(P 378) MULTIOMICS PROFILING OF EPITHELIOID SARCOMA UNVEILS EPIGENETIC CIRCUITRIES THAT UNDERLIE THE PATHOBIOLOGY OF PROXIMAL AND CLASSIC SUBTYPES: RESULTS FROM THE ISG EPISOBS OBSERVATIONAL STUDY

Objective: Epithelioid sarcoma (ES) is an ultra-rare sarcoma hallmarked by loss of SMARCB1/INI1 expression. Based on cytomorphology, ES is distinguished into classic (C-ES) and proximal (P-ES) subtypes, irrespective of anatomic site. This distinction has clinical relevance, as P-ES is associated with a more aggressive behavior and poorer outcome. Whether this behavior is an intrinsic biological feature of P-ES or rather reflects its often proximal and deeper anatomic location remains to be understood. Apart from the SMARCB1/INI1 loss, little is known about the biology of ES and the determinants of the distinctive clinicopathologic features of the two subtypes. To shed light on this we performed a multiomics analysis of a cohort of 24 ES, in the framework of the Italian Sarcoma Group (ISG) observational study focused on ES EpisObs (NCT03099681).

Methods: Clinically annotated samples were collected from patients in the retrospective and prospective cohorts of the EpisObs ISG study, selecting primary, untreated ES, with available material suitable for molecular analyses. Diagnosis was confirmed through a centralized pathology review by expert pathologists, according to the 2020 WHO. FFPE sections with a tumor cell fraction ≥ 60% were selected for multiomics profiling, including targeted-DNA, RNA and miRNA NGS sequencing, and DNA methylation (methyl-DNA) profiling. NGS DNA, RNA, and miRNA libraries were run on Illumina platforms and Illumina Infinium MethylationEpic arrays were used for methyl-DNA analysis. Methyl-DNA:RNA and miRNA:RNA data were anticorrelated using Spearman correlation test. Selected differentially expressed genes were orthogonally validated by IHC. HSES2R, VAESBJ and Epi544 cell lines were used for in vitro miRNA validations and drug response.

Results: The study cohort was composed of 14 C-ES and 10 P-ES, all INI1 negative at IHC (Table 1). Apart from SMARCB1 deletion, no recurrent mutation events were detected. Conversely, P-ES and C-ES clearly differed for RNA, methyl-DNA, and miRNA profiles, forming distinct clusters in unsupervised analyses (Figure 1). Pro-aggressive transcriptional signatures were tendentially enriched in P-ES, which featured upregulation of genes involved in chromatin metabolism or cell cycle, and downregulation of genes associated with mesenchymal features. This was in line with methyl-DNA results that indicated that DNA methylation was higher in P-ES and mostly affected genes involved in cell migration and vasculogenesis. Functional annotation of the genes anticorrelated with the cognate miRNAs indicated that miRNAs also play an important role in shaping ES transcriptome, and miRNA and RNA data integration unveiled novel circuits contributing to the transcriptional features of the two subtypes. Among these, a complex circuit connected miR-137 and miR-24 with HELLS, EZH2, MYC, and E2F, and sustained cell proliferation and deregulation of chromatin metabolism in P-ES. In this circuit HELLS, which is regulated by both miRNAs and promotes DNA methylation-mediated gene silencing, acts as link between these two epigenetic mechanisms of transcriptional regulation. Finally, some differentially expressed genes whose expression may discriminate P-ES and C-ES were validated by IHC analysis.

Conclusion: This multiomics study supports the notion that P-ES and C-ES are distinct molecular entities, characterized by a different degree of phenotypic drift form the mesenchymal makeup and marked by different DNA methylation and miRNA:gene circuitries. The transcriptional profile of P-ES suggests that biology rather than anatomic location sustains its poorer outcome. Our study unveiled also novel immunohistochemical markers that might aid in the differential diagnosis of P-ES and C-ES, thus serving as potential tools of patient stratification in future studies.