Molecular subtyping and insights into sarcoma biology and prognosis.

Abstract

11536 Background: Sarcomas represent a diverse group of mesenchymal malignancies, with over 200 subtypes recognized by the WHO. This heterogeneity, combined with rarity, complicates diagnosis. Here, comprehensive molecular profiling and unsupervised clustering were applied to define sarcoma subtypes associated with differentiation, genomic events, and prognosis. Methods: Whole exome and transcriptome sequencing was performed on 1,046 sarcoma samples via the BostonGene Tumor Portrait test. Hierarchical expression-based clustering was integrated with somatic mutations, fusions, copy number alterations, genome segmentation, and signatures, validated on 23 external datasets (n = 1,678). Results: Nine molecular clusters (C1-C9) reflecting distinct biological patterns were identified (Table). Stable, lineage-restricted clusters included leiomyosarcoma (C2) marked by smooth muscle differentiation pathways and TP53 , ATRX , and RB1 mutations; vascular sarcoma (C7) with vascular signaling and angiogenic amplifications; synovial sarcoma (C6) defined by WNT activation and SS18-SSX fusions; and Ewing sarcoma (C9) characterized by neuronal development signatures and EWSR1-FLI1 fusions. Clusters progressing toward undifferentiated pleomorphic sarcoma (UPS)-like phenotypes included liposarcoma (C3) progressing from adipocyte differentiation with MDM2/CDK4 amplifications; chondrosarcoma/chordoma (C5) transitioning from cartilage development pathways with TGFβ activation; and bone and UPS-like (C1) characterized by angiogenesis signatures, genomic instability, TP53 mutations, and the poorest survival (logrank p < 0.05). Molecular clusters predicted survival regardless of the specific diagnosis. For example, liposarcoma and leiomyosarcoma patients in cluster C1 had a worse prognosis than those in C2 with the same diagnosis. Conclusions: The proposed molecular subtypes simplify sarcoma classification, providing unique prognostic insights missing in the current diagnostic framework. These findings suggest molecular subtypes could offer valuable knowledge for sarcoma research and patient treatment. Сharacterization of molecular sarcoma subtypes. Cluster Predominant diagnoses Undifferentiation progression Biology Alterations (chi-sq p < 0.01) C1 Bone and UPS-like Progressing Undifferentiated or nearly undifferentiated TP53 mutation (35%) C2 Leiomyosarcoma Stable Smooth muscle differentiation TP53 (53%), ATRX (18%), RB1 (15%) mutations C3 Liposarcoma Progressing Adipocyte differentiation MDM2/CDK4 amplifications (45%) C4 Progressing Biologically diverse C5 Chondrosarcoma, chordoma Progressing Cartilage development C6 Synovial sarcoma Stable WNT activation SS18 - SSX fusions (68%) C7 Vascular sarcoma Stable Vascular signalling, angiogenesis C8 Progressing Biologically diverse C9 Ewing sarcoma Stable Neuronal development signatures EWSR1-FLI1 fusions (92%)