SWI/SNF-deficient Sinonasal Carcinomas

Abstract

The classification of poorly differentiated sinonasal carcinomas and their nonepithelial mimics has experienced tremendous developments during the last 2 decades. These recent developments paved the way for an increasingly adopted approach to a molecular-based or etiology-based refined classification of the many carcinoma variants that have been historically lumped into the sinonasal undifferentiated carcinoma category. Among these new achievements, recognition of carcinoma subtypes driven by defects in the Switch/Sucrose nonfermentable (SWI/SNF) chromatin remodeling complex represents a major highlight. This resulted in a new definition of 4 sinonasal entities driven solely or predominantly by Switch/Sucrose nonfermentable complex deficiency: (1) SMARCB1(INI1)-deficient sinonasal carcinoma (lacking gland formation and frequently displaying a non-descript basaloid, and less frequently eosinophilic/oncocytoid morphology, but no features of other definable subtypes), (2) SMARCB1-deficient sinonasal adenocarcinoma (with unequivocal glands or yolk sac-like pattern), (3) SMARCA4-deficient undifferentiated (sinonasal undifferentiated carcinoma-like) carcinoma (lacking glandular or squamous immunophenotypes), and (4) SMARCA4-deficient subset (~80%) of sinonasal teratocarcinosarcoma. Fortunately, diagnostic loss of all these proteins can be detected by routine immunohistochemistry, so that genetic testing is not mandatory in routine practice. This review summarizes the main demographic, clinicopathological, and molecular features of these new entities.

Genetic defects leading to inactivation of components (subunits) of the Switch/Sucrose nonfermentable (SWI/SNF) chromatin remodeling complex have been increasingly identified as central molecular events driving the pathogenesis (initiation and/or progression) of a variety of benign and malignant neoplasms across all the histogenetic lineages (epithelial, mesenchymal, meningeal, glioneural, hematological and melanocytic) and occurring in different organs.¹ Among the >20 genes of the SWI/SNF complex, SMARCB1(INI1), SMARCA4, SMARCA2, and ARID1A are the genes that have been extensively studied in recent years.^1,2 The overall frequency of SWI/SNF gene mutations across all cancers approximates 20%, being similarly frequently mutated as TP53.² In line with a tumor suppressor function of the SWI/SNF complex, it is inactivating mutations or deletions resulting in loss of the respective gene product (protein) that drive these mostly highly aggressive malignancies.² Availability of well-functioning commercial immunohistochemical antibodies against these SWI/SNF proteins represent an emerging powerful tool to diagnose and screen for these genotypes and were the mainstay of the recent recognition of these new carcinoma variants which have now been included in the current World Health Organization classification of sinonasal neoplasms.

In line with the remarkable site-specific clustering of the different SWI/SNF gene mutations among the different body organs (eg, predominance of ARID1A mutations in the gynecologic organs, SMARCB1 in neoplasms of the soft tissue and SMARCA4 mutations in lung cancer),^3,4 the role of SWI/SNF alterations in tumors of the sinonasal tract has been limited to SMARCB1 (INI1) and SMARCA4 subunits.⁵ This review summarizes the main demographic, clinicopathological and molecular features of SWI/SNF-related sinonasal carcinomas.

SMARCB1-DEFICIENT SINONASAL CARCINOMAS

The SWI/SNF-related, matrix-associated, actin-dependent regulator of chromatin (SMARC) Subfamily B, Member 1 (SMARCB1; AKA: integrase interactor 1=INI1; BAF47, hSNF5) is a core subunit of the SWI/SNF complex mapped to chromosome 22q11.2.^6,7 This evolutionarily highly conserved tumor suppressor is universally expressed in the nuclei of all normal human cell types and in neoplasms lacking inactivating SMARCB1 alterations.^6,7 Loss of SMARCB1 results from biallelic inactivation of the SMARCB1 gene. SMARCB1 deficiency has emerged as a defining feature and a diagnostic tool for a variety of neoplastic entities, the prototypes of which are epithelioid sarcoma of soft tissue, intracranial and extracranial malignant rhabdoid tumors (so-called atypical teratoid/rhabdoid tumors=AT/RT), and renal medullary carcinoma.^3,4,6,7 In addition, a variable fraction of other epithelial and mesenchymal entities may show primary or secondary SMARCB1 loss with/or without associated dedifferentiation.^3,4,6,7

SMARCB1-deficient Sinonasal Carcinoma

SMARCB1 deficiency in a subset of poorly or undifferentiated sinonasal carcinoma has been first recognized in 2014 by 2 independent groups,^8,9 followed by a few additional case series, the largest multi-institutional series comprised 39 patients.^8–13 To date, <200 cases have been reported in the literature,¹⁴ although the tumor is more common than the reported cases imply (own unpublished data).

SMARCB1-deficient sinonasal carcinoma shows a slight male predominance.^8–13 Patients present at a mean age of 52 years (range, 19 to 89). Most cases present with a locally advanced disease (cT4).^8,9,11 Among the paranasal sinuses, the ethmoid sinuses are mostly affected while the nasal cavity is variably involved in addition to sinuses, but isolated involvement of the nasal cavity is uncommon.

SMARCB1-deficient sinonasal carcinoma is defined by (1) lack of features of any other defined sinonasal carcinoma type, (2) absence of morphologic squamous or glandular differentiation, and (3) complete loss of SMARCB1 expression.^5,8,9,11 Like most SMARCB1-deficient neoplasms,^3,4,6,7 these tumors display a monomorphic cell morphology, and they lack significant nuclear and cellular pleomorphisms, smudgy bizarre-looking nuclei and tumor giant cells.^5,8,9,11 Most cases (60%) show a basaloid cell morphology that, on occasion, might be indistinguishable from conventional non-keratinizing basaloid squamous cell carcinoma (SCC).^5,8,9,11 As stated above, frankly squamous features and keratinization are definitionally absent. Moreover, the basaloid cells frequently merge with anaplastic-looking larger cells.^5,8,9,11 The cells are usually disposed into large well demarcated and frequently branching solid nests and sheets. Around 30% of cases show eosinophilic cells with frequent plasmacytoid and occasionally rhabdoid cell morphology.^5,8,9,11 Scattered plasmacytoid and/or rhabdoid cells are present in the basaloid tumors as well, but often are only identifiable upon scrutiny. Focal glands have been reported in a few cases in the original description,^8,11 but gland-forming tumors are now considered separately as variants of SMARCB1-deficient sinonasal adenocarcinoma (see below). Foci of clear cells are seen in a few cases, but spindle cells and sarcomatoid features are very rare.¹¹ Surface dysplasia and carcinoma in situ (CIS) have not been observed in any of the reported cases, although variable degree of surface epithelial colonization and pagetoid speared may closely mimic surface dysplasia or CIS.¹¹

Immunohistochemically, the neoplastic cells of SMARCB1-deficient sinonasal carcinoma are uniformly keratin-positive.^5,11 Most cases show variable degree of squamous-like immunophenotype with partial or focal/patchy to diffuse expression of CK5 (64%) and p63/p40 (55%).¹¹ Almost half of the cases express CK7 (48%).¹¹ All reported cases have been negative for high-risk Human Papillomavirus (HPV), Epstein-Barr virus (EBV), and NUT.^5,8,9,11 Focal reactivity for different neuroendocrine immunomarkers has been reported.¹¹ Although SMARCB1 loss is definitional, co-loss of other SWI/SNF subunits is rare; SMARCA2 (but not SMARCA4 and ARID1A) may be detected in a few cases.¹¹Figures 1 and 2 illustrate the morphologic spectrum of these tumors.

FIGURE 1: Representative examples of SMARCB1-deficient sinonasal carcinoma. A, Basaloid (dark-stained) monomorphic cells growing along and replacing the surface epithelium and preexisting mucosal glands resulting in an exophytic-endophytic papillary growth pattern at low-power. B, Cohesive monomorphic basaloid cells with variable peripheral palisading. C, Variable replacement of normal respiratory glands may mimic in-situ tumor or a preexisting papilloma. D, This tumor showed more squamoid poorly cohesive small invasive nests embedded within desmoplastic stroma with strong expression of CK5 (E) and loss of SARCB1 (F). Dramatic (exaggerated) response to platinum chemotherapy justified seeking second opinion in this case (initially classified as squamous cell carcinoma).

FIGURE 2: Representative examples of less common features in SMARCB1-deficient sinonasal carcinoma. A, Eosinophilic (pinkish) tumor cells with copious cytoplasm forming variably sized cellular aggregates and nests, reminiscent of myoepithelial neoplasms. B, Note nuclear uniformity at high power. C, Lymph node metastasis of another eosinophilic tumor might be mistaken for non-keratinizing squamous cell carcinoma. D, Strong expression of EMA is valuable in those rare keratin-poor tumors. E, moderate to focal p40 expression is seen in 50% to 60% of cases. F, Complete (clean) loss of SMARCB1 in the one lymph node metastasis (same case as in C). G, Variable cytoplasmic cross-reactivity with the rhabdoid cell inclusions may be seen in occasional cases and should not be misinterpreted as retained SMARCB1 expression.

Molecular Background of SMARCB1-deficient Sinonasal Carcinoma

Biallelic (homozygous) or monoallelic (heterozygous) deletions affecting the SMARCB1 gene locus are detectable by FISH in the majority of cases.^5,9,11 Cases tested with larger gene panels revealed no additional driver mutations, in particular, IDH2 mutations were absent underlining their distinctness from SNUC.^15–17

Treatment and Prognosis

SMARCB1-deficient sinonasal carcinoma follows an aggressive clinical course with 56% of patients died of disease at a median of 16 months.^8–13 Cause of death was mostly locally advanced unresectable disease, but regional and distant metastases have been reported as well.^8,9,11 In a recent literature review comprising 128 patients with detailed clinical findings and a median follow-up of 17.5 months (range, 0.3 – 149), the 1-, 2- and 3-year overall survival was 84.3% (95% CI: 77.6%-91.0%), 62.9% (95% CI: 53.1-72.7), and 51.8% (95% CI: 40.8-62.8), respectively.¹⁴ The median overall survival was 39.0 months (95% CI: 28.5-49.5). In that review, male sex (P=0.029) and T4b disease stage (P=0.013) were significant determinants of overall survival in univariate, but only T4b disease (P=0.017) remained a significant prognostic factor in the multivariate analysis.¹⁴

Although no detailed prospective studies on histology-tailored therapy for this tumor type are available, an exaggerated response to platinum-based chemotherapy has been observed in anecdotal cases, indicating the need to recognize these tumors correctly, so that their most suitable therapeutic strategies can be assessed reliably in the future.¹¹

Differential Diagnosis

The differential diagnosis is mainly defined by the predominant cytoarchitectural histologic pattern in a given case.¹¹ Historically, most cases have been included in the spectrum of SNUC or were named basaloid nonkeratinizing SCC and myoepithelial carcinoma. Rare myoepithelial carcinomas showing rhabdoid cell features may display SMARCB1 loss as well, mimicking SMARCB1-deficient sinonasal carcinoma.¹⁸ However, myoepithelial markers (S100, SOX10, calponin, and smooth muscle actin) are usually absent in SMARCB1-deficient sinonasal carcinoma.¹¹ Block-type expression of p16 is uncommon in SMARCB1-deficient sinonasal carcinoma and should suggest positivity of HPV-associated basaloid SCC.^5,8,9,11 By definition, SMARCB1-deficient sinonasal carcinoma lacks expression of the NUT protein and are negative for oncogenic HPV and EBV infections. Miscellaneous primary and metastatic SMARCB1-deficient neoplasms may rarely present in the head and neck and should be considered in the differential as well.¹⁹

SMARCB1-deficient Sinonasal Adenocarcinoma

This rare variant of SMARCB1-deficient sinonasal carcinoma is essentially defined by the presence of unequivocal glandular differentiation or by presence of features characteristic of adenocarcinoma.²⁰ Less than 20 cases have been described to date but this tumor type might be underreported or included in the spectrum of unspecified SMARCB1-deficient sinonasal carcinoma.^20–24 They display a higher predilection for males (5:1) compared with undifferentiated (nonglandular) SMARCB1-deficient sinonasal carcinoma.²⁰ The median age is however similar for both tumor types (52 vs. 57 y; age range: 19 to 89 vs. 21 to 82, for undifferentiated and gland-forming tumors, respectively).^11,20 Most tumors originate in the main nasal cavity while some involve multiple sinonasal sites. Les commonly, the ethmoid or the maxillary sinuses may be involved selectively.

Histologically, SMARCB1-deficient sinonasal adenocarcinoma is characterized by a predominantly oncocytoid (more pinkish stained) or plasmacytoid cell morphology which is in sharp contrast to its predominantly basaloid (blue stained) undifferentiated counterpart.^11,20 Unequivocal gland formation (tubules, cribriform glands, intracellular or luminal mucin) and/or the presence of architectural features characteristic of glandular differentiation such as yolk sac tumor-like foci define this tumor variant as adenocarcinoma.^20–24 While the yolk sac-like foci may conceivably display any of the well-known histologic patterns seen in genuine yolk sac tumors, it is the sieve-like reticular-microcystic pattern, the secretory endometrium pattern and the intestinal-like glandular patterns that have been mostly recognized.^20–23 Although these patterns usually represent minor microscopic foci of the tumor, the sieve-like reticular/microcystic yolk sac pattern may represent the predominant or, rarely, the sole pattern throughout the tumor.^21–24 The latter can be diagnostically challenging if not thought of. The tumor cells of SMARCB1-deficient adenocarcinoma are uniformly high-grade with brisk mitotic activity. Foci of necrosis are frequent seen. Rare cases, however, in particular, those with sieve-like yolk sac pattern, may show a deceptively bland cytology.^20–24

Compared with undifferentiated (non-glandular) SMARCB1-deficient carcinoma, this glandular variant usually shows more diffuse and strong expression of CK7 (83%) and less frequent and usually patchy expression of p40 (33%).²⁰ Notably, variable expression of yolk sac-type markers (glypican-3 in 90%, SALL4 in 50%, HepPar-1 in 50%, PLAP in 11%, and AFP in 10% of cases) is frequent, irrespective of the presence or absence of yolk sac tumor-like features.²⁰ Moreover, CDX2 (27%) and CK20 (25%) are focally positive.²⁰ SMARCB1 loss is definitional for these tumors. Anecdotal co-loss of SMARCA4 has been described in a single case but the illustrated image suggests possible weak expression.²⁴ Representative images of SMARCB1-deficient sinonasal adenocarcinoma are illustrated in Figure 3.

FIGURE 3: Representative examples of gland-forming SMARCB1-deficient sinonasal carcinoma. A, Invasive glands with variable sieve-like and pseudocribriforming pattern embedded within abundant desmoplastic stroma. B, Solid and glandular growth with high-grade morphology and apocrine-like cytology mimicking apocrine salivary duct carcinoma and high-grade non-intestinal adenocarcinoma. C–E, Pure reticular-microcystic and sieve-like growth pattern in this tumor closely mimicked yolk sac tumor. The tumor cells expressed AFP in a patchy pattern (F) and SALL4 diffusely (G). H, Loss of SMARCB1 is the defining feature of these tumors.

Treatment and Prognosis

Surgery with or without irradiation was the treatment in most cases. Most reported tumors had no follow-up data due to their consult nature, so that the most appropriate treatment modality and the outcome of these tumors remain to be defined.

Differential Diagnosis

Original diagnosis in most cases was high-grade nonintestinal adenocarcinoma and, less clear cell carcinoma and SNUC.²⁰ Accordingly, exact subtyping should include primary and metastatic high-grade salivary-type adenocarcinoma (mainly salivary duct carcinoma and adenocarcinoma NOS),²⁵ metastatic adenocarcinomas from diverse other sites, germ cell tumors, multiphenotypic HPV-associated carcinomas (especially in biopsies),²⁶ SMARCB1-deficient undifferentiated or basaloid carcinomas and others. The occasional yolk sac-like patterns seen in some tumors combined with frequent immunoreactivity for the germ cell markers represent a pitfall. Genuine yolk sac tumors are vanishingly rare in the sinonasal tract and mainly are diseases of children. Accordingly, reported yolk sac tumors in adult that have predating the description of SMARCB1-deficient sinonasal carcinomas likely have mostly represented this variant and not genuine germ cell neoplasms. To date, no SMARCB1 loss has been reported in genuine germ cell tumors. On the other hand, isochromosome 12 may be of help in equivocal cases and in cases with unevaluable SMARCB1 staining due to artifacts or poor tissue preservation. It is noteworthy that the yolk sac-like pattern seen in many SMARCB1-deficient sinonasal adenocarcinomas also represents a frequent feature of renal medullary carcinoma, another SMARCB1-deficient carcinoma type.²⁷ However, this pattern is otherwise exceptionally rare in SMARCB1-deficient malignancies and it has been only recently reported in rare cases of pediatric atypical teratoid/rhabdoid tumors^28,29 and rare subsets of SMARCB1-deficient vulvar tumors.³⁰ It remains unexplained, why this pattern is likely overrepresented at sinonasal sites among SMARCB1-deficient neoplasia?

SMARCA4-DEFICIENT SINONASAL NEOPLASMS

SMARCA4-deficient Sinonasal Carcinoma

Like SMARCB1, the catalytic subunit of the SWI/SNF chromatin remodeling complex, SMARCA4/BRG1 (mapped to 19p13.2) has been increasingly recognized as primary driver in a variety of mostly highly aggressive malignancies originating in different organs in different age groups.^31–35 With few exceptions, these SMARCA4-deficient malignancies are unified by a large cell or rhabdoid cell undifferentiated epithelioid pattern. They have been described across different organs including the ovaries, the uterus, the thorax (mediastinum and lung), different parts of the digestive tract, the genitourinary tract, and very rarely the soft tissues.^31–35

Since first detailed description by Agaimy and Weichert,³⁶ no more than 22 cases of SMARCA4-deficient sinonasal carcinoma have been reported in the literature.^36–40 These highly aggressive tumors represented 4% of all poorly/undifferentiated sinonasal carcinomas, 9% of all sinonasal undifferentiated carcinomas (SNUC), and 20% of those SNUCs lacking IDH2 mutations.^37–40 Their major demographic and clinicopathological characteristics are comparable to their SMARCB1-deficient counterparts. They however display a higher predilection for males and were diagnosed at a lower median age (35 to 44 vs. 57 y; range, 20 to 70).^37–40 The majority of SMARCA4-deficient carcinomas developed in the nasal cavity and they involved multiple sinonasal sites in a subset of cases. Because of their large monotonous cell morphology and their frequent neuroendocrine immunotraits, they have been frequently initially misclassified as large cell or small cell neuroendocrine carcinomas.³⁹

Histologically, SMARCA4-deficient carcinomas are undifferentiated, hence closely akin to SNUC. They display sheets of large anaplastic epithelioid cells disposed into irregularly communicating nests and lobules or trabeculae within a sparse to prominent reactive edematous or desmoplastic intervening stroma. In contrast to SMARCB1-deficient carcinomas, the basaloid (blue cell) pattern is uncommon in SMARCA4-deficient carcinomas.³⁹ Also, the rhabdoid cells are not conspicuous and tumors composed predominantly of rhabdoid cells are rare. High-grade nuclear features, brisk mitotic activity, and prominent foci of coagulative necrosis characterize these tumors.^39,40 Similar to SNUC, SMARCA4-deficient sinonasal carcinomas are devoid of squamous cells or glandular features.¹⁷

However, foci with abortive neuroendocrine-like features or neuroectodermal-type pseudorosettes may be encountered in rare cases.³⁹ In contrast to teratocarcinosarcoma (TCS) (see below), embryonal-type (squamous, glandular, or respiratory) epithelia, mesenchymal/stromal elements and unequivocal neuroepithelial elements are lacking.

In line with their undifferentiated epithelial origin, SMARCA4-deficient sinonasal carcinomas express only low-molecular weight keratins but lack squamous cell markers (CK5, p63, and p40). Rarely, CK7 may be positive in a patchy manner. Moreover, p16 and NUT are negative. A variable (usually patchy and weak) expression of one or more neuroendocrine markers is observed frequently including synaptophysin (90%), CD56 (60%) and chromogranin-A (40%).³⁹ Rare cases display olfactory-like immunophenotypic features limited to the periphery of the tumor cell lobules in rare suggesting limited (abortive) hybrid neuroepithelial differentiation in line with a histogenetic origin from the olfactory membrane and highlighting that these tumors may share some characteristics with other olfactory-derived malignancies (olfactory neuroblastoma and olfactory carcinoma).^39,41 However, diffuse expression of specific neuroendocrine markers and calretinin is limited to olfactory neuroblastoma and is not seen in SMARCA4-deficient carcinomas. Global loss of SMARCA4 in the neoplastic cells is definitional for SMARCA4-deficient carcinomas. Co-loss of SMARCA2 may be seen in a few cases but SMARCB1 is retained in all cases.³⁹ Representative images of these tumors are illustrated in Figure 4.

FIGURE 4: Representative examples of SMARCA4-deficient sinonasal carcinoma. A, Submucosal growth with irregular lobulation and hemorrhages. B, High power of same case showing large undifferentiated epithelioid and rhabdoid cells. C, Prominent foci of necrosis and packaging mimicking large cell neuroendocrine carcinoma. D (same case as C), Transition from more basaloid-looking small cells (left) to larger epithelioid cells (right). E, This tumor (negative for neuroendocrine markers) showed prominent organoid nests surrounded by calretinin+ and synaptophysin+ peripheral cells indicating partial olfactory-like differentiation (shown is calretinin stain). F, Complete loss of SMARCA4.

Prognosis and Treatment

SMARCA4-deficient sinonasal carcinomas are highly aggressive, similar to their extra-sinonasal counterparts in other organs. They are only next to the uniformly lethal NUT carcinoma in their aggressiveness with 2/3 of affected patients succumbing to their disease within 1 year.^39,40

Differential Diagnosis

SMARCA4-deficient sinonasal carcinoma may closely overlap with a variety of poorly/undifferentiated sinonasal carcinomas that lack a squamous immunophenotypes (SNUC, large cell neuroendocrine carcinoma, p40-negative subsets of NUT carcinoma and CK5/p40-negative SMARCB1-deficient carcinomas). Accordingly, SMARCB1, NUT and neuroendocrine markers are mandatory components of the immunopanel to be used. Although significant overlap may be observed in some cases, the presence of unequivocal neuroendocrine cell features combined with diffuse expression of specific neuroendocrine markers are mandatory to diagnose and separate large cell neuroendocrine carcinoma.

In limited biopsies, separation from multiphenotypic sinonasal carcinoma may be difficult as well, but the uniform presence of salivary-like immune-phenotypes (in particular, uniform expression of SOX10 combined with p16) are helpful in solving the issue if SMARCA4 staining is not readily assessable. Olfactory neuroblastoma (Hyams grade 4) and the recently proposed olfactory carcinoma are other major considerations.⁴¹ Presence of unequivocal epithelial differentiation in a neuroblastoma-like background enables the separation of olfactory carcinoma from both high-grade neuroblastoma and SMARCA4-deficient carcinoma.

SMARCA4-deficient Sinonasal Teratocarcinosarcoma

This rare multiphenotypic (trilineage) and highly aggressive site-specific sinonasal malignancy is still defined by morphology in the current World Health Organization classification but it merits mentioning here due to overlapping molecular features with SWI/SNF-deficient sinonasal carcinomas. TCS is defined by a triphasic growth of teratoma-like (embryonal epithelium of diverse types, neuroectodermal differentiation, and primitive neuroepithelium), carcinoma-like (either malignant differentiated epithelial elements or keratin-positive poorly differentiated proliferations), and sarcomatous stromal/mesenchymal elements with (mainly rhabdomyoblastic and rare osteochondroblastic) or without heterologous mesenchymal elements.⁴² Despite their original description and conceptualization as “malignant teratoma,” a “germ cell origin,” could never be proven for TCS. Notably, recent developments have enabled more insight into the molecular tumorigenesis and histogenesis of many multilineage malignancies, historically considered “malignant teratoma”. Surprisingly, it turned out that many of these highly heterogeneous aggressive malignancies are indeed “single gene diseases”. So “malignant teratomas” of the thyroid could be redefined as DICER1-associated somatic teratoid (blastomatous) malignancies.^43,44 On the same line, our group has recently identified biallelic inactivation of SMARCA4 as the sole driver molecular event in the majority of sinonasal TCS with 68% of cases showing global loss in all tumor components and 14% displaying heterogeneous clonal deficiency.⁴⁵ These observations suggested the possibility that TCS and SMARCA4-deficient carcinoma might belong to the same entity. However, histologic uniformity of SMARCA4-deficient carcinomas as opposed to the significant heterogeneity of TCS argues for the notion that the 2 represent 2 independent diseases unified by similar molecular pathogenesis. This is not surprising given the growing evidence that the mere genotype does not preclude separate classification of histogenetically and biologically diverse entities sharing the same gene defect.⁴ Moreover, transitional areas between TCS and SMARCA4-deficient undifferentiated carcinoma-like foci are paradoxically rare in the experience of the author and are hence not enough to explain the morphologic uniformity observed in reported SMARCA4-deficient undifferentiated carcinomas. One such rare occurrence is illustrated in Figure 5.

FIGURE 5: Representative examples of a rare sinonasal teratocarcinosarcoma showing a carcinoma-like area. A, Abrupt transition from primitive small cell neuroepithelial component (left) to a large cell solid carcinoma-like area (right). B, Higher magnification of the carcinoma-like area shows subtile similarity to SMARCA4-deficient carcinoma but with unusual syncytial growth pattern in contrast to the dyscohesive cell growth of the latter. C, Higher magnification of the neuroepithelial component showing scattered embryonal-type mature squamous foci. D, Higher magnification of neuroepithelial-type rosettes, note small cellular aggregates budding off the rosettes and the intervening mesenchymal stroma. E, Calretinin stain showing similar pattern as depicted in Figure 4. F, SMARCA4 loss is observed both in the neuroepithelial and the stromal component of the same case (all images in this figure are from the same tumor).

Therapeutic Implications of SWI/SNF Deficiency

To date, there exists no specific therapy for these rare and extremely aggressive diseases. However, their varying aggressiveness and the anecdotally reported enhanced response to specific chemotherapeutic regimens strongly argue for classifying them separately. In addition to the above-mentioned enhanced responsiveness of some tumors in the SMARCB1-deficient and SMARCA4-deficient groups to platinum-based regimens,^11,46 promising results of emerging therapeutic opportunities targeting EZH2, CDK4/6 and other SWI/SNF complex vulnerabilities have been shown in clinical trials.^47,48 Finally, evidence is accumulating that at least a (yet not reproducibly identifiable) subset of SWI/SNF-deficient malignancies may respond well to immune checkpoint inhibition therapies.^49,50 All these observations argue for precisely classifying these rare malignancies. Only this approach would likely enable exploration and optimization of morphomolecular-tailored specific therapeutic strategies for them.⁵¹