Introduction

Ewing sarcoma (ES) is a highly aggressive neoplasm that most commonly affects the deep soft tissues of the extremities of adolescents and young adults, and is characterized by translocations fusing EWSR1 with one of the ETS family of transcription factors, most frequently t(11;22)(q24;q12) resulting in EWSR1-FLI1 fusion (in approximately 90% of cases) [1]. Recently, NKX2.2—a homeodomain transcription factor that plays a crucial role in central nervous system development, oligodendrocyte differentiation, and neuroendocrine differentiation in the central nervous system, gastrointestinal tract, and pancreas [2,3,4,5]—has emerged as a new marker for ES that is reportedly both sensitive and specific [6, 7]. NKX2.2 has also been reported to be a downstream target of EWSR1-FLI1 that is necessary for oncogenic transformation via transcription repression, resulting in the block of mesenchymal features in ES [8,9,10]. Furthermore, the combination of CD99 and NKX2.2 has been reported to be highly specific for Ewing sarcoma [11].

NKX2.2 has not yet been investigated specifically in the sinonasal small round blue cell tumor (SRBCT) differential diagnosis which includes many tumors specific to that site. Indeed, the differential diagnosis for sinonasal SRBCTs is broad and includes epithelial (squamous cell carcinoma and variants, NUT carcinoma, lymphoepithelial carcinoma, small cell neuroendocrine carcinoma, teratocarcinosarcoma, salivary gland carcinomas, sinonasal undifferentiated carcinoma), neuroectodermal (olfactory neuroblastoma, malignant mucosal melanoma, Ewing sarcoma), soft tissue (rhabdomyosarcoma, desmoplastic small round cell tumor, synovial sarcoma), hematopoietic (NK/T cell lymphoma, diffuse large B cell lymphoma), and secondary neoplasms (either by direct extension such as nasopharyngeal carcinoma, or by metastasis) [12, 13]. Furthermore, NKX2.2 has not been investigated in the newly-recognized “adamantinoma-like” variant of Ewing sarcoma, a particularly challenging entity with significant morphologic and immunohistochemical overlap with other tumors in the differential diagnosis [14,15,16]. This study aimed to evaluate the utility of NKX2.2 as an immunohistochemical marker for Ewing sarcoma versus other small round blue cell tumors of the sinonasal tract.

Methods

We performed NKX2.2 on formalin fixed and paraffin embedded tissue blocks from 170 poorly differentiated sinonasal neoplasms were retrieved from the Surgical Pathology archives of The Johns Hopkins Hospital. The cases included 73 SCCs (67 poorly differentiated, non-keratinizing, or basaloid types and 6 nasopharyngeal carcinomas), 46 olfactory neuroblastomas, 8 SNUCs, 6 melanomas, 7 Ewing sarcomas, 6 SMARCB1-deficient carcinomas, 6 teratocarcinosarcomas, 5 alveolar rhabdomyosarcomas, 4 solid adenoid cystic carcinomas, 4 NK/T cell lymphomas, 3 NUT midline carcinomas, and 2 small cell carcinomas. 35 of the cases were tested on whole slides, while the remaining 133 tumors were present on previously constructed tissue microarrays [17,18,19]. Three of the four sinonasal adamantinoma-like Ewing sarcomas have been previously published [15, 20]. Immunohistochemistry for NKX2.2 (74.5A5 monoclonal antibody, BD Biosciences, San Diego, CA, 1:100 dilution) was performed on five-micron sections utilizing standard protocols on a Ventana Benchmark XT autostainer (Ventana Medical Systems, Inc. Tucson, AZ). The distribution and intensity of staining was noted. Nuclear staining in <50% of tumor cells was regarded as ‘‘focal’’ while positivity in ≥50% was regarded as “diffuse”.

Results

The results are summarized in Table 1. NKX2.2 was positive in 7 of 7 (100%) Ewing sarcomas (Fig. 1) (all diffuse, 5 strong and 2 weak), including four adamantinoma-like variants (Fig. 2). Among tumors that were not ES, NKX2.2 was also positive in 5 of 6 (83%) teratocarcinosarcomas (always strong and focal), 12 of 46 (26%) olfactory neuroblastomas (all diffuse, 2 strong and 10 weak), 4 of 6 melanomas (2 diffuse, 2 focal, all weak), and 1 of 2 small cell carcinomas (diffuse and strong) (Fig. 3). It was negative in 73 SCCs (67 poorly differentiated, non-keratinizing, or basaloid types and 6 nasopharyngeal carcinomas), 3 NUT carcinomas, 6 SMARCB1-deficient carcinomas, 8 SNUC, 4 solid adenoid cystic carcinomas, 4 NK/T cell lymphomas, and 5 alveolar rhabdomyosarcomas.

Table 1 NKX2.2 immunohistochemistry in sinonasal neoplasms
Full size table
Fig. 1
figure 1

Ewing sarcoma demonstrating typical morphology (a, H&E stain, X400). All cases of Ewing sarcoma were diffusely positive for NKX2.2 by immunohistochemistry (b, X400)

Full size image
Fig. 2
figure 2

Adamantinoma-like Ewing sarcoma often exhibits squamous differentiation on routine histology (a, X400) and is diffusely positive for pan-cytokeratin (b, X400). However, adamantinoma-like Ewing sarcoma is also consistently positive for CD99 (c, X400) and NKX2.2 (d, X400)

Full size image
Fig. 3
figure 3

Some sinonasal tumors that were not Ewing sarcoma were also positive for NKX2.2, including olfactory neuroblastoma (a, b, X400), malignant melanoma (c, d, X400), and teratocarcinosarcoma (e, f, X200)

Full size image

Discussion

Diagnostic pathology of the sinonasal tract poses a particular challenge due to the diverse number of neoplasms that can occur at this site, many of which have overlapping morphological features (so-called “small round blue cell tumors”), and because of their rarity, comprising <1% of all malignancies and only 3% of head and neck malignancies [12, 21]. Further complicating the differential diagnosis is the complex anatomy of the sinonasal tract which often results in suboptimal biopsies that are small, fragmented, crushed, and/or predominantly blood clot, and secondary extension of tumors from adjacent nearby locations such as the nasopharynx [12, 22]. As such, ancillary tests such as immunohistochemistry and molecular techniques play a crucial role in reaching the correct diagnosis [23]. Moreover, there are a number of recently described tumor types that many pathologists may not yet be familiar with, including NUT carcinoma [19], SMARCB1-deficient sinonasal carcinoma [24], HPV-related carcinomas [17], and adamantinoma-like Ewing sarcoma [15, 20].

ES is particularly challenging to diagnose in the sinonasal tract given its morphologic and immunohistochemical heterogeneity, which often overlaps with other tumors in the differential diagnosis [13, 21, 25, 26]. Despite the challenges in diagnosing ES in the sinonasal tract, accurate tumor classification is critical for both prognosis and treatment, as ES is typically treated with specific chemotherapy protocols [27, 28]. The gold standard for diagnosing ES involves demonstrating EWSR1 gene fusions by either reverse-transcription PCR (RT-PCR) or break-apart fluorescence in situ hybridization (FISH) [1, 29]. However, given the higher cost and increased turnaround time of molecular testing, immunohistochemistry has become an appealing alternative. When diffuse and membranous, CD99 (O13) is a highly sensitive marker staining almost all Ewing sarcomas, but it is unfortunately non-specific and stains a significant number of other small round blue cell tumors [1, 30]. Additionally, the FLI1 antibody has been used for diagnosing ES, but it too suffers from poor specificity and in fact appears to be less sensitive than CD99 [16, 30, 31]. The adamantinoma-like variant of ES poses an especially difficult challenge in the sinonasal tract as it demonstrates overt epithelial differentiation morphologically (i.e., squamous pearls, intracellular bridges) and/or immunophenotypically (i.e., diffuse p40 and/or high-molecular weight cytokeratin), thus mimicking squamous cell carcinoma, a much more common malignancy in the sinonasal tract. Nevertheless, adamantinoma-like ES cases harbor the classic EWSR1-FLI1 fusions of ES [14,15,16]. Consequently, an additional marker of ES would be of value in the work-up of small round blue cell tumors, particularly in the sinonasal tract.

Examining 170 poorly differentiated sinonasal neoplasms with a SRBCT appearance, we demonstrated that NKX2.2 is a highly sensitive marker for ES including the adamantinoma-like variant (100% of cases). However, much like CD99, NKX2.2 is not entirely specific for ES. Perhaps not surprisingly given that NKX2.2 is involved in central nervous system development and neuroendocrine differentiation [2,3,4], NKX2.2 was positive in a subset of olfactory neuroblastomas (12 of 46), small cell carcinomas (1 of 2), and teratocarcinosarcomas (5 of 6). Low-grade forms of olfactory neuroblastoma may closely resemble ES, but immunostains for synaptophysin and chromogranin—typically focal or negative in ES—are generally strongly positive in olfactory neuroblastoma. Moreover, olfactory neuroblastoma lacks CD99 staining and often exhibits a S100-positive sustentacular tumor cell population. Small cell carcinoma is typically cytokeratin-positive, but often shows a dot-like distribution not seen in adamantinoma-like ES. In addition, small cell carcinoma is negative for CD99 and, unlike adamantinoma-like ES, is focal or negative for p40. Teratocarcinosarcomas—a rare and unusual neoplasm of the sinonasal tract that demonstrates a heterologous morphology with varying proportions of benign and malignant epithelial, mesenchymal, and neuroepithelial elements [32,33,34]—often show an overt, though often focal, neuroepithelial component, which corresponds to the focal, strong NKX2.2 staining seen in our study. Recognizing the various elements of teratocarcinosarcoma is crucial in separating it from ES and other tumor types, and therefore proper tissue sampling is of utmost importance. Finally, NKX2.2 was positive in 4 of 6 cases of malignant mucosal melanoma, another neuroectodermal neoplasm that has a notoriously wide range of histomorphologic appearances. As a result, if melanoma is a consideration, melanocytic markers such as S100, SOX10, HMB45, and Melan-A must be included in the immunohistochemical panel.

In summary, NKX2.2 is a very sensitive marker for Ewing sarcoma including the adamantinoma-like variant, and is thus useful in the small round blue cell differential diagnosis. At the same time, this marker is not entirely specific, as it will be positive in a subset of other neuroendocrine/neuroectodermal tumors. As a result, it must be used in a panel with other markers, especially cytokeratins, melanocytic markers, and CD99.