Detection of rare and novel gene fusions in patients with diffuse glioma: An institutional retrospective study

Le résumé fourni par la source

To the Editor: Gene fusions are increasingly recognized as important oncogenic drivers that define molecular subgroups and inform targeted strategies. Gene fusions are typically arising from chromosomal translocation, deletion, or inversion and resulting in the creation of a chimeric transcript through the fusion of parts of the 2 genes. These include recurrent KIAA1549::BRAF fusion, which has constitutive BRAF kinase activity, in pilocytic astrocytomas,1 FGFR::TACC fusion in a subset of infiltrating gliomas2 and BCOR fusion in pediatric gliomas.3,4 Reverse transcriptase polymerase chain reaction (RT-PCR) and fluorescence in situ hybridization (FISH) have predominantly been used for detecting gene fusions/rearrangements in routine clinical testing. Although these methods are highly sensitive, RT-PCR requires a priori knowledge of the fusion partner to design specific primer sets, and a break-apart probe used for FISH can only detect one of the rearranged genes. Therefore, both methods are unable to identify novel fusion gene events. Missing rare gene fusions in clinically relevant genes may lead to misdiagnosis or missed opportunities to apply targeted therapy. We present our experience in detecting a novel ROS1 gene fusion from an infantile glioma case using whole transcriptome and targeted RNA-sequencing technology (Archer FusionPlex Sarcoma Kit). Whole transcriptome sequencing is an unbiased method designed to survey all expressed genes whereas targeted methods like FusionPlex interrogate selected transcripts with much higher sensitivity. FusionPlex is a unique, anchored-multiplex PCR-based assay that uses uni-directional PCR amplification to interrogate specific genes without a priori knowledge of the potential fusion partner.5 We first employed whole transcriptome sequencing to identify fusions, then performed FusionPlex and FISH assays for orthogonal validation. We also reviewed results of whole transcriptome sequencing from 238 brain tumor cases enrolled in our precision medicine cohort in order to identify novel oncogenic gene fusions. The oncogenic potential of the novel fusion genes were predicted by Oncofuse6 and AGFusion7; 3 databases (Quiver-Archer, TCGA [The Cancer Genome Atlas], and Cosmic) were interrogated to determine the novelty of fusion candidates.8 Whole transcriptome nominated 5 novel fusions involving oncogenes in our precision medicine database: GPSM2::NTRK2, ATP10D::TEK, NTRK2::CLU, MDM4::GLI1, and PHIP::ROS1. Of these, the only in-frame fusion was a PHIP::ROS1 detected in a pediatric glioblastoma. To support identification of these fusion candidates, we further performed targeted RNA-seq on the 5 tumors using RNA extracted from archival formalin-fixed, paraffin-embedded (FFPE) tissues. A total of 34 genes known to be involved in tumor-associated fusion events were tested using a custom Archer FusionPlex Sarcoma Kit (Invitae, San Francisco, CA, United States) (Table 1). Of the 5 fusion candidates identified by untargeted RNA-seq (whole transcriptome), targeted RNA-seq identified the PHIP::ROS1 gene fusion with supporting evidence including 10 junction reads and 6 spanning fragments. The fusion product comprises exons 1-27 of PHIP, (with the break occurring at codon 1069 in exon 27 at position chr6:79679552), and exon 35-43 of ROS1, (with the break occurring at codon 1883 at position chr6:117642557) (Figure 1A). The PHIP::ROS1 fusion product was predicted to conserve the ROS1 tyrosine kinase domain (Figure 1A). FISH confirmed ROS1 rearrangement and found ROS1 copy number gain (Figure 1B). ROS1 immunohistochemistry (IHC) was performed but did not show overexpression. As reported previously, ROS1 fusion might not result in overexpression but in overactivation.9,10 Therefore, pERK IHC was done to investigate the downstream effect of ROS1 fusion in activating the MAPK/ERK pathway, using a semiquantitative approach, as reported by Deland et al.11 The case with ROS1 fusion showed pERK overexpression with an H-score of 240 (3 for intensity and 80% positive area) compared to another glioblastoma that did not show ROS1 mutations or rearrangements by whole genome sequencing and had an H-score of 100 (2 for intensity and 50% positive area) (Figure 1C and D). Description and validation of PHIP::ROS1 fusion in glioblastoma and radiological and histological characteristics of the index infantile glioma case. (A) Structure and functional domains of PHIP::ROS1 fusion product with the black line indicating the fusion point. The fusion product retains the entire ROS1 kinase domain. (B) Fluorescence in situ hybridization (FISH) assay for ROS1 in the index case. Break-apart green and red signals for ROS1 demonstrate gene rearrangement. FISH incidentally found a copy number gain of ROS1. (C) Hematoxylin and eosin (H&E) stain of frozen tumor tissue that underwent sequencing studies (100× original magnification). (D) Immunohistochemistry (IHC) demonstrates overexpression of pERK (top), compared to a ROS1 wild-type glioblastoma (bottom). (E) Preoperative brain MRI demonstrated a mass involving the right frontal lobe. Target genes by Archer panel. Target genes by Archer panel. The patient was a 5-month-old female presenting with a right scalp mass. MRI revealed a 7.7 × 6.4 × 6.7 cm lesion, which was a heterogeneously enhancing mass centered in the right parietal region (Figure 1E). She underwent a biopsy of the tumor, followed by a gross total resection. The pathology showed hypercellular proliferation of atypical astrocytic cells with pseudopalisading necrosis, microvascular proliferation, and scattered calcification, which resulted in a diagnosis of pediatric glioblastoma (Figure 1C). Whole exome sequencing of the biopsy sample revealed 8p gain; however, no major oncogenic mutations were identified. After surgical resection, she was treated with chemotherapy. Fortunately, no tumor relapse was observed during 6 years of follow-up. Currently, single-gene assays such as FISH, IHC, or RT-PCR are still used for fusion detection in routine clinical practice. However, because of their lack of multiplexing ability, they may miss rare but clinically actionable fusions. The advent of whole transcriptome sequencing has contributed to the discovery and detection of novel gene fusions in brain tumors by facilitating the assessment of multiple genes in 1 assay.12 However, clinical application of whole transcriptome is challenging because whole transcriptome requires relatively large amounts of high-quality RNA obtained from fresh-frozen (FF) tissues. To overcome this limitation, targeted RNA-seq approach is increasingly recognized as an efficient tool to detect clinically actionable fusion transcripts. The FusionPlex Sarcoma Kit used in this study enables detection of gene fusions from small RNA input obtained from archival FFPE blocks.8,13 Investigation of gene fusions using this assay is clinically feasible even when quality and/or quantity of input material is limited. ROS1 inhibitors have been approved for the treatment of a diversity of ROS1 fusion-positive tumors including pediatric high-grade gliomas.14 ROS1 protein is a receptor tyrosine kinase that exerts the oncogenic role through activation of cellular pathways known to be involved in cell growth and cell proliferation.14 Similar to previously reported druggable ROS1 fusions, the PHIP::ROS1 fusion was predicted to be in-frame with preservation of the entire ROS1 kinase domain, suggesting that it may be susceptible to tyrosine kinase inhibitors. If the tumor relapses, ROS1 inhibitors will be selected for this patient. Considering previous reports on oncogenic function of PHIP protein, it is possible that the fusion event involving PHIP potentially promotes aggressiveness of glioma. PHIP gene encodes pleckstrin homology domain-interacting protein that was initially identified as a positive regulator of pancreatic islet cells.15 In tumor biology, PHIP drives cell proliferation and invasive behavior in

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.