Distinct cellular origins and differentiation process account for distinct oncogenic and clinical behaviors of leiomyosarcomas

Abstract

AbstractDifferentiated phenotypes in solid tumors mostly lead to a better outcome than in their more immature counterparts. In leiomyosarcoma (LMS), a very aggressive tumor for which neither chemotherapy nor targeted or immune therapies have demonstrated efficacy, a relatively transcriptionally uniform subgroup of well differentiated tumors has been described and is associated with poor survival. In this disease, differentiation does not systematically hinder tumor progression, so it is a good model for studying the relation between these two mechanisms. By applying ad hoc bioinformatic methods, we conducted a comprehensive analysis in order to detect subgroups of leiomyosarcoma and finely investigate their transcriptomic and genomic profiles. We first used unsupervised hierarchical clustering to identify the most transcriptionally homogeneous LMS subgroup, which we named ‘hLMS’. We derived a transcriptional signature that could consistently stratify patients with similar clinical and genomic characteristics in TCGA and ICGC cohorts. Functional analysis of the signature showed over-expression of SRF and E2Fs targets harboring histone modifications that were specifically identified in smooth muscles in hLMS. By integrating multi-omics data, we identified punctual and structural variations as well as the potential cellular origins of each group of LMS using GTEX normal tissue gene expression. Altogether, our results suggest that hLMS, but not oLMS (other LMS), may originate from vascular smooth muscle cells in which oncogenesis is driven by the MYOCD/SRF axis, which is essential for hLMS survival. We propose that the phenotypic plasticity of vascular smooth muscle cells coupled with MYOCD/SRF pathway overactivation could explain how hLMS balance this uncommon interplay between differentiation and aggressiveness.