Fusion transcripts in high-grade serous carcinoma
Mie Jareid1, Bjarne Johannesen1, Anita Sveen1,2, Katharina Bischof3,4, Kjetil Taskén2,4, Anne Dørum3, Ben Davidson2,5, Ragnhild A. Lothe1,2, Rolf I. Skotheim1,6.
1Department of Molecular Oncology, Institute for Cancer Research, Oslo University Hospital, Oslo, Norway; 2Institute of Clinical Medicine, University of Oslo, Oslo, Norway; 3Department of Gynecological Oncology, Oslo University Hospital - Radiumhospitalet, Oslo, Norway; 4Department of Cancer Immunology, Institute for Cancer Research, Oslo University Hospital, Oslo, Norway; 5Department of Pathology, Oslo University Hospital, Oslo, Norway; 6Department of Informatics, University of Oslo, Oslo, Norway
Background
High-grade serous tubo-ovarian carcinoma (HGSC) is marked by extensive genomic instability. Chromosomal rearrangements result in structural variants including gene fusions, which can be oncogenic drivers or diagnostic, prognostic, or therapeutic targets. In HGSC, fusion genes remain poorly characterized. We aim to investigate fusion transcripts in HGSC and explore their potential as biomarkers in the context of tumor heterogeneity, using multiple samples per patient.
Methods
Twenty-three patients diagnosed with advanced HGSC between 2002 and 2012 provided 108 samples for this study. Samples were collected from ovary and 1-3 extraovarian sites. Total RNA was sequenced using stranded libraries and 2x101 bp paired-end reads. STAR-Fusion was used for fusion predictions and FusionInspector for exclusion of artifacts and fusions previously predicted in normal tissues. We prioritized fusion transcripts expressed in all samples from the respective patient and ≥0.1 fusion fragments per million RNA-seq fragments in at least 1 sample. Validation of selected targests is ongoing.
Results
A total of 3787 different fusion transcripts were predicted (18163 including recurrences and variants). Of these, 141 passed filters and multisite criteria. We further prioritized putatively translated fusions: in-frame fusions (n=28) and protein family motifs in both gene partners (n=16). Of the 16 candidates, 14 were private to a single patient and two were recurrent in one sample from one additional patient each. These fusions, FLNB-SLMAP and RAD21-EIF3H, were expressed at a low level concurrent with high expression of the non-fused partner transcripts.
Discussion
We focused on fusions with a putative protein product. The two recurrent fusions have previously been reported in breast and liver adenocarcinomas. Our results indicate that fusion transcripts are recurrently expressed also in HGSCs, although with a low prevalence when requiring expression across tumor sites in each patient.