Uveal melanoma (UM) is a rare but serious eye cancer. It has a high chance of metastatic spread, especially in advanced stages, which significantly worsens survival. In recent years, eye-sparing treatments, such as targeted radiation therapy, have become more common than enucleation (complete removal of the eye). However, these treatments are not always available to low-income patients or those who live in rural areas, where ophthalmologists or ocular melanoma experts may be limited.

Uveal melanoma (UM) is the most common intraocular malignancy in adults and is known to have a poor prognosis. Although the genetic profiling of UM is well characterized, there is emerging evidence regarding other factors that can influence the presentation and prognostication of UM. Other factors, such as social determinants of health (SDOH) and environmental factors, play an important role in contextualizing not only a patient’s presentation at diagnosis, but also the prognosis and potential management of their care. Thus, such factors are important to identify and investigate to further understand how to tailor treatment and management for UM patients.

Currently, the research behind the relationship between SDOHs, environmental factors, and UM presentation is limited. With regards to SDOHs, mainly socioeconomic burden has been investigated and reported to correlate with a worse presentation of UM. However, other factors, such as limited healthcare availability or lack of internet access, have not yet been explored. Furthermore, many environmental factors, such as water or air pollution, that have been shown to have a correlation with other types of cancer have not yet been investigated or explored with regards to UM. We hypothesize that such SDOHs and environmental factors will have a correlation with the presentation of UM.

The investigation of both SDOHs and environmental factors will allow for further knowledge and understanding of how such factors can impact the presentation and treatment of UM. Furthermore, the information revealed by this investigation will provide context that leads to a more comprehensive clinical picture of UM presentation and management. Such knowledge could influence public health initiatives and provide information for targeted screening for vulnerable populations, allowing for improved patient outcomes and more equitable high-quality care.

Melanocytes are a type of pigment-producing cell found in skin, hair, and eyes, but can form melanoma, a highly aggressive and deadly type of cancer. Over 40% of patients with advanced-stage melanoma do not respond to immune checkpoint inhibitor therapy, a highly effective treatment that leverages the body’s immune system to fight off cancer cells. The Wellstein Lab at Georgetown University is using cell-free DNA, or DNA circulating in the blood from dead or dying cells, to distinguish between patients who are or are not responding to immune checkpoint inhibitor therapy from blood draws during treatment. Although DNA from each cell in the human body is identical, specific patterns of methylation marks on DNA distinguish each type of cell. In this proposal, I aim to better improve our understanding of the DNA methylation patterns unique to melanocytes. I also seek to compare DNA methylation between melanoma tumor cells and normal melanocytes. Successful completion of the aims of this proposal will provide a deeper understanding of DNA methylation in melanocytes and melanoma and provide a way to use simple, non-invasive blood draws to monitor patient response to treatment.

Uveal melanoma (UM) is a rare but lethal cancer of the eye. Despite initial successful treatment, 50% of patients experience spread of the cancer to other organs (called metastasis). When this happens, it is almost universally fatal. Currently, there are no blood tests to predict disease severity or if the patient is at risk of disease spread. We discovered a new cancer cell that has both tumor and immune cell properties and is detected in the blood of cancer patients. We identified these cancer-immune hybrid cells in patients with different types of tumors, including cutaneous and uveal melanoma. We also published that they can be used to identify patients with metastatic disease that is undetected by whole body imaging (in pancreatic cancer). While much of our advances have been in solid tumors, we recently demonstrated that these cancer-immune hybrid cells are increased in patients with greater disease burden from their uveal melanoma. In this study, I propose to establish an assay to identify uveal melanoma patients with high risk for disease spread by extending the specificity of the tumor-immune hybrid cell and combining detection and enumeration of these cells with a cutting-edge imaging approach of the tumor. Successful completion of this study has the potential to identify and stratify the patients with highest risk for metastasis, which would impact their post-treatment surveillance and candidacy for enrollment onto clinical trials directed to patients with metastasis—and has high potential for impacting survival from uveal melanoma.

Melanomas of the conjunctiva (CM) are very deadly. Knowledge regarding the social factors, especially insurance status such as Medicaid enrollment, that can affect CM mortality is limited. We will be using a national database, called SEER-Medicaid, to investigate whether being enrolled in Medicaid affects survival rates. We will be analyzing all patients diagnosed with CM between the years 1999-2008 and categorizing them as Medicaid or not in Medicaid. We will only analyze adults between the age of 18 and 65, as children may have different Medicaid benefits and elderly individuals may also be enrolled in Medicare. We will used advanced statistics to construct five-year survival curves of CM patients and to evaluate the following variables and how they can affect risk of death: age, race, sex, marital status, duration of Medicaid enrollment, therapy, poverty level, region of the United States, tumor location, rural/urban, tumor size, county English proficiency, county education level, county % minority population, reasons for surgery or no surgery, and other non-standard therapy. Because Medicaid patients have historically been observed to have worse health-care access and survival across many cancer types, we expect that CM patients on Medicaid to also have worse outcomes and survival.

Uveal Melanoma (UM), a deadly cancer arising in the eye, is the most common eye cancer in adults. Early detection of UM is important due to the cancer’s ability to spread to the rest of the body early and because effective treatments are available to reduce its spread. Despite the availability of effective treatments, more than half of patients’ cancer spreads to the rest of the body, suggesting that UM may spread before the time of treatment. Choroidal nevi are benign tumors that are commonly seen in patients’ eyes and rarely can turn into cancer. Choroidal nevi can look like UM, which makes the diagnosis of these eye tumors challenging. Therefore, there is a need to identify and treat small UM to minimize the number of tumors that are observed and subsequently grow during the observation period. However, current screening methods face inherent limitations, particularly in regions with limited access to specialized eye cancer doctors. Machine learning (ML) is a field of study in artificial intelligence that can be used to assist in disease diagnosis. ML offers a promising approach to improve the identification and evaluation of eye tumors, thereby providing a potential tool for eye doctors both in the community and who specialize in eye cancer. Despite the significant research being done with ML in medical imaging, few studies have worked towards an ML tool for eye tumors like choroidal nevi and UM. Our previous work used ML to screen images of patients’ eyes to successfully find choroidal nevi and UM. We also used ML to evaluate images and ultrasound of patients’ eyes to assess choroidal nevi and UM for their ability to spread to the rest of the body. The objective of this proposed project is to assess the ability of ML to diagnose choroidal nevi and UM. We will use a large collection of images from the University of Illinois Chicago, which includes multiple different types of images taken of many patients with choroidal nevi and UM. We propose to develop and then improve ML tools for the screening and diagnosis of choroidal nevi and UM. We will use additional information about the patient’s medical history to help the tools perform better. We will also annotate the images before providing them to the ML tools to see if this improves the performance in screening and diagnosing choroidal nevi and UM. After identifying the best-performing ML tool in this research study, we will test this tool to perform other tasks related to the choroidal nevi and UM. We will test the ML tool’s performance in predicting the likelihood that a choroidal nevus will grow and the likelihood that a UM will spread to the rest of the body. We expect these results to provide a better understanding of how ML can be developed into a clinically useful tool to inform and guide management decisions for choroidal nevi and UM in community eye clinics and eye cancer specialist clinics to potentially save patient lives.

Career Development Award – Xiaonan Xu, MD, PhD