Update on Sinonasal Tract Malignancies: Advances in Diagnostic Modalities

Although squamous cell carcinoma remains one of the main differential diagnoses for sinonasal tract malignancies, there are several entities with squamoid differentiation that warrant consideration. The first we will address is that of human papillomavirus (HPV)–related multiphenotypic sinonasal carcinoma. Given its morphology, this tumor was previously denoted as HPV-related carcinoma with adenoid cysticlike features; however, recent advances in the field have updated the name to more accurately reflect its varied histologic presentation due to the multiple patterns of cellular differentiation that might be present in this tumor.⁶  These tumors affect the adult population with a female predominance.^7–10  Most patients present clinically with epistaxis or nasal obstruction and, once diagnosed, these tumors are usually determined to have a high tumor stage; however, despite the high tumor stage, they do not often metastasize to lymph nodes or distant sites.¹¹  Although there is approximately a 35% recurrence rate, no cancer-related deaths have been reported to date.^11,12  On histologic examination, these tumors can be found growing in solid sheets or cribriform nests with a somewhat basaloid appearance (Figure 1, A); high-grade squamous dysplasia of surface epithelium may also commonly be seen in conjunction with the frankly invasive components of this tumor (Figure 1, B). Importantly, these neoplasms show strong expression of p16 and are positive for high-risk HPV via in situ hybridization studies (Figure 1, C). The high-risk HPV strain associated with this entity is usually type 33, although other high-risk types may be less commonly encountered.^13,14  As such, it is prudent to include evaluation of a variety of high-risk HPV strains in hospital assays when entertaining this diagnosis.

A more aggressive squamoid lesion found within the sinonasal tract is that of nuclear protein in testis (NUT) carcinoma. These cases are found in young adults within the midline structures, with approximately 57% of head and neck cases of NUT carcinoma located within the sinonasal tract.^15,16  Like other neoplasms within this location, these patients present with symptoms, including pain, nasal obstruction, and headache; many patients often have metastatic lymph nodes found at the time of tumor presentation, and distant metastases, including to bone, are not uncommon.^16–18  The survival rate is abysmal, at 30%, with many patients dying 7 months after diagnosis.^15,16  Morphologically, these tumors show sheets and nests of monomorphic small round blue cells with abundant mitoses and frequent evidence of tumor necrosis. A prominent feature of this tumor seen in about one-third of cases is abrupt squamous differentiation (Figure 2, A and B).¹⁵  Immunohistochemistry can be very useful in the diagnosis of this lesion because the morphology is not specific toward etiology. Specifically, the NUT antibody stain is positive in most cases and is highly specific (Figure 2, C).¹⁹  Cytokeratins and p63 may also be positive, but neuroendocrine, melanocytic, and smooth muscle markers are negative. At the molecular level, NUT carcinoma is defined by a chromosomal translocation and fusion between NUT midline carcinoma family member 1 (NUTM1) and bromodomain-containing protein 4 (BRD4), although other fusion partners, including BRD3 and others, have also been described.^20–24 

Like its counterpart in the nasopharynx, lymphoepithelial carcinoma of the sinonasal tract is an Epstein-Barr virus–related malignancy with a male predominance and increased incidence in Southeast Asia that often tends to involve the nasal cavity.^25–27  These lesions can have an undifferentiated appearance with a dense lymphoplasmacytic infiltrate (Figure 3, A).^25,28  Necrosis and distinct keratinized areas are usually not identified.²⁶  These lesions express cytokeratins, and squamous markers, including p40, p63, and cytokeratin (CK) 5/6, are typically positive. p16 may show patchy expression, which can be a pitfall during the workup of these lesions.²⁶  Given its association with Epstein-Barr virus, in situ hybridization for Epstein-Barr virus–encoded small RNA (EBER) is positive (Figure 3, B), although occasional EBER-negative cases have been noted.^26,27  Overall, they exhibit a favorable prognosis even when nodal disease is present^25,26,29 ; however, because this disease is quite rare, the survival data are limited given the paucity of patients for long-term follow-up.

Switch/sucrose nonfermentable (SWI/SNF)–deficient carcinomas of the sinonasal tract comprise a group of poorly differentiated tumors characterized by a SWI/SNF complex mutation demonstrating a loss of either switch/sucrose nonfermentable–related matrix-associated actin-dependent regulator of chromatin subfamily B member 1 (SMARCB1) (integrase interactor 1 [INI-1]) or switch/sucrose nonfermentable–related matrix-associated actin-dependent regulator of chromatin subfamily a member 4 (SMARCA4) (Brahma-related gene 1, BRG-1).^30,31  Interestingly, the SWI/SNF complex mutation is typically the lone alteration identified in these neoplasms.^30–33  Although both tumor types affect a wide age range, the SMARCA4-deficient tumors tend to be found in younger patients, with a median age of 44 years.^30,31  These tumors also show a predilection for male patients and usually present at an advanced stage.^30,31,34  SWI/SNF-deficient carcinomas can demonstrate variable histologic appearances, perhaps leading to the diagnosis of many of these tumors as sinonasal undifferentiated carcinomas before the advent of immunohistochemistry for SMARCB1 and SMARCA4. SMARCB1-deficient tumors can show basaloid, plasmacytoid, or rhabdoid morphology (Figure 4, A). Some tumors may also show an oncocytic appearance with glandular formation. SMARCA4-deficient tumors tend to show a more undifferentiated appearance with only rare occurrences of rhabdoid and basaloid morphology. Immunohistochemistry is key for diagnosis because their staining pattern mimics that of other sinonasal malignancies (positivity for cytokeratins, and p63 and p40 in SMARCB1-deficient tumors) but clear loss of either SMARCB1 (INI-1) or SMARCA4 (BRG-1) protein helps render a diagnosis in the absence of molecular studies (Figure 4, B and C).^31,32,35  Because this tumor is extremely aggressive, prognosis is quite poor, with many patients dying from disease within 2 years of diagnosis.^33,36 

First described in the long bones of the skeleton,^37,38  adamantinoma-like Ewing family tumor (EFT) of the head and neck is an unusual tumor of the Ewing family which can also have squamous differentiation. Although the most common extraosseous location for this malignancy is the head and neck, the tumor is most commonly found within the parotid gland, followed by the thyroid, and then the sinonasal tract.^39,40  The average age of onset is 37 years with a slight male predominance.³⁹  Morphologically, EFTs can exhibit a nested pattern of growth with a basaloid appearance (Figure 5, A). Fibrosis and hyaline basement membrane material may be seen, and mitoses and necrosis are usually readily identified (Figure 5, A). As might be expected, these tumors express CD99 and friend leukemia integration 1 (FLI1), as well as squamous markers with variable synaptophysin and chromogranin expression (Figure 5, B and C). They lack expression of S100, smooth muscle markers, NUT, and HPV, helping to differentiate them from other tumors with similar histologic appearances. Like other tumors of the Ewing family, EFTs also share an EWS RNA-binding protein 1 (EWSR1) gene rearrangement. Given that other tumors harbor similar molecular signatures involving EWSR1,^41–43  using an EWSR1 break-apart probe is not satisfactory for diagnosis; further studies to identify the gene partner are necessary, and, in the case of adamantinoma-like EFTs, the gene partner is typically FLI1.⁴⁴ 

As molecular diagnostics become more sophisticated, novel fusions are continuing to be identified, leading to the subcategorization of previously defined tumor classes. One such entity is the defective kernel 1 (DEK)–rearranged carcinoma, usually located within the middle ear, temporal bone, skull base, orbit, and sinonasal tract.^45,46  This tumor type is quite rare, with only approximately 20 cases reported in 3 articles available in the current literature. DEK is a proto-oncogene, and DEK::NUP214 (nuclear pore complex protein 214) rearrangements have previously been well characterized in acute myeloid leukemia.⁴⁷  The fusion partner in these tumors, AFF2, encodes a transcriptional activator, ALF transcription elongation factor 2,⁴⁸  and although the specific mechanism for oncogenesis is still uncertain, the DEK::AFF2 fusion likely drives tumorigenesis in some way related to uncontrolled cell growth by upregulating pro-proliferative pathways. Morphologically, these tumors can grow in a nested pattern or show papillary architecture, sometimes even appearing papilloma-like (Figure 6). They can also demonstrate an oncocytic appearance (Figure 6). Clinically, these neoplasms may act aggressively where studies show high local recurrence rates, nodal involvement, and distant metastases, including 2 patients reportedly dying from disease.^45,46  However, despite this aggressive behavior, 1 patient with such a tumor showed exquisite sensitivity to immune checkpoint inhibitor therapy,⁴⁵  suggesting a subset of these tumors may respond favorably with timely treatment using these drugs. Whether this treatment response is a characteristic of all tumors with this rearrangement is yet to be determined and will require further investigation.

Although adenocarcinoma of the sinonasal tract is currently divided into intestinal type and nonintestinal type, there exist unique entities as subdivisions, particularly of the nonintestinal type. The vast majority of these nonintestinal sinonasal adenocarcinomas demonstrate seromucinous differentiation in terms of expression of S100, discovered on GIST1 (DOG1), and SRY-Box transcription factor 10 (SOX10).⁴⁹  Despite showing similar seromucinous features, the molecular profile of these tumors is diverse, including mutations in β-catenin (CTNNB1) and v-raf murine sarcoma viral oncogene homolog B1 (BRAF), as well as a DnaJ homolog subfamily B member 1/protein kinase CAMP-activated catalytic subunit α (DNAJB1::PRKACA) fusion.^50,51 

One emerging entity with 11 cases reported in the literature includes sinonasal adenocarcinoma with ETS variant transcription factor 6/neurotrophic tyrosine receptor kinase 3/rearranged during transfection (ETV6::NTRK3/RET) fusions. These tumors are typically low grade,⁵²  but the characteristic molecular finding can be a useful therapeutic target for this patient population.

With only fewer than 20 cases reported, one of the newest entities in this class is sinonasal renal cell–like adenocarcinoma. These low-grade tumors show a female predominance with no reported metastases or recurrences.^53,54  Histologically, these tumors exhibit clear cells, much like those seen in clear cell renal cell carcinoma, without any evidence of perineural or lymphovascular invasion, necrosis, or overt nuclear atypia^53,55  (Figure 7). Of note, these patients do not have any history of renal cell carcinoma. These tumors do express CK7 and CAIX with variable paired-box gene 8 (PAX8) expression⁵⁶ ; therefore, due diligence is necessary to rule out a metastatic tumor of renal origin.

Olfactory neuroblastoma is a malignant neoplasm of neuroectodermal origin likely originating from olfactory epithelium. These tumors show nests of round blue cells with stippled nuclear chromatin and often exhibit both true and pseudorosettes. Rhabdomyoblastic differentiation may be observed in select cases. Immunohistochemistry reveals strong expression of neuroendocrine markers. Grading ranges from I to IV, and these tumors show fairly robust survival rates at 5 and 10 years, at 81% and 64%, respectively, for low-grade tumors, and 61% and 41% for high-grade tumors. Although this entity has been documented for decades, a newer tumor subtype, olfactory carcinoma, is becoming increasingly recognized. Olfactory carcinoma can look morphologically similar to olfactory neuroblastoma, including showing rosette formation, but it often also demonstrates clear cells and a more distinctly epithelial phenotype in terms of immunohistochemical profile, including keratin and epithelial membrane antigen positivity with total or partial loss of neuroendocrine markers (Figure 8, A through C).⁵⁷  Moreover, the sustentacular network as demonstrated by S100 staining is often lost in these higher-grade tumors.⁵⁷  The clinical behavior is still under investigation because these tumors are exceedingly rare. Some reports suggest these tumors respond well to induction chemotherapy; however, other reports show that 48% of patients develop persistent, recurrent, or metastatic disease.⁵⁸  Additionally, a recent study has shown that somatostatin receptor 2 (SSTR2) is almost always expressed in olfactory neuroblastoma, regardless of grade, providing a therapeutic target given the prevalence of somatostatin analogs already in clinical use.⁵⁹  In fact, a few metastatic cases of olfactory neuroblastoma were treated with SSTR2-targeted therapy in a recent clinical trial, with excellent clinical results and few side effects in patients.⁶⁰ 

Although sinonasal neuroendocrine carcinoma is no longer a distinct entity in the current World Health Organization classification given the overlap with other tumor types in terms of neuroendocrine marker expression, it is nonetheless important to distinguish olfactory neuroblastoma/carcinoma. Although there can be significant histologic overlap among the entities, bona fide neuroendocrine tumors of the sinonasal tract are typically positive for keratins with a perinuclear dotlike staining pattern.⁶¹  Moreover, in our experience, true neuroendocrine tumors do not show expression for neurofilament, whereas we have seen that in cases of olfactory carcinoma (Figure 8, D). Given that olfactory carcinoma can also express keratins, molecular diagnostics may be key to differentiating these tumors from true neuroendocrine carcinomas. Although olfactory carcinoma is quite rare, mutations in kinase insert domain receptor (KDR), myelocytomatosis (MYC), SIN3 transcription family member B (SIN3B), and NLR family CARD domain containing 4 (NLRC4) were reported in a metastatic high-grade olfactory neuroblastoma.⁶²  Moreover, high-grade olfactory neuroblastomas tend to show more genetic abnormalities than low-grade tumors in terms of chromosomal losses and gains.⁶³  Therefore, additional study in this area is necessary to help parse out the differences between these tumors and guide diagnosis.

Although most round blue cell malignancies of the sinonasal tract are true carcinomas or have a carcinomatous component, one must also consider mesenchymal lesions in the differential diagnosis. One group of tumors is that of the biphenotypic sinonasal sarcoma, a neoplasm with a strong female predominance that is usually found in the nasal cavity.⁶⁴  These lesions are composed of infiltrative spindle cells arranged in a herringbone architecture (Figure 9, A). Less cellular areas can be observed, and it is not uncommon to see blood vessels resembling those seen in hemangiopericytoma. These tumors express smooth muscle, desmin, and S100, with focal nuclear β-catenin staining also present (Figure 9, B and C). At the molecular level, these tumors show paired-box gene 3/mastermind-like protein 1 (PAX3::MAML3) fusions,⁶⁵  although alternative fusion partners include nuclear receptor coactivator 1/2 (NCOA1/2), forkhead box protein 1 (FOXO1), WW domain–containing transcription regulator protein 1 (WWTR1), and NTRK3.^66–68  Prognosis is relatively good, with only a few local recurrences reported many years after the first resection.^69,70 

As previously described, many sinonasal malignancies with an undifferentiated appearance can now be classified based on immunohistochemical profile and molecular diagnostics. The diagnosis of sinonasal undifferentiated carcinoma (SNUC) is now all but a diagnosis of exclusion. One additional subgroup of undifferentiated tumors with fewer than 50 reported in the literature that is becoming increasingly better characterized is the category of tumors with isocitrate dehydrogenase (IDH) mutations, specifically mutations in IDH2, with p.R172S being the most common mutation identified.^71,72 IDH-mutant tumors are hypermethylated and can be separated from their IDH–wild-type counterparts in a distinct class via methylation profiling.⁷³  Like in the central nervous system, identifying an IDH mutation is critical because patients with tumors harboring such mutations tend to have better survival rates than those lacking this genetic change.⁷¹ 

To emphasize the importance of molecular diagnostics, a very recent study by Jurmeister et al⁷⁴  using DNA methylation–based profiling has unraveled the shift from SNUC to more discrete categories.⁷⁴  This study showed that undifferentiated sinonasal tumors could be classified into 4 distinct tumor types, including those with neuroendocrine differentiation (IDH2 mutated or SMARCA4/AT-rich interaction domain 1A [ARID1A] mutated), SMARCB1-deficient tumors, and previously misclassified adenoid cystic carcinomas.⁷⁴  These groups of tumors also corresponded with patient outcome where the IDH2- and SMARCA4/ARID1A-mutated tumors had favorable prognoses, whereas SMARCB1-mutated tumors acted in a more aggressive manner.⁷⁴  Thus, given the widespread use of molecular diagnostic modalities, including methylation profiling, it is not unreasonable to see the loss of an undifferentiated category, and thus we must emphasize the importance of exhausting all avenues of investigation before rendering a diagnosis of sinonasal undifferentiated carcinoma.

Because the histology of these entities can be variable and show significant overlap with other tumor types, the diagnostic workup should start with a broad panel of immunohistochemical stains. Five key groups of stains should be considered. First, markers for carcinoma, including multiple cytokeratins as well as p40 and p63, should be included. In addition, we should investigate expression of neuroendocrine markers (such as chromogranin, synaptophysin, CD56, and INSM1), melanoma markers (including SOX10, S100, Melan-A, and human melanoma black [HMB-45]), and stains to rule out mesenchymal lesions (including CD99, desmin, myogenin, and myogenic differentiation antigen 1 [MyoD1]). Depending on these results, we can move to more specific surrogate markers, including those involving the SWI/SNF pathway (INI-1 in cases of focal rhabdoid differentiation and BRG1 if truly undifferentiated appearance with lack of expression of p63 and p40), translocation markers (NUT, particularly in cases of abrupt squamous differentiation, FLI1, WT1), and stains related to viral-related tumors (p16) as well as corresponding in situ hybridization studies (HPV and EBER). See the Table for a compilation of the most common morphologic and immunohistochemical features of the discussed entities. If the answer is still unclear, additional molecular studies, including fluorescence in situ hybridization to assess specific genetic rearrangements and next-generation sequencing and/or methylation profiling, may be warranted. If, and only if, all possible options are exhausted, a diagnosis of SNUC may be rendered.

Overall, malignancies of the sinonasal tract come in a variety of flavors for which careful investigation is necessary given the large amount of overlap in morphology among these entities. As a pathologist, due diligence often involves using all tools available: clinical data, radiology, immunohistochemistry, and molecular assays. The provided table is a helpful tool for such cases and provides a compiled comparison of morphology and immunohistochemical profile for the work-up of these neoplasms.

Additional Information: Coauthor Virginia LiVolsi, MD, died March 7, 2024.