Primary Cutaneous SMARCB1-deficient Carcinoma

Abstract

Background:

SMARCB1-deficient malignancies can arise in various sites. We describe a novel primary SMARCB1-deficient carcinoma of skin (SDCS) and characterize SMARCB1 mutations in non-melanoma skin cancers (NMSC).

Methods:

Cases underwent immunophenotyping and targeted exome sequencing (MSK-IMPACT) assay interrogating somatic mutations in 468 cancer-related genes. The MSK-IMPACT database from 2014–2020 encompassing 55,000 cases was searched for NMSC with SMARCB1 mutations.

Results:

SDCS arose on the scalp of an 18-year-old woman showing homozygous SMARCB1 deletion with a LATS2 G963E variant. Another case arose on the temple of a 76-year-old man harboring a SMARCB1 W206* mutation associated with loss-of-heterozygosity (LOH), 59 concurrent mutations, and a UV mutation signature (UV-MS). Both tumors exhibited INI1-loss, positive CK5/6, p40, p63, and claudin-4 with negative CD34. Of 378 NMSC cases, including 370 carcinomas, 7 SMARCB1-mutated tumors were identified: 3 squamous cell, 3 Merkel cell, and one basal cell carcinoma. Six showed UV-MS. Five INI1-interrogated cases retained protein expression suggesting they were SMARCB1-proficient.

Conclusions:

SDCS can be clinically aggressive, harbor SMARCB1 homozygous deletions or truncating SMARCB1 mutations associated with LOH, and can occur with or without UV-MS. Overall, SMARCB1 mutations in NMSC are rare with most being of undetermined significance and associated with retained INI1 and UV-MS.

Introduction

The BAF complex represents the mammalian analogue to the yeast SWI/SNF chromatin remodeling complex. Dysregulation of components in this complex may be found in up to 20% of human cancers in the form of both driver and passenger mutations¹. SMARCB1 (BAF47/INI1) is a subunit of the BAF complex and plays a role in tumor suppression via the p16-Rb, Wnt, and sonic hedgehog pathways among others. SMARCB1 loss and deficiency in its activity has been implicated in a growing number of malignancies including epithelioid sarcoma (ES), malignant rhabdoid tumors of the kidney, medullary renal carcinoma, and SMARCB1-deficient sinonasal carcinoma among others^{2, 3, 4, 5}.

SMARCB1 loss is primarily associated with ES in malignant primary cutaneous tumors. Classic-type ES typically shows mildly atypical cells with abundant cytoplasm often arranged in a granuloma-like pattern with central necrosis. Proximal-type ES exhibits rhabdoid morphology with nuclear pleomorphism arranged in sheets or nodules. Immunohistochemically, ES often expresses CD34 as well as a variety of cytokeratins and BAF47 immunohistochemistry demonstrates loss of nuclear expression. The main differential diagnosis in the skin is squamous cell carcinoma (SCC) which shows retention of BAF47 expression and typically stains for p63, p40 and CK5/6. In contrast, only a small minority of ES can show labeling for CK5/6 in rare tumor cells, and p63 and p40 expression is consistently absent in ES^{4, 6, 7}. Molecular studies would demonstrate genetic alterations in SMARCB1 in ES and a UV-mutation signature (UV-MS) in SCCs⁸. Epithelial neoplasms with alterations in SMARCB1 and INI1 protein loss arising primarily in skin have not been described.

Herein, we present two cases of a novel primary SMARCB1-deficient carcinoma of skin (SDCS) delineating their clinical, pathological and molecular features, and explore the type and frequency of SMARCB1 somatic mutations within a large clinical sequencing cohort of non-melanoma skin cancers (NMSC).

Methods

Cases

The study was approved by the Memorial Sloan Kettering Cancer Center institutional research board (IRB). Pathology slides and molecular results on the 2 SMARCB1-deficient cutaneous carcinomas were reviewed by all authors including two trained dermatopathologists (YH and TJH) and two molecular and head and neck pathologists (PC and SD). The clinical MSK-IMPACT molecular cohort (>55,000 cases) was screened for the presence of SMARCB1 genetic alterations in NMSC profiled from 2014–2020.

Immunohistochemistry (IHC)

IHC was performed using clinical protocols with primary antibodies listed in (Table 1).

Table 1:

Immunohistochemistry

Primary antibody	Type	Clone	Manufacturer
Cytokeratin	mouse monoclonal	Cam 5.2	Becton-Dickinson, Franklin Lakes, NJ
BAF-47	mouse monoclonal	25/BAF47	BD Bioscience, San Jose, CA
BRG-1	mouse monoclonal	G-7	Santa Cruz Biotechnology, Dallas, TX
BRST2	mouse monoclonal	D6	Biolegend, San Diego, CA
p40	mouse monoclonal	BC28	Biocare Medical, Concord, CA
p63	mouse monoclonal	VS38C	Agilent Dako, Santa Clara, CA
Cytokeratin	mouse monoclonal	AE1/AE3	Agilent Dako, Santa Clara, CA
CD34	mouse monoclonal	QBend-10	Ventana Medical Systems, Oro Valley, AZ
CK7	mouse monoclonal	OV-TL-12/30	Agilent Dako, Santa Clara, CA
34betaE12	mouse monoclonal	34betaE12	Ventana Medical Systems, Oro Valley, AZ
CK 5/6	mouse monoclonal	D5/16B4	Ventana Medical Systems, Oro Valley, AZ
TTF1	mouse monoclonal	8G7G3/1	Ventana Medical Systems, Oro Valley, AZ
PAX8	rabbit polyclonal	10336–1-AP	Proteintech Group, Rosemont, IL
chromogranin	mouse monoclonal	LK2H10	Ventana Medical Systems, Oro Valley, AZ
ER	mouse monoclonal	6F11	Leica Biosystems, Buffalo Grove, IL
PR	mouse monoclonal	16	Leica Biosystems, Buffalo Grove, IL
NUT	rabbit monoclonal	C52B1	Cell Signaling, Danvers, MA
HMB45	mouse monoclonal	HMB45	Agilent Dako, Santa Clara, CA
S100	rabbit polyclonal	S100	Agilent Dako, Santa Clara, CA

Molecular profiling and copy number alterations (CNA) analysis

Formalin-fixed paraffin-embedded (FFPE) tissue was submitted for tumor DNA extraction and corresponding germline DNA was obtained from peripheral blood samples. All cases were clinically tested by a targeted hybridization capture-based massively parallel sequencing assay, Memorial Sloan Kettering-Integrated Mutation Profiling of Actionable Cancer Targets (MSK-IMPACT™), a clinically validated molecular assay interrogating somatic variants in 468 cancer-related genes as previously described^{9, 10}. Sequence coverage of targeted regions in a tumor sample were compared relative to a standard diploid normal sample to identify CNA. CNA were expressed as the log₂ transformed tumor/normal ratio and a cutoff of 2.0-fold change was required to consider a gene amplified or deleted^{9, 10}. Copy number/zygosity assessment was assessed by Fraction and Allele-Specific Copy Number Estimates from Tumor Sequencing (FACETS) as previously described¹¹ and was performed on all SMARCB1 alterations and on coexisting mutations if designated as likely or predicted oncogenic. The OncoKB annotation in cBioPortal was utilized to evaluate oncogenicity¹².

Results

a). SMARCB1-deficient carcinoma of skin – two case reports

Case 1

Clinical course and pathologic findings

An 18-year-old female with no exposures and a prior history of left parietal alopecia presented with a subcutaneous nodule. The initial clinical impression was that of a benign cyst and she was observed for one year. She then returned with multiple nodules in the occipital region, scalp and left posterior neck skin and lymph nodes and underwent surgical resection (Table 2). Review of the pathology slides on her initial resection revealed an infiltrative, undifferentiated, non-keratinizing malignancy involving the dermis with no connection to the overlying epidermis noted in the provided biopsy material (Figure 1). The tumor cells were positive for AE1/AE3 and p63 and negative for ER, PR, BRST2, Melan A and TTF-1 by immunohistochemistry. She was diagnosed with a poorly differentiated carcinoma involving dermis, metastatic to lymph nodes and was further treated with radiation. No other primary tumors were found on imaging. After one year, she recurred with left neck and right axillary lymph node metastasis when she presented to our institution for further management. Her axillary lymph node metastasis was morphologically similar to the primary tumor and was positive for AE1/AE3, Cam5.2 and p40 and negative for SOX10. The concurrent cytologic preparations showed clusters and single polygonal tumor cells with indistinct nuclear borders and moderate to scant, delicate cytoplasm. Nuclei were irregular, with coarse to vesicular chromatin and prominent nucleoli. A molecular assay showed SMARCB1 deletion and prompted an additional IHC work-up of the tumor revealing nuclear loss of SMARCB1 protein, expression of claudin-4 and CK5/6, and lack of CD34 expression (Figure 1). The patient was treated with chemotherapy and immunotherapy but recurred within 7 months with cutaneous metastases on the chest showing an interstitial arrangement of the tumor cells morphologically similar to the primary scalp carcinoma. In addition, imaging work-up identified pulmonary nodules suspicious for metastases. The patient continued with the immunotherapy but succumbed to her disease 3 years and 10 months after her initial clinical presentation.

Table 2.

Clinical course of patients with SMARCB1-deficient poorly differentiated carcinoma of skin.

Case 1	Year	2015	2016	2017	2018
	DX	Left parietal skin nodularity (clinically)	PDCA in dermis	Metastatic SDCS	SDCS in the dermis
	Site	Left parietal skin	Left parietal, occipital, inferior and superior scalp, left posterior neck skin, left neck lymph nodes	Left neck lymph nodes, right axillary lymph node	Right chest skin; lung nodules (clinically consistent with metastasis)
	RX	Observation	Surgical excision and RT	Chemo, IT	IT
Case 2	Year	2010	2012	2013	2014	2015	2016	2017
	DX	BCC	BCC, SCC	SCC	SCC	Poorly differentiated SCC	Poorly differentiated malignant neoplasm (cytology)	SDCS
	Site	Right temple	Nose, left neck, upper back (BCC), Left scalp (SCC)	Left scalp	Left scalp	Right temple, parotid, LVI, PNI, right neck lymph node metastasis with extranodal extension (stage pT2 pN2b pMx)	Right upper neck	Right submental soft tissue, right retroauricular area, right abdominal skin
	RX	Surgical excision	Surgical excision	Surgical excision with skin graft	Surgical excision with reconstruction	Surgical resection, chemo, RT	RT, IT	RT, IT

Abbreviations: DX, diagnosis; RX, treatment; PDCA, poorly differentiated carcinoma; SDCS, SMARCB1-deficient carcinoma of skin; RT, radiation therapy; IT, immunotherapy, BCC, basal cell carcinoma; SCC, squamous cell carcinoma; LVI, lymphovascular invasion; PNI, perineural invasion.

Figure 1.

SMARCB1-deficient carcinoma involving scalp of an 18-year-old female. The tumor cells infiltrate dermis in the form of nests and cords intersected by variably thick, dense collagenous stroma and is associated with lymphoid aggregates at the tumor periphery [A, H&E, scanning magnification]. The tumor cells surround adnexal structures and no connection to the overlying epidermis was identified [B, H&E, 200x magnification]. Perineural invasion is seen (arrow). The tumor cells are of intermediate size and show smooth cell borders. No prominent rhabdoid features are seen although in areas, they do have a subtle plasmacytoid appearance. The nuclear/cytoplasmic ratio is high and the nuclei are round with finely dispersed chromatin and one or multiple conspicuous nucleoli [C, H&E, 400x magnification]. A cytologic preparation reveals clusters of cells with indistinct nuclear borders and scant delicate cytoplasm, vesicular chromatin and prominent nucleoli [D, Papanicolaou stain, 600x magnification]. Immunohistochemistry revealed loss of nuclear INI1 expression [E, 400x magnification] and positive staining for claudin-4 [F, 400x magnification], p40 [G, 400x magnification], with patchy CK 5/6 expression throughout the tumor [H, 400x magnification].

A CNA plot of Case 1 shows homozygous deletion of SMARCB1 (green arrow). The Y-axis depicts copy number changes expressed as the log2 transformed tumor/normal ratio by their genomic positions indicated on the x-axis. The blue dots represent individual exons and red dots indicate a ≥2-fold tumor/normal ratio [I]. FACETS analysis shows deletion of both SMARCB1 alleles as indicated by the total integer copy number 0 (black line, y-axis). A red line indicates a minor allele. SMARCB1 genomic position on chromosome 22 is indicated by the vertical green line [H].

Abbreviations: FACETS, Fraction and Allele-Specific Copy Number Estimates from Tumor Sequencing; CNA, copy number alterations.

Molecular diagnosis

A targeted exome sequencing of the axillary lymph node metastasis revealed a SMARCB1 deletion and a single coexistent mutation in LATS2 of unknown functional significance (Table 3). FACETS analysis confirmed homozygous SMARCB1 deletion and an otherwise relatively quiet genome without many notable gains or losses (Figure 1).

Table 3.

Genetic alterations in SMARCB1-deficient carcinomas of skin.

Case	Functional significance	Gene	cDNA	AA	Mutation type	Zygosity/ploidy status	Variant allele frequency
18 F	Likely oncogenic	SMARCB1	Whole gene deletion	N/A	Deep deletion	Homozygous deletion	N/A
	Unknown	LATS2	c.2888G>A	p.G963E	Missense		0.35
75 M	Likely or predicted oncogenic	SMARCB1	c.617G>A	p.W206*	Nonsense	LOH	0.44
		KDM6A	c.3878+2T>G	p.X1293_splice	Splice site	LOH	0.62
		EP300	c.4192C>T	p.L1398F	Missense	LOH	0.39
		TP53	c.529_546delCCCCACCATGAGCGCTGC	p.P177_C182del	In frame deletion	LOH	0.39
		TP53	c.867_868delinsTT	p.R290C	Missense	LOH	0.38
		FAT1	c.9568G>T	p.E3190*	Nonsense	LOH	0.38
		KMT2D	c.3810_3811delAC	p.L1271Ifs*14	Frameshift deletion	Diploid	0.27
		CIC	Inversion of exons 16–20	N/A	Structural variant	N/A	N/A
		CDK12	Intragenic deletion of exon 2	N/A	Structural variant	N/A	N/A
		CDKN2A/p16INK4A	Whole gene deletion	N/A	Deep deletion	Homozygous deletion	N/A
		CDKN2A/p14ARF	Whole gene deletion	N/A	Deep deletion	Homozygous deletion	N/A
		CDKN2B	Whole gene deletion	N/A	Deep deletion	Homozygous deletion	N/A
		PTPRD	Whole gene deletion	N/A	Deep deletion	Homozygous deletion	N/A
	Unknown	KDM5C	c.1318T>C	p.Y440H	Missense		0.57
		EIF4A2	c.307C>T	p.Q103*	Nonsense		0.49
		FGFR1	c.2378C>T	p.S793F	Missense		0.44
		MST1R	c.4189C>T	p.P1397S	Missense		0.40
		INPP4B	c.1672G>A	p.G558R	Missense		0.40
		NOTCH1	c.517C>T	p.P173S	Missense		0.40
		AGO2	c.1276G>A	p.A426T	Missense		0.37
		U2AF1	c.598C>T	p.R200W	Missense		0.36
		JAK2	c.2780T>C	p.L927S	Missense		0.36
		MEF2B	c.980C>G	p.P327R	Missense		0.32
		IKBKE	c.571C>T	p.R191*	Nonsense		0.31
		ASXL2	c.2060G>A	p.G687E	Missense		0.31
		RET	c.3095G>A	p.G1032D	Missense		0.30
		MDM4	c.272C>T	p.S91F	Missense		0.30
		CIC	c.338C>T	p.P113L	Missense		0.29
		EPAS1	c.1558G>A	p.D520N	Missense		0.29
		TERT	c.1247G>T	p.R416L	Missense		0.29
		MDC1	c.5498C>T	p.S1833F	Missense		0.29
		NSD2	c.2089C>T	p.L697F	Missense		0.29
		KIT	c.447_448delinsAA	p.G150R	Missense		0.29
		SF3B1	c.104C>T	p.S35F	Missense		0.29
		ROS1	c.2822C>T	p.S941F	Missense		0.28
		SDHA	c.589G>A	p.G197S	Missense		0.28
		ZFHX3	c.4027G>A	p.D1343N	Missense		0.28
		POLD1	c.1139_1140delinsTT	p.A380V	Missense		0.28
		KMT2D	c.4388_4389delinsTT	p.T1463I	Missense		0.28
		ASXL1	c.3500C>T	p.P1167L	Missense		0.28
		FGFR2	c.1268C>T	p.P423L	Missense		0.28
		ROS1	c.1855C>T	p.P619S	Missense		0.28
		CENPA	c.263C>T	p.A88V	Missense		0.27
		MGA	c.775C>T	p.R259W	Missense		0.27
		NOTCH2	c.3707T>C	p.L1236P	Missense		0.27
		ERBB4	c.716C>T	p.S239L	Missense		0.26
		CBL	c.1282C>T	p.P428S	Missense		0.26
		ARID2	c.1418C>T	p.S473F	Missense		0.25
		CDC73	c.251C>T	p.P84L	Missense		0.25
		BLM	c.2186C>T	p.S729F	Missense		0.25
		PREX2	c.1673C>T	p.P558L	Missense		0.24
		IL7R	c.801–1G>T	p.X267_splice	Splice site		0.24
		RPS6KA4	c.2300G>A	p.G767D	Missense		0.24
		EPHB1	c.2732C>T	p.T911M	Missense		0.23
		ARID5B	c.1759_1760delinsAT	p.E587I	Missense		0.22
		TGFBR1	c.1183T>C	p.S395P	Missense		0.18
		PAK5	c.1259_1260delinsTT	p.S420F	Missense		0.12
		PTCH1	c.1322G>A	p.R441H	Missense		0.10
		MDC1	Intragenic deletion of exons 10–12	N/A	Structural variant		0
		TEK	Whole gene deletion	N/A	Deep deletion		N/A

Abbreviations: AA, amino acid; LOH, loss of heterozygosity; N/A, not available

Case 2

Clinical course and pathology findings

A 76-year-old male with a prior history of prostatic adenocarcinoma presented with a 6-year history of multiple basal cell carcinomas (BCC) and SCC of the scalp, face and neck skin treated with surgical excisions, chemoradiation and immunotherapy. His clinical course and pathologic diagnoses are summarized in Table 2. He presented to our hospital in 2017 for further management. Pathology material before 2015 was not available for our review and comparison. Representative sections of the right temple and parotid resection specimen slides (2015) were reviewed. No other primary sites were suggested on imaging studies. The tumor involved the dermis and subcutis comprised of nests, cords and sheets of undifferentiated tumor cells within a collagenous and myxoid stroma. Connection to the overlying epidermis, keratinization, or definitive morphologic evidence of squamous differentiation was not seen in representatively submitted sections. However, in view of positive p63 and 34BE12 cytokeratin protein expression along with negative SOX10, p16, chromogranin, CK7 and CK20, the diagnostic interpretation was most consistent with poorly differentiated SCC of skin (Figure 2). At the same time, the patient underwent a submental soft tissue and retroauricular mass biopsy. On microscopic examination, the undifferentiated tumor cells were morphologically similar to the tumor of the right temple (2015). An extensive diagnostic work-up of a new biopsy revealed the tumor cells were positive for 34BE12, CK5/6, and p40, while negative for CD34, CK7, CK20, TTF-1, PAX8, NUT, S-100, chromogranin and HMB45. In addition, the tumor showed nuclear loss of SMARCB1 (INI1), retained nuclear SMARCA4 (BRG1) and positive labeling for claudin-4. A diagnosis of poorly differentiated carcinoma, SMARCB1-deficient was rendered. The tumor cells in the concurrent cytology specimen were similar, also showing multinucleation (Figure 2). Molecular testing results supported this interpretation (Table 3). Subsequently, he developed abdominal cutaneous metastases. In the view of molecular findings, he was considered for targeted therapy, but his performance status precluded eligibility for a clinical trial. He underwent proton therapy but expired from progression of his disease 61 months after his first skin cancer diagnosis and 24 months after his diagnosis of poorly differentiated carcinoma on the right temple.

Figure 2.

SMARCB1-deficient carcinoma involving the temple skin of a 75-year-old man. Infiltrative undifferentiated tumor cells in dermis are surrounded by collagenous and myxoid stroma [A, H&E, scanning magnification, inset 200x magnification]. Areas of necrosis were present, and apoptotic bodies and mitotic figures are notable. The tumor cells are variably sized, show high nuclear/cytoplasmic ratio, small to moderate amount of eosinophilic, focally very dense/pink cytoplasm [B, H&E, 200x magnification]. The undifferentiated tumor cells show no distinct rhabdoid features. Nuclei are round to oval with open or speckled chromatin, with one or multiple, focally prominent nucleoli [C, H&E, 400x magnification]. A cytologic preparation shows dishesive cells with moderate cytoplasm, prominent nucleoli and multinucleation [D, Diff-Quik stain, 400x magnification]. Immunohistochemistry also revealed loss of nuclear INI1 expression [E, 400x magnification] and positive staining for claudin-4 [F, 400x magnification], p40 [G, 400x magnification], and CK 5/6 [H, 400x magnification].

A CNA plot for Case 2 shows multiple gains and losses including homozygous CDKN2A deletion on chromosome 9p [red dots, I], which is confirmed by FACETS as the total copy number 0 [blue arrow, J]. LOH in SMARCB1 (green line) is indicated as the total copy number of 1 [J].

Abbreviations: FACETS, Fraction and Allele-Specific Copy Number Estimates from Tumor Sequencing; CNA, copy number alterations; LOH, loss of heterozygosity.

Molecular diagnosis

The submental biopsy was subjected to the targeted exome sequencing assay revealing 60 somatic genetic alterations including a truncating SMARCB1 W206* variant at a 44% variant allele frequency (VAF) (Table 3). FACETS analysis showed loss of heterozygosity (LOH) in SMARCB1 consistent with INI1 protein loss (Figure 2). Twelve additional likely or predicted oncogenic variants were detected including TP53 mutations associated with LOH and CDKN2A homozygous deletion. An abundance of G>A and C>T transitions were compatible with a UV mutation signature (UV-MS) (Table 3).

c). Molecular epidemiology of SMARCB1 genetic alterations in NMSC

Prompted by the identification of these two SDCS cases described above, the MSK-IMPACT database was interrogated for somatic SMARCB1 mutations in NMSC excluding ES of skin. A total of 378 patients NMSC, mostly with carcinomas (n=370, 98%) comprised of 406 specimens (239 primaries and 167 metastases) were genotyped by MSK-IMPACT. These included Merkel cell carcinoma (MCC, n=122), squamous cell carcinoma (n=119), extramammary Paget disease (n=65), basal cell carcinoma (n=48), skin adnexal carcinomas (n=40), dermatofibrosarcoma protuberans (n=8), aggressive digital papillary adenocarcinoma (n=2) and the two SDCS described herein. Seven additional cases with SMARCB1 genetic alterations including 3 SCC, 3 MCC and one BCC were identified. In addition to a single G>T transversion, six of 8 SMARCB1 genetic alterations were either C>T, G>A, CC>TT or GG>AA transitions and all but one case showed UV-MS. INI1 IHC performed on all cases with possible loss-of-function mutations with available material (n=5) was retained precluding the diagnosis of SMARCB1-deficient carcinoma in any of 7 additional cases (Table 4).

Table 4.

SMARCB1 mutations of undetermined significance in carcinomas of skin.

Age/sex	Diagnosis	cDNA	AA	Mutation type	Zygosity/ploidy status	Functional significance	INI1 IHC	# Mut	UV-S
71 M	SCC of scalp, BCC of forehead, arm and shoulder met to LN	AMP	N/A	AMP	AMP	Unknown	N/A	38	Yes
53 M	MCC met to parotid	c.1102C>T	p.Q368*	Nonsense	Diploid	Likely oncogenic	Retained	53	Yes
79 M	MCC met to chest wall	c.1130G>A	p.R377H	Missense	Diploid	Predicted oncogenic	N/A*	41	Yes
83 M	SCC of chest and shoulder met to LN	c.1130G>A	p.R377H	Missense	Diploid	Predicted oncogenic	Retained	7	No
63 M	SCC met to parotid	c.494C>T	p.P165L	Missense	Heterozygous gain	Unknown	Retained	68	Yes
66 F	BCC met to LN	c.868_869delinsTT	p.P290L	Missense	LOH	Unknown	Retained	48	Yes
68 M	MCC met to LN	c.378G>T	p.K126N	Missense	Diploid	Unknown	N/A	46	Yes
		c.1143_1144delinsAA	p.A382T	Missense	Diploid	Unknown	N/A

Abbreviations: SCC, squamous cell carcinoma; BCC, basal cell carcinoma; MCC, Merkel cell carcinoma; LN, lymph node; met, metastasis; PD, poorly differentiated; AMP, amplification; AA, aminoacid; LOH, loss of heterozygosity; N/A, not available; # Mut, number of mutations in the sample; UV-S, UV signature.

^*not performed but presumed retained

Discussion

We describe 2 cases of SDCS and examined the frequency and the type of SMARCB1 mutations in a large cohort of NMSC, which is enriched for metastatic and advanced cases, predominantly carcinomas, profiled by massive parallel sequencing. We found that in NMSC, SMARCB1 mutations are rare and most are variants of unknown functional significance. Retained INI1 protein expression in most cases and the presence of UV-MS evidenced by C>T, G>A, CC>TT, or GG>AA transitions affecting SMARCB1 suggest these are UV-induced passenger mutations. However, true SMARCB1-deficient carcinomas can arise in skin and can harbor a SMARCB1 homozygous deletion or truncating SMARCB1 mutations associated with LOH and clinically aggressive behavior.

Both our cases were histopathologically undifferentiated, morphologically similar and involved the dermis without demonstrated connection to the overlying epidermis in representatively submitted specimens and/or limited biopsy material. Unlike other SMARCB1-deficient tumors, rhabdoid morphology was inconspicuous in both cases with tumor cells instead displaying a plasmacytoid appearance⁴. Despite having an immunophenotype commonly seen in SCC, such as expression of CK5/6, p40 and p63, absence of squamous differentiation features, such as keratinization, can aid in distinguishing these cases from typical cutaneous SCC. Moreover, SCC arising in the scalp of a very young woman (Case 1) with biologically aggressive behavior similar to that in SMARCB1-deficient sinonasal carcinoma^{13, 14} would be unusual. While her age, clinical presentation and the tumor genotype would be more in line with ES, positive p40 and claudin-4, and negative CD34 expression argue against this being ES and support epithelial rather than mesenchymal differentiation. Indeed, Schaefer et al. reported the use of claudin-4 expression to distinguish SWI/SNF complex-deficient undifferentiated carcinomas from sarcomas. They demonstrated no expression of claudin-4 in ES whereas 80% of carcinomas harboring SWI/SNF complex deficiency were positive¹⁵.

In Case 2, the patient’s age, tumor location, prior history of multiple recurrent SCC and BCC along with a finding of UV-MS^{16, 17} could be compatible with another SCC. A positive p40/p63, diffuse CK 5/6 expression and negative CD34 would further support squamous differentiation and reinforced an immunophenotype that was incongruous with ES. Nevertheless, while the clinical history and the immunophenotype in Case 2 could justify a diagnosis of poorly differentiated SCC, we suggest that such cases should be separated from a variety of cutaneous SCC and recognized as SDCS for the following reasons: First, a pathogenic SMARCB1 variant consistent with loss of INI1 represents an opportunity to apply novel targeted therapies. Second, the tumor morphology does not support a conventional cutaneous SCC. Third, this tumor appears to be clinically more aggressive than most SCC.

EZH2 inhibitors are novel therapeutic agents for SMARCB1-deficient malignancies. The SWI/SNF complex which includes SMARCB1 is a regulator of the Polycomb Repressive Complex 2 (PRC2) of which EZH2 is a member. Accordingly, loss of SMARCB1 results in deregulation of EZH2 and PRC2-mediated gene silencing that support retention of a stem cell phenotype and, in turn, oncogenesis^{18, 19}. In SDCS, deficiency in SMARCB1 may render them sensitive to EZH2 inhibitors. Of note, our second patient had been considered for this therapy as part of a clinical trial after an initial course of chemotherapy but did not meet the qualifications of the study. The FDA has recently approved tazemetostat as the first EZH2 inhibitor for the treatment of advanced or metastatic ES²⁰. With this approval, detection of SMARCB1 alterations in these carcinomas may now have theragnostic implications.

At the molecular level, the cases were distinct. Case 1 showed a homozygous SMARCB1 deletion and very few coexisting genetic alterations of unknown functional significance. Such molecular profile would be similar to those seen in most other SMARCB1 deficiency-driven malignancies^{5, 21, 22, 23} and emphasize the oncogenic role of SMARCB1 protein loss. In contrast, Case 2 harbored a truncating SMARCB1 W206* variant associated with LOH coexisting with multiple other, likely oncogenic variants involving mainly tumor suppressor genes. While the question of a putative driver in the latter case remains open, a relatively high SMARCB1 VAF suggests its presence in the majority of tumor cells arguing against this being a subclonal event present only in a minor subset of tumor cells. Furthermore, the SMARCB1 variant in Case 2 occurred in the context of UV-MS and was also G>A transition suggesting a UV-induced etiology.

Finally, the tumor phenotype and molecular alterations detected also lent credence to our conclusion that these tumors represent a novel category of malignant neoplasm that should be recognized and distinguished from its mimics. SMARCB1-deficient tumors have been described arising in skin of special sites including the eyelid but these involved the lacrimal gland²⁴. Folpe et al. reviewed a series of malignant vulvar tumors exhibiting SMARCB1-loss classifying them into proximal-type ES, distal-type ES, and malignant myoepithelioma categories²⁵. However, the cases in the latter studies have not been characterized by p40, p63 or CK5/6 and is difficult to make comparison with the SDCS cases in the current study.

In summary, we present two cases of a novel malignant neoplasm with epithelial phenotype carrying genetic alterations in SMARCB1 resulting in INI1 loss arising in the skin that exhibits an aggressive clinical course. An IHC work-up including INI1 and claudin-4 should be performed in any poorly differentiated/undifferentiated cutaneous malignancy exhibiting a squamous immunophenotype but lacking any morphologic evidence of squamous differentiation. An accurate, timely diagnosis of SDCS can appropriately direct disease management and be critical in qualifying these patients for novel targeted therapeutic agents.

Ethics statement:

This study was approved by the MSKCC institutional research review board to be compliant with ethical standards.