Abstract A015: Targetable gene dependencies in Ewing sarcoma subtypes

Abstract Objective: Ewing sarcoma (EwS) is a bone and soft tissue cancer primarily driven by a FET::ETS fusion protein, most commonly EWS::FLI1. While localized tumors have a five-year survival rate of 70–80% with multimodal treatment, the prognosis for relapsed or metastatic disease remains dismal, with survival rates below 30%. Starting from a large, multi-institutional cohort of patient-derived transcriptomes, this study aims to identify prognostic markers and therapeutic vulnerabilities associated with transcriptionally distinct classes of EwS. Methods: To characterize the transcriptional heterogeneity of EwS tumors, we applied RACCOON, an unsupervised hierarchical clustering approach, to define distinct transcriptional subtypes. We then used OTTER, a tumor classifier, to assign tumors from validation cohorts and preclinical models to these subtypes. Single-cell RNA sequencing and commonly used in vitro EwS cell lines were analyzed to assess transcriptional variability at single-cell resolution and in experimental systems. Clinical outcome stratification was evaluated using independent validation cohorts. Subtype-specific gene dependencies were inferred by integrating transcriptional subtype classifications with genome-wide CRISPR screening data from the DepMap project. Results: Three distinct transcriptional subtypes of EwS were identified: EWS::FLI1-high, mesenchymal-like, and muscle-like. The muscle-like group was more distinct from the other subtypes, potentially due to myofiber infiltration. The remaining tumors exhibited a continuum of transcriptional states between EWS::FLI1-high and EWS::FLI1-low/mesenchymal-like subtypes. Transcriptional variability driven by fusion activity was further observed to differ across single-cell datasets and preclinical models. In the validation cohort, mesenchymal-like tumors were associated with the poorest prognosis (<35% five-year overall survival) and the highest metastatic potential. In vivo studies in patient-derived xenografts supported these findings. Integration of CRISPR screening data from DepMap revealed subtype-specific gene dependencies: EWS::FLI1-high tumors showed sensitivity to Fanconi anemia pathway disruption, whereas mesenchymal-like tumors relied on distinct transcriptional regulators. Conclusion: We define three transcriptional subtypes of Ewing sarcoma, EWS::FLI1-high, mesenchymal-like, and muscle-like, with distinct gene dependencies, metastatic potential, and clinical outcomes. Our findings provide a foundation for new diagnostic approaches and molecular subtype-specific treatment strategies in EwS. After further validation, this approach has the potential to inform clinical decision-making and enable more personalized management of patients with EwS. Citation Format: Dusan Pesic, Joshua O Nash, Pedro Lemos Ballester, Timmy Wen, Livia Garzia, David Malkin, Adam Shlien. Targetable gene dependencies in Ewing sarcoma subtypes [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Discovery and Innovation in Pediatric Cancer— From Biology to Breakthrough Therapies; 2025 Sep 25-28; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2025;85(18_Suppl_2):Abstract nr A015.