Clinical Implications of Ultrasound-Based Morphology in Choroidal Melanoma.

OBJECTIVE: To describe the frequency of different B-scan morphologies and their association with clinical features and outcomes. DESIGN: Cohort study of patients enrolled in the Prospective Ocular Tumor Study from January 2000 to January 2024 initially seen at Mayo Clinic Rochester. PARTICIPANTS: Consecutive inclusion of patients with posterior uveal melanoma. METHODS: B-scan ultrasounds were performed by an experienced technician and treatment modalities were implemented by the attending oncologist. MAIN OUTCOMES AND MEASURES: Tumors were classified by shape as observed on B-scan. Enucleation-, metastasis, -and overall survival (EFS, MTS, and OS) rates were analyzed using Cox-regression models and Kaplan-Meier curves. RESULTS: Among 1021 cases of uveal melanoma, 739 (72.4%) were dome-shaped, 119 (11.7%) mushroom-shaped, 85 (8.3%) multilobulated, 77 (7.5%) minimally elevated, and 1 (0.1%) diffuse. The median follow-up duration after presentation was 37 months (3-324). The macula was more commonly involved in minimally elevated tumors compared to the other groups (63.6% vs. 13.8%, p<0.001). These tumors also exhibited a larger proportion of high internal reflectivity (13% vs. 2.3%, p<0.001). The multilobulated group exhibited a significantly larger diameter at baseline (median 15 mm, IQR 6.1-30), whereas the mushroom-shaped group had greater thickness (median 7.9 mm, IQR 1.3 - 17.3) compared to the other groups (p<0.001). EFS at 36 months was lower for mushroom-shaped [60.1% (95% CI, 47.7-70.3)] and multilobulated tumors [71.1% (95% CI, 55.7-82.7)]. At 36 months, multilobulated tumors had lower MFS [68.2% (95%, CI 55-78.2)] and OS [73.9% (95%, CI 59.9-83.64)]. On multivariate analysis adjusted for tumor thickness and diameter, multilobulated melanomas had a higher risk of metastasis (HR 2.08, p=0.003) and death (HR 2.38, p<0.001). CONCLUSION: Choroidal melanoma configuration by B-scan can vary from minimally elevated to dome-shaped to mushroom-shaped or multilobulated. Independent of presenting tumor size, multilobulated morphology was identified as a predictor for metastasis and death. Multilobulated melanomas, identified by a readily available tool such as ultrasonography, warrant a vigilant approach and close monitoring due to a potential association with poor prognosis.

FACTORS ASSOCIATED WITH DELAYED DIAGNOSIS IN PATIENTS WITH PRIMARY VITREORETINAL LYMPHOMA.

PURPOSE: To identify demographic and clinical factors associated with delayed diagnosis in patients with primary vitreoretinal lymphoma (VRL). METHODS: Retrospective, tertiary referral center-based cohort study of all patients at Mayo Clinic in Rochester, Minnesota, with a biopsy-proven diagnosis of VRL from January 1, 2000, to October 31, 2022. RESULTS: There were 87 patients included during the 22-year study period with 73 patients (83.9%) diagnosed with VRL upon initial evaluation at the tertiary center, with the other 14 patients (16.1%) diagnosed later. The median referral time was 4.8 months (range: 0-113 months). Patients who received an initial diagnosis of inflammatory uveitis or another incorrect diagnosis elsewhere were referred slower than those initially diagnosed with VRL (P = 0.04). The most common incorrect initial diagnosis from an outside institution was inflammatory uveitis (n = 35, 40.2%). When patients were split into four groups based on referral time, prior use of corticosteroids was associated with a significant delay in referral (P = 0.03). CONCLUSION: Diagnosing VRL continues to be challenging, as months-long delays from initial evaluation to expert referral center evaluation are common. Prior use of corticosteroids was associated with delay in diagnosis and referral time, underscoring the need to increase awareness regarding differences between VRL and uveitis.

Population-Based Incidence and Clinical Characteristics of Ocular Adnexal Tumors in Olmsted County, Minnesota.

PURPOSE: To determine the incidence and clinical characteristics of ocular adnexaltumors in Olmsted County, Minnesota. METHODS: Retrospective population-based cohort study of all patients residing in Olmsted County, Minnesota diagnosed with any ocular tumor from January 1, 2006, to December 31, 2015. The medical records of all patients with an incident diagnosis of any ocular adnexal tumor were reviewed using the Rochester Epidemiology Project medical record linkage system for patient demographics, tumor type, and histopathologic confirmation. Incidence rates were calculated per 100,000 person-years. Poisson regression analysis was used to analyze changes in incidence over time. RESULTS: There were 717 patients diagnosed with ocular adnexal tumors during the 10-year study period, yielding an age- and sex-adjusted incidence rate of 59.7 per 100,000 (95% CI 55.4 to 64.0, p < 0.05) per year. In total, 764 tumors were diagnosed. Most tumors were eyelid lesions (N = 756, 99.0%), which were mostly benign (N = 512, 67.8%) with epidermal inclusion cysts (N = 275, 36.0%), hidrocystoma (N = 70, 9.2%), and eyelid sebaceous cysts (N = 46, 6.1%) accounting for the majority. Malignant eyelid lesions (N = 244, 31.9%) were relatively common with basal cell carcinoma (N = 184, 24.1%) and squamous cell carcinoma (N = 49, 6.4%) having the highest frequencies. Orbital tumors (N = 8, 1.0%) were infrequent. Of the orbital tumors, the most common was lacrimal gland adenoidcystic carcinoma (N = 2, 25.0%). CONCLUSIONS: In a population-based setting, most ocular adnexal tumors were benign eyelid lesions. Understanding the epidemiology of ocular adnexal tumors is important to aid providers in diagnosing and facilitating appropriate referrals of potentially vision- and life-threatening malignancies.

Accuracy and Completeness of Large Language Models About Antibody-Drug Conjugates and Associated Ocular Adverse Effects.

PURPOSE: The purpose of this study was to assess the accuracy and completeness of 3 large language models (LLMs) to generate information about antibody-drug conjugate (ADC)-associated ocular toxicities. METHODS: There were 22 questions about ADCs, tisotumab vedotin, and mirvetuximab soravtansine that were developed and input into ChatGPT 4.0, Bard, and LLaMa. Answers were rated by 4 ocular toxicity experts using standardized 6-point Likert scales on accuracy and completeness. ANOVA tests were conducted for comparison between the 3 subgroups, followed by pairwise t-tests. Interrater variability was assessed with Fleiss kappa tests. RESULTS: The mean accuracy score was 4.62 (SD 0.89) for ChatGPT, 4.77 (SD 0.90) for Bard, and 4.41 (SD 1.09) for LLaMA. Both ChatGPT (P = 0.03) and Bard (P = 0.003) scored significantly better for accuracy when compared with LLaMA. The mean completeness score was 4.43 (SD 0.91) for ChatGPT, 4.57 (SD 0.93) for Bard, and 4.42 (SD 0.99) for LLaMA. There were no significant differences in completeness scores between groups. Fleiss kappa assessment for interrater variability was good (0.74) for accuracy and fair (0.31) for completeness. CONCLUSIONS: All 3 LLMs had relatively high accuracy and completeness ratings, showing LLMs are able to provide sufficient answers for niche topics of ophthalmology. Our results indicate that ChatGPT and Bard may be slightly better at providing more accurate answers than LLaMA. As further research and treatment plans are developed for ADC-associated ocular toxicities, these LLMs should be reassessed to see if they provide complete and accurate answers that remain in line with current medical knowledge.

Optical coherence tomography features in vitreoretinal lymphoma compared with non-infectious uveitis.

BACKGROUND: Vitreoretinal lymphoma (VRL) is a rare intraocular malignancy that poses a diagnostic challenge due to the non-specific clinical presentation that resembles uveitis. The use of spectral domain optical coherence tomography (SD-OCT) has emerged as a valuable imaging tool to characterize VRL. Therefore, we sought to determine the specific OCT features in VRL compared to the uveitides. METHODS: Retrospective chart review of patients who were seen at Mayo Clinic from January 1, 2010 through December 31, 2022. The medical records and SD-OCT images at time of initial presentation were reviewed in patients with biopsy-proven VRL, intermediate uveitis, or biopsy-confirmed sarcoid posterior uveitis. Patients with VRL or similar uveitides including intermediate uveitis or sarcoid posterior uveitis were included. RESULTS: There were 95 eyes of 56 patients in the VRL group and 86 eyes of 45 patients in the uveitis group, of whom 15 (33.3%) were diagnosed with intermediate uveitis and 30 (66.7%) with sarcoid chorioretinitis. The SD-OCT features more commonly seen at initial presentation in VRL patients (vs. uveitis) included preretinal deposits (31.6% vs. 9.3%, p = 0.002), intraretinal infiltrates (34% vs. 3.5%, p < 0.001), inner retinal hyperreflective spots (15.8% vs. 0%, p < 0.001), outer retinal atrophy (22.1% vs. 2.3%, p < 0.001), subretinal focal deposits (21.1% vs. 4.7%, p = 0.001), retinal pigmented epithelium (RPE) changes (49.5% vs. 3.5%, p < 0.001), and sub-RPE deposits (34.7% vs. 0%, p < 0.001). Features more frequently seen in uveitis included epiretinal membrane (ERM) (82.6% vs. 44.2%, p < 0.001), central macular thickening (95.3% vs. 51.6%, p < 0.001), cystoid macular edema (36% vs. 11.7%, p < 0.001), subretinal fluid (16.3% vs 6.4%, p = 0.04), and subfoveal fluid (16.3% vs. 3.2%, p = 0.003). Multivariate regression analysis controlling for age and sex showed absence of ERM (OR 0.14 [0.04,0.41], p < 0.001) and absence of central macular thickening (OR 0.03 [0,0.15], p = 0.02) were associated with VRL as opposed to uveitis. CONCLUSION: OCT features most predictive of VRL (vs. uveitis) included absence of ERM and central macular thickening.

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.

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 Responses to Retina Questions from Either Experts, Expert-Edited Large Language Models, or Expert-Edited Large Language Models Alone.

Objective: To assess the quality, empathy, and safety of expert edited large language model (LLM), human expert created, and LLM responses to common retina patient questions. Design: Randomized, masked multicenter study. Participants: Twenty-one common retina patient questions were randomly assigned among 13 retina specialists. Methods: Each expert created a response (Expert) and then edited a LLM (ChatGPT-4)-generated response to that question (Expert + artificial intelligence [AI]), timing themselves for both tasks. Five LLMs (ChatGPT-3.5, ChatGPT-4, Claude 2, Bing, and Bard) also generated responses to each question. The original question along with anonymized and randomized Expert + AI, Expert, and LLM responses were evaluated by the other experts who did not write an expert response to the question. Evaluators judged quality and empathy (very poor, poor, acceptable, good, or very good) along with safety metrics (incorrect information, likelihood to cause harm, extent of harm, and missing content). Main Outcome: Mean quality and empathy score, proportion of responses with incorrect information, likelihood to cause harm, extent of harm, and missing content for each response type. Results: There were 4008 total grades collected (2608 for quality and empathy; 1400 for safety metrics), with significant differences in both quality and empathy (P < 0.001, P < 0.001) between LLM, Expert and Expert + AI groups. For quality, Expert + AI (3.86 ± 0.85) performed the best overall while GPT-3.5 (3.75 ± 0.79) was the top performing LLM. For empathy, GPT-3.5 (3.75 ± 0.69) had the highest mean score followed by Expert + AI (3.73 ± 0.63). By mean score, Expert placed 4 out of 7 for quality and 6 out of 7 for empathy. For both quality (P < 0.001) and empathy (P < 0.001), expert-edited LLM responses performed better than expert-created responses. There were time savings for an expert-edited LLM response versus expert-created response (P = 0.02). ChatGPT-4 performed similar to Expert for inappropriate content (P = 0.35), missing content (P = 0.001), extent of possible harm (P = 0.356), and likelihood of possible harm (P = 0.129). Conclusions: In this randomized, masked, multicenter study, LLM responses were comparable with experts in terms of quality, empathy, and safety metrics, warranting further exploration of their potential benefits in clinical settings. Financial Disclosures: Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of the article.

Trends in the Clinical Presentation of Uveal Melanoma During the COVID-19 Pandemic.

BACKGROUND AND OBJECTIVE: The objective was to evaluate factors associated with clinical presentation of uveal melanoma (UM) during the initial two years of the coronavirus disease 2019 pandemic. PATIENTS AND METHODS: This was a multi-site, retrospective cohort study of patients treated for uveal melanoma during the first (early) and second (late) year of the pandemic compared with the year prior (control). RESULTS: A total of 48, 67, and 75 patients were in the control, early, and late cohorts, respectively. The early cohort had a higher frequency of large tumors (control: 29.2%, early: 40.3%, late: 29.3%; P < 0.001) at presentation. Both the early and late cohorts had higher rates of enucleation (control: 8.33%, early: 20.9%, late: 18.67%; P ≤ 0.0338) compared to the control cohort. CONCLUSIONS: While there was an increase in large tumors along with a rise in enucleation during the first year of the pandemic, enucleation rates remained elevated even while tumor sizes normalized. [Ophthalmic Surg Lasers Imaging Retina 2024;55:278-284.].

Intraocular sarcoid-like reaction in patients with chronic lymphocytic leukemia.

OBJECTIVE: To investigate the frequency and clinical features of intraocular paraneoplastic sarcoid-like reaction (SLR) in patients with chronic lymphocytic leukemia (CLL). METHODS: Retrospective review of patients with CLL from January 1, 1980, to December 31, 2020. Eye examinations were searched for 22 keywords suggestive of SLR, and charts were manually reviewed. RESULTS: Of 4209 unique patients with CLL, 1021 (24%) had at least 1 eye examination on record, and 324 (8%) had 1 or more keyword eye examination findings. After manual review, 12 patients (<1%) were identified as having probable SLR with characteristic features (n = 7), possible but not classic (n = 1), or suspect but less likely (n = 4). All patients (n = 8) with probable or possible SLR were White, and half (n = 4) were male. Intraocular SLR was diagnosed a mean of 49.7 months after the CLL diagnosis (n = 7) or preceded the CLL diagnosis by 1 month (n = 1). Involvement was bilateral in 5 patients, with 13 total affected eyes and mean presenting Snellen visual acuity of 20/50. Common characteristic features on initial examination included vitreous cell (n = 13), anterior-chamber cell (n = 10), keratic precipitates (n = 9), posterior synechiae (n = 6), chorioretinal lesions (n = 5), and vitreous haze (n = 5). Treatment included topical corticosteroids alone (n = 5), with sub-Tenon corticosteroids (n = 1), or with steroid-sparing immunosuppressive agents (n = 1) or oral corticosteroids alone (n = 1). After a mean follow-up of 19.8 months, final mean visual acuity was 20/30. CONCLUSION: Intraocular SLRs affect fewer than 1% of patients with CLL. SLR should be on the differential diagnosis list for any CLL patient with ocular complaints, and most patients can be managed successfully with corticosteroids.

Quantitative Biomarkers Derived from a Novel, Contrast-Free Ultrasound, High-Definition Microvessel Imaging for Differentiating Choroidal Tumors.

Angiogenesis has an essential role in the de novo evolution of choroidal melanoma as well as choroidal nevus transformation into melanoma. Differentiating early-stage melanoma from nevus is of high clinical importance; thus, imaging techniques that provide objective information regarding tumor microvasculature structures could aid accurate early detection. Herein, we investigated the feasibility of quantitative high-definition microvessel imaging (qHDMI) for differentiation of choroidal tumors in humans. This new ultrasound-based technique encompasses a series of morphological filtering and vessel enhancement techniques, enabling the visualization of tumor microvessels as small as 150 microns and extracting vessel morphological features as new tumor biomarkers. Distributional differences between the malignant melanomas and benign nevi were tested on 37 patients with choroidal tumors using a non-parametric Wilcoxon rank-sum test, and statistical significance was declared for biomarkers with p-values < 0.05. The ocular oncology diagnosis was choroidal melanoma (malignant) in 21 and choroidal nevus (benign) in 15 patients. The mean thickness of benign and malignant masses was 1.70 ± 0.40 mm and 3.81 ± 2.63 mm, respectively. Six HDMI biomarkers, including number of vessel segments (p = 0.003), number of branch points (p = 0.003), vessel density (p = 0.03), maximum tortuosity (p = 0.001), microvessel fractal dimension (p = 0.002), and maximum diameter (p = 0.003) exhibited significant distributional differences between the two groups. Contrast-free HDMI provided noninvasive imaging and quantification of microvessels of choroidal tumors. The results of this pilot study indicate the potential use of qHDMI as a complementary tool for characterization of small ocular tumors and early detection of choroidal melanoma.

Multimodal imaging risk factors predictive of small choroidal melanocytic lesion growth to melanoma: An educational study and pictorial guide.

BACKGROUND: Risk factors for small choroidal melanocytic lesion growth to melanoma have been redefined using multimodal imaging. We explored provider ability to recognize risk factors for small choroidal melanocytic lesion growth to melanoma before and after image-based education and with and without multimodal imaging. METHODS: Providers were invited to participate in a survey assessing ability to identify risk factors for small choroidal melanocytic lesion growth to melanoma using either fundus imaging or multimodal imaging. Risk factors included thickness >2 mm on ultrasonography, subretinal fluid on optical coherence tomography, presence of orange pigment by autofluorescence, acoustic hollowness by ultrasonography, and diameter >5 mm by fundus imaging. Performance was assessed before and after reviewing an educational PowerPoint providing pictorial examples of risk factors. Comparison between groups was conducted using two-tailed Fisher's exact test. RESULTS: Thirty and 26 providers completed the pre-education and post-education assessments, respectively. Post-education participants were more accurate within ±1 risk factor for lesions with zero risk factors (77% vs. 100%, p = 0.01) or two risk factors (79% vs. 91%, p = 0.03). Following education, participants presented with multimodal imaging more often correctly identified lesions with four (12% vs. 42%, p = 0.03) or five (4% vs. 39%, p = 0.004) risk factors, demonstrated lower mean level of concern for lesions with zero risk factors (2.0 vs. 1.4, p < 0.001), and expressed higher level of concern for lesions with 5 risk factors (2.4 vs. 3.6, p < 0.001). CONCLUSION: Use of multimodal imaging may be more beneficial than education itself to improve accuracy of risk factor identification for small choroidal melanocytic lesions.

Central subfield thickness predicts visual acuity outcomes in plaque-irradiated eyes with choroidal melanoma.

OBJECTIVE: To determine the association between pre-operative central subfield thickness (CST) and post-radiotherapy visual acuity (VA), cystoid macular edema (CME), and intravitreal anti-vascular endothelial growth factor (VEGF) requirement. DESIGN: Single-center retrospective study. PARTICIPANTS: Patients with plaque-irradiated extramacular choroidal melanoma treated between 11/11/2011 and 4/30/2021. Pre-operative CST difference between the affected and unaffected eye was used. Kaplan-Meier analysis and hazard ratios were calculated. RESULTS: Of 85 patients, pre-operative CST was greater in the melanoma-affected eye (vs. fellow eye) by mean of 20.4 µm (median 14.0, range - 60.0-182.0). Greater CST at presentation (vs. fellow eye) was associated with larger tumor diameter (p = 0.02), greater tumor thickness (p < 0.001), and more frequent tumor-related Bruch's membrane rupture (p = 0.006). On univariate analysis of outcome data, greater CST at presentation (vs. fellow eye) was associated with higher 5-year risk (1.09 [1.02-1.17], p = 0.02) of VA 20/200 or worse and increased (1.10 [1.01-1.20], p = 0.03) likelihood for anti-VEGF injections after plaque irradiation. There was no significant association with CME. The association between CST and VA outcome remained significant on multivariate analysis accounting for impact of tumor thickness and radiation dose to optic disc, while tumor distance to fovea was the only significant factor on multivariate analysis for anti-VEGF injections. CONCLUSION: Greater CST at presentation (vs. fellow eye) was associated with worse VA outcome following plaque radiotherapy for choroidal melanoma. Large-sized tumors may contribute to a higher intraocular VEGF burden, potentially leading to greater preoperative CST, which correlates with poor VA outcome post-plaque radiotherapy.

Predictive Value of the International Classification of Diseases, 9th Revision Codes for Identifying Ocular Oncology Diagnoses.

Introduction: The aim of this study was to determine the predictive value of International Classification of Diseases, 9th Revision (ICD-9) billing codes for identifying ocular oncology diagnoses. Methods: Population-based retrospective cohort study of all Olmsted County, Minnesota residents with any ocular neoplasm-related ICD-9 code from January 1, 2006 to October 1, 2015. All medical records were reviewed for confirmation of ocular neoplasm. Diagnoses with ≥5 cases confirmed via a medical record review were compared to corresponding ICD-9 codes. Main outcome measures included positive predictive value (PPV), negative predictive value (NPV), sensitivity, and specificity of ICD-9 codes. Results: Among 3,932 subjects with ≥1 ocular neoplasm-related ICD-9 code, 21 diagnoses met study criteria. The most frequent intraocular, extraocular/orbital, and ocular surface diagnoses were choroidal nevus (n = 824), epidermal inclusion cyst (n = 263), and conjunctival nevus (n = 74), respectively. PPVs ranged from 1.2% to 73.8%, NPVs from 96.9% to 100%, sensitivity from 0% to 100%, and specificity from 85.7% to 100%. Among malignant neoplasms, PPV ranged from 0% to 73.8%: ocular surface squamous neoplasia (PPV: 0%), choroidal melanoma (PPV: 25.0%), eyelid squamous cell carcinoma (PPV: 46.7%), and eyelid basal cell carcinoma (PPV: 73.8%). Among benign neoplasms, PPV ranged from 1.2% (dermoid cyst) to 61.6% (choroidal nevus). Conclusion: There was a wide variation in a predictive value of ocular neoplasm-related ICD-9 billing codes, which suggests that ocular oncology-related claims data alone may overestimate the true number of ocular oncology diagnoses.

Ocular Whipple Disease: Cases Diagnosed Over Four Decades.

PURPOSE: To describe ocular involvement in subjects with Whipple's disease (WD). METHODS: Retrospective review of documented WD cases seen at Mayo Clinic between 1980 and 2021 with ocular involvement. RESULTS: Of 217 patients with WD, 30 had eye exams and four (two female, median age 58.5 years) had ocular involvement. Findings included anterior/intermediate uveitis (n = 2), intermediate uveitis and phlebitis (n = 1), and chorioretinitis with vitritis (n = 1). The diagnosis was confirmed by vitreous biopsy in three of four cases. In two cases, WD diagnosis was unconfirmed prior to the ocular diagnosis. Systemic manifestations included gastrointestinal symptoms in all patients, synovitis (n = 3), weight loss (n = 2), and pericarditis (n = 1). Mean time from onset of ocular symptoms to ocular diagnosis was 11 months (range 2-28 months). Prior systemic symptoms were present as long as 3 years. CONCLUSIONS: WD is uncommon and ocular involvement is even more rare. However, WD should be considered in the differential for all patients with chronic recalcitrant uveitis, especially in the setting of polyarthralgias and/or gastrointestinal symptoms. Vitreous biopsy is a reliable method to diagnose ocular WD.Abbreviations and Acronyms: Whipple's disease (WD), intestinal lipodystrophy (IL), polymerase-chain reaction (PCR), periodic acid-Schiff (PAS), trimethoprim/sulfamethoxazole (TMP/SMX).