DNA Methylation Profiling Identifies Distinct Clusters in Angiosarcomas

Abstract Purpose: DNA methylation profiling has previously uncovered biologically and clinically meaningful subgroups within many tumor types, but was not yet performed in angiosarcoma. Angiosarcoma is a rare sarcoma with very heterogeneous clinical presentations, which may be based on differences in biological background. In this exploratory study, DNA methylation profiling of 36 primary angiosarcoma samples from visceral, deep soft tissue, radiation-induced, and UV-induced localizations was performed. Experimental Design: Primary angiosarcoma formalin-fixed paraffin-embedded samples from visceral, soft tissue, radiation-induced, and UV-induced origin were collected from a nationwide search for angiosarcoma in the Netherlands. DNA was extracted for methylation profiling with the Illumina Infinium MethylationEPIC array. Quality control assessment and unsupervised hierarchical clustering were performed. Copy-number profiles were generated and analyzed for chromosomal stability. Clinical data were obtained from the Netherlands Cancer Registry. Results: DNA methylation profiling by unsupervised hierarchical clustering of 36 angiosarcoma samples (6 visceral, 5 soft tissue, 14 radiation-induced, 11 UV-induced) revealed two main clusters (A and B), which were divided into four subclusters. The clusters largely corresponded with clinical subtypes, showing enrichment of UV-induced cases in cluster A1 and radiation-induced cases in cluster A2. Visceral and soft tissue cases almost exclusively fell into cluster B. Cluster A showed significantly increased chromosomal instability and better overall survival (22 vs. 6 months, P = 0.046) compared with cluster B. Conclusions: In this novel methylation profiling study, we demonstrated for the first time four different angiosarcoma clusters. These clusters correlated with clinical subtype, overall survival, and chromosomal stability.