Detecting hotspots of intra- and transchromosomal fusions in liposarcomas by RNA sequencing.

Abstract

11561 Background: Liposarcoma (LPS) is characterized by unstable genomes and high occurrence of gene fusions. Hotspots of both recurrent and non-recurrent gene fusions can provide information about structural alterations in certain LPS subtypes. For instance, myxoid (M) LPS expresses the oncogenic FUS-DDIT3 protein and is known to be fusion-driven. Moreover, disrupted locus 12q13-15 is an important feature of well-dedifferentiated (WD) and dedifferentiated (DD) LPS, and is the site of copy number alterations (CNAs) and gene fusions. Here, we used RNA sequencing to uncover hotspots of intra- and transchromosomal gene fusions in LPS patient samples, identifying potentially clinically relevant events in certain chromosomal regions. Methods: The BostonGene internal LPS cohort (n=150) was analyzed by bulk whole-transcriptome sequencing, using STAR-fusion for sequence calling. Quality control was performed using FastQC, FastQ Screen, RSeQC, and MultiQC. Tumor purity was assessed via pathological and bioinformatics examination with a threshold of 20%. One sample Poisson rate test was used to evaluate statistical significance of gene fusion hotspots. Results: We identified 4,080 gene fusions among four LPS subtypes (DDLPS, WDLPS, MLPS, and PLPS - pleomorphic LPS). Of those, 2,302 (56.4%) were intrachromosomal and 1,778 (43.6%) were transchromosomal. Over half of these fusions (1,263/2,302 intrachromosomal fusions, 54.9%; 1,047/1,778 transchromosomal fusions, 58,9%) were detected on chromosome 12. Most identified fusions occurred in the 12q13-15 region (q-value < 0.001), with q15 being especially prevalent in transchromosomal fusions (q-value < 0.001). The most prevalent recurrent fusion across our LPS cohort was FUS-DDIT3 (N=21, MLPS samples). Another notable recurrent fusion was TRIO-TERT (N=4, DDLPS). We also identified extended gene fusion hotspots in regions containing important oncogenes such as MDM2 and FRS2 in LPS subtypes (DDLPS, WDLPS, PLPS) that are not considered fusion-driven (Table). These findings suggest an oncogenic role of such fusions in these LPS subtypes, along with known CNAs like MDM2 amplification. Conclusions: Our comprehensive transcriptomic analysis of gene fusions in LPS samples uncovered both new and established hotspots of chromosomal rearrangements. Identification of such hotspots improves our understanding of LPS oncogenesis and thus can enhance the diagnostic accuracy and discovery of new biomarkers. Hotspots of gene fusions in liposarcoma. Hotspot cytoband Diagnosis q-value Important genes 1q23.3 DDLPS < 0.001 ATF6 1q24.3 DDLPS, WDLPS < 0.001 DNM3 12q13.3 MLPS < 0.001 DDIT3 12q14.1 DDLPS, WDLPS < 0.001 CDK4 12q14.3 DDLPS, WDLPS < 0.001 HMGA2, YEATS4 12q15 DDLPS, WDLPS, PLPS < 0.001 MDM2, FRS2, CPM 16p11.2 MLPS < 0.001 FUS