Introduction

Ewing’s sarcoma family of tumours (ESFT) comprises a group of small round cell tumours (SRCT) mainly affecting the bone and soft tissues of children and young adults [12, 16–18, 20, 37]. ESFT are genetically defined by specific chromosomal translocations resulting in a fusion of the EWSR1 gene with a member of the ETS family of transcription factors: FLI1 (11q24), ERG (21q22), FEV (2q33), ETV1 (7p22) and ETV4 (17q12). In addition, other, less frequent, gene fusions have been described in ESFT, including FUS/ERG, FUS/FEV, EWSR1/POU5F1, EWSR1/PATZ1, EWSR1/SP3, EWSR1/NFATc2, BRD4/NUT and CIC/DUX4 [3, 7, 20, 24, 32, 35].

The histopathology fluctuates between undifferentiated tumours (Ewing’s sarcoma (ES)) and tumours with neuroectodermal differentiation/rosette formation (PNET) [12, 17, 18, 20, 32, 37]. Additionally, several histological varieties of ESFT are considered, including large cell ES, ES with epithelioid cells, clear cells, ES with spindle cells, vascular-like ES with haemangioendothelial features, ES with neuroectodermal features (isolated pseudorosettes), synovial sarcoma-like PNET, sclerosing PNET and adamantinoma-like ES [12, 17, 18, 20, 27, 32, 37]. In order to homogenize these ESFT subtypes, the tumours have been regrouped into three larger categories according to their predominant morphological criteria: conventional/classical/typical ES, PNET with neuroectodermal features and atypical ES comprising all subtypes distinct from the other two types [12, 20].

Atypical ESFT are the most challenging of this group of tumours, and the differential diagnosis with other SRCT of bone and soft tissue is difficult since these subtypes can resemble other neoplasms [12, 17, 18, 20, 21, 27]. The present article describes nine cases of genetically confirmed, atypical ESFT presenting unusual structural findings at morphological and immunohistochemical (IHC) levels associated with conventional or more uncommon clinical presentations. These characteristics may complicate the differential diagnosis with other sarcomas, carcinomas and lymphomas, thus creating problems in formulating the therapeutic approach.

Clinical data

Age, gender and tumour location, histopathology findings, immunohistochemical, differential diagnosis and genetic results are summarized in Table 1.

Table 1 Clinicopathologic and genetic characteristics of the present series
Full size table

Material and methods

The whole tumour specimen was fixed in buffered formalin, embedded in paraffin and processed for pathological investigation. Sections, 4 μm thick, were cut for haematoxylin/eosin (H/E), IHC and fluorescence in situ hybridization (FISH) studies. The antibodies employed, with their source and dilution, are summarized in Table 2. The immunostaining score was classified as non-informative (scant material, artefact and tumour necrosis), negative (mild in <5%), 1+ (mild in 5–10%), 2+ (moderate in 10–50%) and 3+ (intense in >50%). Positive nuclear staining was considered for Fli-1, myogenin and INI1 and cytoplasm/membranous staining was considered for CD99, CAV-1, HNK-1, osteonectin, osteocalcin, LCA, CD20, CD3, CD138, CK (AE1/AE3), CK20, EMA, SMA, desmin, synaptophysin, chromogranin, NB84, TH, HMB-45, Melan-a and CD117. EWSR1 (22q12) (Vysis) and LSI SYT (18q11.2) (Vysis) break apart probes were used for FISH analysis. FISH was performed according to the conventional protocol for paraffin-embedded tumours in whole sections [24, 31, 34]. In normal cells two fused or very close (red-green) signals were observed. In cells with translocations, two types of signals were observed: one fused (red-green) signal, and another signal split into green and red. An average of 200 well-separated nuclei was counted in each case. A positive result was defined as more than 15% of cells with translocation. For RT-PCR analysis, total RNA was extracted using conventional protocols [24, 26]. The primer sets (Table 3) were applied according to the incidence of typical gene fusions for specific SRCT.

Table 2 Immunohistochemical study. Antibodies, source and dilution
Full size table
Table 3 Primers used for RT-PCR
Full size table

Results

  1. Case 1

    The histopathology revealed the presence of mature cortical and trabecular bone tissue, as well as soft tissue infiltrated by a malignant neoplasia of small round cells. The cytoplasm was abundant, slightly eosinophilic and occasionally vacuolate. Cells were distributed in strands, nests and small sheets within a richly vascularized, fibrohyaline stroma, and with perivascular deposits of a mantle of fibrillar, hyaline eosinophilic material, interposed among the neoplastic cells corresponding to reactive osteoid formation probably secondary to bone destruction (Fig. 1a, b). IHC study revealed positivity for CD99+++, FLi-1+++, HNK-1++,CAV-1+++, osteonectin+ (Fig. 1c), CK+++, EMA++ and INI1++. The tumour was negative for other markers including S-100, HMB-45, CD45, CD20, SMA and desmin. FISH analysis revealed a positive EWSR1 break apart, and with RT-PCR an EWSR1/Fli-1 fusion was demonstrated. The final diagnosis was atypical ESFT mimicking small cell osteosarcoma.

    Fig. 1
    figure 1

    a Case 1, atypical ESFT with hyaline eosinophilic material, interposed among the neoplastic cells [haematoxylin and eosin (H&E ×40)]. b Case 1, atypical large cell ESFT (H&E ×40). c Case 1, osteonectin cytoplasmic staining (+++) in reactive bone and poor positivity in ESFT neoplastic cells. d Case 2, atypical ESFT with extensive desmoplastic stroma (H&E ×20). e Case 2, atypical ESFT with cells distributed in nests or strands embedded in a highly desmoplastic stroma (H&E ×40). f Case 2, caveolin-1 membranous staining (+++) in neoplastic cells. g Case 3, atypical ESFT with small cell tumour showing slightly eosinophilic cytoplasm (H&E ×40), h case 3, atypical ESFT associated with prominent spindle cell proliferation (H&E ×40). i Case 3, CD99 membranous staining (+++) in tumour cells

    Full size image
  2. Case 2

    Histopathologic studies revealed a proliferation of small round cell cells with scant eosinophilic cytoplasm and ill-defined cell borders, oval to round nuclei and moderate mitotic index. The cells were distributed in nests or strands embedded in a high desmoplastic stroma (Fig. 1d, e). Occasionally, pseudorosette structures were observed. The IHC showed strongly immunoreactive for CD99, Fli-1, CAV-1 (Fig. 1f) and CK (dot-like), moderately positive for HNK-1 and stained weakly for S100. EMA, desmin and SMA were negative. FISH analysis revealed EWSR1 break apart positive and RT-PCR detected EWSR1/FlI-1 gene fusion. The final diagnosis was atypical ESFT with desmoplasia, and the differential diagnosis was desmoplastic small round cell tumour (DSRCT).

  3. Case 3

    The histopathology showed the presence of a small cell tumour associated with prominent spindle cells proliferation (Fig. 1g, h). The cytoplasm was scant and slightly eosinophilic. Neither rosette formation nor glandular structures were detected. IHC study revealed positivity for CD99+++ (Fig. 1i), Fli-1+++, HNK-1++, CAV-1+++, CK+++ and EMA+. The tumour was negative for other markers including S-100, SMA and desmin. FISH analysis revealed a positive EWSR1 break apart and negative SYT translocation and the RT-PCR detected EWSR1/ERG gene fusion. The final diagnosis was atypical ESFT with spindle cell proliferation mimicking synovial sarcoma.

  4. Case 4

    The histopathology revealed the presence of soft tissue infiltrated by a malignant neoplasm of small round cells. The cytoplasm was prominent, eosinophilic with rhabdoid and rhabdomyoblast morphology (Fig. 2a, b). Cells were distributed in large sheets within a richly vascularized stroma. IHC study revealed positivity for CD99+++, Fli-1+++, HNK-1+++, CAV-1+++ (Fig. 2c), CK+++ and INI1++. The tumour was negative for EMA, SMA, desmin and myogenin. FISH analysis revealed a positive EWSR1 break apart and RT-PCR detected EWSR1/FLi-1 fusion, although PAX/FKHR fusion genes were not demonstrated. The final diagnosis was atypical ESFT with rhabdoid cells resembling adult pleomorphic rhabdomyosarcoma or malignant tumour with rhabdoid phenotype.

    Fig. 2
    figure 2

    a Case 4, atypical ESFT with prominent eosinophilic cytoplasm (H&E ×40). b Case 4, atypical ESFT, the cytoplasm showing rhabdoid and rhabdomyoblast morphology (H&E ×40). c Case 4, caveolin-1 cytoplasmic staining (+++) in tumour cells. d Case 5, atypical ESFT with prominent vascular formation (H&E ×40). e Case 5, atypical ESFT with haemosiderin deposition and vascular formation (H&E ×40). f Case 5, CD99 membranous staining (+++). g Case 6, atypical ESFT with undifferentiated morphology (H&E ×20). h Case 6, atypical ESFT, scant cytoplasm, undifferentiated morphology and vascular formation (H&E ×40). i Case 6, atypical ESFT, negative for NB84

    Full size image
  5. Case 5

    The neoplasm showed an SRCT morphology, but with prominent vascular formation adopting a haemangioendothelial pattern (Fig. 2d, e). The tumour cell cytoplasm was scant and focal spindle cell morphology was observed. Neither rosette nor pseudorosette formation was detected. IHC study revealed positivity for CD99 (+++) (Fig. 2f), Fli-1 (+++), CAV1 (+++), HNK-1 (+) and CK. EMA, desmin, myogenin, SMA, S100, chromogranin and synaptophysin were negative. FISH analysis revealed positive translocation (break apart) for EWSR1 gene and negative results for SYT. RT-PCR showed the EWSR1/FEV fusion transcript. No other gene fusions involved in specific SRCT (synovial sarcoma, extraskeletal myxoid chondrosarcoma, round/myxoid liposarcoma, DSRCT, clear cell sarcoma) were detected. The final diagnosis was atypical vascular ESFT resembling soft tissue tumour with vascular formation.

  6. Case 6

    Histopathologic studies revealed a proliferation of small round cell with scant cytoplasm and undifferentiated morphology, abundant apoptotic cells were observed and the mitotic index was low (Fig. 2g–i). Occasionally, pseudorosette structures were observed. The IHC showed positivity for CD99 (+++), FLI-1 (+++), CAV-1 (++), HNK-1 (+) and CK (+). EMA, desmin, SMA, NB84, TH, CD45, CD20, chromogranin and synaptophysin were negative. FISH analysis revealed EWSR1 break apart positive and no MYCN gene amplification. RT-PCR detected EWSR1/Fli-1 gene fusion. The final diagnosis was atypical ESFT (large cells) mimicking undifferentiated neuroblastoma.

  7. Case 7

    The neoplasm showed and SRCT morphology. The cytoplasm fluctuated from scant to abundant and clear or eosinophilic appearance was observed. Tumour cells were arranged in several patterns including diffuse SRCT, tubular, cord-like, clear cells, epithelioid and pseudo-glandular. In several areas the tumour cells adopted a palisading architecture, being sustained by a collagen basal-like membrane (Fig. 3a, b). This basement membrane was homogeneous, eosinophilic and PAS positive. IHC study revealed positivity for CD99 (+++) (Fig. 3c), Fli1(+++), CAV1 (+++) and HNK-1 (++). Poor cytoplasmic expression for EMA was observed. CK, desmin, SMA, S100 and myogenin were negative. FISH analysis revealed positive translocation (break apart) for EWSR1 gene and negative results for SYT. RT-PCR showed the EWSR1/FEV fusion transcript. No other gene fusions involved in specific SRCT (synovial sarcoma, extraskeletal myxoid chondrosarcoma, round/myxoid liposarcoma, DSRCT or clear cell sarcoma) were detected. The final diagnosis was atypical ESFT (adamantinoma-like) resembling soft tissue myoepithelial carcinoma.

    Fig. 3
    figure 3

    a Case 7, atypical ESFT with adamantinoma-like structures (H&E ×40). b Case 7, atypical ESFT, cord-like, clear cells, epithelioid and pseudo-glandular patterns. The tumour cells adopted a palisading architecture, being sustained by a collagen basal-like membrane (H&E ×40). c Case 7, CD99 membranous and cytoplasmic staining (+++) in tumour cells. d Case 8, atypical cutaneous ESFT with neoplastic cells restricted to the dermis (H&E ×20). e Case 8, superficial atypical ESFT (H&E ×40). f Case 8, CD99 membranous staining (+++) in tumour cells and negative in cutaneous epithelial structures. g Case 9, atypical ESFT with undifferentiated morphology, and scant eosinophilic and plasma cells (arrow; H&E ×40). h Case 9, atypical ESFT, incohesive cells and undifferentiated morphology mimicking lymphoblastic lymphoma (H&E ×40). i Case 9, atypical ESFT, FISH/EWSR1 with break apart (arrows), positive for EWSR1 translocation

    Full size image
  8. Case 8

    The round cell neoplasm was restricted to the dermis (Fig. 3d, e). The tumour was composed of sheets and lobules of small- to medium-sized round cells showing scant cytoplasm. Mitotic index was low and neither necrosis nor pseudorosette formation were observed. IHC study revealed positivity for CD99 (+++) (Fig. 3f), Fli1 (++), CAV1 (++) and HNK-1 (+). CK, CK20, EMA, desmin, SMA, S100 and myogenin were negative. FISH analysis revealed positive translocation for EWSR1 gene and RT-PCR showed the EWSR1/Fli-1 fusion transcript. No other gene fusions involved in specific SRCT (synovial sarcoma, extraskeletal myxoid chondrosarcoma, round/myxoid liposarcoma, DSRCT or clear cell sarcoma) were detected. The final diagnosis was atypical superficial ESFT mimicking Merkel cell carcinoma.

  9. Case 9

    Histopathologic studies revealed a proliferation of small round cell cells with very incohesive cells and no intervening stroma mimicking lymphoma neoplasm (Fig. 3g, h). The cells showed scant cytoplasm and the mitotic index was low. No pseudorosette structures were observed. The IHC showed positivity for CD99, FLI-1, CAV-1 and HNK-1. CK, EMA, desmin, CD45, CD20, CD3 and CD138 were negative. FISH analysis revealed EWSR1 break apart positivity (Fig. 3i) and RT-PCR detected an EWSR1/Fli-1 gene fusion. The final diagnosis was atypical ESFT resembling non-Hodgkin lymphoma (patient with previous history of non-Hodgkin lymphoma).

Discussion

Atypical variants of ESFT have been described mostly by Nascimiento et al. [27], Llombart-Bosch et al. [17, 18, 20] and Folpe et al. [12]. The criteria used for these histological subtypes include the high degree of cell heterogeneity, the existence of unusual patterns or the presence of pseudovascular structures [12, 20]. Cell heterogeneity of atypical ESFT is consistent with the presence of large, clear, spindle, epithelioid or rhabdoid cells in wide areas of the tumour [12, 20, 27]. The existence of unusual patterns is characterized by components such as desmoplasia, sclerosis and adamantinoma-like areas [6, 12, 13, 20, 27, 37]. Pseudovascular structures offer in atypical ESFT a vascular-like haemangiomatous appearance [20]. In these cases, the correlation between the clinical data (tumour location, age and sex), morphological, IHC and genetic findings is mandatory [12, 20]. Additionally, epithelial differentiation has been demonstrated in 20% of genetically confirmed ESFT [15]. Particularly, the atypical variant of ESFT frequently shows epithelial differentiation at histological and IHC level, further complicating the differential diagnosis with other SRCT of bone and soft tissue with CK immunoexpression (adamantinoma, synovial sarcoma, rhabdomyosarcoma, epithelioid sarcoma, DSRCT, malignant rhabdoid tumour (MRT) or tumour with rhabdoid phenotype, myoepithelial carcinoma, neuroendocrine carcinomas and metastatic carcinomas). Furthermore, many of these soft tissue tumours (DSRCT, myoepithelial carcinoma and neuroendocrine carcinoma) can display an EWSR1 rearrangement by fluorescence in situ hybridization which complicates the differential diagnosis with atypical ESFT [4, 5, 25, 29]. Therefore, the use of RT-PCR analysis to detect specific gene fusions in the latter cases is mandatory.

In the present series, we describe nine cases of SRCT with a genetically confirmed final diagnosis of atypical ESFT, resembling sarcomas, carcinomas and lymphomas due to the presence of unusual cell morphology or atypical patterns. Six cases (1–6) mimic sarcomas with or without osteoid formation. Age and tumour location in (case 1) were concordant with the diagnosis of ESFT; however, osteoid deposition and EMA IHC positivity were unexpected characteristics. The osteoid deposits complicated the differential diagnosis with small cell osteosarcoma (SCO); and the EMA positivity with undifferentiated synovial sarcoma. Molecular studies revealing EWSR1 translocation and EWSR1/Fli1 gene fusion, a finding inconsistent with synovial sarcoma, defined the diagnosis of atypical ESFT in case 1. It seems reasonable that in any atypical SRCT with osteoid deposition, CD99 positivity and EWSR1 rearrangement, the confirmation of the translocation should be carried out by two different molecular methods. On the other hand, ESFT can reveal osteoid deposition and IHC osteonectin positivity due to reactive new bone formation. Moreover, in cases with presumptive diagnosis of ESFT in bone, but negative for the specific translocations or gene fusions, the possibility of SCO should be excluded and several histopathological sections from the paraffin block are mandatory to exclude malignant osteoid formation [2, 10, 23]. Definitively, despite the morphological link between ESFT and SCO, malignant osteoid formation remains the hallmark of SCO, while the presence of an EWSR1 translocation with the specific gene fusions are diagnostic of ESFT [23]. Despite the CK and EMA positivity in the present tumour, the diagnosis of synovial sarcoma was excluded since molecular studies did not reveal SYT rearrangement. As far as we know, EMA expression has not so far been described in genetically confirmed ESFT, so this finding complicated the differential diagnosis with other SRCT with EMA expression (synovial sarcoma, MRT, epithelioid sarcoma and myoepithelial carcinoma). To our knowledge, loss of INI1 expression has not yet been reported in ESFT; however, this is a frequent event in MRT, and occasionally, epithelioid sarcoma, myoepithelial carcinoma, epithelioid malignant peripheral nerve sheath tumour and synovial sarcoma can reveal loss of INI1 expression.

Findings resembling DSRCT (desmoplastic stroma; case 2) are unusual in ESFT [12, 20, 37]. Nevertheless, an unusual subtype of ESFT with desmoplasia has been described by Folpe et al. [12]. In case 2, the retroperitoneal location, desmin negativity and EWSR1/Fli1 gene fusion by RT-PCR are more consistent with ESFT than with DSRCT. However, DSRCT with EWSR1/Fli1 or EWSR1/ERG gene fusion have already been described, but with evident IHC desmin expression [28]. Desmin expression is very rare in ESFT, nevertheless sporadic studies have reported this finding in which the tumours were classified as biphenotypic sarcomas with or without EWSR1 rearrangement [9, 33]. Indeed, in desmin-positive SRCT, the possibility of ESFT remains very unlikely and the possibility of other SRCT with muscular differentiation (rhabdomyosarcoma and SCO) should be exhaustively excluded. Furthermore, in a recent study performed by Llombart-Bosch et al. in more than 500 genetically confirmed ESFT, desmin expression was absent in all tumours (data unpublished). The term ‘biphenotypic sarcoma’ has been used for tumours with either neural and muscular differentiation, or neural and osteoid production, although the actual existence of these tumours is still under debate. In fact, in the post-molecular era only very sporadic reports regarding biphenotypic sarcoma have been published, and the existence of this particular neoplasm is becoming doubtful.

A sarcoma without osteoid production and with predominant spindle cells expressing CK made the differential diagnosis between atypical ES and undifferentiated synovial sarcoma difficult (case 3). In this case, tumour location and age are suggestive of both tumours; however, EMA negativity and the absence of SYT translocation and SYT/SSX transcript made the existence of synovial sarcoma improbable [9, 11, 19]. Additionally, the finding of EWSR1 translocation and EWSR1/ERG gene fusion confirms the diagnosis of ESFT, in this case atypical ESFT with spindle cells.

Tumour location (inguinal region), the morphologic characteristics of tumour cells (abundant eosinophilic cytoplasm) made the differential diagnosis between atypical ES and adult rhabdomyosarcoma difficult [11, 19, 30] (case 4). FISH analysis revealed EWSR1 translocation and the RT-PCR, EWSR1/Fli1 transcript confirmed the diagnosis of ESFT. In addition, desmin, myogenin and SMA were negative, making the possibility of adult rhabdomyosarcoma unlikely. Despite the age of the patient, the possibility of alveolar rhabdomyosarcoma was excluded since the RT-PCR failed to detect PAX-FKHR gene fusion.

In atypical ESFT, vascular-like areas have been reported similar to case 5 in which the tumour cells display a pseudoendothelial pattern forming vascular or lacunar spaces, lined by tumour cells that progressively elongate and cover lacunae filled with erythrocytes and plasma material [12, 18, 20]. Tumour cells, lining the surface of the vessel-like spaces lack basal membrane and are in continuity with the more solid cellular fields [12, 18, 20]. Abundant vascular formation with erythrocyte lysis can generate haemosiderin deposits, misinterpreted clinically as the melanic pigment observed in melanoma or clear cell sarcoma.

Tumour location (paravertebral area), patient age and absence of pseudorosette formation can make the differential diagnosis between neuroblastoma and atypical ESFT difficult (case 6). Children below 5 years of age with SRCT of bone and soft tissue should be carefully evaluated to exclude metastatic neuroblastoma [8, 31]. The present case showed IHC findings suggestive of ESFT instead of neuroblastoma, and additionally the EWSR1 translocation was demonstrated by FISH and RT-PCR.

Differential diagnosis of atypical ESFT with carcinomas is quite complex in some tumours (cases 7 and 8). The soft tissue tumour location and the trabecular, tubular and pseudo-glandular patterns at histological level made the differential diagnosis with adamantinoma-like EFST and soft tissue myoepithelial carcinoma confusing [5, 14] (case 7). Myoepithelial carcinoma displays epithelial and neural differentiation with CK and S100 expression [5, 14] and both markers were negative in case 7, making the diagnosis of myoepithelial carcinoma improbable. The differential diagnosis has become more complex since Brandal et al. [5] described the EWSR1 rearrangement in myoepithelial carcinomas with (EWSR1/ZNF444) gene fusion. Furthermore, Antonescu et al. [1] described the same finding in a large series of myoepithelial carcinomas. The finding of an EWSR1/FEV gene fusion is probably the principal event in this case, and it is unlikely that another gene fusion involving EWSR1 would be detected in the same tumour despite EWSR1 translocation by FISH. On the other hand, due to the partial epithelial nature (EMA positive) of the present tumour, metastatic carcinoma should also be considered in the differential diagnosis, including neuroendocrine carcinoma that may harbour EWSR1 translocation, as recently described [25]. This patient had no previous history of carcinoma and neuroendocrine markers were negative. Although the morphological features (cord-like, tubular epithelioid and pseudo-glandular patterns) were uncommon for ESFT, case 7 was finally diagnosed as adamantinoma-like ESFT.

The superficial location (dermis) of the tumour and round cell morphology complicated the differential diagnosis between atypical ES and Merkel cell carcinoma [22, 32, 36] (case 8). However, Merkel cell carcinoma usually displays CK20 expression, while the present case was negative for this marker. EWSR1 translocation (FISH and RT-PCR) confirmed the diagnosis of superficial atypical ESFT. Nevertheless, ESFT can occur as a cutaneous tumour, making the differential diagnosis with Merkel cell tumours complex [22, 32, 36], especially when 20% of ESFT have epithelial differentiation (CK+) and in several cases Merkel cell neoplasms express CD99 [22, 32, 36]. Additionally, CK20 immunoreactivity is consistent with Merkel cell carcinoma, while its expression is absent in ESFT, as confirmed by our working group in more than 500 genetically confirmed ESFT tested with CK20 with lacking expression in all tumours (unpublished data).

Atypical ESFT can be confused with haematological disease [16–18], particularly in patients with a previous history of lymphomas (case 9). Lymphoid markers (CD45, CD20 and CD138) were negative in this case, thus excluding lymphoma/myeloma, and genetic studies revealed EWSR1 translocation confirming the diagnosis of ESFT.

In conclusion, the cases herein reported represent unusual histological variants of ESFT mimicking sarcomas, carcinomas and lymphomas. These results strongly suggest that tumours showing overlapping morphological and immunohistochemical features with other SRCT of soft tissue and bone can be readily differentiated using complementary molecular techniques. In SRCT with an EWSR1 translocation demonstrated by FISH, the RT-PCR analysis of specific sarcoma-related gene fusions can offer important clues when diagnosing specific entities, especially in tumours with atypical histopathology and/or IHC findings. However, for the time being the integration of clinical, histopathological, IHC and genetic data remains the corner stone for the diagnosis of bone and soft tissue sarcomas in which genetic rearrangements with reciprocal translocations are present.