Novel characteristics for immunophenotype, FISH pattern and molecular cytogenetics in synovial sarcoma

As a rare and highly aggressive soft tissue sarcoma, the new immunophenotype, atypical FISH pattern and relevant molecular cytogenetics of synovial sarcoma (SS) remain less known, although it is characteristically represented by a pathognomonic chromosomal translocation t (X; 18) (p11.2; q11.2). Methodologically, the morphology was retrospectively analysed by using H&E staining, and immunohistochemical features were investigated by using markers that have been recently applied in other soft tissue tumors. Moreover, FISH signals for SS18 and EWSR-1 break-apart probes were examined. Finally, cytogenetic characteristics were analysed via RT-PCR and Sanger sequencing. Consequently, nine out of thirteen cases that were histologically highly suspected as SS were finally identified as SS via molecular analysis. Histologically, nine SS cases were divided into monophasic fibrous SS (4/9), biphasic SS (4/9) and poorly differentiated SS (1/9). Immunohistochemically, SOX-2 immunostaining was positive in eight cases (8/9) and PAX-7 immunostaining was diffusely positive in the epithelial component of biphasic SS (4/4). Nine cases showed negative immunostaining for NKX3.1 and reduced or absent immunostaining for INI-1. Eight cases showed typically positive FISH signalling for the SS18 break-apart probe, whereas one case exhibited an atypical FISH pattern (complete loss of green signalling, case 2). Furthermore, the SS18-SSX1 and SS18-SSX2 fusion genes were identified in seven cases and two cases, respectively. The fusion site in 8 out of 9 cases was common in the literature, whereas the fusion site in case 2 was involved in exon 10 codon 404 in SS18 and exon 7 codon 119 in SSX1 (which has not been previously reported), which notably corresponded to the complete loss of green signalling in the FISH pattern. Additionally, FISH analysis of the EWSR-1 gene in nine SS cases demonstrated aberrant signalling in three cases that were recognized as a monoallelic loss of EWSR-1 (1/9), an amplification of EWSR-1 (1/9) and a translocation of EWSR-1 (1/9). In conclusion, SS18-SSX fusion gene sequencing is obligatory for a precise diagnosis of SS when dealing with a confusing immunophenotype and atypical or aberrant FISH signalling for SS18 and EWSR-1 detection.

Histological evaluation. The 3-μm tissue sections from the formalin-fixed and paraffin-embedded (FFPE) tumor specimens were stained by using routine haematoxylin and eosin (H&E). The H&E-stained slides were re-reviewed by three surgical pathologists (LLZ, GXH and FT) under a multi-head microscope and preliminary diagnoses were confirmed. Moreover, the histological subtypes of synovial sarcoma were divided into monophasic, biphasic and poorly differentiated types based on the criteria of Soft Tissue and Bone Tumors, WHO Classification of Tumors, 5th Edition 2 . Tumor grade was determined by using the Fédération Nationale des Centres de Lutte contre le Cancer (FNCLCC) grading system 13 . Immunohistochemistry. The 3-μm whole tissue sections from FFPE tumor specimens with both negative and positive controls were automatically retrieved and immunostained using the EnVision technique in the Ventana BenchMark XT instrument (Ventana Medical Systems, Tucson, AZ, USA), followed by a light haematoxylin counterstain. For the indicated antibodies, the procedure of antigen heat retrieval was 100 °C for 30 min by using the EDTA buffer (pH = 8.4). The procedure of primary antibody incubation was 37 °C for 1 h. Information on commercially available antibodies against cytokeratin (AE1/AE3), EMA, CD99, Bcl-2, TLE1, NKX3.1, SOX2, PAX-7 and INI-1 was listed in Table 1. In principle immunohistochemical staining was independently evaluated by two investigators (LLZ and GXH). In the face of inconsistency between investigators, all investigators in this project would be requested to evaluate and discuss until a consensus was reached. Furthermore, the intensity of staining was scored as "-" (negative), "1+" (weak positive), "2+" (moderate positive) or "3+" (strong positive). The extent of staining was evaluated as focal (< 10% of tumor cells), diffuse (> 75% of tumor cells) or specific percentage (approximately 10-75% of tumor cells).

Fluorescence in situ hybridization (FISH). Commercially available SS18 and EWSR-1 Break Apart
Rearrangement Probes (cat#F.01083 and cat#F.01194) were purchased from Guangzhou Lbp Medicine Science & Technology Co. (Guangzhou, Guangdong, China). For the SS18 break-apart probe, one end of the probe was labelled with the red spectrum (telomeric, 5′ to SS18, 649 kb), and the other end was labelled with the green spectrum (centromeric, 3′ to SS18, 925 kb); additionally, the probes were separated by a gap of 131 kb within the SS18 gene. For the EWSR-1 break apart probe, one end of the probe was labelled with the green spectrum (telo- www.nature.com/scientificreports/ meric, 5′ to EWSR-1, 826 kb), and the other end was labelled with the red spectrum (centromeric, 3′ to EWSR-1, 439 kb); additionally, the probes were separated by a gap of 152 kb within the EWSR-1 gene. FISH analysis was performed according to the manufacturer's protocol. FISH assays were performed on 3-μm-thick FFPE tissue sections. The sections were deparaffinized in xylene twice for 30 min and dehydrated in 100% ethanol twice for 5 min. Additionally, the tissue sections were pretreated with high temperature and high pressure, after which they were digested with pepsin solution (4.0 mol/ml). After washing, tissue sections were denatured at 85 °C and hybridized with the indicated probe overnight at 37 °C in a humidified chamber. The slides were then washed with 2 × SSC at 37 °C for 10 min and again washed with 0.1% NP40/2 × SSC at room temperature for 5 min, followed by dehydrated using graded ethanol. Tissue slides were counterstained with 2.0 μg/ml DAPI. Fifty nonoverlapping tumor nuclei (which were clearly identified and contained unequivocal signals) were counted for each case. A split signal was considered positive for gene rearrangement if the distance between the green and red signals was greater than 2 signal diameters. Moreover, according to our laboratory practice experience and previous studies 14, 15 , a case was considered positive for gene rearrangement when at least 15% of the tumor cells exhibited split-apart signals. In principle counting of the cases was independently performed by two investigators (LLZ and ZCW). Furthermore, the way to resolve the inconsistency between investigators was the same as that under immunohistochemical evaluation.
Reverse transcription polymerase chain reaction (RT-PCR). All tumor specimens with enough material were analysed for the presence of the fusion genes SS18-SSX1, SS18-SSX2 or SS18-SSX4 by using RT-PCR. Total RNA was isolated from two 10-μm tissue sections of FFPE specimen blocks by using the RNeasy FFPE Kit (Qiagen, Valencia, CA) according to the manufacturer's instructions. To eliminate the contamination of genomic DNA, RNA samples were treated with DNase I. RNA reverse-transcription into cDNA was performed by using the Thermo RevertAid™ First Strand cDNA Synthesis Kit (Thermo Fisher Scientific Inc.) for 1 h at 42 °C and 5 min at 70 °C by using SSX-1/2-B reverse primer: 5′-cattttgtgggccagatgc-3′, as has been previously described 10 . Moreover, PCRs were performed for 35 cycles by using a QIAGEN Multiplex PCR Kit (Qiagen, Valencia, CA) with the following cycle conditions: denaturation at 94 °C for 30 s, annealing at 58 °C for 90 s and extension at 72 °C for 90 s. The primers were used in the following combinations: SS18-SSX consensus forwards primer: 5′-agaccaacacagcctggaccac-3′; SS18-SSX1-specific reverse primer: 5′-acactcccttcgaatcattttcg-3′; SS18-SSX2-specific reverse primer: 5′-gcacttcctccgaatcatttc-3′; and SS18-SSX4-specific reverse primer: 5′-gcacttccttcaaaccattttct-3′. cDNA from angiomatoid fibrous histiocytoma tissue was used as the negative control. Furthermore, glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as the reference gene, and human GAPDH primers were obtained from Sangon Biotech Shanghai Co. Ltd. (Cat no. B661104). The products of the classic fusion gene and human GAPDH reference gene were 111 bp and 138 bp, respectively. The RT-PCR products were further fractionated on 3.0% agarose gels and visualized via GoldView staining and ultraviolet illumination.
Sequence analysis. PCR products were subjected to Sanger sequencing by Sangon Biotech Shanghai Co.
Ltd. To accurately sequence the whole fragments, all of the PCR products were purified and then subjected to TA cloning.

Results
Clinical characteristics. The clinical information of the 9 patients is summarized in Table 2. In this case series, the ages of the patients ranged from 11 to 65 years old (median 45 years), and males were slightly more predominant than females (6:3). The maximum tumor diameter ranged from 1.2 to 8 cm (median 5 cm). Additionally, tumors were located in joints or near the joints in 6 cases and other uncommon sites in 3 cases, including the neck, abdomen and thigh. All of the patients received surgical tumorectomy, and 6 patients subsequently received adjuvant chemotherapy and/or radiotherapy. Follow-up information was available for 6 patients and the follow-up time ranged from 13 to 82 months (median 37 months). Lung metastasis occurred in 2 patients, and they died of disease within 23-35 months of initial diagnosis. Moreover, recurrence was identified at 24 months in 1 case; however, the patient was still alive with disease at 44 months. Three patients had no evidence of disease at 13, 39 and 82 months.
Histopathological features. On the basis of morphological features of synovial sarcoma, we classified the tumor as monophasic fibrous synovial sarcoma (MFSS) (4/9, 44.5%), biphasic synovial sarcoma (BSS) (4/9, 44.5%) and poorly differentiated synovial sarcoma (PDSS) (1/9, 11%). The histological features of the nine cases are summarized in Table 3. MFSS is commonly composed of uniform and delicate spindle tumor cells with fascicle or dense sheet patterns. Moreover, spindle tumor cells often have sparse cytoplasm and hyperchromatic nuclei with inconspicuous nucleoli. Variable amounts of hyalinized collagen were observed in the tumor stroma ( Fig. 1a). BSS was composed of mixed spindle and epithelial tumor cells (Figs. 2a, 3a). The spindle components resembled those of MFSS. Furthermore, the epithelial tumor cells were often arranged in glandular, tubular, nest or cord patterns, with occasional alveolar or papillary architecture. In the glandular area, the epithelial tumor cells were cuboidal and had ovoid nuclei and pale eosinophilic cytoplasm with intraluminal secretions (Fig. 2a).
In the solid cord area, the epithelial tumor cells exhibited clear cytoplasm and a vague transition to spindle cells  www.nature.com/scientificreports/ (Fig. 3a). PDSS was consistently composed of large epithelial cells with staghorn-shaped vessels. Furthermore, the epithelial cells had round or ovoid vesicular nuclei and were arranged in a sheet pattern. Mitotic figures were brisk in PDSS (Fig. 4a).
Unlike SOX-2, PAX-7 expression was more frequent and extensive in the epithelial component than in the spindle component of BSS (4/4) (Figs. 1d, 2d). Interestingly   Tables 4 and 5. By using the SS18 break-apart probe for FISH detection of these cases, we observed the classical red and green break-apart signal and fusion signal (1F/1R/1G), which demonstrated the SS18 gene arrangement in 8 out of 9 cases (Fig. 5a). Notably, 83% of tumor cells showed one fusion signal and one red signal (1F/1R) accompanied by the complete loss of the green signal in case 2 (Fig. 5b). The complete loss of green signal was unusual and was classified as representing an atypical FISH signal pattern, thus making it difficult for us to identify the SS18 gene arrangement in case 2. For the detection of EWSR-1 gene arrangement, no gene alteration was observed in 6 out of 9 cases (Fig. 6a). Unexpectedly, EWSR-1 gene monoallelic loss, EWSR-1 translocation and EWSR-1 amplification were observed in case 4, case 6 and case 8, respectively (Fig. 6b-d). These data indicate that EWSR-1 gene alteration is occasionally accompanied by SS18 gene arrangement in SS. Therefore, the simultaneous detection of the SS18 and EWSR-1 genes seems essential for the differential diagnosis of SS.

RT-PCR and sequencing. We detected the SS18-SSX fusion gene in all cases, including the SS18-SSX1
fusion gene in 7 cases and the SS18-SSX2 fusion gene in 2 cases (Table 4). Subsequent cDNA sequencing demonstrated that the gene fusion site in 8 out of 9 cases involved exon 10 codon 410 in SS18 and exon 6 codon 111 in  www.nature.com/scientificreports/ SSX1 or SSX2, which is the typical fusion site (as has been previously reported) 14 . Notably, the SS18-SSX1 fusion gene was unequivocally detected in case 2, which showed a complete loss of the green FISH signal (Fig. 7a). Furthermore, cDNA sequencing demonstrated that the gene fusion site involved exon 10 codon 404 in SS18 and exon 7 codon 119 in SSX1 in case 2 (Fig. 7b). The product length of the fusion gene was 92 bp in case 2, which was obviously shorter than that in the other SS cases (Table 4). Furthermore, the missing fragment in case 2 may be involved in the binding sites of the FISH probe labelled with green signal, explaining the possible reason for the complete loss of the green FISH signal. Consequently, a novel SS18-SSX1 fusion site that was not previously reported was identified in case 2.

Discussion
In this study, we first analysed the clinicopathological features and morphology subtypes in nine SS cases. The details of the clinical information are listed in Table 2. Histologically, 4 cases were classified as MFSS with uniform and delicate spindle tumor cells, as well as fascicle or dense sheet patterns, and 4 cases were identified as BSS with mixed spindle and epithelial tumor cells, as well as an occasionally vague borderline between spindle and epithelial tumor cells. Moreover, PDSS with consistently large epithelial tumor cells was identified in case 3. SOX-2 is a transcription factor that is essential for maintaining embryonic and neural stem cells and has been documented to be a marker for cancer stem cells in various cancer types, such as squamous cell carcinoma, pancreatic cancer, breast cancer, glioblastoma, colorectal cancer and prostate cancer 20,21 . Although SOX-2 has been reported to be expressed in 58% of SS cases, the relationship between histological subtype and SOX-2 expression remains unknown 16 . PAX-7 is transcriptionally required for the specific development of skeletal muscle stem cells and has been proven to be expressed in rhabdomyosarcoma, Ewing sarcoma and PDSS; however, the expression Table 4. The molecular and cytogenetic features of SS18-SSX fusion gene in nine patients who were genetically diagnosed as SS. SS synovial sarcoma, FISH Fluorescence in situ hybridization, F fusion signaling, R red signaling, G green signaling, NCBI Reference Sequence: NM_001007559.3 (SS18 transcript), NM_001278691.2 (SSX1 transcript) and NM_003147.6 (SSX2 transcript).   6,7 . With respect to NKX3.1, which was implicated in EWSR1-NFATC2 sarcoma and mesenchymal chondrosarcoma, no expression of NKX3.1 was observed in our series of cases 18,19 .
A previous study demonstrated that partial loss of green signalling was observed as an atypical FISH for the SS18 break-apart probe 23 , whereas we observed a complete loss of green signalling in case 2. The novel atypical FISH pattern (case 2) was further verified as a SS18-SSX1 fusion gene by using RT-PCR. Subsequently, a novel gene fusion site involved exon 10 codon 404 in SS18 and exon 7 codon 119 in SSX1 was discovered in case 2. The other cases in our series showed a typical FISH pattern for the SS18 break-apart probe and common fusion gene site. The sequencing assay demonstrated that the fusion gene product length in case 2 was shorter than that in the other cases. We speculated that the missing product fragment included the site to which the green probe may bind, thus resulting in the complete loss of the green signal in case 2. Due to the fact that the FISH probe information was confidential, we could not further determine the reason for the complete loss of the green signal.
To explore the accompanying gene change, EWSR-1 gene arrangement was selected for detection via the FISH assay. Unexpectedly, EWSR-1 gene monoallelic loss (1/9, case 4), translocation (1/9, case 6) and amplification (1/9, case 8) were discovered in our case series. Notably, the EWSR-1-translocated case possessing morphologic subtype that was BSS exhibited an ambiguous transition between epithelial and spindle tumor cells, which was different from other BSS cases. EMA immunostaining was negative in the EWSR-1-translocated case, but positive in the other eight cases. However, the details of EWSR-1 translocation still need to be further investigated via sequencing. Previous studies have also demonstrated EWSR-1 gene changes in SS, including monoallelic losses of EWSR-1, EWSR-1-NR4A3 and EWSR1-SSX1 gene fusion 15,[24][25][26] . Therefore, the misinterpretation of the change in the EWSR-1 gene may be a pitfall in diagnosing SS.

Conclusions
In conclusion, a panel of SOX-2, PAX-7, INI-1 and NKX3.1 immunohistochemical markers (combined with classical markers, such as TLE-1, CK-pan, EMA, CD99 and BCL-2) can be used as an ancillary tool for the differential diagnosis of SS. Due to the atypical FISH pattern for the SS18 break-apart probe and EWSR-1 gene change which occasionally occur in SS, SS18-SSX gene sequencing analysis was obligatory for a precise diagnosis of SS when dealing with the above mentioned situation. www.nature.com/scientificreports/

Data availability
The raw picture from DNA gel electrophoresis in case 2 was provided as supplementary information (Supplementary Fig. 1). The sequence data of SS18-SSX fusion site in each SS case were < 200 bp and not suitable for uploading the INSDC database, so the data were also provided as supplementary information (Supplementary Table 1). The datasets, not otherwise specified, used and/or analyzed during the current study available from the corresponding author on reasonable request.