Beyond SMARCB1 Loss: Recent Insights into the Pathobiology of Epithelioid Sarcoma

Epithelioid sarcoma (ES) is a very rare and aggressive mesenchymal tumor of unclear origin and uncertain lineage characterized by a prevalent epithelioid morphology. The only recurrent genetic alteration reported in ES as yet is the functional inactivation of SMARCB1 (SWI/SNF-related matrix-associated actin-dependent regulator of chromatin subfamily B member 1), a key component of the SWI/SNF (SWItch/Sucrose Non-Fermentable) chromatin remodeling complexes. How SMARCB1 deficiency dictates the clinicopathological characteristics of ES and what other molecular defects concur to its malignant progression is still poorly understood. This review summarizes the recent findings about ES pathobiology, including defects in chromatin remodeling and other signaling pathways and their role as therapeutic vulnerabilities.


Introduction
Epithelioid sarcoma (ES) is a very rare (incidence~0.02-0.05/100,000) mesenchymal neoplasm of uncertain lineage characterized by a prevalent epithelioid morphology [1]. It was first described by Enzinger in 1970 as a "peculiar form of sarcoma that has been repeatedly confused with a chronic inflammatory process, a necrotizing granuloma and a squamous cell carcinoma" [2]. It primarily affects adolescents and young adults, accounting for less than 1% of all sarcomas in adults and 4-8% of pediatric non-rhabdomyosarcomatous sarcomas [3,4].
Histopathologically, ES is typically composed of nodules of cells with epithelioid morphology characterized by the co-expression of mesenchymal and epithelial markers. Specifically, ES cells are positive for vimentin, EMA (epithelial membrane antigen), and low-molecular-weight cytokeratins (CK8 and CK19), whereas they are usually negative for high-molecular-weight cytokeratins (CK5 and CK6) [22,23]. Over 50% of ES are positive for CD34, which is useful in the differential diagnosis with carcinomas [22,24,25]. ERG is also commonly expressed, which may lead to a misclassification with endothelial tumors [26][27][28]. As better described below, the loss of SMARCB1/INI1 protein expression is an essential diagnostic marker of ES [29].
According to the 2020 WHO classification of tumors of soft tissues and bone, ES is classified into two distinct subtypes: the classic or conventional type (C-ES, also called the "distal type") and the proximal type (P-ES, also called the "large-cell type"), first described in 1997 [24,29]. Originally, the terms distal and proximal indicated the anatomical localization of the tumor. Nowadays, the classification in C-ES and P-ES has histological significance, irrespective of the site of the tumor, although C-ES mainly occurs in the distal upper extremities and P-ES primarily develops in the proximal axial regions of the body. A common feature of C-ES is the presence of cellular nodules of epithelioid and spindle cells surrounding a central area of degeneration and/or necrosis that resembles a granuloma [29]. Instead, P-ES are composed of large epithelioid, carcinoma-like cells often showing rhabdoid features and growing in a multi-nodular, sheet-like pattern [29]. The two ES variants differ also for demographic and clinical characteristics. C-ES is mainly observed in adolescents and young adults (20-40 years of age), with males being affected twice as often as females. C-ES presents as superficial, slowly growing and often ulcerated nodules that resemble a cutaneous squamous cell carcinoma [30], mostly affecting the distal upper extremities (hand, forearm and arm) [31]. P-ES is most frequently diagnosed in young to middle-aged adults (20-65 years of age), with a male-to-female ratio of 1.6:1. Deep soft tissues of proximal limbs, limb girdles, and midline of the trunk are the most common sites of P-ES development [24]. P-ES is considered more aggressive than C-ES, due to the higher rate of recurrences and earlier development of metastases [24,32].

SMARCB1 Loss, an ES Molecular Hallmark
A molecular hallmark of ES is the loss of SMARCB1/INI1 protein expression, although extremely rare cases compatible with ES diagnosis but retaining SMARCB1/INI1 immunostaining (SMARCB1-proficient ES) are described in the literature [29]. SMARCB1 (SWI/SNF-related matrix-associated actin-dependent regulator of chromatin subfamily B member 1), a.k.a. INI1 (Integrase Interactor 1), is a subunit of the mammalian SWI/SNF (SWItch/Sucrose Non-Fermentable) ATP-dependent chromatin remodeling complexes. SWI/SNF complexes, also known as BRG1/BRM-associated factor (BAF) complexes, are central regulators of nucleosome remodeling. By promoting sliding or ejection of nucleosomes, they facilitate the access of the transcriptional machinery to DNA [33][34][35].
Very rare ES associated with hereditary, cancer predisposing SMARCB1 alterations have been reported. A case of homozyogously deleted ES occurring in the setting of SMARCB1 constitutional deletion was reported by Le Loarer and coworkers in a 25-yearold ES patient without prior familial or personal history of cancer [57]. More recently, an ES arisen in a patient affected by a rhabdoid tumor predisposition syndrome due to a constitutional intron/exon SMARCB1 variant (c.501-1G > A), likely responsible for aberrant mRNA splicing, has been documented [60].
In the fraction of ES that, although negative for SMARCB1/INI1 immunostaining, lack obvious SMARCB1 genetic defects (SMARCB1-intact ES), it has been suggested that SMARCB1 functional inactivation may be achieved through epigenetic mechanisms. Promoter silencing was ruled out as no promoter methylation was detected in SMARCB1intact ES [61,62], and decitabine treatment failed to restore SMARCB1/INI1 expression [62]. Instead, several miRNAs have been proposed as SMARCB1 epigenetic inactivators. Based on the differential expression between SMARCB1-intact ES and SMARCB1-deleted MRT samples, miR-193a-5p, miR-206, miR-381, and miR-671-5p were suggested to be possibly implicated in the inactivation of SMARCB1 this subset of ES [63][64][65]. Unfortunately, the capacity of these miRNAs to actually inhibit SMARCB1 in the context of ES cells was not confirmed in other studies [62]. Therefore, the mechanism underlying the loss of SMARCB1/INI1 expression in SMARCB1-intact ES remains to be clarified.
Similar to what reported for SMARCB1-negative MRT cells [66][67][68], restoration of SMARCB1 expression in SMARCB1-deleted ES cells (i.e., the VA-ES-BJ cell line) induces cell cycle arrest and impairs anchorage-independent growth and cell migration, substantiating its tumor suppressive role [69]. Moreover, the reintroduction of SMARCB1 in MRT cells correlated with re-activation of tissue-specific lineage-determining genes, Very rare ES associated with hereditary, cancer predisposing SMARCB1 alterations have been reported. A case of homozyogously deleted ES occurring in the setting of SMARCB1 constitutional deletion was reported by Le Loarer and coworkers in a 25-yearold ES patient without prior familial or personal history of cancer [57]. More recently, an ES arisen in a patient affected by a rhabdoid tumor predisposition syndrome due to a constitutional intron/exon SMARCB1 variant (c.501-1G > A), likely responsible for aberrant mRNA splicing, has been documented [60].
In the fraction of ES that, although negative for SMARCB1/INI1 immunostaining, lack obvious SMARCB1 genetic defects (SMARCB1-intact ES), it has been suggested that SMARCB1 functional inactivation may be achieved through epigenetic mechanisms. Promoter silencing was ruled out as no promoter methylation was detected in SMARCB1-intact ES [61,62], and decitabine treatment failed to restore SMARCB1/INI1 expression [62]. Instead, several miRNAs have been proposed as SMARCB1 epigenetic inactivators. Based on the differential expression between SMARCB1-intact ES and SMARCB1-deleted MRT samples, miR-193a-5p, miR-206, miR-381, and miR-671-5p were suggested to be possibly implicated in the inactivation of SMARCB1 this subset of ES [63][64][65]. Unfortunately, the capacity of these miRNAs to actually inhibit SMARCB1 in the context of ES cells was not confirmed in other studies [62]. Therefore, the mechanism underlying the loss of SMARCB1/INI1 expression in SMARCB1-intact ES remains to be clarified.
Similar to what reported for SMARCB1-negative MRT cells [66][67][68], restoration of SMARCB1 expression in SMARCB1-deleted ES cells (i.e., the VA-ES-BJ cell line) induces cell cycle arrest and impairs anchorage-independent growth and cell migration, substantiating its tumor suppressive role [69]. Moreover, the reintroduction of SMARCB1 in MRT cells correlated with re-activation of tissue-specific lineage-determining genes, indicating that failure of differentiation underpins tumor development in a context of SMARCB1 deficiency [70]. It is worth mentioning that the silencing of other components of the SWI/SNF complex in ES and MRT cells further affects proliferation, indicating that, despite the loss of SMARCB1/INI1, these tumors retain a residual SWI/SNF activity [62].
The mechanism of action of SMARCB1 as a tumor suppressor relies on the intersection with several pathways, including cell proliferation and survival [71] (Figure 2).
indicating that failure of differentiation underpins tumor development in a context of SMARCB1 deficiency [70]. It is worth mentioning that the silencing of other components of the SWI/SNF complex in ES and MRT cells further affects proliferation, indicating that, despite the loss of SMARCB1/INI1, these tumors retain a residual SWI/SNF activity [62].
The mechanism of action of SMARCB1 as a tumor suppressor relies on the intersection with several pathways, including cell proliferation and survival [71] (Figure 2). SMARCB1 negatively controls cyclin D1, E2F, and AURKA expression, and the loss of SMARCB1 in tumors was associated with an upregulation of these targets and cell cycle perturbation [68,[72][73][74]. Moreover, SMARCB1 was demonstrated to directly bind MYC and to interfere with MYC-mediated transcriptional regulation [75,76]. Accordingly, SMARCB1 loss was correlated with enhanced MYC activity and increased DNA replication in diverse SMARCB1-negative cell lines, including MRT and ES cells [77]. Since BET bromodomain inhibitors have been shown to lead to repression of MYC-driven transcription [78], this class of compounds could represent a therapeutic avenue to be explored. SMARCB1 was reported to potentiate p53 transactivation activity by direct binding [79], and to favor nucleotide excision repair by interacting with several components of this machinery, including BRCA1, BARD1 and XPC [80,81]. This suggests that SMARCB1 deficiency may result in an alleviation of p53 tumor suppressive pathway and impaired control over genome stability. Functional interactions of SMARCB1 with Wnt/β-catenin and sonic hedgehog (SHH) signaling pathways have also been documented [82,83]. In particular, SMARCB1 has been shown to bind GLI1, a SHH effector, thereby inhibiting the SHH signaling. The loss of SMARCB1 results in GLI1 hyperactivation and increased tumorigenicity, a fact that has led to the hypothesis that GLI1 targeting may be therapeutic for SMARCB1-deficient tumors [82]. Finally, a role for SMARCB1 in regulating the expression of IL6 has been recently proposed [84].
Notably, SMARCB1 was shown to repress EZH2 and, accordingly, high levels of EZH2 have been reported in SMARCB1-deficient tumors, including ES [61], AT/RT [85], MRT [86] and chordomas [87]. EZH2 is the catalytic subunit of PRC2 complexes (Polycomb Repressive Complex 2) that mediate the transcriptional repression of target genes though methylation of histone H3 on lysine 27 (H3K27). [86]. PRC2 complexes are functional antagonists of the SWI/SNF complexes, and the balance between the two is key for cellular homeostasis [35, 86,88,89] (Figure 3). SMARCB1 negatively controls cyclin D1, E2F, and AURKA expression, and the loss of SMARCB1 in tumors was associated with an upregulation of these targets and cell cycle perturbation [68,[72][73][74]. Moreover, SMARCB1 was demonstrated to directly bind MYC and to interfere with MYC-mediated transcriptional regulation [75,76]. Accordingly, SMARCB1 loss was correlated with enhanced MYC activity and increased DNA replication in diverse SMARCB1-negative cell lines, including MRT and ES cells [77]. Since BET bromodomain inhibitors have been shown to lead to repression of MYC-driven transcription [78], this class of compounds could represent a therapeutic avenue to be explored. SMARCB1 was reported to potentiate p53 transactivation activity by direct binding [79], and to favor nucleotide excision repair by interacting with several components of this machinery, including BRCA1, BARD1 and XPC [80,81]. This suggests that SMARCB1 deficiency may result in an alleviation of p53 tumor suppressive pathway and impaired control over genome stability. Functional interactions of SMARCB1 with Wnt/β-catenin and sonic hedgehog (SHH) signaling pathways have also been documented [82,83]. In particular, SMARCB1 has been shown to bind GLI1, a SHH effector, thereby inhibiting the SHH signaling. The loss of SMARCB1 results in GLI1 hyperactivation and increased tumorigenicity, a fact that has led to the hypothesis that GLI1 targeting may be therapeutic for SMARCB1-deficient tumors [82]. Finally, a role for SMARCB1 in regulating the expression of IL6 has been recently proposed [84].
Notably, SMARCB1 was shown to repress EZH2 and, accordingly, high levels of EZH2 have been reported in SMARCB1-deficient tumors, including ES [61], AT/RT [85], MRT [86] and chordomas [87]. EZH2 is the catalytic subunit of PRC2 complexes (Polycomb Repressive Complex 2) that mediate the transcriptional repression of target genes though methylation of histone H3 on lysine 27 (H3K27). [86]. PRC2 complexes are functional antagonists of the SWI/SNF complexes, and the balance between the two is key for cellular homeostasis [35,86,88,89] (Figure 3). Concomitant inactivation of EZH2 and SMARCB1 is synthetic lethal: EZH2 silencing in SMARCB1-deficient MRT cells significantly impaired cell proliferation and triggered cell senescence in vitro and, in mouse models, it prevented the formation of tumors driven by SMARCB1 loss [86]. Similarly, EZH2 pharmacological inhibition induced strong antiproliferative effects in SMARCB1-deleted cells and determined a complete regression of xenografts in mouse [90,91].
Based on these findings, EZH2-inhibitors have been considered as a potential therapeutic strategy in SMARCB1-deficient tumors. In 2013, an open-label, multicenter, phase I/II trial to evaluate the EZH2 inhibitor tazemetostat as a single agent in subjects with advanced solid tumors or with B-cell lymphomas was launched [92]. The study showed that tazemetostat had a safety profile and showed antitumor activity in a subset of patients that included ES. The study was followed by a phase II, multicenter study in adult subjects with SMARCB1/INI1-negative tumors or relapsed/refractory synovial sarcoma (NCT02601950) in 2015 [58]. Intriguingly, of the 62 ES patients enrolled in this trial, 15% (9/62) showed durable objective response after 32 weeks of treatment, and 21% (13/62) remained progression-free at 1 year [58]. These encouraging results led to the accelerated approval by the USA Food and Drug Administration (FDA) of tazemetostat for the treatment of adults and adolescents over 16 years of age with locally advanced or metastatic ES not eligible to complete surgical resection. In addition, based on preclinical evidence showing synergy between tazemetostat and doxorubicin [93], the therapeutic value of this combination is currently being evaluated as a frontline therapy in ES patients (NCT04204941). Other PRC2 components are also considered for therapeutic targeting. For instance, the clinical role of an inhibitor of EED (APG-5918/EEDi-5273), another core component of the PRC2 complexes, is going to be evaluated in different neoplasms, including ES (NCT05415098).
Intriguingly, a large fraction of SMARCB1/INI1-negative tumors show some degree of immune infiltration and PD-L1 expression [94,95], and EZH2 inhibition has been shown to have immunologic effects in both regulatory T cells and tumors [96]. These facts suggest that immune checkpoint inhibitors, as single agents or in combination with tazemetostat, may be promising therapeutic options. Accordingly, a number of clinical trials have been Concomitant inactivation of EZH2 and SMARCB1 is synthetic lethal: EZH2 silencing in SMARCB1-deficient MRT cells significantly impaired cell proliferation and triggered cell senescence in vitro and, in mouse models, it prevented the formation of tumors driven by SMARCB1 loss [86]. Similarly, EZH2 pharmacological inhibition induced strong antiproliferative effects in SMARCB1-deleted cells and determined a complete regression of xenografts in mouse [90,91].
Based on these findings, EZH2-inhibitors have been considered as a potential therapeutic strategy in SMARCB1-deficient tumors. In 2013, an open-label, multicenter, phase I/II trial to evaluate the EZH2 inhibitor tazemetostat as a single agent in subjects with advanced solid tumors or with B-cell lymphomas was launched [92]. The study showed that tazemetostat had a safety profile and showed antitumor activity in a subset of patients that included ES. The study was followed by a phase II, multicenter study in adult subjects with SMARCB1/INI1-negative tumors or relapsed/refractory synovial sarcoma (NCT02601950) in 2015 [58]. Intriguingly, of the 62 ES patients enrolled in this trial, 15% (9/62) showed durable objective response after 32 weeks of treatment, and 21% (13/62) remained progression-free at 1 year [58]. These encouraging results led to the accelerated approval by the USA Food and Drug Administration (FDA) of tazemetostat for the treatment of adults and adolescents over 16 years of age with locally advanced or metastatic ES not eligible to complete surgical resection. In addition, based on preclinical evidence showing synergy between tazemetostat and doxorubicin [93], the therapeutic value of this combination is currently being evaluated as a frontline therapy in ES patients (NCT04204941). Other PRC2 components are also considered for therapeutic targeting. For instance, the clinical role of an inhibitor of EED (APG-5918/EEDi-5273), another core component of the PRC2 complexes, is going to be evaluated in different neoplasms, including ES (NCT05415098).
Intriguingly, a large fraction of SMARCB1/INI1-negative tumors show some degree of immune infiltration and PD-L1 expression [94,95], and EZH2 inhibition has been shown to have immunologic effects in both regulatory T cells and tumors [96]. These facts suggest that immune checkpoint inhibitors, as single agents or in combination with tazemetostat, may be promising therapeutic options. Accordingly, a number of clinical trials have been designed to address this hypothesis. For instance, a phase II clinical trial aimed to test the safety and effectiveness of the combination of nivolumab (anti-PD-1 antibody) and ipilimumab (anti-CTLA-4 antibody) in SMARCB1/INI1-negative tumors is ongoing (NCT04416568) and a phase I/II study to assess the value of combining these immune checkpoint inhibitors (nivolumab and ipilimumab) with tazemetostat has just been launched (NCT05407441) for the treatment of ES, MRT, AT/RT, chordoma and other SMARCB1/INI1-or SMARCA4deficient tumors. An additional trial (NCT05286801) will determine the role of tiragolumab (anti-TIGIT antibody) and atezolizumab (anti-PD-L1 antibody) in relapsed or refractory SMARCB1-or SMARCA4-deficient tumors.

Other Players in ES Pathobiology
Beyond the loss of SMARCB1/INI1 expression, very little is known about the biology of ES. A Medline search was done using the following string "Epithelioid sarcoma AND (biology OR genetics)". By 30 June 2022, 163 papers matched the search. The following section summarizes the main results of these works and the papers quoted therein.
Very few recurrent alterations have been detected as yet in ES, in addition to SMARCB1 inactivation. Functional inactivation of SMARCB1 seems to be paralleled by lack of expression of PBRM1/BAF180, another subunit of the SWI/SNF complexes. In fact, Li and coworkers reported that over 90% (17/18) of SMARCB1/INI1-negative ES fail to express also PBRM1/BAF180, suggesting a synergic role of these two proteins in ES pathogenesis [97]. Loss of the expression of other SWI/SNF subunits, namely SMARCA4/BRG1, SMARCC1/BAF155, SMARCC2/BAF170 [98], or ARID1A/BAF250A [99], has been claimed as causative for the extremely rare SMARCB1-proficient ES cases in which SMARCB1/INI1 protein expression is retained ( Figure 2). Recently, amplification of BIRC3/YAP1 with trisomy of chromosome 2 was reported in an infantile SMARCB1-proficient ES [100], and a SS18-NEDD4 fusion was detected in a very aggressive cutaneous neoplasm with pathological characteristics resembling ES but retaining SMARCB1/INI1 expression [101]. It must be emphasized that it is still controversial whether SMARCB1-proficient ES actually exist or rather represent other, ES-like entities.
Besides impaired chromatin metabolism, ES features alterations in several signaling pathways, including EGFR, c-MET and AKT/mTOR. Cascio and colleagues reported strong and homogeneous membrane expression of EGFR in 73% (11/15) clinical ES samples analyzed by immunohistochemistry, although none of these cases showed evidence of EGFR amplification or activating mutations of the gene [102]. Marked expression of EGFR, associated with Tyr 1173 EGFR phosphorylation, was confirmed by Xie and colleagues. The authors also showed the involvement of the PI3K-AKT-mTOR signal transduction pathway in ES, with activation of mTOR (as inferred by the expression of the phosphorylated form of 4EBP1 and SRP) in all samples and loss of PTEN expression in 40% of the cases. In addition, they demonstrated that chemical inhibition of EGFR (erlotinib) impairs proliferation, migration, invasion and triggers apoptosis in ES cell models (Epi544 and VA-ES-BJ) [103,104]. Notably, the combination of EGFR blockade and mTOR inhibition (rapamycin) showed synergistic effects [103]. Synergistic effects were also observed for the combination of MET with EGFR [69] or mTOR inhibitors [69,104]. These effects were justified by the elevated levels of MET expression in both ES samples and cell lines [69,104].
ES also feature perturbation in the regulation of cell cycle. Jamshidi and colleagues demonstrated significantly reduced immunostaining for the CDK inhibitor p16 in about 1/3 of the samples analyzed [62]. The homozygous deletion of the CDKN2A locus, which encodes p16 and p14, was detected also in ES cell lines (VA-ES-BJ and HE-ES) [62,69]. Lualdi et al. reported copy number gain of the MYC gene in a large fraction of ES [54], but this result was not confirmed in other series [58,62].