SALL-4 and Beta-Catenin Expression in Sinonasal Teratocarcinosarcoma

Abstract

Sinonasal teratocarcinosarcoma (SNTCS) is a rare, aggressive malignancy that displays a heterogeneous combination of malignant blastema-like, epithelial and mesenchymal components. Its exact histogenesis is unknown with hypotheses ranging from true germ cell derivation to origin from pluripotent stem cells. However, despite this tumor’s multiphenotypic histology, which includes frequent glandular, squamous, and neuroectodermal differentiation similar to adnexal germ cell tumors, SNTCS appears to have some differences from adnexal teratomas. For example, unlike adnexal teratomas, SNTCS has never been described as a component in a mixed germ cell tumor. Accurate recognition of SNTCS is difficult due to its rarity and histologic overlap with other sinonasal tumors. It is even more problematic on biopsy, since not all elements may be present in small samples. SNTCS can also share similar staining patterns with other neoplasms in the differential diagnosis. A recent study found nuclear β-catenin expression in a single TCS, but this has yet to be confirmed in additional cases. SALL-4, a marker of germ cell tumors, has not been examined. We performed β-catenin and SALL-4 immunohistochemistry on whole sections of 7 SNTCS and 19 other sinonasal neoplasms to assess whether β-catenin and SALL-4 are of utility in establishing a diagnosis of SNTCS. Intensity of expression and percentage of staining was noted for each tumor. For SNTCS, distribution of staining within each histologic component (immature neuroectodermal, epithelial, and mesenchymal) was also documented. Nuclear β-catenin expression was not identified in any SNTCS, with all cases demonstrating membranous expression (6 cases) or cytoplasmic and membranous expression (1 case). SALL-4 immunohistochemistry, however, was relatively sensitive (85.7%) and specific (89.5%) for SNTCS. SALL-4 expression was also identified in one poorly differentiated neuroendocrine carcinoma and one case of sinonasal undifferentiated carcinoma. SALL-4 appears to have utility in distinguishing SNTCS from other high grade sinonasal tumors.

Introduction

Malignant sinonasal tumors comprise less than 5% of all head and neck malignancies, with an annual incidence of 0.83 cases per 100,000 people [1]. Sinonasal teratocarcinosarcoma (SNTCS), in particular, is an extremely rare, high grade subtype that frequently recurs locally, can metastasize, and has a disease specific mortality of 60–70% at 5 years [2–5]. Composed of a mixture of immature/primitive and mature mesenchymal, epithelial, and neuroepithelial tissues, this complex mixture of elements makes it difficult to distinguish from the many other sinonasal tumors in the differential diagnosis. Moreover, due to differences in prognosis and treatment, separation of these entities is crucial, but may be challenging, particularly on limited material [6].

Immunohistochemistry provides an inexpensive and rapid way to discriminate between SNTCS and other tumors of the sinonasal tract. Two potential markers that may be of value are β-catenin and SALL-4. Birkeland and colleagues recently identified an activating β-catenin p.S45F mutation in one case of SNTCS and further demonstrated that the index case and one other were reactive for β-catenin in tumor cell nuclei by immunohistochemistry [7]. Rooper et al. also demonstrated nuclear β-catenin localization by immunohistochemistry in an additional 6 cases [8]. β-Catenin is a protein that plays a role in gene transcription as part of the Wnt signaling pathway [9]. In addition, β-catenin also interacts with e-cadherin to regulate cell-cell adhesion [10]. In normal tissue, β-catenin immunohistochemistry demonstrates membranous reactivity, reflecting the role of β-catenin in normal cell adhesion. Activating β-catenin mutations or alterations in the Wnt signaling pathway result in accumulation of the protein within the cell nucleus, which can be detected by immunohistochemistry. Nuclear localization of β-catenin has been demonstrated in a number of different tumors, including colorectal adenocarcinoma, endometrioid adenocarcinoma, and solid pseudopapillary tumor of the pancreas [11–13]. While initial studies suggest that β-catenin immunohistochemistry may be useful in the diagnosis of SNTCS, β-catenin expression has not been evaluated in any significant number of SNTCS nor systematically evaluated in other malignant sinonasal tumors.

SALL-4 is a transcription factor that is not normally expressed in post-embryonic tissues, but may be reactivated in malignancies, most notably in germ cell tumors, including post-pubertal testicular teratomas [14]. As the immature elements in SNTCS share many of the same histomorphologic features as teratomas, it is possible that SALL-4 may be positive in SNTCS. SALL-4 expression has not previously been evaluated in SNTCS or other sinonasal tumors.

As further elucidation of the immunophenotype of SNTCS may aid in their recognition, we studied the pattern of β-catenin and SALL-4 expression in a cohort of SNTCS, olfactory neuroblastoma (ONB), sinonasal undifferentiated carcinoma (SNUC), poorly differentiated neuroendocrine carcinoma (PD-NEC), and undifferentiated non-keratinizing nasopharyngeal carcinoma (NPC) to determine their diagnostic utility in this setting.

Materials and Methods

This study was approved by the Vanderbilt University Medical Center Institutional Research board. Resection materials from seven SNTCS were obtained from four large academic medical centers (Vanderbilt University Medical Center, Nashville TN; Massachusetts General Hospital/Harvard Medical School, Boston, MA; Emory University, Atlanta, GA; and The University of Chicago, Chicago, IL). Cases of ONB (n = 5), SNUC (n = 6), PD-NEC (n = 4), and NPC (n = 4) were identified by a query of the Vanderbilt laboratory information system (Cerner Copath, North Kansas City, MO, USA). Cases with no residual viable tumor following neoadjuvant treatment were excluded. The hematoxylin and eosin-stained slides, pathology reports, and ancillary molecular data (if applicable) were reviewed independently by two pathologists (MC, KE) and the diagnoses confirmed. SNTCS was diagnosed based on histomorphologic criteria outlined in the World Health Organization (WHO) classification of Head and Neck Tumors and required the presence of mixed epithelial, mesenchymal and primitive neuroepithelial elements (Fig. 2). Although we realize that in practice cases may not contain all three elements, we selected slides that featured triphasic morphology to evaluate for differential staining in each histologic element [2]. Review of immunohistochemistry performed at the time of diagnosis demonstrated triphasic differentiation in the SNTCS cases, including expression of cytokeratins such as AE1/AE3 and CAM5.2 within the epithelial components, expression of desmin and/or smooth muscle actin in mesenchymal elements, and S100, synaptophysin or CD56 expression within immature neuroectodermal elements. Cases of ONB, SNUC, PD-NEC, and NPC were selected by using the WHO criteria for each respective entity. Confirmatory immunohistochemistry for each entity included CD56, calretinin, and sustentacular S100 expression (ONB); CK8/18 and CAM5.2 expression with absence of squamous, melanocytic, neuroendocrine, and hematolymphoid markers and retention of INI1 (SNUC); synaptophysin, chromogranin, and CD56 expression with absent calretinin and S100 expression (PD-NEC); and positive EBV in situ hybridization (NPC). Because the cases were selected from archival material, BRG1 (SMARCA4) immunostain was not performed on cases of SNUC, as the significance of this marker had not yet been identified at the time of diagnosis.

Fig. 2.

Expression patterns of SALL-4 and β-catenin in sinonasal tumors with primitive/poorly differentiated morphology (×200 magnification). SNTCS sinonasal teratocarcinosarcoma, SNUC sinonasal undifferentiated carcinoma, PD-NEC poorly differentiated neuroendocrine carcinoma, ONB olfactory neuroblastoma, NPC nasopharyngeal carcinoma

Immunostaining for β-catenin and SALL-4 was performed at the Vanderbilt Translational Pathology Shared Resource (TPSR) immunohistochemistry core laboratory using standard protocols. The following antibodies were used: β-catenin (Clone: β-Catenin-1 M3539, Agilent) and SALL-4 (Clone: polyclonal HPA015291, Sigma-Aldrich). Immunohistochemistry was performed on full slide sections rather than microarray to assess for possible heterogeneity in staining.

Slides were evaluated by two study pathologists (MC and KE), with the pattern of β-catenin expression (nuclear, membranous, cytoplasmic or combination) and intensity of staining noted for each tumor. The percentage of tumor cells positive for SALL-4 and intensity of staining (scale 0–3+) was also scored. For SNTCS cases, the localization of β-catenin and SALL-4 staining was also evaluated, whether in the epithelial, mesenchymal, and/or primitive components of the tumor.

Results

A total of 7 SNTCS, 6 SNUC, 4 PD-NEC, 4 NPC and 5 ONB cases were identified in which sufficient tumor tissue was present for immunohistochemical analysis. Results of SALL-4 and β-catenin immunohistochemistry are shown in Table 1.

Table 1.

Percentage of sinonasal tumors expressing SALL-4 and β-catenin by immunohistochemistry

Tumor type (n)	SALL-4% positivity (n)	β-catenin % positivity
Teratocarcinosarcoma (7)	85.7% (6)	Nuclear: 0 Cytoplasmic: 0 Membranous: 85.7% (6) Combination: 14.3% (1) Total: 100% (7)
Sinonasal undifferentiated carcinoma (6)	16.7% (1)	Nuclear: 0 Cytoplasmic: 33.3% (2) Membranous: 50.0% (3) Combination: 16.7% (1) Total: 100% (6)
Poorly differentiated neuroendocrine carcinoma (4)	25.0% (1)	Nuclear: 0 Cytoplasmic: 0 Membranous: 75.0% (3) Combination: 25.0% (1) Total: 100% (4)
Non-keratinizing nasopharyngeal carcinoma (4)	0	Nuclear: 0 Cytoplasmic: 0 Membranous: 100% (4) Combination: 0 Total: 100% (4)
Olfactory neuroblastoma (5)	0	Nuclear: 0 Cytoplasmic: 1 Membranous: 2 Combination: 2 Total: 100% (5)

The seven SNTCS exhibited cytoplasmic-membranous immunoreactivity for β-catenin but lacked nuclear staining. The epithelial elements in all seven SNTCS cases demonstrated strong 3+ (4 cases) to moderate 2+ (3 cases) membranous expression of β-catenin. The mesenchymal elements demonstrated moderate 2+ cytoplasmic localization in three cases. The blastema-like immature elements showed patchy moderate 2+ membranous localization of β-catenin in one case. The other sinonasal tumors (n = 19) including ONB, SNUC, NPC, and PD-NEC carcinoma also displayed membranous β-catenin expression, although concurrent cytoplasmic expression was also seen in 2 SNUCs, 1 PD-NEC, and 1 ONB. The only tumor with β-catenin nuclear staining (strong, 100% of cells, with patchy cytoplasmic staining) was a case classified as SNUC. Immunohistochemistry performed at the time of initial diagnosis of this case was positive for AE1/AE3, negative for p40, CK5/6, desmin, Melan-A, chromogranin, synaptophysin, and CD99, and had retained expression of INI1, an immunoprofile similar to the other cases of SNUC evaluated in this study. However, unlike these cases, this SNUC also demonstrated strong diffuse nuclear SALL-4 positivity. While this case was tested for INI1 expression (SMARCB1 gene product), BRG1 immunostain (SMARCA4 gene product) was not performed at the time of initial diagnosis as the role of this protein in sinonasal malignancies had not yet been identified.

SALL-4 was positive in the majority (n = 6, 85.7%) of SNTCS (Fig. 1). The epithelial components demonstrated weak 1+ staining in one case, moderate 2+ staining in one case, and strong expression in 3 cases. Significant staining was not seen in the mesenchymal/stromal component of any of the SNTCS. The primitive/blastema-like elements demonstrated moderate 2+ expression in 2 cases and strong 3+ expression in 3 cases. SALL-4 was not entirely specific for SNTCS, as it also was positive in one case each of PD-NEC (moderate 2+ staining in 20% of cells) and SNUC (strong 3+ staining in 90% of cells). Overall, SALL-4 was found to have a sensitivity of 85.7% and specificity of 89.5% for the diagnosis of SNTCS.

Fig.1.

a At low power, teratocarcinosarcoma is characterized by a polymorphic appearance with the presence of multiple tissue types (H&E, ×4 magnification). These tissue types include b epithelial, which may be glandular, or, as pictured here, characterized by a distinctive mature glycogenated squamous epithelium c a mesenchymal component which is characterized by bland to frankly malignant spindle cells and d primitive/immature elements resembling blastemal/fetal tissues (H&E, ×20 magnification)

Discussion

SNTCS is a rare and aggressive tumor of the sinonasal tract [2, 3]. On resection specimens, it is relatively straightforward to diagnose due to its characteristic triphasic morphology. However, in small biopsy samples, the immature components may be difficult to distinguish from other sinonasal tumors with small round blue cell morphology. Rather than being related to chromosome 12p alterations like mediastinal or gonadal teratomas, SNTCS appear to have a distinct molecular pathogenesis [15]. It has recently been demonstrated that SNTCS may be characterized by loss of expression of the SMARCA4 (BRG1) protein, a component of the SWI/SNF (SWItch/Sucrose Non-Fermentable) family of proteins [16]. The discovery that SNTCS represents a molecularly distinct entity underscores the need for accurate diagnosis, given the potential therapeutic implications. Several targeted therapies including inhibitors for Cyclin-Dependent Kinase 4/6 (CDK4/6), Aurora kinase A (AURKA), and Enhancer of Zeste Homolog 2 (EZH2) appear to have a potential role in the treatment of tumors with SMARCA4 loss [17–19]. It has also been shown that, despite poor prognosis, SMARCA4 loss in lung tumors predicted sensitivity to platinum-based chemotherapy [20].

Prior immunophenotypic characterization of SNTCS demonstrated a relatively non-specific immunophenotype, including positivity for cytokeratins AE1/AE3 and CAM 5.2 in the epithelial components and CD56, chromogranin, INSM-1, and synaptophysin positivity within the neuroepithelial component [21]. NKX2.2 is positive in SNTCS, but Ewing sarcoma, melanoma, small cell carcinoma, and ONB all express NKX2.2 as well, limiting its clinical utility for differentiating tumors with small round blue cell morphology [22]. Two series have described β-catenin mutations in SNTCS, as well as nuclear localization of β-catenin by immunohistochemistry [7, 8]. However, we did not find that β-catenin immunohistochemistry reliably distinguished SNTCS from other high grade sinonasal tumors, nor did we identify any cases with nuclear β-catenin expression in our cohort.

One possible explanation for the disparate β-catenin expression results between our study and prior studies may be due to the small sample size. Since the Wnt/β-catenin pathway was shown to be altered in only 38% of cases in the study by Rooper et al., it is possible that our study does not include a case with this molecular alteration [8]. Another possible reason that our study did not replicate the prior findings on β-catenin expression may be due to the complex relationship between the BRG1 protein and the Wnt/β-catenin pathway. SMARCA4/BRG1 alterations appear to be a common event in SNTCS, as evidenced by the loss of expression of BRG1 in 82% of SNTCS cases [16]. The SMARCA4/BRG1 protein appears to interact with the Wnt signaling pathway in two different ways: by regulating transcription of target genes and by stabilizing β-catenin [23]. It is possible that the presence or absence of nuclear β-catenin expression by immunohistochemistry in SNTCS may be related to the portion of the Wnt signaling pathway that is impacted by SMARCA4 alterations. Although it is outside the scope of the current study, molecular profiling of the cases in our cohort may help clarify the relationship between the molecular pathogenesis and the immunohistochemical profile of SNTCS.

In contrast to β-catenin immunohistochemistry, SALL-4 immunohistochemistry appears to be relatively sensitive and specific for the diagnosis of SNTCS. SALL-4 is a zinc finger transcription factor that is essential to early embryogenesis and is expressed by testicular, ovarian, and extragonadal germ cell tumors [14]. It has a broader expression profile than other germ cell markers such as OCT4, particularly in primitive germ cell tumors [24] and is expressed by a subset of SMARCA4-deficient thoracic sarcomatoid tumors, aggressive thoracic neoplasms that demonstrate the same underlying alteration in SWI/SNF proteins that has been described in SNTCS [25, 26]. By using genome-wide promoter analysis, Kidder et al. demonstrated that the SWI/SNF chromatin-remodeling complex interacted with a number of pluripotency-related genes, which include the gene encoding for SALL-4 as well as other genes encoding for proteins such as Oct4, Nanog, and SOX2 [27]. Additional work has shown that SALL-4 interacts with the SWI/SNF chromatin remodeling complex through Baf60a in response to double stranded DNA damage [28]. Given the mechanistic association of SALL-4 and SWI/SNF complex as well as our finding of SALL-4 expression in SNTCS, it appears that SALL-4 may act as a surrogate marker for alterations in SWI/SNF complex proteins, including SMARCA4 loss. Although outside of the scope of our current study, analyzing the relationship between BRG1 loss and SALL-4 expression by immunohistochemistry may help elucidate the relationship between these two proteins. It is notable that SALL-4 expression was not entirely specific for SNTCS in our study, as it was also found in one case of SNUC and one case of PD-NEC. Although we attempted to avoid errors in tumor categorization by restricting our study to resection specimens, one potential explanation for these results is that these two cases may represent cases of SNTCS in which all three morphologic elements had not been sampled due to subtotal resection or lack of development of all three elements. Another possibility is that it is possible that the two tumors with SALL4 expression may fall into the recently described category of SMARCA4-deficient sinonasal carcinoma subset of undifferentiated high grade sinonasal tumors have been found to also be associated with SMARCA4 inactivation [29]. This again reinforces the need for further study of the relationship between loss of expression of members of the SWI/SNF protein complex, including SMARCA4, and the presence of SALL-4 expression by immunohistochemistry.

In summary, while β-catenin may not always have clinical utility as an immunohistochemical marker for SNTCS, SALL-4 immunohistochemistry appears to be sensitive and moderately specific for distinguishing SNTCS from other high grade primitive sinonasal tumors. Expression of SALL-4 in a sinonasal tumor should prompt an evaluation for alterations in SWI/SNF proteins, including SMARCA4.

Author Contributions

MLC: Design, case selection, data collection and analysis, manuscript drafting, review and final approval of manuscript. JSL: Case selection, review and final approval of manuscript. WCF: Case selection, review and final approval of manuscript. NAC: Case selection, review and final approval of manuscript. QS: Case selection, review and final approval of manuscript. KAE: Design, case selection, data collection and analysis, manuscript drafting, review and final approval of manuscript.

Funding

No external funding was received. The research was performed using internal Vanderbilt faculty funds provided by author KAE.

Data Availability

Raw data will be shared upon request.

Declarations

Conflict of interest

The authors declared that they have no conflict of interest.

Consent to Participate

Waiver of consent (research performed on anonymized archival materials).

Consent for Publication

Waiver of consent (research performed on anonymized archival materials).

Ethical Approval

Approved by Vanderbilt IRB board.