Ort: MFPL, Dr.-Bohr-Gasse 9, 1030 Wien, Seminarraum 2. Stock (2.202)

Ewing sarcoma is a highly aggressive malignant tumor that occurs in the bone or soft tissues and mainly affects children, adolescents, and young adults. It is driven by a fusion gene, most frequently EWS-FLI1, which epigenetically reprograms the cell at specific regions and creates a highly unique epigenetic signature. The strength of this signature varies between patients and is linked to clinical data. The primary tumor is frequently located in the bone, and thus biopsies are often invasive and cannot be performed every time the patient presents at the clinic. Liquid biopsies, the analysis of cell-free DNA (cfDNA) or RNA in body liquids such as blood, constitute a novel, promising way of obtaining higher-resolution data and can thus allow for early detection of relapse and treatment response. Previous studies of liquid biopsies of Ewing sarcoma patients have focused on the detection of the fusion gene. This allows for the quantification of tumor content, but does not inform about tumor heterogeneity, an emerging theme whose importance is being increasingly recognized in the field. In the present study, I utilized the recently established property of cfDNA as a nucleosome footprint to show, for the first time, that the characteristic epigenetic signature of Ewing sarcoma, previously reported for cell lines and tumor tissue, can be recapitulated in whole-genome sequencing data of liquid biopsy samples. I further show that the signal, a drop in coverage around Ewing-sarcoma-specific DNaseI hypersensitivity sites (EsDHSs), is confounded by biasing factors (GC content, mappability, and di- and trinucleotide frequencies). I developed a tool, LIQUORICE, that takes the particularities of cfDNA data into account and corrects for these biases. After bias correction and quantification of the signal, 26 out of 27 samples with non-zero tumor content estimations had signal intensities outside the range of controls. Signal intensities generally correlated with tumor content estimations, but varied markedly between some samples with similar tumor content. Further studies will show how much of this variation in signal strength is attributable to the underlying tumor heterogeneity, i.e., different signature strengths in the tumor. Eventually, the bias-corrected coverage around EsDHSs could serve as a biomarker for tumor presence and could inform about heterogeneity, delivering clinically and scientifically relevant information from a simple blood draw.