Integrated differential DNA methylation and gene expression of formalin-fixed paraffin-embedded uveal melanoma specimens identifies genes associated with early metastasis and poor prognosis.

PURPOSE Uveal melanoma (UM) is an aggressive malignancy, in which nearly 50% of the patients die from metastatic disease. Aberrant DNA methylation is recognized as an important epigenomic event in carcinogenesis. Formalin-fixed paraffin-embedded (FFPE) samples represent a valuable source of tumor tissue, and recent technology has enabled the use of these samples in genome-wide DNA methylation analyses. Our aim was to investigate differential DNA methylation in relation to histopathological classification and survival data. In addition we sought to identify aberrant DNA methylation of genes that could be associated with metastatic disease and poor survival. METHODS FFPE samples from UM patients (n = 23) who underwent enucleation of the eye in the period 1976-1989 were included. DNA methylation was assessed using the Illumina Infinium HumanMethylation450 array and coupled to histopathological data, Cancer Registry of Norway- (registered UM metastasis) and Norwegian Cause of Death Registry- (time and cause of death) data. Differential DNA methylation patterns contrasting histological classification, survival data and clustering properties were investigated. Survival groups were defined as "Early metastasis" (metastases and death within 2-5 years after enucleation, n = 8), "Late metastasis" (metastases and death within 9-21 years after enucleation, n = 7) and "No metastasis" (no detected metastases ≥18 years after enucleation, n = 8). A subset of samples were selected based on preliminary multi-dimensional scaling (MDS) plots, histopathological classification, chromosome 3 status, survival status and clustering properties; "Subset Early metastasis" (n = 4) vs "Subset No metastasis" (n = 4). Bioinformatics analyses were conducted in the R statistical software. Differentially methylated positions (DMPs) and differentially methylated regions (DMRs) in various comparisons were assessed. Gene expression of relevant subgroups was determined by microarray analysis and quantitative reverse-transcription polymerase chain reaction (qRT-PCR). RESULTS DNA methylation analyses identified 2 clusters that separated the samples according to chromosome 3 status. Cluster 1 consisted of samples (n = 5) with chromosome 3 disomy (D3), while Cluster 2 was comprised of samples (n = 15) with chromosome 3 monosomy (M3). 1212 DMRs and 9386 DMPs were identified in M3 vs D3. No clear clusters were formed based on our predefined survival groups ("Early", "Late", "No") nor histopathological classification (Epithelioid, Mixed, Spindle). We identified significant changes in DNA methylation (beta FC ≥ 0.2, adjusted p < 0.05) between two sample subsets (n = 8). "Subset Early metastasis" (n = 4) vs "Subset No metastasis" (n = 4) identified 348 DMPs and 36 DMRs, and their differential gene expression by microarray showed that 14 DMPs and 2 DMRs corresponded to changes in gene expression (FC ≥ 1.5, p < 0.05). RNF13, ZNF217 and HYAL1 were hypermethylated and downregulated in "Subset Early metastasis" vs "Subset No metastasis" and could be potential tumor suppressors. TMEM200C, RGS10, ADAM12 and PAM were hypomethylated and upregulated in "Subset Early metastasis vs "Subset No metastasis" and could be potential oncogenes and thus markers of early metastasis and poor prognosis in UM. CONCLUSIONS DNA methylation profiling showed differential clustering of samples according to chromosome 3 status: Cluster 1 (D3) and Cluster 2 (M3). Integrated differential DNA methylation and gene expression of two subsets of samples identified genes associated with early metastasis and poor prognosis. RNF13, ZNF217 and HYAL1 are hypermethylated and candidate tumor suppressors, while TMEM200C, RGS10, ADAM12 and PAM are hypomethylated and candidate oncogenes linked to early metastasis. UM FFPE samples represent a valuable source for methylome studies and enable long-time follow-up.