A transcriptomic analysis of retroperitoneal well differentiated liposarcoma (WDLPS) and well differentiated (WD) and dedifferentiated (DD) components of dedifferentiated liposarcoma (DDLPS).

Abstract

e23520 Background: Retroperitoneal DDLPS is an aggressive tumor characterized by a WD component, morphologically similar to a WDLPS, and a DD component that drives patient prognosis. We aimed at unraveling whether WDLPS and DDLPS are distinct entities or entail a progressive evolution from WDLPS to DDLPS through a transcriptomic analysis. Methods: A transcriptomic analysis was performed in a retrospective series of 107 patients with primary retroperitoneal WDLPS (N = 68, 63.5%) or DDLPS (N = 39, 36.5%) who underwent surgery (2011-15). Paired DD, WD, and normal fat (NF) components were sampled in DDLPS, while paired WD and NF components were sampled in WDLPS. RNA-Seq data were normalized according to the trimmed mean of M-value (TMM) algorithm and differential expression was evaluated with the voom method implemented into the edgeR package. Enrichment in hallmark gene sets from Molecular Signatures Database (MSigDB) was evaluated with gene sets enrichment analysis (GSEA), by using t-statistic as measure of ranking. A false discovery rate (FDR) adjusted p-value < 0.05 was considered for statistical significance. Results: Differential expression analysis revealed marked transcriptional changes within paired components of DDLPS (DD, WD, NF) and WDLPS (WD, NF). Changes of WD and NF components between WDLPS and DDLPS were not statistically significant. Gene sets were analyzed to evaluate the ensemble and overcome the FDR correction applied to single genes. Hallmarks deregulated in WD component of DDLPS compared to WDLPS were detected also in their paired DD component. Among them, G2M checkpoint and mitotic spindle were up-regulated, while adipogenesis, fatty acid metabolism, cholesterol homeostasis, oxidative phosphorylation, and peroxisome were down-regulated. These differences persisted also when NF components were compared with their paired WD component of DDLPS and WDLPS. G2M checkpoint and mitotic spindle gene sets did not differ between NF of DDLPS and NF of WDLPS, suggesting these two hallmarks as tumor-specific. Conversely, expression of adipogenesis, fatty acid metabolism, and oxidative phosphorylation was up-regulated together with other gene sets related to metabolism in NF of DDLPS. Tumor inflammation and interferon response were up-regulated in WD components compared to their paired NF components. Also, interferon response was down-regulated in WD component of DDLPS compared to WDLPS, and inflammation was down-regulated in DD component compared to both WD components. Conclusions: Transcriptomic changes that distinguished WDLPS and WD component of DDLPS increased progressively also in the paired DD component of DDLPS, supporting the hypothesis of a progression from WDLPS to DDLPS in some liposarcomas. Changes observed in NF may suggest a paracrine effect sustaining tumor dedifferentiation.