The Journal of Molecular Diagnostics, Vol. 20, No. 1, January 2018

jmd.amjpathol.org

Development and Evaluation of a Pan-Sarcoma Fusion Gene Detection Assay Using the NanoString nCounter Platform

Kenneth T.E. Chang,*y Angela Goytain,z Tracy Tucker,x Aly Karsan,zx Cheng-Han Lee,zx{ Torsten O. Nielsen,zx{ and Tony L. Ngzx{

From the Department of Pathology and Laboratory Medicine,* KK Women’s and Children’s Hospital, Singapore, Singapore; the Duke-NUS Medical School,y Singapore, Singapore; the Department of Pathology,z University of British Columbia, Vancouver, British Columbia, Canada; the Department of Pathology,x British Columbia Cancer Agency, Vancouver, British Columbia, Canada; and the Department of Pathology,{ Vancouver General Hospital, Vancouver, British Columbia, Canada

Accepted for publication September 22, 2017. The NanoString nCounter assay is a high-throughput hybridization technique using target-specific probes that can be customized to test for numerous fusion transcripts in a single assay using RNA from formalin- Address correspondence to fi fi Tony L. Ng, M.D., Ph.D., xed, paraf n-embedded material. We designed a NanoString assay targeting 174 unique fusion junctions Department of Pathology and in 25 sarcoma types. The study cohort comprised 212 cases, 96 of which showed fusion gene expression Laboratory Medicine, Vancou- by the NanoString assay, including all 20 Ewing sarcomas, 11 synovial sarcomas, and 5 myxoid lip- ver General Hospital, 855 W osarcomas tested. Among these 96 cases, 15 showed fusion expression not identified by standard clinical 12th Ave, Vancouver, BC V5Z assay, including EWSR1-FLI1, EWSR1-ERG, BCOR-CCNB3, ZC3H7B-BCOR, HEY1-NCOA2, CIC-DUX4, COL1A1- 1L3, Canada. E-mail: tony. PDGFB, MYH9-USP6, YAP1-TFE3,andIRF2BP2-CDX1 fusions. There were no false-positive results; howev- [email protected]. er, four cases were false negative when compared with clinically available fluorescence in situ hybridi- zation or RT-PCR testing. When batched as six cases, the per-sample reagent cost was less than conventional techniques, such as fluorescence in situ hybridization, with technologist hands-on time of 1.2 hours per case and assay time of 36 hours. In summary, the NanoString nCounter Sarcoma Fusion CodeSet reliably and cost-effectively identifies fusion genes in sarcomas using formalin-fixed, paraffin- embedded material, including many fusions missed by standard clinical assays, and can serve as a first- line clinical diagnostic test for sarcoma fusion gene identification, replacing multiple individual clinical assays. (J Mol Diagn 2018, 20: 63e77; https://doi.org/10.1016/j.jmoldx.2017.09.007)

Sarcomas are malignant mesenchymal tumors arising from sarcomas in the latter category can be classified on the basis soft tissues and bones. Diagnosis of sarcomas can be chal- of pathognomonic chromosomal translocations. These lenging for pathologists because of the large number of en- generate fusion genes encoding chimeric transcription factors tities, many with overlapping histomorphologies and that alter the epigenome and dysregulate transcription of immunophenotypes. Nonetheless, accurate diagnosis is crit- target genes or function as chimeric protein kinases or ical because certain diagnoses require specifictreatment overexpressed growth factors.2 The importance of molecular strategies or affect eligibility for clinical trials. Subtype genetic testing for accurate diagnosis of sarcomas was diagnoses confer important prognostic information that affects immediate local management and follow-up strate- Supported by the Canadian Cancer Society grant 701582 (T.O.N.). gies. From a genetic perspective, sarcomas can be classified K.T.E.C. was a visiting fellow in the Terry Fox Foundation Strategic Health into two main groups. Approximately three-quarters of sar- Research Training Program in Cancer Research at the Canadian Institutes of comas are karyotypically complex and lack recurrent genetic Health Research (TGT-53912). The work in Singapore was funded by the VIVA-KKH Pediatric Brain and Solid Tumor Program grant alterations.1 The other quarter (including most pediatric, KRDUK16V0100 (K.T.E.C.). adolescent, and young adult tumors) have simple genetic Disclosures: T.O.N. has a licensing agreement with NanoString Tech- 2 alterations that often define specific sarcoma entities. Many nologies for the Prosigna assay, not part of the current study.

Copyright ª 2018 American Society for Investigative Pathology and the Association for Molecular Pathology. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0). https://doi.org/10.1016/j.jmoldx.2017.09.007 Chang et al demonstrated in the Clinical Effect of Molecular Methods in sequences within a given sample.10 These probes are cus- Sarcoma Diagnosis (GENSARC) study, in which the histo- tomizable and are assembled into pooled sets, termed logic diagnosis of 53 of 384 sarcomas (13.8%) was amended CodeSets. This technique requires neither nucleic acid by molecular findings, with implications for patient man- library preparation nor determination of sequence data, uses agement and prognostication.3 However, even when no nucleic acid amplification steps, is tolerant of poorer- molecular techniques optimized for formalin-fixed, paraffin- quality nucleic acid, and is, therefore, well suited to embedded (FFPE) pathology tissue are available, diagnostic routine diagnostic pathology settings in which FFPE mate- identification of such fusion genes can be challenging in the rial is all that is usually available. A recent publication has clinical setting for the following reasons. First, for some described use of the NanoString technology to detect sarcoma types, such as Ewing sarcoma, multiple fusion specific gene fusions in nonesmall-cell lung cancer, in combinations involving several partner genes are possible.4 which this platform performed well on diagnostic tumor Second, the specific break point in each partner gene can specimens, demonstrably improving clinical sensitivity and be variable, resulting in a variety of exon-exon fusion com- specificity compared with standard techniques.11 binations at the transcript level. Third, current standard In this study, we investigate the accuracy and feasibility clinical assays lack multiplex capabilities, potentially entail- of a NanoString-based assay using a custom-designed ing multiple costly consecutive investigations with long net CodeSet composed of target-specific probes complement- turnaround times when identifying unusual gene fusions. ing a list of sarcoma fusions that detect all common and There is, therefore, a clinical need for a single pan-sarcoma most uncommon gene translocations in bone and soft tissue gene fusion identification assay that is both cost-effective tumors. Our specific aims are as follows: first, to demon- and has a short turnaround time that meets patient care strate that the NanoString assay is at least as sensitive and requirements. specific as FISH/RT-PCR testing for identification of fusion Current standard molecular diagnostic methods for iden- oncogenes on clinical specimens; second, to determine per- tifying gene fusions in use in hospital laboratories include sample cost and turnaround time; and third, to assess the conventional cytogenetics, fluorescence in situ hybridization capability of the assay to improve diagnosis by detecting (FISH), and RT-PCR.5 Conventional cytogenetics is a well- fusion genes in cases in which no fusion gene was identified established technique that provides a gross pan- through conventional molecular techniques. chromosomal survey and can detect large chromosomal al- fi 6 terations without prior knowledge of speci c alterations. Materials and Methods However, cytogenetic analysis requires fresh tissue and cell culture techniques, is labor intensive, and can be technically Probe Design challenging, with a fairly lengthy turnaround time and an inability to detect chromosomal changes beyond the resolu- An in-depth literature search was performed for gene tion of 10 to 15 megabases. For these reasons, it is rapidly translocations involved in sarcomas. The sequence across becoming obsolete in the face of technological advances. the fusion break was collected from relevant case reports FISH can be applied to fresh/frozen or FFPE tissue. Fusion when reported.12e94 When the exact sequence was not translocation probes and break-apart probes are available, reported, the junction sequence was deduced from reported which identify gene fusions or rearrangements by means of exon/exon break points. The RNA sequences across the differentially labeled probes flanking the gene break points of junction site of the gene translocations were submitted to interest.7,8 FISH is not amenable to multiplexing, and multi- the NanoString Bioinformatics group (NanoString Tech- ple iterations of individual FISH assays (each requiring nologies, Seattle, WA) for the design of a NanoString additional tissue, technical time, and quality control proced- Elementsebased assay CodeSet. The CodeSet is composed ures) may need to be performed to identify rare fusion genes of target-specific probes, each with three oligonucleotide in tumors with multiple possible partner genes. RT-PCR probes per fusion site: a capture probe, a reporter probe, likewise can be performed on fresh/frozen or FFPE tissue, and a protector probe (Figure 1). Each reporter and capture although it is less sensitive on the latter.9 A single RT-PCR probe is approximately 50 nucleotides long, and each assay can determine the presence of a translocation using protector probe is approximately 32 nucleotides long. A primer pairs targeting each partner gene. However, the unique optical barcode tag of six fluorescent signals binds possibility of multiple partner genes and permutations of the 30 end of each reporter probe, and a universal biotin tag possible exonic break points in several sarcoma types make binds the 50 end of each capture probe. The tagged reporter RT-PCR even less sensitive than FISH in this setting. probe binds across the RNA fusion junction site between The advent of high-throughput molecular techniques has the 50 and 30 partner exons of the translocation. The 50 opened up new possibilities for the diagnosis of sarcomas biotinylated capture probe binds the RNA sequence bearing gene fusions. One option, the NanoString nCounter directly 30 of the reporter probe on the 30 partner exon of analysis system, uses nucleic acid hybridization probes with the translocation. The protector probe binds the reporter specific and distinguishable fluorescent bar code tags that probe when it is not bound to RNA, in a manner that in- allow detection of multiple specific short nucleic acid creases specificity.

64 jmd.amjpathol.org - The Journal of Molecular Diagnostics NanoString Sarcoma Fusion Detection

RNA Extraction

RNA was extracted from scrolls or unstained sections (5 mm thick; 2 to 3 for excisional/incisional biopsy specimens and 4 to 6 for core biopsy specimens) of FFPE tissue with a High Pure FFPET RNA isolation kit (Roche, Laval, QC, Canada), according to the manufacturer’s instructions. Briefly, on unstained slides, the tissue was moistened with 10 mL xylene per slide, then scraped from the slide with a clean razor blade, and transferred to an Eppendorf tube (Eppendorf, Hamburg, Germany) containing 900 mL xylene for deparaffinization. The tissue was dehydrated, and re- sidual xylene was removed with two washes using absolute ethanol. The dehydrated tissue was proteinase K treated to release nucleic acids. DNA was enzymatically removed, and the RNA was column purified and eluted in 30 to 40 mL elution buffer to maximize the RNA concentration.

Figure 1 Schematic of the probe design for fusion gene detection using the NanoString nCounter Elements platform. In the CodeSet, each fusion CodeSet Assembly and RNA/Probe Hybridization junction is represented by three oligonucleotide probes. Probe A binds the RNA transcript across the fusion junction. Probe B binds directly 30 of probe High-performance liquid chromatographyepurified oligo- A on the RNA transcript. A protector probe binds part of probe A when it is nucleotide probes (Integrated DNA Technologies, San not bound to an RNA transcript for increased specificity. A reporter tag Diego, CA) were combined in an equimolar pool for each fi containing a unique optical barcode speci c to the fusion break point is probe type at 5 nmol/L (reporter probes), 25 nmol/L (cap- attached to each probe A for multiplexed analysis. A biotin tag is attached to each probe B for attachment to a streptavidin-coated cartridge. ture probes), or 10 nmol/L (protector probes) in TE buffer (10 mmol/L Tris, pH 8, and 1 mmol/L EDTA) to assemble the CodeSet. Before each hybridization, an aliquot of each Case Selection probe pool was further diluted 50 in TE buffer/0.1% Tween 20 (MilliporeSigma, Oakville, ON, Canada). Optical Ethics approval was obtained from the appropriate com- barcode tags and universal capture tags were bound to the mittees, as required (University of British Columbia probe oligonucleotides, and the probe pools were combined Research Ethics Board approval number H06-00013, Van- according to the manufacturer’s protocol for 6 or 12 cases couver Coastal Health Research Institute approval number per run. The final concentrations of each probe in the hy- V06-0266, and SingHealth Centralized Institutional Review bridization mixture were 20 pmol/L (reporter probes), 100 Board approval number 2012/450/F). A total of 217 exci- pmol/L (capture probes), and 40 pmol/L (protector probes). sion and core biopsy cases were selected from FFPE A total of 100 to 300 ng of RNA was hybridized to the preserved sarcoma and sarcoma-like tumor samples banked tagged probe set for 18 hours at 67 C, followed by at Vancouver General Hospital (Vancouver, BC, Canada) incubation at 4 C for up to 1 hour. The hybridized RNA/ and KK Women’s and Children’s Hospital (Singapore, probe mixture was cleaned and bound to a streptavidin- Singapore). This included 212 cases used in the final eval- coated cartridge in a NanoString Prep Station (NanoString uation study of the Sarcoma Fusion CodeSet and five Technologies) on the high-sensitivity setting. Unbound additional cases of low-grade fibromyxoid sarcoma/scle- probes and RNA were removed via two rounds of magnetic rosing epithelioid fibrosarcoma used in the preliminary bead purification, leaving only tripartite complexes contain- CodeSet study that were not reevaluated with the final ing RNA with both a reporter and a capture probe bound. The CodeSet. Among these cases, 74 were collected prospec- hybridized RNA/probe tripartite complexes were bound to a tively from Vancouver General Hospital and four additional streptavidin-coated cartridge, then immobilized, and the op- institutions in Canada: BC Cancer Agency (Vancouver, BC, tical barcodes were aligned for analysis. Immediately after Canada), BC Children’s Hospital (Vancouver, BC, Canada), the preparation station completion, the cartridge was sealed Tom Baker Cancer Centre (Edmonton, AB, Canada), and and transferred to a NanoString Digital Analyzer device Royal Alexandra Hospital (Edmonton, AB, Canada). These (NanoString Technologies) for data collection. cases were assessed in batches of six cases per run as real-time clinical samples. Hematoxylin and eosin slides and Data Analysis prior immunohistochemistry and FISH results were reviewed by a subspecialty sarcoma pathologist (T.L.N.). The optical barcodes were counted at a maximum sensitivity of Each case was assigned an anonymized study number 555 fields of view in a second-generation NanoString Digital before RNA extraction and probe hybridization. Analyzer (NanoString Technologies), with each barcode

The Journal of Molecular Diagnostics - jmd.amjpathol.org 65 Chang et al counted equivalent to one hybridized RNA/probe triplex. Raw Miniprep System (Promega, Madison, WI), according to the counts of the optical barcodes for each sample were normalized manufacturer’s instructions. The tissue sections were to a linear standard curve of positive control probes in each deparaffinized in mineral oil heated to 80C for 1 minute to sample with nSolver software version 2.5 (NanoString Tech- melt the paraffin. The tissue was digested with proteinase K, nologies). The normalized count values were exported to Excel and total nucleic acid was column purified. The nucleic acid 2010 software (Microsoft, Redmond, WA), in which a probe was eluted from the column in 40 mL nuclease-free water. median was determined, representing the median normalized Possible sarcoma-related translocations were assessed by count for each probe from the first 24 cases run on a code set. targeted RNA sequencing using the Archer Sarcoma Gene Once the probe median was determined, it remained a fixed Panel (ArcherDX, Boulder, CO), according to the manu- value for all future samples analyzed with this probe. A probe facturer’s instructions. In brief, first-strand cDNA was ratio was calculated by dividing the normalized value by the prepared from 250 ng total nucleic acid with random respective probe median (probe ratio Z normalized count/ primers. After second-strand synthesis, the cDNA was end probe median). The sample median represented the median of repaired and a poly(A) tail was added. Archer M Bar Code the probe ratios for each sample. Finally, a sample ratio (SR) adaptors for Ion Torrent (ArcherDX) were annealed to the was calculated by dividing the probe ratio for each sample by cDNA library. Archer Adaptor primers were ligated to the the sample median for that sample (SR Z probe ratio/sample barcoded cDNA. The cDNA was cleaned up with AMPure median). A sample ratio of 5.0 was considered positive for XP beads (Beckman Coulter, Brea, CA). Two rounds of the corresponding gene fusion. A sample ratio <5.0 was nested PCR were performed with gene-specific primers of considered negative for that fusion. the Archer Sarcoma Gene Panel (ArcherDX) with AMPure XP bead purification after each round of PCR. The cDNA RNA Quality Assessment library was quantitated with a KAPA Library Quantification Kit for Ion Torrent (Kapa Biosystems, Wilmington, MA). RNA quality was assessed by comparison of the expression The quantitated libraries were pooled at equimolar con- of four control genes (ACTB, GAPDH, SDHA, and UBC) centrations and loaded into an enriched PCR at 13 pmol/L included in the CodeSet. The normalized optical barcode using an Ion Torrent Personal Genome Machine Template counts for each control gene were compared with the OT2 200 kit (ThermoFisher Scientific). The combined li- average normalized count for each control gene from the brary was sequenced on an Ion Torrent PGM 318 chip first 24 cases run on each CodeSet. An average normalized (ThermoFisher Scientific). The Ion Torrent Sequencing data expression percentage of 30% for all four control genes was were analyzed with Archer Analysis software version 5.0 required for a case to pass the minimum RNA quality (ArcherDX). assessment. Results RT-PCR Confirmation Proof-of-Concept Study to Establish Assay and Quality Positive fusion calls were confirmed by RT-PCR and Sanger Metric Parameters sequencing across the fusion junction. Cases that were determined fusion positive by FISH, but designated fusion Our ultimate objective was to develop a pan-sarcoma negative with the NanoString assay, were assessed by RT- NanoString CodeSet with panel coverage of most fusion PCR for the most likely fusion variants involving the FISH variants described in all but the rarest types of translocation- positive gene. For each case, cDNA was prepared from 300 associated sarcomas. The assay design approach that was to 500 ng RNA with random hexamers and Superscript IV chosen uses NanoString nCounter junction probes that span (ThermoFisher Scientific, Waltham, MA), according to the unique fusion break point junctions, incorporating the use of manufacturer’s protocol. RT-PCR was performed at either toehold exchange technology for increased specificity, 57Cor60C annealing for 35 cycles. The PCRs were enabling the identification of specific fusion genes. The analyzed on 2% agarose/1 triseborate-EDTA gels at 125 design of the CodeSet was constrained by the NanoString V. For at least one case for each fusion variant, the resulting nCounter Elements technology, which has an upper limit of PCR bands were extracted from the gel with a Qiagen gel 192 probes within a single panel (inclusive of four required extraction kit (Qiagen, Hilden, Germany). The purified PCR normalization genes). bands were then Sanger sequenced. The sequences were As an initial proof-of-concept study toward this objective, aligned by Nucleotide BLAST (https://blast.ncbi.nlm.nih. a preliminary CodeSet that covered 170 fusion variants gov) to confirm the gene fusion junction site. across 27 sarcoma types was developed (Supplemental Table S1). This was developed to establish assay condi- Targeted Next-Generation Sequencing tions and standard operating protocols, including minimal RNA criteria required for the assay, and to determine Total nucleic acid was isolated from 5 sections of FFPE minimum assay readout cut points for calling a positive tissue (10 mm thick) with a Promega ReliaPrep FFPE gDNA fusion. A coverage of >90% of literature-reported sarcoma

66 jmd.amjpathol.org - The Journal of Molecular Diagnostics NanoString Sarcoma Fusion Detection

Table 1 Summary of the Sarcoma Fusion CodeSet Coverage Sarcoma type Fusion Variants, n Alveolar soft part sarcoma ASPSCR1-TFE3 2 Alveolar rhabdomyosarcoma PAX3-FOXO1, PAX7-FOXO1, PAX3-FOXO4, PAX3-NCOA1, 9 PAX3-NCOA2, PAX3-AFX, PAX3-INO80D Aneurysmal bone cyst CDH11-USP6, COL1A1-USP6, OMD-USP6, TRAP150-USP6, 9 ZNF9-USP6 Angiomatoid fibrous histiocytoma EWSR1-ATF1, EWSR1-CREB1, EWSR1-CREM, FUS-ATF1 1 BCOR-rearranged sarcoma BCOR-CCNB3, ZC3H7B-BCOR 2 Biphenotypic sinonasal sarcoma PAX3-MAML3 1 CIC-rearranged sarcoma CIC-DUX4, CIC-FOXO4 3 Clear-cell sarcoma EWSR1-ATF1, EWSR1-CREB1, EWSR1-CREM, FUS-CREM 10 Congenital fibrosarcoma ETV6-NTRK3, EML4-NTRK3 3 Dermatofibrosarcoma protuberans COL1A1-PDGFB 23 (tested using preliminary CodeSet) Desmoplastic small round cell tumor EWSR1-WT1 10 Epithelioid hemangioendothelioma WWTR1-CAMTA1, YAP1-TFE3 3 Ewing sarcoma EWSR1-ERG, EWSR1-FLI1, EWSR1-ETV1, EWSR1-ETV4, 31 EWSR1-FEV, FUS-ERG, FUS-FEV Ewing-like sarcoma (rare variants) EWSR1-NFAT2, EWSR1-SMARCA5, EWSR1-SP3 5 Extraskeletal myxoid chondrosarcoma EWSR1-NR4A3, RBP56-NR4A3, TAF15-NR4A3, TCF12-NR4A3, 8 TFG-NR4A3 High-grade endometrial stromal sarcoma YWHAE-NUTM2, ZC3H7B-BCOR 2 Inflammatory myofibroblastic tumor ATIC-ALK, CARS-ALK, CLTC-ALK, PPFIBP1-ALK, RANBP2-ALK, 17 SEC31A-ALK, TPM3-ALK, TPM4-ALK, NAB2-PDGFRB, TFG-ROS1, YWHAE-ROS1, EML4-ALK, PRKAR1A-ALK, LMNA-ALK, TFG-ALK Lipoblastoma COL1A2-PLAG1, HAS2-PLAG1 2 Low-grade endometrial stromal sarcoma JAZF1-SUZ12, JAZF1-PHF1, MEAF6-PHF1, EPC1-PHF1 7 Mesenchymal chondrosarcoma HEY1-NCOA2, IRF2BP2-CDX1, NUPIO7-LRG5 3 Myoepithelial tumor of soft tissue EWSR1-POU5F1, EWSR1-PBX1, EWSR1-ZNF444 4 Myxoid liposarcoma FUS-DDIT3, EWSR1-DDIT3 16 Nodular fasciitis HMGA2-LPP, MYH9-USP6 5 Pericytoma of bone ACTB-GLI1 5 Synovial sarcoma SS18-SSX1, SS18-SSX2, SS18-SSX4, SS18L1-SSX1, 17 SS18-RESP2-SSX1 Tenosynovial giant-cell tumor COL6A3-CSF1 3

fusion variants was aimed, although this preliminary included in the CodeSet. More important, for the three CodeSet did not include the less common exon-exon junc- low-grade fibromyxoid sarcoma/sclerosing epithelioid tion variants for certain fusion genes. This CodeSet was fibrosarcoma cases, it was shown in each case that there evaluated using a set of FFPE tissues for 33 sarcoma cases, were additional highly variable inserted sequences at the 16 of which were previously demonstrated to be fusion fusion exon-exon junction, a phenomenon seen commonly positive by clinically validated FISH assays (data not in FUS-CREB3L2 fusion genes.95,96 There was one false- shown). Fusion gene expression in 11 of the 16 FISH- negative case each of Ewing sarcoma and clear-cell positive cases was demonstrated, including three of four sarcoma, with both found to express less common variants Ewing sarcomas, three of three synovial sarcomas, two of of the EWSR1-FLI1 (exon 7 to exon 7) and EWSR1-ATF1 two myxoid liposarcomas, one of two clear-cell sarcomas, (exon 8 to exon 4) fusions, respectively, that were not one of one high-grade endometrial stromal sarcoma, one of covered by the probes in the preliminary CodeSet. one myoepithelial tumor of soft tissue, and none of three The minimum input requirements acceptable for sub- low-grade fibromyxoid sarcoma/sclerosing epithelioid mitting a sample for the assay were determined. On the fibrosarcoma cases. RT-PCR confirmed the presence of the basis of the general requirements for the NanoString specific fusion variants identified by NanoString in these 11 nCounter assay, samples in which at least 100 ng of RNA cases. For the remaining five cases, RT-PCR demonstrated could be obtained at a concentration of 10 ng/mLwereused that these false negatives were related to absence of the (although samples with lower RNA quantity were also specific fusion variant in the CodeSet rather than failure tested). For all cases, two to three scrolls or unstained of the NanoString assay to identify a fusion type that was slides (5 mm thick) derived from excisions or incisional

The Journal of Molecular Diagnostics - jmd.amjpathol.org 67 Chang et al

Table 2 Evaluation Study of the Sarcoma Fusion CodeSet Using a Cohort of 212 Cases NanoString fusion positive NanoString fusion negative RNA failure Prior Prior Prior Prior testing No prior testing þ testing Total testing þ (true positive testing (false (true No prior Prior Prior No prior Diagnosis cases (true positive) unexpected) (true positive) negative) negative) testing testing þ testing testing Ewing sarcoma 20 16 4 Nodular fasciitis 14 2 1 4 51 1 Synovial sarcoma 11 8 3 Dermatofibrosarcoma 11 2 6 11 1 protuberans Low-grade endometrial 12 5 2 32 stromal sarcoma High-grade endometrial 43 1 stromal sarcoma Alveolar soft part 52 21 sarcoma Clear-cell sarcoma 95 13 Myoepithelioma of 52 1 2 soft tissue Myxoid liposarcoma 55 BCOR-rearranged sarcoma 3 3 CIC-rearranged sarcoma 4 1 1 11 Infantile fibrosarcoma 33 Alveolar 55 rhabdomyosarcoma Desmoplastic small 22 round cell tumor Angiomatoid fibrous 2 1 1 histiocytoma Mesenchymal 2 11 chondrosarcoma Epithelioid 7 11 4 1 hemangioendothelioma Extraskeletal myxoid 11 chondrosarcoma Secretory breast 11 carcinoma Low-grade sarcoma, not 1 1 otherwise specified Undifferentiated spindle 11 1 54 1 cell sarcoma Undifferentiated round 6 6 cell sarcoma Additional fusion- 8 8 associated mesenchymal tumors* Malignant mesenchymal 32 14 15 12 tumors, not fusion associatedy Malignant 14 66 2 nonmesenchymal tumorsz (table continues)

68 jmd.amjpathol.org - The Journal of Molecular Diagnostics NanoString Sarcoma Fusion Detection

Table 2 (continued) NanoString fusion positive NanoString fusion negative RNA failure Prior Prior Prior Prior testing No prior testing þ testing Total testing þ (true positive testing (false (true No prior Prior Prior No prior Diagnosis cases (true positive) unexpected) (true positive) negative) negative) testing testing þ testing testing Benign mesenchymal 9 23 13 tumors, not fusion associatedx Nonneoplastic/reactive 5 12 2 lesions{ Total cases 212 63 15 18 4 38 55 4 5 10

All data are expressed as n. *Aneurysmal bone cyst (n Z 3), lipoblastoma (n Z 1), inflammatory myofibroblastic tumor (ALK negative) (n Z 2), tenosynovial giant-cell tumor (n Z 1), and malignant solitary fibrous tumor (n Z 1). yOsteosarcoma (n Z 4), nonalveolar rhabdomyosarcoma (n Z 3), chondrosarcoma (n Z 3), poorly differentiated spindle cell sarcoma (intracranial) (n Z 3), epithelioid sarcoma (n Z 2), rhabdoid tumor of kidney (n Z 2), perivascular epithelioid cell tumor (n Z 2), undifferentiated pleomorphic sarcoma (n Z 2), low-grade myofibroblastic sarcoma (n Z 2), high-grade spindle cell sarcoma of genital origin (n Z 1), malignant peripheral nerve sheath tumor (n Z 1), uterine smooth muscle tumor of uncertain malignant potential (n Z 1), dedifferentiated liposarcoma (n Z 1), mixed-type liposarcoma (n Z 1), leiomyosarcoma (n Z 1), adamantinoma (n Z 1), atypical teratoid/rhabdoid tumor (n Z 1), and uterine adenosarcoma (n Z 1). zMelanoma (n Z 4), poorly differentiated neuroendocrine malignancy (n Z 2), poorly differentiated germ cell malignancy (n Z 1), poorly differentiated sarcomatoid carcinoma (n Z 1), poorly differentiated epithelioid malignancy (pituitary carcinoma) (n Z 1), poorly differentiated carcinoma (n Z 1), olfactory neuroblastoma (n Z 1), endometrial carcinoma (n Z 1), adenoid cystic carcinoma (n Z 1), and mesothelioma (n Z 1). xDesmoplastic fibroblastoma (n Z 1), kaposiform hemangioendothelioma (n Z 1), angiofibroma of soft tissue (n Z 1), benign fibrous histiocytoma (n Z 1), benign fibrous histiocytoma of bone (n Z 1), fibrohistiocytic tumor not otherwise specified (n Z 1), ALK-positive histiocytosis not otherwise specified (n Z 1), desmoid-type fibromatosis (n Z 1), and endometrial stromal nodule (n Z 1). {Normal lymph node (n Z 1), necrotic tissue (status after resection of Ewing sarcoma) (n Z 1), reactive fibroblastic proliferation (n Z 1), inflamed synovial tissue (n Z 1), and splenic hamartoma (n Z 1). þ, Present; , absent. biopsy specimens were submitted, whereas four to six In all assay-positive cases, the normalized SR metrics used scrolls/unstained slides were submitted from core biopsy as the assay readout were significantly higher than background specimens. When possible, the scrolls or unstained slides values for the remaining probes in the assay, and consistently were obtained under RNase-free conditions, although in higher than the SR of 5 used as our cutoff for an unequivocally many instances samples were taken under noneRNase-free positive case. No cases expected to be fusion negative conditions in a clinical laboratory. demonstrated an SR >5 for any probe pair; hence, our assay did In this proof-of-concept phase, samples, regardless of not read out any false-positive results. the RNA quantity or concentration, were submitted, with an RNA quantity/quality metric determined from the assay Design of the Sarcoma Fusion CodeSet data. This metric was based on the quantified RNA expression of the normalization genes in the assay (SDHA, On the basis of the results of the above proof-of-concept UBC, GAPDH,andACTB)incomparisontotheaverage study, a Sarcoma Fusion CodeSet with improved coverage normalization gene expression in our study. This value of fusion variants was designed for the most common and ranged from 4% to 586% among the samples. As a quality clinically important tumor types (Table 1). However, control cutoff, samples showing >30% averaged normali- because of the current ceiling of 192 probes in a single zation gene expression were considered as acceptable NanoString Elementsebased CodeSet panel, coverage for (based on average expression observed across all tested certain diagnoses needed to be removed to accommodate samples). However, if a fusion gene was detected in such added probes. Therefore, probes for low-grade fibromyx- samples with <30% averaged normalization gene oid sarcoma and dermatofibrosarcoma protuberans were expression, these cases were still considered to be fusion removed because of the genomically complex positive. Indeed, fusion gene expression (with SR >5) was FUS-CREB3L2 fusions and high variability of COL1A1- detected in three cases, including one synovial PDGFB fusion break points in low-grade fibromyxoid sarcoma, one myoepithelioma of soft tissue, and one der- sarcoma cases and the lack of clinical need for ancillary matofibrosarcoma protuberans. In addition to these three diagnostics in most dermatofibrosarcoma protuberans cases, two additional cases showed <30% averaged cases. In their place, additional probes were added tar- normalizationgeneexpressionwithnofusiongene geting rare variants of certain fusions already in the panel, expression detected; these were deemed as RNA failure including variant EWSR1-FLI1 splice sites and related and considered uninterpretable. Ewing sarcoma fusions (increasing from 20 to 31 probes),

The Journal of Molecular Diagnostics - jmd.amjpathol.org 69 Chang et al clear-cell sarcoma fusions (from 4 to 10 probes), desmo- additional FISH studies for FLI1). One case of plastic small round cell tumor fusions (from 6 to 10 mesenchymal chondrosarcoma demonstrated the HEY1- probes), myxoid liposarcoma fusions (from 8 to 16 NCOA2 fusion despite testing negative for this fusion by probes), and synovial sarcoma fusions (from 15 to 17 targeted next-generation sequencing (NGS). One case of probes). Probes were also added targeting novel fusions angiomatoid fibrous histiocytoma showed expression of the recently described in alveolar rhabdomyosarcoma (eg, EWSR1-CREB1 fusion, with prior RT-PCR testing showing PAX3-INO80D)andinfantilefibrosarcoma (eg, EML4- no expression of the EWSR1-ATF1 fusion. One case of NTRK3). The remaining probes were identical to those nodular fasciitis showed expression of MYH9-USP6, with used in the proof-of-concept CodeSet. The Sarcoma prior break-apart FISH testing showing a single intact USP6 Fusion CodeSet, therefore, covered 174 fusion variants signal, suggestive of monosomy or a deletion event but across 25 sarcoma types. For cases of suspected derma- without diagnostic evidence of USP6 rearrangement. Three tofibrosarcoma protuberans, samples were tested using the cases of undifferentiated round cell sarcoma showed BCOR- preliminary CodeSet. CCNB3 expression. All three cases tested negative for various gene rearrangements by FISH in the routine Evaluation Study of the Sarcoma Fusion CodeSet molecular workup of undifferentiated round cell or spindle cell sarcomas (but had not been subjected to BCOR-CNNB3 In total, 212 cases were tested in an evaluation study of the testing because it was not available at the originating Sarcoma Fusion CodeSet (Table 2). This cohort included institution) (Supplemental Table S2). One case of soft tissue FFPE samples of both adult and pediatric sarcoma/soft myoepithelioma showed expression of the rare tissue tumor cases, obtained from multiple institutions (in IRF2BP2-CDX1 fusion; this case had tested negative for Canada and Singapore). These FFPE blocks were obtained EWSR1 rearrangement by FISH. One case of undifferenti- from 1995 to 2017. Among these cases, 71 were fusion ated spindle cell sarcoma showed ZC3H7B-BCOR expres- positive by routine clinical FISH or RT-PCR testing, 57 sion. One case of a poorly differentiated spindle cell tested as fusion negative based on FISH or RT-PCR sarcoma showing CIC-DUX4 expression was also identified, testing for the presumed gene fusion or gene which had been found to be FISH negative for SS18 rear- rearrangement of interest, and 84 had not undergone any rangements or MDM2 amplification during its original prior molecular testing. Among the cases without prior diagnostic workup. One case of epithelioid hemangioendo- testing or with prior negative testing results, a validated thelioma showed unexpected expression of a YAP1-TFE3 clinical FISH or RT-PCR fusion gene assay was not fusion. Finally, one case of a low-grade sarcoma, type not available. Hence, among the 141 cases with no prior otherwise specified, arising in the pelvis of a male patient, testing or with only negative FISH/RT-PCR results, showed unexpected expression of COL1A1-PDGFB. In all several could still carry fusion genes. of these cases demonstrating previously nonidentified fusion Among the 71 fusion-positive cases by routine clinical gene expression, subsequent in-house RT-PCR and Sanger testing, 4 failed the determined RNA quality metric (ie, sequencing studies confirmed the expression of these <30% averaged normalization gene expression). Of the NanoString-detected fusion genes. remaining 67 cases, the NanoString assay confirmed fusion Therefore, in total, the Sarcoma Fusion CodeSet gene expression in 63 cases (89%), including 16 of 16 demonstrated fusion gene expression in 96 of the 212 cases Ewing sarcomas, eight of eight synovial sarcomas, five of (Table 2). This included all tested cases of Ewing sarcoma, five myxoid liposarcomas, five of five alveolar rhabdo- synovial sarcoma, myxoid liposarcoma, alveolar rhabdo- myosarcomas, and three of three high-grade endometrial myosarcoma, infantile fibrosarcoma, desmoplastic small stromal sarcomas (Table 2). In addition, 19 cases with no round cell tumor, and mesenchymal chondrosarcoma. No prior FISH/RT-PCR clinical testing were fusion positive by false-positive results were observed. NanoString (Table 2). The NanoString assay failed to detect four fusions that The NanoString assay detected fusions in 15 cases were positive by FISH, RT-PCR, or NGS. The assay was negative by clinical FISH/RT-PCR. This included four cases positive in only five of six fusion-proved clear-cell sarcoma of Ewing sarcoma found to be negative for EWSR1 cases, with the NanoString-negative case showing rear- rearrangement by FISH or RT-PCR clinical diagnostic rangement of both EWSR1 and ATF1 by break-apart FISH. assays. In two of these cases, clinical RT-PCR testing was Subsequent Sanger sequencing of the fusion junction negative because they expressed unusual fusion variants showed an unconventional EWSR1 exon 8eATF1 exon 4 (EWSR1 exon 10eFLI1 exon 8 and EWSR1 exon fusion, with a 100-nucleotide deletion spanning a portion of 10eFLI1 exon 6) that are not detectable with the both EWSR1 and ATF1 and a 30-nucleotide duplication of clinically used panel of RT-PCR primers. In the two other ATF1 exon 4 inserted near the junction site (Supplemental cases, clinical FISH testing was noninformative because of Table S1). Three additional cases of clear-cell sarcoma did insufficient material in one case, whereas the other case not demonstrate fusion gene expression either by prior showed a nonrearranged EWSR1 result with two normal FISH/RT-PCR testing or with the NanoString assay, with fused signals (insufficient material was available for the diagnoses made on the basis of detailed review of

70 jmd.amjpathol.org - The Journal of Molecular Diagnostics NanoString Sarcoma Fusion Detection

Table 3 Sensitivity/Specificity Analysis Based on the Evaluation Study of the Sarcoma Fusion CodeSet Prior clinical testing Prior clinical testing Variable fusion positive fusion negative No prior testing Total NanoString fusion 63 (True positives) 15 (True positives; 18 (True positives; 96 positive RT-PCR/Sanger confirmed positive) RT-PCR/Sanger confirmed positive) NanoString fusion 4 (False negatives) 38 (True negatives) 55 97 negative RNA failure 4 5 10 19 Total 71 57 84 212

All data are expressed as n. clinical features and exclusion of malignant melanoma as an endometrial stromal sarcoma cases, the SR value (4.1) alternative diagnosis. In addition, the NanoString assay was was just lower than the threshold cutoff value of 5, sug- positive in only five of six alveolar soft part sarcoma cases, gesting that this may have demonstrated a true-positive with one NanoString-negative case showing TFE3 rear- result if more RNA was available for this sample. Both of rangement by FISH, suggesting that this represents a case these low-grade endometrial stromal sarcoma samples were expressing a novel TFE3-related fusion that is not covered taken from archival tissue blocks >10 years old, which may in this CodeSet. One case of CIC-rearranged sarcoma was have contributed to the low RNA quality. found by RT-PCR to express a fusion variant of CIC-DUX4, On the basis of the 193 cases that met our minimum with a novel break point within DUX4 that results in RNA quality control metrics, the Sarcoma Fusion CodeSet retention of most of the DUX4 gene.97 Finally, one case showed a test sensitivity for fusion gene detection of 96% of angiomatoid fibrous histiocytoma showed expression by [true positives/(true positives þ false negatives)] and a RT-PCR of an EWSR1-ATF1 variant (exon 11 to exon 3) specificity of 100% [true negatives/(true negatives þ false that was not represented in the CodeSet. positives)] (Table 3). Thirty-three cases were core biopsy The assay was also unable to detect a fusion in four of specimens with an RNA input range of 36 to 300 ng. A seven epithelioid hemangioendothelioma cases (with one diagnosis was attainable for 27 (81.8%) of these core further case failing the RNA quality/quantity metric), biopsy cases. although RT-PCR testing also could not definitively demonstrate WWTR1-CAMTA1 fusions in any of these Quality Control Metric Results from Evaluation Study cases (data not shown). A few cases of aneurysmal bone cyst, tenosynovial giant-cell tumor, and lipoblastoma were Among all of the cases tested, the averaged normalization also tested with this CodeSet but were fusion negative, gene expression values ranged from 4% to 1110%. In total, although again RT-PCR testing could not demonstrate 19 cases showed <30% averaged normalization gene fusion gene expression in any of these cases (data not expression. However, fusion gene expression (with shown). SR >5) was detected in three of these cases. This included A total of 32 nonefusion-associated malignant mesen- one case of dermatofibrosarcoma protuberans showing chymal tumors, 9 nonefusion-associated mesenchymal COL1A1-PDGFB expression (SR, 363.2) with averaged tumors, and 14 nonmesenchymal tumors were also tested, normalization gene expression of 15%, one case of representing cases in which a fusion-associated sarcoma was myoepithelioma of soft tissue showing EWSR1-NR4A3 considered in the differential diagnosis at some point. expression (SR, 5.9) with averaged normalization gene Clinical FISH or RT-PCR testing was performed in a large expression of 22%, and one case of synovial sarcoma proportion of these cases to rule out possible gene showing SS18/SSX expression (SR, 11.5) with averaged rearrangements; again, no false-positive results were normalization gene expression of 26%. There were no demonstrated. One case of secretory breast carcinoma false-positive results in cases with <30% averaged demonstrated expression of ETV6-NTRK3 by the Nano- normalization gene expression. String assay, consistent with FISH results showing an ETV6 Further verification of our results was achieved by gene rearrangement (data not shown). testing a subset of nine samples at two different sites Among the 212 cases, 19 were not suitable for molecular (Canada and Singapore) using different reagent lots and testing, with the samples rejected on the basis of poor RNA consumables, and different nCounter analyzer machines quantity or quality using the <30% averaged normalization run independently by different technologists, although the gene expression cutoff. This included four cases that were same CodeSet oligonucleotides were shared between sites. tested as fusion positive, including one nodular fasciitis case Data analysis was also performed in the same manner, with FISH positive for USP6 rearrangement, one CIC-rearranged median probe signal values calculated at the Vancouver sarcoma case positive for CIC-DUX4 by targeted NGS, and site also used to determine the sample ratio values at the two low-grade endometrial stromal sarcoma cases FISH Singapore site. Overall, there was 100% concordance in positive for JAZF1-SUZ12. In one of these two low-grade these results (data not shown).

The Journal of Molecular Diagnostics - jmd.amjpathol.org 71 Chang et al

Cost Analysis NanoString nCounter platform allows for use of RNA of relatively poor quality from FFPE tissues, reflected in our The reagent cost of one NanoString assay run (testing 174 relatively low rate of RNA failure cases in our evaluation fusion variants) is comparable to the cost of one FISH study. Furthermore, the test turnaround time is shorter and analysis.98 However, when batched as six cases per run, the costs are lower than for NGS approaches because neither technologist time required per case for the NanoString assay library preparation nor nucleic acid amplification steps are was 1.2 hours, compared with an estimated 4 hours per case needed. The data analysis complexity is negligible compared for FISH (including scoring time). Therefore, the total per with NGS, which requires specialized bioinformatics pipe- sample cost for the NanoString assay is less than the cost of lines and sequence analysis strategies. The disadvantages a single FISH assay run, particularly when considering the relate to the slightly higher RNA quantity requirement for the total per case cost for FISH because multiple genes often NanoString assay (approximately 100 ng minimum) and, need to be analyzed by FISH to arrive at a molecular more important, the inability to detect unexpected novel diagnosis. Therefore, the NanoString assay appears to be fusion gene variants, including those with high variability at cost effective for sarcoma molecular diagnosis compared the expected fusion gene exon-exon junctions. Targeted NGS with available assays. fusion gene assays are not necessarily always able to detect such fusion gene variants with alternate exon-exon junctions. These can sometimes incorporate large nucleic acid insertions Discussion at the junction that result in amplicons that may exceed the sequencing capabilities of NGS panel sequences that are Herein, we describe the design and evaluation of a Nano- often coupled to a PCR-based target amplification step. In String nCounterebased molecular diagnostic assay for the addition, the Elements-based NanoString assay is flexible and detection of fusion genes in sarcomas. On the basis of the high allows for new probes to be added when novel exon-exon specificity of the assay (with no false-positive results arising junctions are discovered. in our evaluation studies), along with the low false-negative An alternative NanoString-based approach we had also rate, this assay can be of utility in the clinical setting for the considered uses a 50/30 positional gene expression imbalance sarcoma types included in the CodeSet. This includes most of approach, in which probes are designed that compare the the sarcoma types covered by routine molecular diagnostic ratio of gene expression upstream and downstream of the services and many novel sarcoma types not covered by fusion junctions of interest, avoiding the challenges of available molecular testing. In conjunction with the relatively designing probes targeting each fusion break point sequence low assay cost and technologist time required to run the assay, variant. However, this approach was not deemed suitable for along with the quick turnaround time and simple bioinfor- sarcoma fusion gene identification, particularly because a matics analysis, this could play a role as a first-line diagnostic gene expression imbalance of the exons across the fusion test, with the potential to replace many standard molecular junction is not necessarily expected. This is because most FISH and RT-PCR tests with a single more comprehensive sarcoma fusion transcripts do not result from promoter assay. exchange and subsequent overexpression of the 30-fusion In comparison to FISH and RT-PCR, the NanoString partner (as seen with ALK and related lung cancer fusions) assay offers several significant benefits. First, the panel- and, therefore, do not necessarily lead to differential based approach to fusion gene detection alleviates the need expression of exons across the break point for either fusion for running multiple individual FISH or RT-PCR tests for partner. Another drawback of the expression imbalance single genes or single fusion variants. Second, the assay approach is that it is agnostic to the corresponding fusion significantly improves turnaround time and labor costs, partner of the assayed gene of interest, the knowledge of while allowing for complete characterization of the fusion which is important for the correct identification of many gene partners and exon-exon junction that could potentially sarcoma types. An additional technical limitation at the be of importance for clinical management. For example, one present time is the probe capacity of 192 probes within the case of widespread intra-abdominal and pelvic small round NanoString Elements-based CodeSet. The current assay blue cell sarcoma was encountered that by clinical FISH uses 174 probes, with an additional four probes serving as assay was EWSR1 rearrangement positive but WT1 and control probes; hence, there is insufficient capacity for FLI1 rearrangement negative. The NanoString assay iden- probes to identify 50-30 imbalances. As part of future tified an EWSR1-ERG fusion in this case, allowing it to be expansion of the assay, we plan to look into testing whether classified as a Ewing sarcoma rather than a desmoplastic 50-30 imbalances could be sufficient in select genes to detect round cell tumor (a tumor that would not respond to Ewing gene translocations in sarcoma entities known to have a chemotherapy). large degree of variability at the fusion junction (eg, CIC- In comparison to a targeted NGS approach to fusion gene rearranged sarcoma, low-grade fibromyxoid sarcoma, and identification, the NanoString assay offers several advantages sclerosing epithelioid fibrosarcoma). but also some disadvantages. The advantages are mainly Accurate determination of the sensitivity of the assay is related to test efficiency and cost-effectiveness, because the challenging because of the wide range of rare individual

72 jmd.amjpathol.org - The Journal of Molecular Diagnostics NanoString Sarcoma Fusion Detection

Table 4 Summary of Expected Diagnostic Coverage for Fusion-Associated Sarcoma Types Category A: >95% Category B: 75%e95% Category C: <75% Category D: coverage Category E: fusion-associated estimated coverage estimated coverage estimated coverage not yet validated sarcomas not covered by assay Alveolar rhabdomyosarcoma Alveolar soft part sarcoma Epithelioid Aneurysmal bone cyst Low-grade fibromyxoid hemangioendothelioma sarcoma BCOR-rearranged sarcoma Angiomatoid fibrous CIC-rearranged sarcoma Biphenotypic sinonasal Sclerosing epithelioid histiocytoma sarcoma fibrosarcoma Congenital fibrosarcoma Clear-cell sarcoma Ewing-like sarcoma Myxoinflammatory (rare variants) fibroblastic sarcoma Dermatofibrosarcoma Extraskeletal myxoid Inflammatory Pseudomyogenic protuberans chondrosarcoma myofibroblastic tumor hemangioendothelioma Desmoplastic small round Low-grade endometrial Lipoblastoma Solitary fibrous tumor cell tumor stromal sarcoma High-grade endometrial Myoepithelial tumor Pericytoma of bone stromal sarcoma of soft tissue Ewing sarcoma Nodular fasciitis Tenosynovial giant-cell tumor Mesenchymal chondrosarcoma Myxoid liposarcoma Synovial sarcoma

diagnoses that it covers, most of which lack a corresponding these false-negative results was related to the fusion variant clinically validated assay to compare with. Therefore, coverage of our assay rather than any inability of the assay although it would have been most informative to determine to detect fusions targeted by the CodeSet (ie, these are the sensitivity and specificity of the assay for each diag- diagnostic false negatives but not technical false negatives). nostic tumor entity and fusion gene variant, the number of In contrast, there was a case of mesenchymal chon- cases for individual entities is unavoidably low. Instead, a drosarcoma that was fusion positive by the NanoString single sensitivity and specificity value is reported for all assay but fusion negative even by targeted NGS assay, diagnostic types combined. These reported values well suggesting that the detection limit of our NanoString assay approximate the performance of the assay in a true clinical might be lower than NGS in certain instances. setting because the distribution of cases used in our study With regard to the quality control aspects of our assay, reflects the distribution and relative prevalence of the the use of the averaged normalization gene expression various translocation-associated sarcoma types encountered percentage cutoff of <30% is a suitable screen for RNA at pediatric and adult sarcoma centers. quality. Alternative approaches would have been an RNA Determination of the true sensitivity of our assay is also quantity cutoff or quality cutoff (eg, 260/280 ratio). How- limited by the imperfect sensitivity of the orthogonal gold ever, fusion genes were detected in many cases with low standard methods used (namely, FISH and RT-PCR). input RNA quantity, including one case of dermatofi- Indeed, the ideal orthogonal assay to use for our evaluation brosarcoma protuberans demonstrating COL1A1-PDGFB study would have been whole-genome or whole- expression with only 36 ng RNA input. To address cases transcriptome NGS of the cases, but this would not have that show unexpected RNA failure or unexpected absence of been feasible because of the costs for such a study, and does fusion gene expression (which we expect in 5% to 10% of not represent the clinical standard. Overall, the NanoString test cases based on our evaluation study), a targeted or assay was able to identify fusions in 15 cases for which whole-transcriptome NGS approach could be considered as fusions were not identified by clinical FISH and/or RT- areflex assay for such cases. PCR; however, there were four false-negative cases (ie, The identification of specific genetic abnormalities, NanoString panel assay negative but FISH and/or RT-PCR including gene fusions, is an integral aspect of sarcoma positive). In two of these false-negative cases (one case each diagnosis. It has been suggested that molecular testing of angiomatoid fibrous histiocytoma and CIC-rearranged should be routinely used for accurate diagnosis and appro- sarcoma), the fusions were identified by NGS-based assay priate clinical management and prognostication, particularly (eg, anchored multiplex PCR-based sarcoma fusion assay), for nonpleomorphic round and spindle cell sarcomas that are but would not have been detected by currently used clinical otherwise undifferentiated by hematoxylin and eosin molecular diagnostic tests. For each of the false-negative morphology and immunophenotype alone.3 We envision cases, the identified fusion gene variant was not repre- this assay as a potential first-line molecular diagnostic test sented in the tested CodeSet. In other words, the reason for for nonpleomorphic sarcomas, particularly for the tumor

The Journal of Molecular Diagnostics - jmd.amjpathol.org 73 Chang et al types with >75% fusion variant coverage (Table 4). The 3. Italiano A, Di Mauro I, Rapp J, Pierron G, Auger N, Alberti L, diagnostic role for this assay remains to be ascertained for a Chibon F, Escande F, Voegeli AC, Ghnassia JP, Keslair F, Lae M, few tumor types (Table 4), whereas use of alternative Ranchere-Vince D, Terrier P, Baffert S, Coindre JM, Pedeutour F: Clinical effect of molecular methods in sarcoma diagnosis (GEN- diagnostic assays (eg, FISH and NGS) may be better for SARC): a prospective, multicentre, observational study. Lancet Oncol entities in other categories that are known to have a high 2016, 17:532e538 degree of variability at the fusion junction sites. We are 4. Barr FG, Womer RB: Molecular diagnosis of ewing family tumors: actively working to expand the CodeSet to improve too many fusions.? J Mol Diagn 2007, 9:437e440 coverage for these diagnostic types. Overall, on the basis of 5. Demicco EG: Sarcoma diagnosis in the age of molecular pathology. Adv Anat Pathol 2013, 20:264e274 our evaluation study, there will be a 4% false-negative rate; 6. Turc-Carel C, Pedeutour F, Durieux E: Characteristic chromosome 99,100 this is similar to the sensitivity of FISH. An ideal abnormalities and karyotype profiles in soft tissue tumors. Curr Top approach to such cases would be secondary testing with Pathol 1995, 89:73e94 FISH or a targeted NGS-based approach, such as an 7. Lee W, Han K, Harris CP, Shim S, Kim S, Meisner LF: Use of FISH anchored multiplex-PCR assay for fusion gene detection, to detect chromosomal translocations and deletions: analysis of chromosome rearrangement in synovial sarcoma cells from paraffin- because this would comprehensively assess for the partic- embedded specimens. Am J Pathol 1993, 143:15e19 ularly rare variant translocations that cannot be efficiently 8. Taylor C, Patel K, Jones T, Kiely F, De Stavola BL, Sheer D: covered by an Elements-based NanoString CodeSet. Diagnosis of Ewing’s sarcoma and peripheral neuroectodermal Because of the cost effectiveness of the NanoString assay tumour based on the detection of t(11;22) using fluorescence in situ e combined with its short assay turnaround time and ease of hybridisation. Br J Cancer 1993, 67:128 133 9. Chen QR, Vansant G, Oades K, Pickering M, Wei JS, Song YK, data analysis, this single assay can replace most clinically Monforte J, Khan J: Diagnosis of the small round blue cell tumors available FISH and RT-PCR assays as the initial molecular using multiplex polymerase chain reaction. J Mol Diagn 2007, 9: diagnostic test for sarcomas. The assay is also amenable to 80e88 iterative updates based on the NanoString Elements 10. Geiss GK, Bumgarner RE, Birditt B, Dahl T, Dowidar N, chemistry so that newly described fusion variants can be Dunaway DL, Fell HP, Ferree S, George RD, Grogan T, James JJ, Maysuria M, Mitton JD, Oliveri P, Osborn JL, Peng T, Ratcliffe AL, easily brought into the CodeSet without the need to Webster PJ, Davidson EH, Hood L, Dimitrov K: Direct multiplexed regenerate the entire panel. From a practical perspective, measurement of gene expression with color-coded probe pairs. Nat sarcoma specimens can also be batched with other Biotechnol 2008, 26:317e325 NanoString assays performed in parallel in the same 11. Suehara Y, Arcila M, Wang L, Hasanovic A, Ang D, Ito T, Kimura Y, Drilon A, Guha U, Rusch V, Kris MG, Zakowski MF, run. In summary, our custom-designed NanoString fi e Rizvi N, Khanin R, Ladanyi M: Identi cation of KIF5B-RET and Elements based sarcoma assay is highly sensitive and GOPC-ROS1 fusions in lung adenocarcinomas through a compre- specific, making it suitable for prospective evaluation in a hensive mRNA-based screen for tyrosine kinase fusions. Clin Cancer clinical setting for sarcoma molecular diagnosis. Res 2012, 18:6599e6608 12. Ladanyi M, Lui MY, Antonescu CR, Krause-Boehm A, Meindl A, Argani P, Healey JH, Ueda T, Yoshikawa H, Meloni-Ehrig A, Acknowledgments Sorensen PH, Mertens F, Mandahl N, van den Berghe H, Sciot R, Dal Cin P, Bridge J: The der(17)t(X;17)(p11;q25) of human alveolar soft We thank Drs. Malcolm Hayes and Sona Sihra (British part sarcoma fuses the TFE3 transcription factor gene to ASPL, a e Columbia Cancer Agency and University of British novel gene at 17q25. Oncogene 2001, 20:48 57 13. Williams A, Bartle G, Sumathi VP, Meis JM, Mangham DC, Columbia, Vancouver, BC, Canada), Lien Hoang and Grimer RJ, Kindblom LG: Detection of ASPL/TFE3 fusion tran- Andrew Churg (Vancouver General Hospital and University scripts and the TFE3 antigen in formalin-fixed, paraffin-embedded of British Columbia, Vancouver, BC, Canada), Victor Lee tissue in a series of 18 cases of alveolar soft part sarcoma: useful (National University Health System, Singapore, Singapore), diagnostic tools in cases with unusual histological features. Virchows e and Sittampalam Kesavan and Sathiya Selvarajan Arch 2011, 458:291 300 14. Moller E, Mandahl N, Mertens F, Panagopoulos I: Molecular iden- (Singapore General Hospital, Singapore, Singapore) for tification of COL6A3-CSF1 fusion transcripts in tenosynovial giant contributing cases toward this study. cell tumors. Genes Chromosomes Cancer 2008, 47:21e25 15. Barr FG: Gene fusions involving PAX and FOX family members in alveolar rhabdomyosarcoma. Oncogene 2001, 20:5736e5746 Supplemental Data 16. Barr FG, Qualman SJ, Macris MH, Melnyk N, Lawlor ER, Strzelecki DM, Triche TJ, Bridge JA, Sorensen PH: Genetic hetero- Supplemental material for this article can be found at geneity in the alveolar rhabdomyosarcoma subset without typical https://doi.org/10.1016/j.jmoldx.2017.09.007. gene fusions. Cancer Res 2002, 62:4704e4710 17. Sumegi J, Streblow R, Frayer RW, Dal Cin P, Rosenberg A, Meloni- Ehrig A, Bridge JA: Recurrent t(2;2) and t(2;8) translocations in References rhabdomyosarcoma without the canonical PAX-FOXO1 fuse PAX3 to members of the nuclear receptor transcriptional coactivator family. 1. Guillou L, Aurias A: Soft tissue sarcomas with complex genomic Genes Chromosomes Cancer 2010, 49:224e236 profiles. Virchows Arch 2010, 456:201e217 18. Wang X, Bledsoe KL, Graham RP, Asmann YW, Viswanatha DS, 2. Mertens F, Antonescu CR, Hohenberger P, Ladanyi M, Modena P, Lewis JE, Lewis JT, Chou MM, Yaszemski MJ, Jen J, Westendorf JJ, D’Incalci M, Casali PG, Aglietta M, Alvegard T: Translocation- Oliveira AM: Recurrent PAX3-MAML3 fusion in biphenotypic related sarcomas. Semin Oncol 2009, 36:312e323 sinonasal sarcoma. Nat Genet 2014, 46:666e668

74 jmd.amjpathol.org - The Journal of Molecular Diagnostics NanoString Sarcoma Fusion Detection

19. Oliveira AM, Hsi BL, Weremowicz S, Rosenberg AE, Dal Cin P, pathological, and genetic (COL1A1-PDGFB) study with therapeutic Joseph N, Bridge JA, Perez-Atayde AR, Fletcher JA: USP6 (Tre2) implications. Histopathology 2009, 54:860e872 fusion oncogenes in aneurysmal bone cyst. Cancer Res 2004, 64: 34. Maire G, Martin L, Michalak-Provost S, Gattas GJ, Turc-Carel C, 1920e1923 Lorette G, Pedeutour F: Fusion of COL1A1 exon 29 with 20. Oliveira AM, Perez-Atayde AR, Dal Cin P, Gebhardt MC, Chen CJ, PDGFB exon 2 in a der(22)t(17;22) in a pediatric giant cell Neff JR, Demetri GD, Rosenberg AE, Bridge JA, Fletcher JA: fibroblastoma with a pigmented Bednar tumor component: evi- Aneurysmal bone cyst variant translocations upregulate USP6 tran- dence for age-related chromosomal pattern in dermatofi- scription by promoter swapping with the ZNF9, COL1A1, TRAP150, brosarcoma protuberans and related tumors. Cancer Genet and OMD genes. Oncogene 2005, 24:3419e3426 Cytogenet 2002, 134:156e161 21. Antonescu CR, Dal Cin P, Nafa K, Teot LA, Surti U, Fletcher CD, 35. O’Brien KP, Seroussi E, Dal Cin P, Sciot R, Mandahl N, Fletcher JA, Ladanyi M: EWSR1-CREB1 is the predominant gene fusion in Turc-Carel C, Dumanski JP: Various regions within the alpha-helical angiomatoid fibrous histiocytoma. Genes Chromosomes Cancer domain of the COL1A1 gene are fused to the second exon of the 2007, 46:1051e1060 PDGFB gene in dermatofibrosarcomas and giant-cell fibroblastomas. 22. Pierron G, Tirode F, Lucchesi C, Reynaud S, Ballet S, Cohen- Genes Chromosomes Cancer 1998, 23:187e193 Gogo S, Perrin V, Coindre JM, Delattre O: A new subtype of bone 36. Patel KU, Szabo SS, Hernandez VS, Prieto VG, Abruzzo LV, sarcoma defined by BCOR-CCNB3 gene fusion. Nat Genet 2012, 44: Lazar AJ, Lopez-Terrada D: Dermatofibrosarcoma protuberans 461e466 COL1A1-PDGFB fusion is identified in virtually all dermatofi- 23. Puls F, Niblett A, Marland G, Gaston CL, Douis H, Mangham DC, brosarcoma protuberans cases when investigated by newly devel- Sumathi VP, Kindblom LG: BCOR-CCNB3 (Ewing-like) sarcoma: a oped multiplex reverse transcription polymerase chain reaction and clinicopathologic analysis of 10 cases, in comparison with conven- fluorescence in situ hybridization assays. Hum Pathol 2008, 39: tional Ewing sarcoma. Am J Surg Pathol 2014, 38:1307e1318 184e193 24. Gambarotti M, Benini S, Gamberi G, Cocchi S, Palmerini E, 37. Yokoyama Y, Shimizu A, Okada E, Ishikawa O, Motegi S: A rapid Sbaraglia M, Donati D, Picci P, Vanel D, Ferrari S, Righi A, Dei and efficient newly established method to detect COL1A1-PDGFB Tos AP: CIC-DUX4 fusion-positive round-cell sarcomas of soft tis- gene fusion in dermatofibrosarcoma protuberans. Biochem Biophys sue and bone: a single-institution morphological and molecular Res Commun 2012, 425:353e356 analysis of seven cases. Histopathology 2016, 69:624e634 38. Wang J, Hisaoka M, Shimajiri S, Morimitsu Y, Hashimoto H: 25. Graham C, Chilton-MacNeill S, Zielenska M, Somers GR: The CIC- Detection of COL1A1-PDGFB fusion transcripts in dermatofi- DUX4 fusion transcript is present in a subgroup of pediatric primitive brosarcoma protuberans by reverse transcription-polymerase chain round cell sarcomas. Hum Pathol 2012, 43:180e189 reaction using archival formalin-fixed, paraffin-embedded tissues. 26. Italiano A, Sung YS, Zhang L, Singer S, Maki RG, Coindre JM, Diagn Mol Pathol 1999, 8:113e119 Antonescu CR: High prevalence of CIC fusion with double- 39. Antonescu CR, Gerald WL, Magid MS, Ladanyi M: Molecular var- homeobox (DUX4) transcription factors in EWSR1-negative undif- iants of the EWS-WT1 gene fusion in desmoplastic small round cell ferentiated small blue round cell sarcomas. Genes Chromosomes tumor. Diagn Mol Pathol 1998, 7:24e28 Cancer 2012, 51:207e218 40. de Alava E, Ladanyi M, Rosai J, Gerald WL: Detection of chimeric 27. Kawamura-Saito M, Yamazaki Y, Kaneko K, Kawaguchi N, transcripts in desmoplastic small round cell tumor and related Kanda H, Mukai H, Gotoh T, Motoi T, Fukayama M, Aburatani H, developmental tumors by reverse transcriptase polymerase chain re- Takizawa T, Nakamura T: Fusion between CIC and DUX4 up- action: a specific diagnostic assay. Am J Pathol 1995, 147: regulates PEA3 family genes in Ewing-like sarcomas with t(4; 1584e1591 19)(q35;q13) translocation. Hum Mol Genet 2006, 15:2125e2137 41. Werner H, Idelman G, Rubinstein M, Pattee P, Nagalla SR, 28. Machado I, Cruz J, Lavernia J, Rubio L, Campos J, Barrios M, Roberts CT Jr: A novel EWS-WT1 gene fusion product in desmo- Grison C, Chene V, Pierron G, Delattre O, Llombart-Bosch A: Su- plastic small round cell tumor is a potent transactivator of the insulin- perficial EWSR1-negative undifferentiated small round cell sarcoma like growth factor-I receptor (IGF-IR) gene. Cancer Lett 2007, 247: with CIC/DUX4 gene fusion: a new variant of Ewing-like tumors 84e90 with locoregional lymph node metastasis. Virchows Arch 2013, 463: 42. Nakanishi Y, Oinuma T, Sano M, Fuchinoue F, Komatsu K, Seki T, 837e842 Obana Y, Tabata M, Kikuchi K, Shimamura M, Ohmori K, 29. Sugita S, Arai Y, Tonooka A, Hama N, Totoki Y, Fujii T, Aoyama T, Nemoto N: Coexpression of an unusual form of the EWS-WT1 Asanuma H, Tsukahara T, Kaya M, Shibata T, Hasegawa T: A novel fusion transcript and interleukin 2/15 receptor betamRNA in a CIC-FOXO4 gene fusion in undifferentiated small round cell sar- desmoplastic small round cell tumour. J Clin Pathol 2006, 59: coma: a genetically distinct variant of Ewing-like sarcoma. Am J Surg 1108e1110 Pathol 2014, 38:1571e1576 43. Isphording A, Ali RH, Irving J, Goytain A, Nelnyk N, Hoang LN, 30. Giacomini CP, Sun S, Varma S, Shain AH, Giacomini MM, Gilks CB, Huntsman DG, Nielsen TO, Nucci MR, Lee CH: Balagtas J, Sweeney RT, Lai E, Del Vecchio CA, Forster AD, YWHAE-FAM22 endometrial stromal sarcoma: diagnosis by reverse Clarke N, Montgomery KD, Zhu S, Wong AJ, van de Rijn M, transcription-polymerase chain reaction in formalin-fixed, paraffin- West RB, Pollack JR: Breakpoint analysis of transcriptional and embedded tumor. Hum Pathol 2013, 44:837e843 genomic profiles uncovers novel gene fusions spanning multiple 44. Micci F, Walter CU, Teixeira MR, Panagopoulos I, Bjerkehagen B, human cancer types. PLoS Genet 2013, 9:e1003464 Saeter G, Heim S: Cytogenetic and molecular genetic analyses of 31. Wang WL, Mayordomo E, Zhang W, Hernandez VS, Tuvin D, endometrial stromal sarcoma: nonrandom involvement of chromo- Garcia L, Lev DC, Lazar AJ, Lopez-Terrada D: Detection and some arms 6p and 7p and confirmation of JAZF1/JJAZ1 gene fusion characterization of EWSR1/ATF1 and EWSR1/CREB1 chimeric in t(7;17). Cancer Genet Cytogenet 2003, 144:119e124 transcripts in clear cell sarcoma (melanoma of soft parts). Mod Pathol 45. Micci F, Panagopoulos I, Bjerkehagen B, Heim S: Consistent rear- 2009, 22:1201e1209 rangement of chromosomal band 6p21 with generation of fusion 32. Knezevich SR, McFadden DE, Tao W, Lim JF, Sorensen PH: A genes JAZF1/PHF1 and EPC1/PHF1 in endometrial stromal sarcoma. novel ETV6-NTRK3 gene fusion in congenital fibrosarcoma. Nat Cancer Res 2006, 66:107e112 Genet 1998, 18:184e187 46. Panagopoulos I, Micci F, Thorsen J, Gorunova L, Eibak AM, 33. Llombart B, Sanmartin O, Lopez-Guerrero JA, Monteagudo C, Bjerkehagen B, Davidson B, Heim S: Novel fusion of MYST/Esa1- Serra C, Requena C, Poveda A, Vistos JL, Almenar S, Llombart- associated factor 6 and PHF1 in endometrial stromal sarcoma. Bosch A, Guillen C: Dermatofibrosarcoma protuberans: clinical, PLoS One 2012, 7:e39354

The Journal of Molecular Diagnostics - jmd.amjpathol.org 75 Chang et al

47. Panagopoulos I, Thorsen J, Gorunova L, Haugom L, Bjerkehagen B, from a t(9;17)(q22;q11.2) translocation in an extraskeletal myxoid Davidson B, Heim S, Micci F: Fusion of the ZC3H7B and BCOR chondrosarcoma. Oncogene 1999, 18:7599e7601 genes in endometrial stromal sarcomas carrying an X;22- 64. Hisaoka M, Ishida T, Imamura T, Hashimoto H: TFG is a novel translocation. Genes Chromosomes Cancer 2013, 52:610e618 fusion partner of NOR1 in extraskeletal myxoid chondrosarcoma. 48. Antonescu CR, Le Loarer F, Mosquera JM, Sboner A, Zhang L, Genes Chromosomes Cancer 2004, 40:325e328 Chen CL, Chen HW, Pathan N, Krausz T, Dickson BC, Weinreb I, 65. Panagopoulos I, Mertens F, Isaksson M, Domanski HA, Rubin MA, Hameed M, Fletcher CD: Novel YAP1-TFE3 fusion Brosjo O, Heim S, Bjerkehagen B, Sciot R, Dal Cin P, defines a distinct subset of epithelioid hemangioendothelioma. Genes Fletcher JA, Fletcher CD, Mandahl N: Molecular genetic char- Chromosomes Cancer 2013, 52:775e784 acterization of the EWS/CHN and RBP56/CHN fusion genes in 49. Patel NR, Salim AA, Sayeed H, Sarabia SF, Hollingsworth F, extraskeletal myxoid chondrosarcoma. Genes Chromosomes Warren M, Jakacky J, Tanas M, Oliveira AM, Rubin BP, Lazar AJ, Cancer 2002, 35:340e352 Lopez-Terrada D, Wang WL: Molecular characterization of epithe- 66. Sjögren H, Wedell B, Meis-Kindblom JM, Kindblom LG, lioid haemangioendotheliomas identifies novel WWTR1-CAMTA1 Stenman G: Fusion of the NH2-terminal domain of the basic helix- fusion variants. Histopathology 2015, 67:699e708 loop-helix protein TCF12 to TEC in extraskeletal myxoid chon- 50. Jeon IS, Davis JN, Braun BS, Sublett JE, Roussel MF, Denny CT, drosarcoma with translocation t(9;15)(q22;q21). Cancer Res 2000, Shapiro DN: A variant Ewing’s sarcoma translocation (7;22) fuses 60:6832e6835 the EWS gene to the ETS gene ETV1. Oncogene 1995, 10: 67. Antonescu CR, Suurmeijer AJ, Zhang L, Sung YS, Jungbluth AA, 1229e1234 Travis WD, Al-Ahmadie H, Fletcher CD, Alaggio R: Molecular 51. Llombart-Bosch A, Pellin A, Carda C, Noguera R, Navarro S, Pey- characterization of inflammatory myofibroblastic tumors with dro-Olaya A: Soft tissue Ewing sarcomaeperipheral primitive neu- frequent ALK and ROS1 gene fusions and rare novel RET rear- roectodermal tumor with atypical clear cell pattern shows a new type rangement. Am J Surg Pathol 2015, 39:957e967 of EWS-FEV fusion transcript. Diagn Mol Pathol 2000, 9:137e144 68. Lovly CM, Gupta A, Lipson D, Otto G, Brennan T, Chung CT, 52. Ng TL, O’Sullivan MJ, Pallen CJ, Hayes M, Clarkson PW, Borinstein SC, Ross JS, Stephens PJ, Miller VA, Coffin CM: In- Winstanley M, Sorensen PH, Nielsen TO, Horsman DE: Ewing sar- flammatory myofibroblastic tumors harbor multiple potentially coma with novel translocation t(2;16) producing an in-frame fusion of actionable kinase fusions. Cancer Discov 2014, 4:889e895 FUS and FEV. J Mol Diagn 2007, 9:459e463 69. Ouchi K, Miyachi M, Tsuma Y, Tsuchiya K, Iehara T, Konishi E, 53. Peter M, Couturier J, Pacquement H, Michon J, Thomas G, Yanagisawa A, Hosoi H: FN1: a novel fusion partner of ALK in an Magdelenat H, Delattre O: A new member of the ETS family fused to inflammatory myofibroblastic tumor. Pediatr Blood Cancer 2015, 62: EWS in Ewing tumors. Oncogene 1997, 14:1159e1164 909e911 54. Shing DC, McMullan DJ, Roberts P, Smith K, Chin SF, Nicholson J, 70. Panagopoulos I, Nilsson T, Domanski HA, Isaksson M, Lindblom P, Tillman RM, Ramani P, Cullinane C, Coleman N: FUS/ERG gene Mertens F, Mandahl N: Fusion of the SEC31L1 and ALK genes in an fusions in Ewing’s tumors. Cancer Res 2003, 63:4568e4576 inflammatory myofibroblastic tumor. Int J Cancer 2006, 118: 55. Sorensen PH, Lessnick SL, Lopez-Terrada D, Liu XF, Triche TJ, 1181e1186 Denny CT: A second Ewing’s sarcoma translocation, t(21;22), fuses 71. Takeuchi K, Soda M, Togashi Y, Sugawara E, Hatano S, Asaka R, the EWS gene to another ETS-family transcription factor, ERG. Nat Okumura S, Nakagawa K, Mano H, Ishikawa Y: Pulmonary inflam- Genet 1994, 6:146e151 matory myofibroblastic tumor expressing a novel fusion, PPFIBP1- 56. Urano F, Umezawa A, Hong W, Kikuchi H, Hata J: A novel chimera ALK: reappraisal of anti-ALK immunohistochemistry as a tool for gene between EWS and E1A-F, encoding the adenovirus E1A novel ALK fusion identification. Clin Cancer Res 2011, 17:3341e3348 enhancer-binding protein, in extraosseous Ewing’s sarcoma. Biochem 72. Kubo T, Matsui Y, Naka N, Araki N, Goto T, Yukata K, Endo K, Biophys Res Commun 1996, 219:608e612 Yasui N, Myoui A, Kawabata H, Yoshikawa H, Ueda T: Expression 57. Sumegi J, Nishio J, Nelson M, Frayer RW, Perry D, Bridge JA: A of HMGA2-LPP and LPP-HMGA2 fusion genes in lipoma: identi- novel t(4;22)(q31;q12) produces an EWSR1-SMARCA5 fusion in fication of a novel type of LPP-HMGA2 transcript in four cases. extraskeletal Ewing sarcoma/primitive neuroectodermal tumor. Mod Anticancer Res 2009, 29:2357e2360 Pathol 2011, 24:333e342 73. Wang L, Motoi T, Khanin R, Olshen A, Mertens F, Bridge J, Dal 58. Szuhai K, Ijszenga M, de Jong D, Karseladze A, Tanke HJ, Cin P, Antonescu CR, Singer S, Hameed M, Bovee JV, Hogendoorn PC: The NFATc2 gene is involved in a novel cloned Hogendoorn PC, Socci N, Ladanyi M: Identification of a novel, translocation in a Ewing sarcoma variant that couples its function in recurrent HEY1-NCOA2 fusion in mesenchymal chondrosarcoma immunology to oncology. Clin Cancer Res 2009, 15:2259e2268 based on a genome-wide screen of exon-level expression data. Genes 59. Yamaguchi S, Yamazaki Y, Ishikawa Y, Kawaguchi N, Mukai H, Chromosomes Cancer 2012, 51:127e139 Nakamura T: EWSR1 is fused to POU5F1 in a bone tumor with 74. Panagopoulos I, Thorsen J, Gorunova L, Micci F, Haugom L, translocation t(6;22)(p21;q12). Genes Chromosomes Cancer 2005, Davidson B, Heim S: RNA sequencing identifies fusion of the 43:217e222 EWSR1 and YY1 genes in mesothelioma with t(14;22)(q32;q12). 60. Wang L, Bhargava R, Zheng T, Wexler L, Collins MH, Roulston D, Genes Chromosomes Cancer 2013, 52:733e740 Ladanyi M: Undifferentiated small round cell sarcomas with rare 75. Brandal P, Panagopoulos I, Bjerkehagen B, Heim S: t(19;22)(q13; EWS gene fusions: identification of a novel EWS-SP3 fusion and of q12) Translocation leading to the novel fusion gene EWSR1-ZNF444 additional cases with the EWS-ETV1 and EWS-FEV fusions. J Mol in soft tissue myoepithelial carcinoma. Genes Chromosomes Cancer Diagn 2007, 9:498e509 2009, 48:1051e1056 61. Agaram NP, Zhang L, Sung YS, Singer S, Antonescu CR: Extra- 76. Agaram NP, Chen HW, Zhang L, Sung YS, Panicek D, Healey JH, skeletal myxoid chondrosarcoma with non-EWSR1-NR4A3 variant Nielsen GP, Fletcher CD, Antonescu CR: EWSR1-PBX3: a novel fusions correlate with rhabdoid phenotype and high-grade gene fusion in myoepithelial tumors. Genes Chromosomes Cancer morphology. Hum Pathol 2014, 45:1084e1091 2015, 54:63e71 62. Antonescu CR, Argani P, Erlandson RA, Healey JH, Ladanyi M, 77. Antonescu CR, Zhang L, Chang NE, Pawel BR, Travis W, Katabi N, Huvos AG: Skeletal and extraskeletal myxoid chondrosarcoma: a Edelman M, Rosenberg AE, Nielsen GP, Dal Cin P, Fletcher CD: comparative clinicopathologic, ultrastructural, and molecular study. EWSR1-POU5F1 fusion in soft tissue myoepithelial tumors: a mo- Cancer 1998, 83:1504e1521 lecular analysis of sixty-six cases, including soft tissue, bone, and 63. Attwooll C, Tariq M, Harris M, Coyne JD, Telford N, Varley JM: visceral lesions, showing common involvement of the EWSR1 gene. Identification of a novel fusion gene involving hTAFII68 and CHN Genes Chromosomes Cancer 2010, 49:1114e1124

76 jmd.amjpathol.org - The Journal of Molecular Diagnostics NanoString Sarcoma Fusion Detection

78. Brandal P, Panagopoulos I, Bjerkehagen B, Gorunova L, Skjeldal S, 90. Nilsson G, Skytting B, Xie Y, Brodin B, Perfekt R, Mandahl N, Micci F, Heim S: Detection of a t(1;22)(q23;q12) translocation Lundeberg J, Uhlen M, Larsson O: The SYT-SSX1 variant of sy- leading to an EWSR1-PBX1 fusion gene in a myoepithelioma. Genes novial sarcoma is associated with a high rate of tumor cell prolifer- Chromosomes Cancer 2008, 47:558e564 ation and poor clinical outcome. Cancer Res 1999, 59:3180e3184 79. Puls F, Arbajian E, Magnusson L, Douis H, Kindblom LG, 91. Przybyl J, Sciot R, Rutkowski P, Siedlecki JA, Vanspauwen V, Mertens F: Myoepithelioma of bone with a novel FUS-POU5F1 Samson I, Debiec-Rychter M: Recurrent and novel SS18-SSX fusion fusion gene. Histopathology 2014, 65:917e922 transcripts in synovial sarcoma: description of three new cases. 80. Antonescu CR, Tschernyavsky SJ, Decuseara R, Leung DH, Tumour Biol 2012, 33:2245e2253 Woodruff JM, Brennan MF, Bridge JA, Neff JR, Goldblum JR, 92. Safar A, Wickert R, Nelson M, Neff JR, Bridge JA: Characterization Ladanyi M: Prognostic impact of P53 status, TLS-CHOP fusion of a variant SYT-SSX1 synovial sarcoma fusion transcript. Diagn transcript structure, and histological grade in myxoid liposarcoma: a Mol Pathol 1998, 7:283e287 molecular and clinicopathologic study of 82 cases. Clin Cancer Res 93. Skytting B, Nilsson G, Brodin B, Xie Y, Lundeberg J, Uhlen M, 2001, 7:3977e3987 Larsson O: A novel fusion gene, SYT-SSX4, in synovial sarcoma. J 81. Bode-Lesniewska B, Frigerio S, Exner U, Abdou MT, Moch H, Natl Cancer Inst 1999, 91:974e975 Zimmermann DR: Relevance of translocation type in myxoid lip- 94. Storlazzi CT, Mertens F, Mandahl N, Gisselsson D, Isaksson M, osarcoma and identification of a novel EWSR1-DDIT3 fusion. Genes Gustafson P, Domanski HA, Panagopoulos I: A novel fusion gene, Chromosomes Cancer 2007, 46:961e971 SS18L1/SSX1, in synovial sarcoma. Genes Chromosomes Cancer 82. Hosaka T, Nakashima Y, Kusuzaki K, Murata H, Nakayama T, 2003, 37:195e200 Nakamata T, Aoyama T, Okamoto T, Nishijo K, Araki N, 95. Mertens F, Fletcher CD, Antonescu CR, Coindre JM, Colecchia M, Tsuboyama T, Nakamura T, Toguchida J: A novel type of EWS- Domanski HA, Downs-Kelly E, Fisher C, Goldblum JR, Guillou L, CHOP fusion gene in two cases of myxoid liposarcoma. J Mol Reid R, Rosai J, Sciot R, Mandahl N, Panagopoulos I: Clinicopath- Diagn 2002, 4:164e171 ologic and molecular genetic characterization of low-grade fibro- 83. Powers MP, Wang WL, Hernandez VS, Patel KS, Lev DC, Lazar AJ, myxoid sarcoma, and cloning of a novel FUS/CREB3L1 fusion gene. Lopez-Terrada DH: Detection of myxoid liposarcoma-associated Lab Invest 2005, 85:408e415 FUS-DDIT3 rearrangement variants including a newly identified 96. Guillou L, Benhattar J, Gengler C, Gallagher G, Ranchere-Vince D, breakpoint using an optimized RT-PCR assay. Mod Pathol 2010, 23: Collin F, Terrier P, Terrier-Lacombe MJ, Leroux A, Marques B, 1307e1315 Aubain Somerhausen Nde S, Keslair F, Pedeutour F, Coindre JM: 84. Bridge JA, Sanders K, Huang D, Nelson M, Neff JR, Muirhead D, Translocation-positive low-grade fibromyxoid sarcoma: clinicopath- Walker C, Seemayer TA, Sumegi J: Pericytoma with t(7;12) and ologic and molecular analysis of a series expanding the morphologic ACTB-GLI1 fusion arising in bone. Hum Pathol 2012, 43: spectrum and suggesting potential relationship to sclerosing epithe- 1524e1529 lioid fibrosarcoma: a study from the French Sarcoma Group. Am J 85. Dahlen A, Mertens F, Mandahl N, Panagopoulos I: Molecular genetic Surg Pathol 2007, 31:1387e1402 characterization of the genomic ACTB-GLI fusion in pericytoma with 97. Loke BN, Lee VK, Sudhanshi J, Wong MK, Kuick CH, t(7;12). Biochem Biophys Res Commun 2004, 325:1318e1323 Puhaindran M, Chang KT: Novel exon-exon breakpoint in CIC- 86. Amary MF, Diss TC, Flanagan AM: Molecular characterization of a DUX4 fusion sarcoma identified by anchored multiplex PCR novel variant of a SYT-SSX1 fusion transcript in synovial sarcoma. (Archer FusionPlex Sarcoma Panel). J Clin Pathol 2017, 70: Histopathology 2007, 51:559e561 697e701 87. Crew AJ, Clark J, Fisher C, Gill S, Grimer R, Chand A, Shipley J, 98. Parker D, Belaud-Rotureau MA: Micro-cost analysis of ALK rear- Gusterson BA, Cooper CS: Fusion of SYT to two genes, SSX1 and rangement testing by FISH to determine eligibility for crizotinib SSX2, encoding proteins with homology to the Kruppel-associated therapy in NSCLC: implications for cost effectiveness of testing and box in human synovial sarcoma. EMBO J 1995, 14:2333e2340 treatment. Clin Med Insights Oncol 2014, 8:145e152 88. Fligman I, Lonardo F, Jhanwar SC, Gerald WL, Woodruff J, 99. Tanas MR, Goldblum JR: Fluorescence in situ hybridization in the Ladanyi M: Molecular diagnosis of synovial sarcoma and charac- diagnosis of soft tissue neoplasms: a review. Adv Anat Pathol 2009, terization of a variant SYT-SSX2 fusion transcript. Am J Pathol 1995, 16:383e391 147:1592e1599 100. Horn H, Allmanritter J, Doglioni C, Marx A, Muller J, 89. Mikami T, Kurose A, Javed F, Takeda Y: Detection of rare variant of Gattenlohner S, Staiger AM, Rosenwald A, Ott G, Ott MM: Fluo- SS18-SSX1 fusion gene and mutations of important cancer-related rescence in situ analysis of soft tissue tumor associated genetic al- genes in synovial sarcoma of the lip: gene analyses of a case and terations in formalin-fixed paraffin-embedded tissue. Pathol Res Pract literature review. J Oral Maxillofac Surg 2015, 73:1505e1515 2014, 210:804e811

The Journal of Molecular Diagnostics - jmd.amjpathol.org 77