PURPOSE: Validated and accurate prognostic testing is critical for precision medicine in uveal melanoma (UM). Our aims were to (1) prospectively validate an integrated prognostic classifier combining a 15-gene expression profile (15-GEP) and PRAME RNA expression and (2) identify clinical variables that enhance the prognostic accuracy of the 15-GEP/ PRAME classifier. MATERIALS AND METHODS: This study included 1,577 patients with UM of the choroid and/or ciliary body who were enrolled in the Collaborative Ocular Oncology Group Study Number 2 (COOG2) and prospectively monitored across 26 North American centers. Test results for 15-GEP (class 1 or class 2) and PRAME expression status (negative or positive) were available for all patients. The primary end point was metastasis-free survival (MFS). RESULTS: 15-GEP was class 1 in 1,082 (68.6%) and class 2 in 495 (31.4%) patients. PRAME status was negative in 1,106 (70.1%) and positive in 471 (29.9%) patients. Five-year MFS was 95.6% (95% CI, 93.9 to 97.4) for class 1/ PRAME (–), 80.6% (95% CI, 73.9 to 87.9) for class 1/ PRAME (+), 58.3% (95% CI, 51.1 to 66.4) for class 2/ PRAME (–), and 44.8% (95% CI, 37.9 to 52.8) for class 2/ PRAME (+). By multivariable Cox proportional hazards analysis, 15-GEP was the most important independent predictor of MFS (hazard ratio [HR], 5.95 [95% CI, 4.43 to 7.99]; P < .001), followed by PRAME status (HR, 1.82 [95% CI, 1.42 to 2.33]; P < .001). The only clinical variable demonstrating additional prognostic value was tumor diameter. CONCLUSION: In the largest prospective multicenter prognostic biomarker study performed to date in UM to our knowledge, the COOG2 study validated the superior prognostic accuracy of the integrated 15-GEP/ PRAME classifier over 15-GEP alone and clinical prognostic variables. Tumor diameter was found to be the only clinical variable to provide additional prognostic information. This prognostic classifier provides an advanced resource for risk-adjusted metastatic surveillance and adjuvant trial stratification in patients with UM.

Background: Wide-field imaging plays an increasingly important role in ocular oncology clinics. The purpose of this review is to describe the commonly used wide-field imaging devices and review conditions seen in ocular oncology clinic that underwent wide-field imaging as part of the multimodal evaluation.Summary of review: Wide-field or wide-angle imaging is defined as greater than 50° field of view. Modern devices can reach far beyond this reporting fields of view up to 267°, when utilizing montage features, with increasingly impressive resolution. Wide-field imaging modalities include fundus photography, fluorescein angiography (FA), fundus autofluorescence (FAF), indocyanine angiography (ICG), spectral domain optical coherence tomography (SD-OCT), and recently wide-field OCT Angiography (OCTA). These imaging modalities are increasingly prevalent in practice. The wide-field systems include laser, optical, and lens based systems that are contact or non-contact lens systems each with its own benefits and drawbacks. The purpose of this review is to discuss commonly used wide-field imaging modalities for retinal and choroidal tumors and demonstrate the use of various widefield imaging modalities in select ocular oncology cases.Conclusions: Clinical examination remains the gold standard for the evaluation of choroidal and retinal tumors. Wide-field imaging plays an important role in ocular oncology for initial documentation, surgical planning, determining the relationship of the tumor to adjacent ocular structures, following tumor size after treatment, and monitoring for recurrence.

As shown in various oncologic conditions, patients from disadvantaged backgrounds often face barriers to care from poor follow-up to delayed diagnosis to difficulty accessing appropriate treatment.^1-4 Overall, research has shown that there are many clinically important differences between patients with uveal melanoma (UM) of different races, ethnicities, and socioeconomic status (SES).⁵ The age-adjusted incidence of UM differs by race (0.31 in black, 0.38 in Asian, 1.67 in Hispanic, and 6.02 in non-Hispanic white populations),^6-10 while the prevalence of choroidal nevi, which have a known association with UM development,¹¹ is 0.6% in black, 2.7% in Hispanic, and 5.6% in non-Hispanic white populations.¹² One study suggested decreased mortality in Hispanic compared with white patients (odds ratio [OR], 0.6),¹³ while another study found no statistically significant difference in metastasis, survival, or mortality between ethnicities.¹ Histopathologic studies have reported that the clinical and genetic course of nonwhite patients with UM may be different than in white patients, with Asian Indian individuals presenting younger,¹⁴ Mexican-Mestizo patients showing a previously unreported polymorphism,¹⁵ and Hispanic patients presenting with unique symptoms and broader-based tumors.² Various case reports have shown that UM occurs in all races.^2,16-20

While color fundus photos are used in routine clinical practice to diagnose ophthalmic conditions, evidence suggests that ocular imaging contains valuable information regarding the systemic health features of patients. These features can be identified through computer vision techniques including deep learning (DL) artificial intelligence (AI) models. We aim to construct a DL model that can predict systemic features from fundus images and to determine the optimal method of model construction for this task. Data were collected from a cohort of patients undergoing diabetic retinopathy screening between March 2020 and March 2021. Two models were created for each of 12 systemic health features based on the DenseNet201 architecture: one utilizing transfer learning with images from ImageNet and another from 35,126 fundus images. Here, 1277 fundus images were used to train the AI models. Area under the receiver operating characteristics curve (AUROC) scores were used to compare the model performance. Models utilizing the ImageNet transfer learning data were superior to those using retinal images for transfer learning (mean AUROC 0.78 vs. 0.65, p-value < 0.001). Models using ImageNet pretraining were able to predict systemic features including ethnicity (AUROC 0.93), age > 70 (AUROC 0.90), gender (AUROC 0.85), ACE inhibitor (AUROC 0.82), and ARB medication use (AUROC 0.78). We conclude that fundus images contain valuable information about the systemic characteristics of a patient. To optimize DL model performance, we recommend that even domain specific models consider using transfer learning from more generalized image sets to improve accuracy.

Uveitis is a chronic disease that often requires long-term medical treatment. Vision-threatening complications may occur because of inadequate suppression of inflammation or as an adverse effect of the treatment itself. Medications to treat uveitis can be delivered by topical drops, local injections (subconjunctival, posterior sub-Tenon, or intravitreal), or systemically. Each of these administration routes has drawbacks and potential complications.^1-10