通过游离DNA分析追踪融合驱动型肉瘤的肿瘤演化和异质性

Background : Tumor evolution and intra-tumoral heterogeneity drive therapy resistance and relapse, yet remain poorly understood and difficult to track in real time. HEROES-AYA, a German multi-center consortium funded by the National Decade Against Cancer, leverages fusion-driven bone and soft tissue sarcomas in children, adolescents, and young adults as a model system to dissect these processes. Defined by diverse oncogenic fusions, these aggressive tumors frequently evade therapy, leading to poor outcomes. In parallel with bulk and single-cell multi-omics profiling of tumor tissue, we employ plasma liquid biopsies as a minimally invasive, serially collectable source of circulating tumor DNA (ctDNA) to map spatiotemporal tumor dynamics. Methods : 112 plasma samples from 60 patients (12 alveolar rhabdomyosarcoma, 11 synovial sarcoma, 8 myxoid liposarcoma, and 29 other fusion-driven sarcomas) were collected retro- and prospectively in several German sarcoma trials (1-15 samples per patient). Plasma from healthy donors served as control. Cell-free DNA (cfDNA) was profiled using an ultra-sensitive enzymatic methylation sequencing assay optimized for pico- to nanogram inputs and sequenced at 10-20x. A multi-layered bioinformatics framework integrated copy number variations (CNVs), single-nucleotide variants (SNVs), and methylation profiles of cfDNA, benchmarked against matched tumor profiles. For 30 patients (58 tumor samples), single-cell multi-omics profiling (RNA, ATAC, and DNA sequencing) of paired tumor tissue was used to decode cfDNA clonal composition. Results : Despite low cfDNA input amounts (median: 3ng; range: below detection limit - 17ng), genome-wide cfDNA profiles were successfully generated for all samples. ctDNA signals showed strong concordance with matched tumor tissue across CNV, SNV, and methylation layers. Longitudinal plasma sampling added molecular time points unavailable from tissue biopsies, revealing newly acquired broad copy number alterations and focal amplifications (affecting, e.g., CDK6 ) at progression. Integration with single-cell data enabled deconstruction of bulk cfDNA profiles, revealing distinct compositions in the plasma with dominant, multiple, or divergent clones relative to tumor tissue. For example, in a case of pediatric alveolar rhabdomyosarcoma, ctDNA at relapse more closely resembled the profiles of the primary tumor rather than those in a concurrent lymph node metastasis, demonstrating how liquid biopsies capture tumor dynamics across anatomical sites. Conclusions : The integration of a multi-layered cfDNA assay with single-cell tumor sequencing provides an improved understanding of clonal dynamics and tumor evolution. By identifying cfDNA markers of heterogeneity and emerging resistance, this approach may establish a translational path for their longitudinal assessment via liquid biopsies in precision oncology.