65 Exploring DNA Methylation Landscapes in Kinase-Driven Sarcomas

Background DNA methylation profiling (DNAmp) has proven useful for classifying sarcomas and identifying novel subtypes. Our aim is to use it to categorize and explore tumors with kinase-encoding gene rearrangements. Design We collected soft tissue tumors with alterations of kinase-encoding genes and characterized their DNAmp with Illumina MethylationEPIC beadchips. We used Heidelberg’s Sarcoma Classifier to assign each sample to a methylation class (MC). To assess the inter-sample methylome-wide similarity, we performed t-distributed stochastic neighbor embedding (t-SNE) with a wide reference cohort of known sarcoma entities. Results We collected 43 cases, with variable locations (limbs, skin, internal organs) and a balanced representation of males (23) and females (20). Patients’ ages ranged from 0 (congenital) to 72 years old. Genetic alterations involved NTRK1/3 (18), ALK (12), BRAF (7), MET (2), RET (2), and ROS1 (2). The Sarcoma Classifier assigned the samples to the following MCs: inflammatory myofibroblastic tumour (IMT, 11), MPNST-like (9), infantile fibrosarcoma (IF, 8), angiomatoid fibrous histiocytoma (AFH, 2), alveolar soft part sarcoma (ASPS, 1), dermatofibrosarcoma protuberans (1), Langerhans cell histiocytosis (1), lipoma (1), RMS-like (1), or no match (8) (Table 1). In t-SNE analysis (Figure 1), samples with either histological diagnosis or MC of IF clustered together, regardless of genetic fusion. Exceptions were two cases that localized closer to AFH, suggesting a tissue contamination by immune cells. Spindle cell tumors with TRK fusions (TRK-SCMT, mostly assigned to MPNST-like MC), both pediatric and adult, clustered together in proximity with IFs; with the exception of one with a peculiar CAPZA2::MET fusion, that behaved as an outlier. Two samples clustered with EGFR-altered spindle cell lesions: one had EGFR duplication, while the other had a LMNA::NTRK1 rearrangement and no known EGFR alterations. Most samples with histological diagnosis of IMT also formed a distinct cluster. Conclusions In conclusion, the application of DNAmp in the context of tumors with kinase-encoding gene rearrangements provided several layers of information, useful both for diagnostic purposes and to explore the correlations between different neoplastic entities.