Novel RGAG1-BCOR gene fusion revealed in a somatic soft tissue sarcoma with a long follow-up

A 54-year-old man presented with a large intramuscular, painless swelling in his lower leg (Fig. 1A). Ultrasound and magnetic resonance imaging (Fig. 1B) revealed a well-demarcated intramuscular tumor (95 × 50 × 40 mm) that was highly indicative of sarcoma. Core biopsy showed a myxoid mesenchymal neoplasia and staging work-up excluded metastatic disease. Complete resection with clear resection margins was performed (Fig. 1C). Reverse transcription-polymerase chain reaction (RT-PCR) and fluorescence in situ hybridization (FISH) were performed and excluded the diagnosis of a myxoid/round cell liposarcoma. NGS was not available as a routine diagnostic tool at this time, and therefore, the diagnosis of exclusion was an unusual cellular EMC. However, we were not able to confirm this diagnosis on the molecular level, since a test for the detection of NR4A3 gene rearrangements was not available. The patient received adjuvant radiotherapy (cumulative 60 Gy), but no chemotherapy. Five years later upon availability, NGS analysis of the tumor was performed detecting a RGAG1-BCOR gene fusion. The patient was still under complete remission after 48 months of follow-up. However, more recently, the patient has been diagnosed with a NRAS gene–mutated melanoma by another institution originating in the right gluteal region with lymph node metastasis, without relation to the sarcoma. He received immunotherapy for melanoma. At 87 months of follow-up since sarcoma diagnosis, there are no signs of recurrence or metastases of both tumors.

The patient signed and agreed to the general patient consent according to ethical standards of the University Hospital Zurich, allowing the use of all tissue, cells, and images in this report.

Staining was performed by an automated IHC system on formalin-fixed and paraffin-embedded tissue of representative areas of the tumor cut into 2-μm-thick sections according to the protocol used in our institute [12]. Samples were loaded on the BenchMark Ultra (Ventana) or the Bond-Max system (Leica). The following antibodies were used: S100 (Dako), synaptophysin (Novocastra), SOX9 (Millipore), cytokeratin (Dako), CD99 (Novocastra), CD34 (Ventana/Roche), smooth muscle actin (Sigma), calponin (Dako), desmin (Dako), HMB45 (Dako), p63 (Ventana / Roche), caldesmon (Dako), epithelial membrane antigen (Dako), Ki-67 (Ventana/Roche), WT1 (Ventana/Roche), BCOR (Santa Cruz Biotechnology), and CCNB3 (SIGMA Chemical Company).

RNA was extracted from paraffin-embedded tissue as previously described [13]. For the detection of fusions typically found in liposarcoma, reverse transcription of the RNA into cDNA and its amplification were performed using the OneStep RT-PCR kit (Qiagen), the GeneAmp PCR system 9700 (Applied Biosystems), and the following primers: EWSR1-DDIT3 type 1 (5′-TCCTACAGCCAAGCTCCAAGTC-3′, -5′-CCGAAGGAGAAAGGCAATGACTCAG-3′), EWSR1-DDIT3 type 2 (5,-GAC CCATGGATGAAGGACC-3′, 5′CCGAAGGAGAAAGGCAATGACTCAG-3′), FUS/TLS-DDIT3 type 1 (5′-GGAAGTGACCGTGGTGGCTT-3 ′, 5′-CCGAAGGAGAAAGGCAATGACTCAG-3′) and FUS/TLS-DDIT3 type 2 (5′-GCAGAACCAGTACAACAGCAGCAGTG-3′, 5′-CCGAAGGAGAAAGGCAATGACTCAG-3′). RT-PCR was performed as previously described by Bode et al. [14].

The probe kit used to detect the DDTI3(CHOP) gene on chromosome 12q13 was the LSI DDTI3 kit (Abbott) containing a 700-kb probe labeled with SpectrumOrange (centromeric) and a 663-kb probe with SpectrumGreen (telomeric). Detection of the EWSR1 gene on chromosome 22q12 was performed using LSI EWS (Abbott) consisting of a 497-kb probe labeled with SpectrumOrange (centromeric) and a 1100-kb probe with SpectrumGreen (telomeric). Both kits were used according to the protocol of Abbott Molecular. 50 cells were examined and the cut-off for positive rearrangement was set at 25% of cells presenting split signals. The protocol corresponded to Wolpert et al. [12].

RNA isolation was performed using the Maxwell 16 LEV RNA FFPE Purification Kit (Promega Corporation) and RNA quality was assessed on a Bioanalyzer using the RNA 6000 Pico Kit (Agilent) according to the manufacturer’s instructions. NGS Libraries were prepared using the “TruSight RNA Pan-Cancer Panel” (Illumina, Inc). Quality and quantity were determined with the DNA 1000 Kit and Bioanalyzer (Agilent). Libraries were sequenced on the MiSeq platform (Illumina, Inc). The RNA fragments were sequenced using the MiSeq system (Illumina, Inc., USA).

The FusionMap software was used to detect RNA fusion reads according to Ge et al. [15].

Resected specimen contained a poorly demarcated, lobulated tumor measuring 5.7 × 4.1 × 10.5 cm, white-greyish with focal hemorrhages (Fig. 1C).

Histologically, the tumor presented as polylobulated and had myxoid stroma with some hyalinized areas. It showed multinodulary invasive growth into the surrounding tissue. The cells were monomorphic with relatively narrow cytoplasm and monomorphic nuclei with finely stippled chromatin (Fig. 2A). The tumor was rich in mitotic Figures (21/10 HPF).

The initial immunoprofile was rather unspecific with focal positivity for S100, synaptophysin and SOX9 (Fig. 2B). Reactions for pan-cytokeratin, CD99, CD34, SMA, calponin, desmin, HMB45, p63, caldesmon, and EMA were negative. The proliferation index of MIB1 was heterogeneous of up to 40% (Fig. 2D). Staining with antibodies against BCOR and CCNB3 was carried out after NGS analysis, showing strong nuclear expression of the BCOR protein (Fig. 2C), while CCNB3 reaction was negative.

No rearrangement of the DDTI3(CHOP) gene was detected, excluding differential diagnosis of liposarcoma. There was no rearrangement of the EWSR1 gene.

The initially performed RT-PCR showed no t(12;16) and (12;22)(FUS-DDIT3 and EWSR1-DDIT3) specific fusion transcripts.

Due to unusual histopathologic features of the tumor, the NGS has been performed upon availability retrospectively. Using the FusionMap, we were able to detect three fusions with RGAG1 and BCOR genes as fusion partners. For the first two fusions, BCOR was found in 5′ position and RGAG1 in 3′ position. The breakpoints of the first fusion sequence (Suppl. Figure 1) were found in exon 2 of BCOR and exon 2 of RGAG1, whereas the breakpoints of the second fusion sequence (Suppl. Figure 2) were found in exons 2 of BCOR and in the promoter region of RGAG1. For both of these fusion products, their potential open reading frames showed a stop codon within the first six amino acid positions from the fusion junction, making the potential protein products most likely non-functional.

The third sequence reported by the FusionMap represented a fusion of RGAG1 in 5′ position and BCOR in 3′ position (Fig. 3A and B). The breakpoint in the RGAG1 gene was found in exon 3, whereas the alignment by FusionMap indicated the position of the breakpoint in the BCOR gene at the 3′-end of intron 2 (two base pairs upstream of the 5′-end of exon 3). Alternatively, the breakpoint in the BCOR gene could be two base pairs upstream of the 3′-end of exon 2. The detection of this fusion was supported by 61 paired reads (i.e., one read in the BCOR and the corresponding paired reads in the RGAG1 gene) and 268 junction-spanning reads. Using RT-PCR (Fig. 3D) and Sanger sequencing, we could confirm the existence of this gene fusion (Fig. 3E and G). However, the exact location of the breakpoint in BCOR remained ambiguous as we analyzed RNA and not DNA. Nevertheless, we predict the expected fusion product to be in frame for both RGAG1 and BCOR (Fig. 3C and G). Furthermore, RT-PCR revealed the existence of a second RGAG1-BCOR-fusion that is 30 bp shorter than the first RGAG1-BCOR-fusion. We were able to confirm this finding by visual inspection of the sequencing reads obtained by the Pan-Cancer panel (Fig. 3F). Also, for this fusion product, we predict a stop codon within 6 AA from the fusion point (Suppl. Figure 3).