Abstract
Purpose :
Uveal Melanoma (UM) is the most common intraocular tumor in adults. Despite effective local treatments, 50% of patients develop metastasis. Consequently, it is necessary to find better ways to determine prognosis, as well as to discover the causative agents behind the transformation of UM. Epigenetic changes in cancer, especially DNA methylation, are one of the most promising targets for the development of powerful diagnostic, prognostic, and predictive biomarkers of disease progression. However, very few studies have looked into methylation events in UM. The purpose of this study is to determine the methylation events that predict poor prognosis in UM.
Methods :
Matched clinical, genetic and methylation data for 80 UM cases was obtained from The Cancer Genome Atlas database. Each case was classified according to its GNAQ/GNA11 and BAP1 mutation status, as well as by clinical features. Methylation β values were sorted to determine the most differentially methylated loci, and the top 3000 sites were selected for further analysis. Hierarchical clustering was performed for these sites using Minfi to separate the cases based on methylation patterns. The top ten differentially methylated loci were then used for correlative analysis of survival, BAP1 mutation status, and clinical features.
Results :
Hierarchical clustering revealed two distinct groups for UM based on methylation events. These classifications were found to effectively separate high and low risk UMs. Kaplan-Meier tests revealed that survival differed very significantly between the high and low risk groups (P < 0.001). Mann-Whitney tests revealed BAP1 mutations were significantly related to the high risk group for the top 10 loci (P = 0.001-0.038), with no BAP1 mutations in the “low risk” group. There were no significant differences in the groups in terms of GNAQ/GNA11 mutations or sex of the patient.
Conclusions :
Hierarchical clustering of methylation values is a promising approach both as an marker of prognosis in UM and as an opportunity to gain insight into potentially reversible changes that drive UM towards a dangerous phenotype. Clustering of methylation provides an intuitive method to help determine the outcome of patients with UM, as it clusters based on prognostically relevant factors and does not differ in features that are irrelevant for prognosis (GNAQ/GNA11 mutations, sex). Further work should be done to elucidate the causative factors behind the observed changes in methylation.
This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.