Abstract 5960: Identification and validation of sarcoma cellular ecosystems associated with prognosis and predictive of immunotherapy response

Abstract

Abstract The tumor microenvironment contributes to tumorigenesis, disease progression, and response to therapy in soft tissue sarcomas (STSs). However, characterizing the malignant and stromal cells that make up STSs and correlating their abundance with clinical outcomes has proven difficult in fixed clinical specimens. We employed a machine learning framework called EcoTyper (Luca et al. Cell 2021) to identify cell type-specific transcriptional states and define tumor ecotypes consisting of co-occurring cell states from bulk transcriptomes. Analyzing 292 previously published STSs profiled by bulk RNA-sequencing (RNA-Seq), we identified 23 transcriptionally-defined sarcoma cell states in malignant, immune, and other stromal cells and validated the majority of these cell states using single cell RNA-Seq. Although sarcomas originate from mesenchymal tissues, we identified four epithelial-like sarcoma cell states and observed epithelial differentiation and expression of epithelial markers within malignant sarcoma cells across histologies. Cell states reflected known and novel cell phenotypes and many were strongly associated with patient outcomes in our RNA-Seq training cohort and an independent validation cohort of STSs profiled by microarray (n=309). By identifying co-occurring cell states across STSs, we discovered 3 sarcoma ecotypes associated with underlying genomic alterations and distinct clinical outcomes. On spatial transcriptomic analysis, we observed co-localization of cell states within the same sarcoma ecotypes and spatial aggregation of ecotypes within STSs, suggesting that sarcoma ecotypes represent distinct functional units within human sarcomas. One ecotype (sarcoma ecotype 3: SE3) defined by CLEC5A/SPP1+ M2-like immunosuppressive macrophages and MYC/MTORC1-activated epithelial-like malignant cells was associated with inferior progression-free survival (PFS) in the RNA-Seq training cohort (P=0.01) and inferior metastasis-free survival in the microarray validation cohort (P=0.002). Remarkably, SE3 was associated with improved PFS in patients with metastatic STSs treated with ipilimumab and nivolumab (n=38, P=0.003) but not patients treated with chemotherapy. Furthermore, SE3 outperformed previously reported predictors of immunotherapy response in STSs, PD-L1 expression and tertiary lymphoid structures (TLS), for predicting response to immunotherapy (AUCs: SE3 0.87; PD-L1 0.81; TLS 0.62). Finally, SE3 similarly predicted response to immunotherapy in an independent validation cohort (n=29, AUC 0.89). In summary, our findings provide a high-resolution cell atlas of STSs to guide the development of novel therapeutic strategies. In addition, we have identified a predictive biomarker of response to immune checkpoint inhibition that may enable personalization of systemic therapy in patients with advanced STSs. Citation Format: Ajay Subramanian, Neda Nemat-Gorgani, Timothy J. Ellis-Caleo, David G.P. van IJzendoorn, Timothy J. Sears, Anish Somani, Bogdan A. Luca, Maggie Y. Zhou, Martina Bradic, Ileana A. Torres, Eniola Oladipo, Christin New, Deborah E. Kenney, Raffi S. Avedian, Robert J. Steffner, Michael S. Binkley, David G. Mohler, William D. Tap, Sandra P. D’Angelo, Matt van de Rijn, Kristen N. Ganjoo, Nam Q. Bui, Gregory W. Charville, Aaron M. Newman, Everett J. Moding. Identification and validation of sarcoma cellular ecosystems associated with prognosis and predictive of immunotherapy response [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 5960.