Clear cell sarcoma of soft tissue, formerly known as malignant melanoma of soft parts, is a poor prognosis neoplasm primarily affecting young adults between twenty and forty years of age.1 This tumor is distinct from the similarly-named clear cell sarcoma of the kidney, a rare paediatric renal tumor with highly variable histological patterns and a predilection for bone metastasis.2 For simplicity, the term "clear cell sarcoma" will be used throughout this review to refer to the soft tissue cancer.
Clear cell sarcoma was first recognized in 1965 by Franz Enzinger as a previously undescribed malignant sarcoma associated with tendons and aponeuroses, morphologically distinct from other malignant tumors of these tissues such as fibrosarcoma and synovial sarcoma.3 Since then, many technological advances – including cytogenetic karyotyping, polymerase chain reaction (PCR), fluorescence in situ hybridization, and tissue microarrays have led not only to improved diagnostic tools, but also to an understanding of the molecular biology and genetics of clear cell sarcoma.4,5 However, many questions remain as to why the genetic mutations present in clear cell sarcomas lead to aggressive disease that is highly resistant to current chemotherapies.6
Chromosomal Translocations In Clear Cell Sarcoma
Translocation-associated sarcomas such as clear cell sarcoma, Ewing family tumors, and desmoplastic small round cell tumor possess non-homologous chromosomal translocations, whereby two different chromosomes interchange parts with each other, that appear to be fundamental in the progression of these diseases. In most cases, chromosomal translocations in these tumors generate chimeric master transcription factors by fusing one transcription factor’s DNA-binding domain to a regulatory domain from a different transcription factor, leading to dysregulated expression of the original target genes. Such events are thought to be fundamental oncogenic driver mutations: though these sarcoma cells often carry relatively few genetic mutations, translocations involving master regulator oncogenes simultaneously dysregulate a number of downstream oncogenes leading to cancer progression. This has been best characterized for synovial sarcoma with the invention of the synovial sarcoma mouse model, wherein the conditional expression of the SS18/SSX fusion oncoprotein as the sole human protein expressed in immature mouse myoblasts leads to development of tumors in mice with the same histology, protein and gene expression profile changes characteristic of human synovial sarcomas.7 Similarly, various human mesenchymal cell lines undergo malignant transformation after expression of the EWSR1/FLI1 fusion oncoprotein consistently found in Ewing family tumors.8,9
As opposed to the often complicated mutational profiles of pleomorphic sarcomas and most carcinomas, translocation-associated sarcomas have genetic changes that are, at least in theory, ideal candidates for targeted therapies. Clear cell sarcoma bears an in-frame t(12;22)(q13;q12) translocation that produces a fusion oncoprotein EWSR1/ATF1. If the effects of this oncoprotein can be reversed by a targeted therapy, tumor growth would be expected to stop, greatly improving prognosis or possibly even curing patients with this disease.
Chromosomal translocations occur when portions from two chromosomes are broken and the different sequences get fused together. By contrast, chromosomal crossover during sexual reproduction involves the interchanging of chromosomes at identical (homologous) regions. Chromosomal translocations may actually be relatively common events, but it is rare that they occur in the middle of genes to produce transcripts encoding fusion protein sequences derived from different chromosomes, as in clear cell sarcoma. It is especially problematic when these abnormal fusions involve "master" transcription factors, which control many pathways of cellular control, permanently fused to new regulatory factors.
Mesenchyme is a form of undifferentiated connective tissue derived from the mesoderm ("middle tissue") during embryogenesis. Mesenchymal cells have the potential to differentiate into various tissues such as muscle, bone, and tendons. By definition, sarcomas are cancers of mesoderm-derived tissues and are considered to involve the oncogenic transformation of mesenchymal cells. Numerous experiments have attempted to artificially reproduce oncogenic transformation in mesenchymal cells by introducing known sarcoma mutations, such as translocations, and observing whether the cells become like the expected sarcoma. Such experiments were successful in forming Ewing family tumors, a group of sarcomas arising in bone and soft tissue usually caused by a EWSR1/FLI1 fusion oncoprotein.
Clinical Features of Clear Cell Sarcoma
Most patients present between twenty and forty years of age with a painless mass arising in the distal extremities, especially the feet and ankles, fixed to tendons and aponeuroses.1 Sites such as the arms, hands, and trunk are also reported.10 Clear cell sarcoma is primarily deep-seated and very rarely originates in the subcutis or lower dermis of the skin.1
Clear cell sarcoma is a locally aggressive neoplasm with a high rate of recurrence and metastasis (up to 50%).1 Five year disease-specific survival rates have been reported as 50-67%, but are not representative of long term survival since many patients develop lung and bone metastases more than five years after initial resection.3 Ten and twenty year disease-specific survival rates are better long term indicators of disease-specific survival with rates reported as low as 33% and 10%, respectively, reflecting the fact that current chemotherapies have limited effectiveness in preventing or curing metastases after resection.1,6
Clear cell sarcoma is commonly characterized using magnetic resonance imaging (MRI). T1-weighted images show a slightly hyperintense, homogeneous tumor signal compared to nearby muscle tissue. T2-weighted images, where water gives a stronger signal than fat tissue, show high intensity after gadolinium contrast, especially compared to surrounding muscle tissue. It is thought that the melanin content in clear cell sarcoma may alter the MRI signal intensities compared with other soft tissue tumors, although these changes are neither dramatic nor specific enough to make a diagnosis by imaging alone.
Although rare overall, clear cell sarcoma has been reported as the second most common malignant soft tissue tumor of the foot and ankle in patients aged 20-40, after synovial sarcoma (and excluding Kaposi sarcoma); therefore, anatomical location and observable association with tendons or aponeuroses can be a valuable diagnostic clue in clear cell sarcoma cases10,11 Necrosis or bony destruction are rarely identified, leading to underestimation of malignant potential prior to biopsy.10
This imaging technique primarily differentiates between tissues of varying water and fat content. T1-weighted images have a bright signal for fatty tissue and low signal where there is high water content. T2-weighted images are the opposite, enhancing the water signal. Bone is dark in both types of MRI image, unlike in plain X-ray or CT images, where bone is white. Since soft tissue tumors usually induce an increased blood supply to grow, they often appear enhanced in T2-weighted images. Gadolinium contrast agent is administered in order to further enhance the signal of water in blood vessels.
Although computed tomography (CT) has limited utility above MRI in imaging primary soft tissue sarcomas, it is useful for monitoring local recurrence and lung metastases.12,13 Distant systemic metastases of sarcomas can be identified by combining a full-body CT scan with positron emission tomography (PET). PET highlights tumor tissues using an injected tracer substance and is visualized as a hyperintense signal in comparison to non-malignant tissue. When a CT scan is overlaid with a PET image, radiologists can locate and monitor clear cell sarcoma lesions throughout the body.
Grossly, clear cell sarcoma tumors are ovoid, variably circumscribed, and have a history of slow growth – belying their high metastatic potential1 They can display red-brown to black pigmented cut surfaces on an otherwise typical grey to tan background1,14 Diagnosis of clear cell sarcoma is currently based on histopathology and immunohistochemistry (IHC), supported by molecular testing (most commonly fluorescence in situ hybridization) to exclude the differential diagnosis of melanoma.5 Clear cell sarcoma tumor specimens display fascicular growth patterns of spindle or polygonal cells15–17 Cytoplasm is either clear or eosinophilic, and bundles of cells are outlined by eosinophilic, fibrous septae.
At the cellular level, the degree of malignancy of clear cell sarcoma can be deceptive, in that mitotic figures are sometimes few with nuclei that are neither hyperchromatic nor pleomorphic, although histologic variants exist.1,16 Uniquely among primary soft tissue tumors, pre-melanosomes are present in almost all cases of clear cell sarcoma,1,18 detectable by electron microscopy. As a result, by IHC, clear cell sarcoma cells are almost always positive for the melanoma markers S-100, HMB45 and melan-A,1 although melanin staining is not always observed.19
Histopathology involves making diagnoses by microscopic examination of surgically-excised tissue specimens. Pathologists are often able to identify specific diseases by the shape and growth pattern of cells. This can be aided by immunohistochemistry, which involves using antibodies to stain tissue specimen slides for the key proteins expressed in diseased tissue in order to distinguish from otherwise microscopically similar diseases. FISH is a molecular genetic (DNA) test that can be applied to microscope slides to detect gene translocations, amplifications, or deletions that occur in diseases such as cancer to further solidify a diagnosis in difficult cases.
Clear cell sarcoma tumors display varied levels of melanocytic differentiation based on the degree of expression of melanocytic IHC markers, the number of pre-melanosomes in the cytoplasm, and the expression of melanin, a pigment normally expressed by melanocytes in the basal layer of the epidermis.1,17 Clear cell sarcoma cells are currently thought to be derived from a neural crest cell precursor in common with melanocytes rather than from more differentiated melanocytes themselves.19,20 Interestingly, evidence supports that clear cell sarcoma melanocytic differentiation and transformation result from over-activation of melanocyte-specific transcription factors, such as microphthalmia-associated transcription factor (MITF), in undifferentiated neural crest-originating mesenchymal cells.18 Again this raises similarities with melanoma, a disease where, in addition to somatic activating mutations of BRAF, MITF gene amplifications and overexpression of MITF have been reported.21 Though MITF amplifications have not been investigated in clear cell sarcoma, the principal oncogenic event is the EWSR1/ATF1 translocation.22 Thus, as opposed to anaplastic cancers where aggressiveness is usually correlated with dedifferentiation, malignant transformation of clear cell sarcoma is possibly linked to increased but dysregulated expression of melanocytic differentiation genes in multipotent melanocyte progenitors.
Neural crest cells are a type of mesenchymal cell originating at the time of neural tube formation during early embryogenesis. Cells of the neural crest migrate throughout the body to give rise to many tissues, including the melanosomes of the skin. Clear cell sarcoma is hypothesized to originate only in neural crest-derived cells that harbor the specific EWSR1/ATF1 oncogenic translocation.
The distinctive genetic aberration most consistently found in clear cell sarcoma is a reciprocal chromosomal translocation of chromosomes 12 and 22 occurring at chromosomal arms q13 and q12, respectively, notated t(12;22)(q13;q12). The translocation produces an in-frame gene fusion of Activating Transcription Factor 1 (ATF1) and Ewing Sarcoma Breakpoint Region 1 (EWSR1). This chromosomal translocation was first described by cytogenetic karyotyping in the early 1990s by Bridge et al.23,24 Some cases of clear cell sarcoma may harbour cryptic fusions or use alternative partners for EWSR1, such as CREB1, in order to cause similar oncogenic changes.25,26 In recent years, EWSR1 translocations with either ATF1 or CREB1 have been confirmed in over 90% of analyzed cases using PCR and dual color, break-apart FISH.16,25,27,28 However, Pierotti et al. claims that a significant number of deep soft tissue cases with no history of melanoma have been misdiagnosed as melanoma based on the absence of a EWSR1-translocation, when many were actually true clear cell sarcoma.29 Both EWSR1-translocation-negative cases and metastatic melanoma cases were consistently found to harbor chromosome 22 amplifications.
The gastrointestinal variant of clear cell sarcoma harbors some of the same translocations present in clear cell sarcoma of soft tissue, although its histology and protein expression have some differences. Gastrointestinal clear cell sarcoma often has a more heterogeneous growth architecture including solid, nested, and pseudo-papillary patterns all within a single tumor.28 Cell morphology is primarily epitheliod with prominent nucleoli and higher numbers of mitotic figures. Gastrointestinal clear cell sarcoma also only rarely shows staining for melanoma markers other than S-100. Taken together, these characteristics of the gastrointestinal variant of clear cell sarcoma suggest it likely represents a disease distinct from the more common soft tissue variant.
Several translocation breakpoints of t(12;22)(q13;q12) give rise to different chimeric EWSR1/ATF1 transcripts, diagrammed in Figure 3, which were characterized by PCR analyses using primer sets specific to various regions of EWSR1/ATF1. The three most common fusion transcript variants were described in detail by Panagopoulos et al. (2002), each with different genomic breakpoints between the EWSR1 and the ATF1 genes.30 pes 1 and 2 are by far the most common fusion transcript variants found in clear cell sarcoma.30,31
To date, two distinct partner genes have been observed to be fused with EWSR1 in clear cell sarcoma, ATF1 being the more common. Antonescu et al. (2006) first described the fusion of EWSR1 with CREB1, a close homolog of ATF1, in gastrointestinal clear cell sarcoma patient specimens.28 Others have described the EWSR1/CREB1 variant arising in sites other than the gastrointestinal tract, including peripheral soft tissues typical of conventional clear cell sarcoma,16,25–27 challenging the original belief that this variant is specific to the gastrointestinal tract. Inversely, the EWSR1/ATF1 variant has been observed in the gastrointestinal tract as a primary location.32 This less common fusion variant in clear cell sarcoma resembles the Type 2 fusion variant of EWSR1/ATF1 described by Panagopoulos et al. Given the homology between ATF1 and CREB1, fusion of EWSR1 to either of these genes may exhibit similar oncologic effects specific to the biology of this disease.
Few other consistent genetic mutations have been reported in clear cell sarcoma. Cytogenetically, cases lacking identifiable t(12;22) have been found to occasionally harbor chromosome 22 amplifications.23,24 Trisomy 8, and less often trisomy 2 and trisomy 7, have also been observed as consistent non-random chromosomal changes in clear cell sarcoma. The t(12;22)(q13;q12) translocation expressing EWSR1/ATF1 has not yet been definitively proven to be responsible for cellular transformation to clear cell sarcoma; in contrast, the Ewing family tumor fusion EWSR1/FLI1, when expressed in bone marrow-derived primary mesenchymal stem cells, is adequate to initiate transformation to cells with Ewing family tumor-like characteristics.8,9 By way of comparison, EWSR1/ATF1 expression in bone marrow-derived mesenchymal progenitor cells is not sufficient for transformation and tumor growth in xenografted mice, suggesting that expression of EWSR1/ATF1 in the correct cellular context (possibly requiring neural crest-derived cells) may be necessary.33
EWSR1/ATF1 Protein Structure and Function
The in-frame t(12;22)(q13;q12) translocation in clear cell sarcoma generates transcripts encoding a characteristic fusion oncoprotein with similarities to those expressed in Ewing family tumors, desmoplastic small round cell tumor, extraskeletal myxoid chondrosarcoma, and in variants of myxoid liposarcoma.34 These tumors all express chimeric proteins where EWSR1 is fused to a DNA-binding domain of another transcription factor. However, little is yet known about the EWSR1/ATF1 fusion oncoprotein in terms of its target genes, with the exception of the MITF gene.
Important insights could come from identifying the protein functions and interactions of EWSR1 and ATF1 in non-malignant normal tissues and in tumor cell lines. Native EWSR1 protein function remains largely elusive, but several studies have shown that it may concurrently act as both a potent transcriptional activator and repressor. EWSR1 protein structure, shown in Figure 4, is composed of a C-terminal RNA-binding domain35,36 and an N-terminal activation domain called EAD (EWSR1 Activation Domain).36 Previously described as "molecular Velcro", the EAD possesses an unfolded, disordered structure reminiscent of some transcriptional activation domains, and is capable of (at least transiently) binding up to one hundred proteins in the human proteome.37
Transcription factors are proteins that bind to promoter regions of DNA to regulate gene expression. Transcription factors have dedicated structures (DNA binding domains) involved in recognizing distinct recognition sequences. EWSR1 is not a transcription factor per se, and it not supposed to bind DNA; however, the EWSR1/ATF1 fusion abnormally gains the ability to bind and regulate ATF1 target genes that it ordinarily should not.
Protein-protein interactions: In addition to binding DNA, proteins also bind other proteins through specific interactions that occur between protein binding domains. Protein-protein interactions have many functions; for instance, proteins bind and interact to transmit signals between the cell membrane and the nucleus in order to stimulate cell growth and development, a commonly over-activated theme in cancer. Some proteins are only functional in a complex made up of several proteins bound together, such as ATF1 and CREB1. Other proteins act as molecular scaffolds that hold closely several proteins that otherwise do not interact, which may be how the EWSR1/ATF1 fusion oncoprotein strongly activates transcription.
Native ATF1, a member of the CREB transcription factor family, is a transcriptional co-activator heterodimerizing with CREB1. Both bind specific DNA sequences called cAMP response elements (CREs) when intracellular cAMP levels increase as a consequence of cell signaling pathways.38,39 The net result is activation of cAMP-inducible genes, which are widespread in the human genome.39 The normal regulatory domain of ATF1 and CREB1 activity is partly missing in clear cell sarcoma fusion oncoproteins, implying they are no longer properly regulated by cAMP levels.38,40,41 The retained portion of ATF1 contains a DNA binding domain (termed bZip) that is responsible for binding to CRE sequences in gene promoters.38,41 bZip domains are widespread in DNA-binding proteins, including those of the ATF/CREB family.38 Effectively, the activation domain of EWSR1 gains the ability to bind CRE-containing promoters through fusion to the DNA binding domain of ATF1.
The EWSR1/ATF1 fusion oncoprotein of clear cell sarcoma has been hypothesized to act as a potent activator of CRE-containing promoters because of the potent transcriptional activation domain of EWSR1, and computational analysis of gene expression profile data supports this concept.42 In normal cells, transcription of ATF1 target genes is regulated by a number of kinases. For example, protein kinase A phosphorylates ATF1 and CREB1 at their kinase inducible domains, which leads to activation of targeted genes through binding with CREB-binding protein (CBP). CBP is able to directly alter the epigenetic code in chromatin by histone acetylation to facilitate gene expression.40,41 Evidently, the control region of the kinase inducible domain of ATF1 is lost in the EWSR1/ATF1 fusion oncoprotein (Figure 4), and it is suspected that the ATF1 region of the fusion is not responsible for binding CBP in EWSR1/ATF1. Similarly, the entire kinase inducible domain of CREB1 is lost in EWSR1/CREB1. The capability of EWSR1/ATF1 to interact with CBP has been confirmed in numerous studies, although the exact site of binding has yet to be confirmed.41,43 Evidence shows that a region within EWSR1 between amino acids 83-227, which contains a lysine acetylation site (GNK; figure 4), is required for CBP interaction. Acetylated lysines are known to interact with protein domains called bromodomains. CBP is known to possess a bromodomain that binds to acetylated lysines (44); thus, it is possible that CBP interacts with EWSR1/ATF1 at this acetylation site.
The role of epigenetics in cancer is a relatively recent area of active research. Proteins such as HDAC, Polycomb, CBP and DNA methyltransferase chemically modify DNA and histones in ways that do not change the DNA sequence. These changes involve histone acetylation and methylation and DNA methylation, and can be thought of as one way cells decide which genes should be turned on or off at a given time in a given tissue. Malfunctioning of epigenetic programming may lead to a variety of diseases, including cancer. Some translocation-associated sarcomas (such as synovial sarcoma) are known to alter the epigenetics of, and thereby dysregulate, important genes that drive oncogenic transformation.
It has been reported that phosphorylation of serine 266 within the EWSR1 portion of EWSR1/ATF1, shown in Figure 4, is key to transcriptional activation.45 A study by Olsen and Hinrichs reports that perturbed phosphorylation of serine 266 reduced the binding of EWSR1/ATF1 to CRE sequences. They conclude that serine 266 is a necessary molecular feature for EWSR1/ATF1 to bind to DNA and transactivate transcription. Understanding these structural features of EWSR1/ATF1 may open the doorway to developing new targeted treatments for clear cell sarcoma, as well as the several related sarcomas with similar EWSR1 translocations.
A Transcription-Dysregulating Protein Complex
EWSR1 has been thought to interact with a large number of transcriptional co-activator proteins.36 This could partially explain the ability of EWSR1/ATF1 to strongly activate cAMP-inducible genes in clear cell sarcoma, acting as a protein scaffold for other transcriptional co-activators. SOX10 and CBP are prime candidates as transcriptional activation complex partners with EWSR1/ATF1 resulting in overactivation of downstream oncogenes. In fact, SOX10 may be the most important protein cofactor for EWSR1/ATF1 required for high expression of MITF in clear cell sarcoma,18 given that SOX10 upregulates MITF expression in non-malignant cells, whereas depletion of SOX10 in clear cell sarcoma leads to a dose-dependent depletion of MITF activity, even in the presence of EWSR1/ATF1. Together, these results suggest that SOX10 is required for MITF expression and that EWSR1/ATF1 acts in conjunction with SOX10 to drive high levels of MITF oncogene expression in clear cell sarcoma.
SOX10 is a critical transcription factor expressed in undifferentiated cells of neural crest lineage.71 Therefore, the fact that clear cell sarcoma cells also express SOX10 supports the theory that clear cell sarcoma originates from neural crest cells. Furthermore, if SOX10 activity could be targeted by a drug, perhaps clear cell sarcoma cells would selectively die while non-malignant cells would remain unaffected in adults.
On the other hand, it is possible that EWSR1/ATF1 could also mediate transcriptional repression by the recruitment of repressor proteins to target promoters, analogous to EGR1 repression by the synovial sarcoma SS18/SSX oncoprotein.46,47 EGR1 is a tumor suppressor gene targeted by SS18/SSX through its interaction with ATF2 (a close homolog of ATF1), with repression mediated by additional interactions with histone deacetylases (HDACs) and Polycomb proteins.46 Though EWSR1/ATF1 has to date only been shown to mediate transcriptional activation, several studies have described similar fusion transcription factors and associated complexes that are able to activate or repress transcription depending on the targeted gene and recruited co-factors. One compelling example is found in acute myelogenous leukemia, which expresses the fusion oncoprotein AML1/ETO.48 Similar to SS18/SSX in synovial sarcoma, AML1/ETO complexes with a DNA-binding protein to localize to gene promoters and mediate transcriptional repression via co-repressors and HDACs. However, as in clear cell sarcoma, the fusion oncoprotein in acute myelogenous leukemia also binds CBP/p300, a transcriptional activator, indicating gene activation. Although an analogous mechanism has yet to be proven in clear cell sarcoma, some CRE-containing targets genes such as the dual specificity phosphatase 1 are known to be downregulated in clear cell sarcoma.
Oncogenic Targets of EWSR1/ATF1
Gene expression profiling and IHC data has indeed identified abnormal up-regulation of multiple well-known oncogenes in clear cell sarcoma, such as those encoding the growth/survival-related receptor tyrosine kinases c-Kit, c-Met, and ERBB3, the anti-apoptotic regulator protein BCL-2, the neural crest-specific SOX10 transcription factor, and the melanocytic master transcription factor MITF.16,18,42,49,50 These genes have been implicated in a variety of cancers as they regulate a common set of cell signalling pathways contributing to uncontrolled growth, invasion, and angiogenesis.
Gene expression profiling allows researchers to observe the level of expression of a large number (1000+) of genes in biopsied or cultured tissue samples using DNA microarrays. The expression profiles from cancer tissue are compared to normal tissue from the same patient and/or the cancer’s original tissue type in order to see how the differences contribute to oncogenesis. This is a very appealing way to discover better diagnostic biomarkers (genetic marks of specific diseases) and apply personalized, targeted therapies.
Few if any of these oncogenes have been proven to be direct targets of EWSR1/ATF1, with the important exception of MITF. Indeed, the MITF promoter contains a CRE that has been confirmed to be bound by EWSR1/ATF1 and a nearby SRY site bound by its SOX10 cofactor, together resulting in greatly increased expression.18 Additionally, these authors showed that MITF expression is necessary for both melanocytic differentiation and clear cell sarcoma tumor cell survival in vitro. By disrupting EWSR1/ATF1 from binding DNA and activating transcription, they observed decreased expression of MITF target genes, decreased levels of tyrosinase activity (the rate-limiting step of melanin biosynthesis), and a drastic decline in clear cell sarcoma survival and proliferation. However, Li et al. found that a synthetic MITF promoter introduced into clear cell sarcoma or melanoma cells was not able to become activated by EWSR1/ATF1, possibly reflecting differences in the epigenetic or cofactor context.51
EWSR1/CREB1 and EWSR1/ATF1 Fusions in Other Tumors
CREB1 and ATF1 are 65% identical in sequence and retain similar exons in oncogenic fusions. Gene expression profiling shows that gastrointestinal variant clear cell sarcoma specimens, expressing the EWSR1/CREB1 fusion, do not express typical melanocytic differentiation genes, although they do express SOX10 at comparable levels to soft-tissue variant clear cell sarcoma control tissues.28
Angiomatoid fibrous histiocytoma, a soft tissue tumor with cellular morphology, gene expression profile, and clinical prognosis dramatically different from clear cell sarcoma has nevertheless been shown to carry EWSR1/ATF1 translocations.52 Angiomatoid fibrous histiocytoma has a very much better prognosis than clear cell sarcoma: while local recurrence occurs in up to 10% of patients, metastases are very rare.
Angiomatoid fibrous histiocytomas are similar to clear cell sarcoma in that they usually arise in distal extremities of young adults, but are histologically distinct storiform spindle cell neoplasms containing blood-filled cystic spaces surrounded by a lymphoid cuff, associated with subcutaneous tissues rather than tendons.52 Furthermore, they do not display significant expression of MITF, GP100, CDK, or MET by IHC, unlike clear cell sarcoma. Histology, IHC and clinical presentation should guide diagnosis towards angiomatoid fibrous histiocytoma even when the presence of a EWSR1/ATF1 or EWSR1/CREB translocation has been identified.
Interestingly, Antonescu et al. subsequently found that the majority of angiomatoid fibrous histiocytomas actually carry EWSR1/CREB1 rather than EWSR1/ATF1 fusions.53 Gene expression data show that angiomatoid fibrous histiocytoma does not express SOX10, unlike clear cell sarcoma. While clear cell sarcoma displays significant expression of melanoma-related genes, neither angiomatoid fibrous histiocytoma nor gastrointestinal clear cell sarcoma express these genes. It is possible that these two genetically similar tumors may be derived from different progenitor cells (non-neural crest) than clear cell sarcoma.
Current and Future Treatment of Clear Cell sarcoma
The present treatment of clear cell sarcoma is limited, in many centres, to wide surgical excision and radiation therapy. Only a small percentage of clear cell sarcoma tumors have at best shown partial response or stable disease following conventional cytotoxic chemotherapy regimens.6 However, the upregulated oncogenes c-Kit, c-Met and ERBB3 identified in clear cell sarcoma suggest that receptor tyrosine kinase inhibitor drugs such as sunitinib, crizotinib, and EGFR inhibitors may be worth investigating.
Chemotherapy Versus Targeted Therapy
It is important to note the difference between conventional chemotherapy and targeted therapy. Chemotherapy drugs affect actively dividing cells and thus do not target cancer cells specifically, although cancer cells actively divide at a much greater rate than most non-malignant cells. Thus, chemotherapies work best on fast growing cancers, but have side effects on actively dividing non-malignant cells, such as blood and immune cells (leading to anemia and immunosuppression), and cells of the hair follicle (leading to alopecia). Targeted therapies are different in that they are designed to act only against specific molecular targets that drive tumorigenesis to reverse their specific oncogenic effects. Such therapies are in theory an improvement over chemotherapy since they are expected to stop and reverse the exact molecular events causing the cancer (eg. constitutively active signaling protein), not simply the result of these events (eg. uncontrolled growth and division).
Receptor Tyrosine Kinase Inhibitors
Receptor tyrosine kinases transmit proliferative signals to the nucleus, which are usually turned on too strongly in cancer cells, causing them to grow and divide much faster. Receptor tyrosine kinase inhibitors suppress growth of cancer cells that are addicted to over-activated proliferative signaling by receptor tyrosine kinases.
Sunitinib, a multi-kinase inhibitor of platelet-derived growth factor receptor (PDGFR), vascular endothelial growth factor receptor (VEGFR), and c-Kit, all receptor tyrosine kinases that contribute to cancer cell proliferation, is FDA approved for the treatment of renal cell carcinoma and gastrointestinal stromal tumor,54 and is still in phase II clinical trials for treatment of a broad range of other solid tumors.55,56 Only one patient with multiple clear cell sarcoma tumors has been documented to have had a partial response (decreased size of most lesions and decreased tumor density in one lesion), following two months of sunitinib treatment.57
In 2012, the FDA approved the PDGFR/VEGFR/c-Kit inhibitor pazopanib for the treatment of soft tissue sarcomas, which has been successful in achieving cell line and xenograft growth inhibition in some models of soft tissue sarcoma as well as partial responses and tolerable toxicities in phase II clinical trials.58–60
Perhaps the most promising receptor tyrosine kinase inhibitors for clear cell sarcoma are those targeting MET, a downstream target of MITF that is upregulated in clear cell sarcoma. One phase II clinical trial using the MET inhibitor ARQ 197 on clear cell sarcoma has completed,61 and a very recent (January 2012) cross-tumoral phase II clinical trial including clear cell sarcoma has begun with the use of the MET inhibitor crizotinib.62
Small Molecule Inhibitors
An attractive alternative to targeting downstream oncoproteins is to develop drugs that target co-activator interactions of EWSR1/ATF1, such as those with SOX10 and CBP.18,41,50,51 There are currently a number of small molecule drugs under investigation that disrupt the interaction between CBP and CREB1, but whether this would also prevent EWSR1/ATF1 or EWSR1/CREB1 from binding CBP is unknown.63,64 Other inhibitors that disrupt CBP/p300 acetyltransferase activity, such as Lys-CoA-Tat and C646, are currently being tested. Wang et al. have shown Lys-CoA-Tat and C646 significantly inhibit growth of acute myelogenous leukemia cell lines as well as reduce the survival of tumor cells transplanted in vivo in mice after drug treatment(48 While such CBP/p300 inhibitors capable of disrupting the oncogenic effects of EWSR1/ATF1 gene activation could also be effective therapeutics in clear cell sarcoma and similar soft tissue sarcomas, no such studies have been published. There are no small molecule drugs that specifically inhibit SOX10 enhancer activity, which in theory could inhibit SOX10 from activating a number of neural crest-specific oncogenic pathways and thus inhibit cancer progression.
In synovial sarcoma, Su et al. have characterized how SS18/SSX drives oncogenesis and have shown that HDAC inhibitors target and reverse the effects of the SS18/SSX oncoprotein, inducing cell death in synovial sarcoma cells in vitro.47 Interestingly, HDAC inhibitors have shown similar or greater effectiveness on clear cell sarcoma cells in vitro, a level of apparent sensitivity greater than that of other tested sarcomas, non-malignant mesenchymal cells, hematopoietic and epithelial malignancies, implying that the fusion oncoprotein in clear cell sarcoma may act via HDAC-mediated epigenetic programming of targeted genes (as in synovial sarcoma).65 HDAC inhibitors have been approved for use in cutaneous T-cell lymphoma, and have demonstrated relatively low toxicity in clinical trials.66–68
What is an HDAC?
Histone Deacetylases (HDACs) are enzymes that remove a chemical structure called an acetyl group from histones. Histones are the main structural proteins of chromatin around which the double helical strands of DNA are wound to compact the genetic code into the nucleus of the cell. The change in acetylation of histones by HDACs is one way that the structure of chromatin may be altered to promote or prevent gene transcription. HDACs have the power drive oncogenesis by removing acetyl groups to shut off tumor suppressor genes.
The HDAC inhibitors MS-275 and romidepsin were both found to cause suppression of EWSR1/ATF1 and its target MITF in three clear cell sarcoma cell lines,65 and others have shown that the HDAC inhibitors sodium butyrate, trichostatin A, and suberoylanalide hydroxamic acid (SAHA; vorinostat) also potently suppress MITF.72 These studies among others provided scientific support for clinical trials testing HDAC inhibitors in sarcomas.
Mechanistic explanations for the particular susceptibility of clear cell sarcoma to HDAC inhibition are currently being investigated. In acute myelogenous leukemia, the AML1/ETO fusion oncoprotein has been shown to repress many targets, but also activate a subset of oncogenes by recruiting the transcriptional co-activator CBP.48 Thus, this fusion oncoprotein exhibits completely opposite transcriptional effects depending on the promoter and proteins it binds. Similarly, the protein complex SP1, generally known as an activator of transcription, has been also found to recruit repressors to some promoters with the aid of HDACs.69 Thus, though EWSR1/ATF1 is generally considered an activator of transcription, the fusion oncoprotein in clear cell sarcoma may actually be involved in repression of critical tumor suppressor genes through alternative binding of transcriptional co-repressors on such promoters; HDAC inhibitors might aid in reactivating these tumor suppressor genes. HDAC inhibitors may alternatively work well in clear cell sarcoma by reactivating regulators of EWSR1/ATF1 itself, secondarily counteracting its oncogenic effects. A third possibility is that a critical function of EWSR1/ATF1 is activation of a repressor of downstream tumor suppressor genes, with this repressor’s action reversed by HDAC inhibitors. This arrangement would be similar to EWSR1/FLI1-mediated up-regulation of the NKX2.2 repressor in Ewing family tumors,70 whereby the net effect of the EWSR1/FLI1 oncoprotein, mediated by NKX2.2 and HDACs, is to repress critical differentiation or tumor suppressor genes that are supposed to be active. Other speculative possibilities exist based on the complexity of chromatin and epigenetic control by histone- and DNA-modifying proteins. Ultimately, detailed molecular experiments should be able to distinguish among these possibilities, explaining the mechanism of HDAC inhibitor action and informing possible improvements in therapeutic strategies.
Clear cell sarcoma is a highly malignant tumor of young adults with poor long term survival rates, in large part due to its lack of response to current chemotherapies. Diagnosis is primarily based on biopsy histology supported by IHC staining, often requiring molecular tests for EWSR1/ATF1 or its variants to rule out the more common differential diagnosis of malignant melanoma. Direct inhibitors of this fusion oncoprotein are not yet available, but molecular insights into the biology of clear cell sarcoma and related translocation-associated sarcomas may lead to targeted therapies, some of which are already under evaluation.