Diagnostic yield of NanoString nCounter FusionPlex profiling in soft tissue tumors

Abstract Diagnostic histopathology of soft tissue tumors can be troublesome as many entities are quite rare and have overlapping morphologic features. Many soft tissue tumors harbor tumor‐defining gene translocations, which may provide an important ancillary tool for tumor diagnosis. The NanoString nCounter platform enables multiplex detection of pre‐defined gene fusion transcripts in formalin‐fixed and paraffin‐embedded tissue. A cohort of 104 soft tissue tumors representing 20 different histological types was analyzed for the expression of 174 unique gene fusion transcripts. A tumor‐defining gene fusion transcript was detected in 60 cases (58%). Sensitivity and specificity of the NanoString assay calculated against the result of an alternative molecular method were 85% and 100%, respectively. Highest diagnostic coverage was obtained for Ewing sarcoma, synovial sarcoma, myxoid liposarcoma, alveolar rhabdomyosarcoma, and desmoplastic small round cell tumor. For these tumor types, the NanoString assay is a rapid, cost‐effective, sensitive, and specific ancillary screening tool for molecular diagnosis. For other sarcomas, additional molecular testing may be required when a translocation transcript is not identified with the current 174 gene fusion panel.


| INTRODUCTION
Soft tissue tumors represent a remarkably heterogeneous group of neoplasms, with many subtypes being exceptionally rare. More than 100 different soft tissue tumors have been described in the latest 2013 WHO classification. 1 The proper histological classification of soft tissue tumors is grounded in the microscopic analysis of tumor growth patterns and their cytological features, which may be a difficult exercise, since many tumors have overlapping morphologic features. Although tumor-associated protein markers may be visualized by ancillary immunohistochemistry (IHC), many tumors show nonspecific, overlapping or absent marker expression. Thus, it may be difficult or impossible to render an objective accurate diagnosis, in particular when studying small biopsy specimens with a limited amount of tumor tissue.
Fortunately, a significant number of soft tissue tumors, in particular those with monomorphic round cell, spindle cell or epithelioid morphology, harbor recurrent gene translocations, which are often tumor-specific. These unique recurrent translocations were first discovered in the early 1990s by chromosomal banding techniques, for example, the t(X;18)(p11;q11) translocation in synovial sarcoma, which results in the tumor specific SS18-SSX fusion genes. 2 At the molecular level, with knowledge of the exon regions involved in fusion genes, RT-PCR and Fluorescence In Situ Hybridisation (FISH, using break-apart probes) methods became available to detect these particular gene fusions and rearrangements.
In the past decade, pathologists have witnessed the rapid development of next generation sequencing (NGS) techniques, which allow simultaneous detection of multiple fusion transcripts. This translated into more accurate classification and also prognostication of soft tissue tumors. 3 At present, the two novel molecular multiplex methods commonly used in Dutch sarcoma centers are the anchored multiplex PCR (AMP)-based NGS (Archer FusionPlex Sarcoma assay) 4 and the NanoString nCounter platform. 5 The Archer AMP PCR method targets exons of 26 genes commonly involved in fusion genes of soft tissue tumors, whereas the NanoString assay is a high-throughput hybridization technique, which uses specific probes that target 174 unique gene fusion junctions in 22 soft tissue tumor types. 6 In this quality control study, we evaluated the sensitivity and specificity of the NanoString nCounter platform for gene fusion detection in 22 different soft tissue tumors, adding our results to the initial report on this method. 7 2 | MATERIALS AND METHODS 2.2 | NanoString gene expression profiling RNA was isolated from four 5-μm-thick formalin-fixed and paraffinembedded tissue sections containing at least 50% tumor cells using the RNeasy mini kit (Qiagen) according to suppliers instructions. Total RNA was quantified with Qubit (ThermoFisher).

| Case selection
The soft tissue and bone tumor probe set as described by Chang et al. 7 was ordered from IDT Technologies (Leuven, Belgium). In contrast to the initial study, our panel did not contain probes for the detection of COL1A1-PDGFB gene fusion transcripts, as can be found in dermatofibrosarcoma protuberans. Probes were hybridized with 100 ng RNA overnight in a thermocycler at 67 C with a heated-lid at 72 C. The RNA-probe complexes were loaded on an nCounter cartridge, and hybridized, washed and read on a nCounter SPRINT platform according to suppliers instructions (NanoString nCounter Technologies, Seattle, WA).

| Data analysis
The platform-generated Reporter Code Count (RCC) files containing the raw data were analyzed. Samples with a geometric mean of the raw counts of the four reference genes (ACTB, GAPDH, SDHA, UBC) of <500 were excluded from the analysis due to low RNA input or poor RNA quality. Subsequent data normalization were performed with the nSolver Analysis Software (NanoString nCounter Technologies) to correct for differences in hybridization efficiency using the respective control probes. Counts were not corrected for RNA input. Following a log 2 transformation of the normalized data, the interquartile range (IQR) of counts for each probe across all samples in the run was calculated. Outliers in each sample, that is, positive signal for a gene fusion transcript, were determined as counts larger than 1.5*IQR, and which exceed the background threshold of 40 counts. The counts were not compared to the median of the counts across all the probes within a sample as reported by Chang et al. 7 A comparison of both methods did not alter the results for the sample set described in this work (data not shown).

| RESULTS
As shown in Figure 1 and Table 1, the NanoString assay detected gene fusions in 60/104 cases suitable for analysis. In 44/60 NanoString positive cases, a similar gene fusion had already been detected by previous alternative molecular testing. In the other 16/60 cases, no previous molecular testing had been performed. The detected fusion genes are summarized in Table S1.
In 44/104 cases, no fusion was detected by the NanoString assay, whereas in 8/44 cases, a gene rearrangement or fusion had been found by prior alternative molecular testing (5 by FISH, 2 by FISH and RT-PCR, and 1 by targeted NGS). Thus, there were no false-positive NanoString results and eight false-negative NanoString results. Overall, fusion gene detection by NanoString had a sensitivity of 85% and specificity of 100%.

| Concordant and discordant (false-negative) cases
Of the 52/104 cases, in which a gene rearrangement or fusion had been detected by prior molecular testing, NanoString was positive
3.3 | The relative value of the NanoString assay is strongly associated with the level of diagnostic evidence in daily practice

| Estimated diagnostic coverage of NanoString in STB tumors
By combining the results of this study ( Desmoplastic small round cell tumor  For other previously identified translocations that could not be confirmed with the current NanoString panel, it is unknown whether this is due to a lack of performance of the fusion gene probes or a lack of probes for other known and unknown fusion. For example, this study included one CIC-rearranged sarcoma in which a CIC rearrangement was demonstrated by FISH previously. However, a F I G U R E 2 Diagnostic value of Nanostring nCounter FusionPlex in different fusion-associated tumor types. α: Two alveolar rhabdomyosarcomas, two aneurysmal bone cysts, three angiomatoid fibrous histiocytomas, one biphenotypic sinonasal sarcoma, three clear cell sarcomas, two infantile fibrosarcomas, three desmoplastic small round cell tumors, two epithelioid hemangioendotheliomas, eight Ewing sarcomas, two extraskeletal myxoid chondrosarcomas, two inflammatory myofibroblastic tumors, one mesenchymal chondrosarcoma, five myxoid liposarcomas, and eight synovial sarcomas. β: One BCOR-rearranged sarcoma, one CIC-rearranged sarcoma, one clear cell sarcoma, three epithelioid hemangioendotheliomas, and two inflammatory myofibroblastic tumors. γ: Two alveolar soft part sarcomas, three alveolar rhabdomyosarcomas, one aneurysmal bone cyst, one angiomatoid fibrous histiocytoma, three mesenchymal chondrosarcomas, two myxoid liposarcomas, and three nodular fasciitis. δ: One alveolar soft part sarcoma, three aneurysmal bone cysts, one angiomatoid fibrous histiocytoma, two biphenotypic sinonasal sarcomas, one infantile fibrosarcoma, one inflammatory myofibroblastic tumor, three lipoblastomas, two mesenchymal chondrosarcomas, two nodular fasciitis, and five tenosynovial giant cell tumors. but is associated with higher costs and longer turnaround times that are comparable to NGS sequencing of large targeted panels.
In conclusion, the NanoString nCounter FusionPlex assay is a screening tool with high sensitivity and specificity [5][6][7]12,13 for the detection of sarcoma-defining fusion gene transcripts in Ewing sarcoma, synovial sarcoma, myxoid liposarcoma, alveolar rhabdomyosarcoma, and desmoplastic small round cell tumors. Its diagnostic yield for rare soft tissue tumors is limited and might require additional or alternative testing.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.

SUPPORTING INFORMATION
Additional supporting information may be found online in the Supporting Information section at the end of this article.