Background
Clear cell sarcoma of soft tissue (CCSST) is a malignant mesenchymal tumor that mostly affects young adults and tends to affect the lower extremities, close to the tendon and aponeuroses [
1]. Histologically, CCSSTs have epithelioid tumor nests accompanied by some spindling areas, and wreath-like multinucleated giant cells. CCSSTs present with a melanocytic differentiation and often express melanocytic markers including S-100 protein, melanoma antigen (Melan-A), human melanoma black 45 (HMB45), microphthalmia-associated transcription factor (MITF), and SRY-Box 10 (SOX-10) on immunohistochemistry (IHC). Ultrastructurally, CCSST has premelanosomes in the cytoplasm of tumor cells and shares some characteristic features with malignant melanomas (MMs). MMs genetically have BRAF mutations, although CCSST lacks this mutation. Clear cell sarcoma-like gastrointestinal tumor (CCSLGT) is also a malignant mesenchymal tumor that shares some pathological features with CCSST and arises from the gastrointestinal tract, such as the small and large intestine, and stomach. CCSLGT was originally reported to be an “osteoclast-rich tumor of the gastrointestinal tract with features resembling clear cell sarcoma of the soft parts” [
2] and the first case of CCSLGT was reported by Alpers et al. [
3] as a “malignant neuroendocrine tumor of the jejunum with osteoclast-like giant cells” in 1985. Subsequently, the term CCSLGT was first used by Kosemehmetoglu et al. [
4] in their review, which included 13 CCSLGT cases. However, some authors have proposed using the term “malignant gastrointestinal neuroectodermal tumor,” because CCSLGTs lack melanocytic differentiation on IHC and ultrastructural examination and appear to have poorer prognosis [
5]. Although CCSLGT has a similar histology to CCSST in some respects, such as a clear cytoplasm and epithelioid cells, there are some differing characteristics. CCSLGT has a pseudo-papillary growth pattern and many osteoclast-type giant cells, and the tumor cells tend to be positive for S-100 protein but show less expression of melanocytic markers on IHC [
6]. Genetically, CCSST and CCSLGT usually have characteristic chimeric fusions of Ewing sarcoma breakpoint region 1 (
EWSR1) with cAMP response element-binding protein (
CREB) gene family members,
EWSR1-activating transcription factor 1 (
ATF1) and
EWSR1-CREB1, which were derived from each translocation of t(12;22)(q13;q12) and t(2;22)(q34;q12), respectively [
7‐
10].
EWSR1-ATF1 fusion is much more frequent than
EWSR1-CREB1 fusion, but
EWSR1-CREB1 fusion of CCSLGT is comparatively often observed.
In this study, we used fluorescence in situ hybridization (FISH) and reverse transcription polymerase chain reaction (RT-PCR) to perform genetic analyses of 22 cases of CCSSTs and CCSLGTs, and compared their different chimeric fusion types.
Methods
Case selection
The study protocol for the collection of tumor samples and clinical information were approved by the Institutional Review Board of Sapporo Medical University Hospital (Sapporo, Japan; No. 292–3012). We selected 22 cases of clear cell sarcoma (CCS) including 16 CCSST and 6 CCSLGT cases from the clinicopathological archive at the Department of Surgical Pathology, Sapporo Medical University Hospital. We reviewed all hematoxylin and eosin-stained sections and confirmed that each case fulfilled the histologic criteria of CCSST and CCSLGT.
Immunohistochemistry
We evaluated previously reported IHC findings of melanocytic markers, including S-100 protein, Melan-A, HMB45, and SOX-10, and assessed their positivity. We also performed additional IHC using representative sections from formalin-fixed and paraffin-embedded tissues in some cases. These tissues were sliced into 3-mm-thick sections and examined with an automated IHC system at Sapporo Medical University Hospital. All slides were loaded into a PT Link module (Agilent Technologies, Santa Clara, CA) and subjected to a heat-induced antigen-retrieval protocol with EnVision FLEX Target Retrieval Solution (Agilent Technologies) before being transferred to the Autostainer Link 48 instrument (Agilent Technologies) and Dako Omnis (Agilent Technologies). We used antibodies against the following antigens: S-100 protein (polyclonal; Agilent Technologies), Melan-A (A103; Agilent Technologies), HMB45 (HMB45; Agilent Technologies), and SOX-10 (N-20; Santa Cruz Biotechnology, Santa Cruz, CA).
Fluorescence in situ hybridization
We performed FISH using the specimens obtained from tumor materials and 4 μm slices on glass slides. First, we selected an area showing representative histology and marked a 5-mm-diameter circle with a marker on the glass slides for FISH analyses. We performed FISH using dual color break apart probe for EWSR1 (Abbott, Abbott Park, IL), ATF1 (Empire Genomics, Buffalo, NY), CREB1 (Empire Genomics), and CREM (Empire Genomics). FISH was conducted as previously described [
11], with the following modifications: baking (1 h at 60 °C), deparaffinization, target gene activation (20 min with 0.2 M HCl followed by 30 min with pretreatment solution at 80 °C), enzyme treatment (60 min with protease solution at 37 °C), re-fixation (10 min in 10% formalin neutral buffer solution), denaturation (5 min with denaturation solution at 72 °C), washing and dehydration (1 min each in 70%, 85%, and 100% ethanol), hybridization with 10 mL DNA probe solution (5 min at 90 °C followed by 48 h at 37 °C), and washing with post-hybridization wash buffer (2 min at 72 °C). As a counterstain, 10 μL 4,6-diamidino-2-phenylindole was added. Slides were coverslipped for viewing under a fluorescence microscope.
To detect the presence of
EWSR1,
ATF1, CREB1, and
CREM rearrangements, we counted 50 nuclei in tumor cells that showed a pair of fused and split signals, and calculated the percentage of split signals. The signals were considered split when the distance between the red and green signals was at least twice the estimated signal diameter. We did not evaluate any truncated and overlapping cells in FISH analysis. We considered the specimen to be “split positive” if split signals were observed in more than 10% of tumor cells [
12].
Reverse transcription-polymerase chain reaction
We detected chimeric fusions by RT-PCR using fresh tumor samples in several available cases. RT-PCR analysis was performed for
EWSR1-ATF1 and
EWSR1-CREB1 fusions. For RT-PCR detection of the
EWSR1-ATF1 and
EWSR1-CREB1 fusions, we used the forward primer EWSex7-F1 with either the CREB1ex7-RevA primer (binds both
CREB1 and
ATF1; sequence: TCCATCAGTGGTCTGTGCATACTG) or the CREB1ex7-RevC primer (specific for
CREB1; sequence: GTACCCCATCGGTACCATTGT) [
1,
7,
13].
Discussion
Although CCSST and CCSLGT share similar pathological findings, there are apparent morphological and immunohistochemical differences between the two tumor types. We confirmed the differences in histology and IHC results in our cohort cases. Histologically, the cytological findings of tumor cells and architectural proliferation pattern differed. The tumor cells of CCSST were polyhedral to epithelioid, and spindle-shaped with round to mildly irregular-shaped nuclei and conspicuous nucleoli. In contrast, the tumor cells of CCSLGT had epithelioid tumor cells with irregular-shaped nuclei showing a coarse chromatin pattern and more eosinophilic cytoplasm. Nucleoli were not remarkable in CCSLGT compared to CCSST. CCSST exhibited sheet-like, solid, and nested tumor cell proliferation. In contrast, CCSLGT additionally showed a pseudo-papillary growth pattern. The existence of scattered osteoclast-type giant cells was also characteristic of CCSLGT. CCSSTs were positive for Melan-A and/or HMB45 melanocytic markers in addition to S-100 protein, as determined by IHC. On the other hand, CCSLGTs were not reactive for any melanocytic markers, with the exception of S-100 protein and SOX-10. As in previously reported studies, EWSR1-CREB1 fusion tended to be detected in CCSLGTs. This genetic tendency might reflect the morphological and immunohistochemical differences between the two tumor types.
The novel finding of the study was that we discovered
CREM rearrangement in a few CCSs. Kao et al. [
14‐
16] stated that an
EWSR1-CREM fusion was previously detected by RNA sequencing in 2 melanoma cell lines (CHL-1 and COLO 699) and proposed that these cell lines may have originated from CCS because of the histological and immunohistochemical overlap between malignant melanoma and CCS. On the other hand,
EWSR1-CREM fusion was found in a unique myxoid mesenchymal tumor that was recently described as a new entity [
14,
17]. This myxoid tumor is thought to have an intracranial location, and 8 cases have previously been reported, of which 7 occurred in intracranial lesions like meninges, brain tumors, and ventricles, and one case arose in the pelvic/perirectal region. A genetic study revealed
EWSR1 fusions with
CREB family genes in all of these tumors. Among the 8 tumors, 3 had
EWSR1-CREM fusion, 4 had
EWSR1-CREB1 fusion, and one tumor showed
EWSR1-ATF1 fusion. However, histological and immunohistochemical findings completely differed between this particular myxoid mesenchymal tumor and CCSST/CCSLGT, and interestingly, these genetic results corresponded to those of CCSST and CCSLGT.
CCSLGT was originally reported as an “osteoclast-rich tumor of the gastrointestinal tract with features resembling clear cell sarcoma of the soft parts” [
2]. However, some authors prefer to refer to CCSLGT as a “malignant gastrointestinal neuroectodermal tumor” (GNET), because these tumors lack evidence of melanocytic differentiation [
5]. A recent review discussed the relationship between clear cell sarcoma of the gastrointestinal tract (CCS-GIT) with GNET. There were differences in morphology and IHC findings between CCS-GIT and GNET. GNET tended to show a wider spectrum of growth patterns, including a pseudo-papillary growth pattern, and exhibited no evidence of melanocytic differentiation. Clinically, CCS-GIT affected males more often than females, and GNET occurred in younger patients although no significant differences existed in their biological behaviors [
18]. It has been reported that GNET has poorer prognosis than CCS-GIT [
5], but additional studies are needed to clarify the relationship between the two entities.
While a previous study revealed that
EWSR1-ATF1 fusion was identified by RT-PCR in 91% of CCSST cases [
1], we detected
EWSR1-ATF1 fusion in 13 of 16 CCSST (81.3%) both by FISH and RT-PCR. Moreover, after excluding 1 case of
EWSR1-CREM fusion, 13 of 15 CCSST cases (86.7%) had
EWSR1-ATF1 fusion. The percentage of positive cases in the present study nearly reached that of the previous study. CCSLGT has been identified as having
EWSR1-ATF1 or
EWSR1-CREB1 fusion, with CCSLGT more frequently having
EWSR1-CREB1 fusion. These fusions have also been detected in angiomatoid fibrous histiocytoma and primary pulmonary myxoid sarcoma despite the morphological and immunohistochemical differences between CCSST/CCSLGT and these tumors. To detect specific fusions, FISH is an effective and useful tool using formalin-fixed, paraffin-embedded sections in routine pathological work. In this study, we successfully detected
EWSR1,
ATF1,
CREB1, and
CREM rearrangements by FISH in the majority of cases. Some cases showed only one specific rearrangement and did not reveal any rearrangement of partner genes. In such cases, it is expected that certain unknown partner genes can form some novel chimeric genes, and that powerful analytic tools such as next-generation sequencing will be useful for detecting these novel fusions.
Acknowledgements
The authors thank the following pathologists for kindly contributing case materials and clinical follow-up information: Akiko Tonooka, Department of Pathology, Tokyo Metropolitan Cancer and Infectious Diseases Center Komagome Hospital, Tokyo, Japan; Misa Ishihara and Kimio Hashimoto, Department of Pathology, Nishi-Kobe Medical Center, Hyogo, Japan; Shigeo Hara, Department of Diagnostic Pathology, Kobe City Medical Center General Hospital, Hyogo, Japan; Koki Moriyoshi, Division of Clinical Pathology, National Hospital Organization Kyoto Medical Center, Kyoto, Japan; Shin Ichihara, Department of Surgical Pathology, Sapporo Kosei General Hospital, Hokkaido, Japan; Yukio Morishita, Department of Pathology, Tokyo Medical University Ibaraki Medical Center, Ibaraki, Japan; and Atsushi Uchida, Department of Pathology, Tsukuba Medical Center Hospital, Ibaraki, Japan.