Gene Expression Changes Associated with Dedifferentiation in Liposarcoma Predict Overall Survival

Abstract

Up to 10% of well-differentiated liposarcoma (WDLS) progress to dedifferentiated liposarcoma (DDLS). We aimed to identify gene expression changes associated with dedifferentiation and whether these were informative of tumour biology of DDLS. We analysed datasets from the Gene Expression Omnibus (GEO, ID = GSE30929) database to identify differentially expressed genes between WDLS (n = 52) and DDLS (n = 39). We validated the signature on whole and laser-capture microdissected samples from patients with tumours consisting of mixed WDLS and DDLS components. A subset of this signature was applied to an independent dataset from The Cancer Genome Atlas (TCGA, n = 58 DDLS) database to segregate samples based on gene expression and compared for recurrence and overall survival (OS). A 15-gene signature consisting of genes with increased expression in DDLS compared to WDLS was generated. This signature segregated WDLS and DDLS samples from patients with mixed component tumours and across multiple recurrences. A further subset of this signature, consisting of five genes (AQP7, ACACB, FZD4, GPD1, LEP), segregated DDLS in a TCGA cohort with a significant difference in OS (p = 0.019) and recurrence-free survival (RFS) (p = 0.061). The five-gene model stratified DDLS into prognostic groups and outperformed clinical factors in existing models in retroperitoneal DDLS.

Genetic, Epigenetic and Transcriptome Alterations in Liposarcoma for Target Therapy Selection

“…Significant differences in the expression of INSM1 in WDLPS and DDLPS may assist in the diagnosis, enriching the diagnostic index system of mesenchymal cancers [43]. Additional chromosomal abnormalities, more exclusive for DDLPS than for WDLPS, are recurrent amplifications of 1p32 and 6q23, in particular, overexpression of ASK1, DDR2, ERBB3, STAT6, FGFR1, MAP3K5, LGR5, MCL1, CALR, AQP7, ACACB, FZD4, GPD1, LEP and ROS1 [21,[44][45][46]. Another set of core genes in DDLPS identified as significantly enriched in microarray profiling generated from DDLPS and normal fat controls include APP, MDM2, CDK1, PCNA, TKT, CDK4, CDC20, BUB1B, BARD1, ADRB2, LGALS3, CAV1, CCNA2 and CDKN2A.…”