Cancer in Children: Clinical Management, 5th Edition

Chapter 20. Ewing sarcoma and peripheral primitive neuroectodermal tumour

Michael Paulussen

Heinrich Kovar

Herbert Jürgens


The Ewing family of tumours (EFT) is a clinically heterogeneous group of malignant tumours including Ewing sarcoma (ES), malignant peripheral primitive neuroectodermal tumour (pPNET) and Askin tumour (ES of the thoracic wall, affecting children, adolescents and young adults). EFT most commonly originate in bone but occasionally arise in soft tissue. Lücke in 1866 and Hildebrand in 1890 first described single cases of this tumour, but James Ewing was the first to recognize Ewing sarcoma as a distinct entity. He published reports between 1921 and 1939 on ‘diffuse endothelioma’ or ‘endothelial myeloma’, tumours sensitive to radiation and considered to be of endothelial origin.1 The exact histogenesis remains unknown, although the current view is that EFT originate from an ubiquitous pluripotent stem cell thought to be of neuroectodermal origin. At the genetic level, EFT demonstrate a tumour-specific chromosomal translocation between the long arms of chromosome 22 and, most commonly, chromosome 11 (85 per cent) or chromosome 21 (10 per cent). These aberrations result in a gene fusion which is diagnostic for EFT and which differentiates them from other childhood malignancies with a similar small round cell phenotype. In James Ewing's time, and up to the late 1960s, >90 per cent of patients died of distant metastases within 2 years of diagnosis despite the well known radioresponsiveness of these tumours. This shows that micrometastases are most likely to be present at diagnosis, or develop very early in the course of disease, in the vast majority of patients. Definitive cure of EFT has only been observed since the introduction of chemotherapy and, with current combinations of multiagent chemotherapy and local therapy, cure rates of >50 per cent can be achieved.


In Caucasian populations, EFT represent the second most common primary malignant bone tumour in childhood and adolescence and account for 10–15 per cent of all primary malignant bone tumours. Annual incidence rates are approximately 3 per million in children aged <15 years, and 2.4 per million for patients aged 15–24 years. The median age at presentation is 15 years, more than half of cases occurring in the second decade of life, and there is a male predominance (1.5:1) which increases with age (Fig. 20.1). EFTare almost unknown in African and Chinese populations. There is no evidence for genetic predisposition and it seems unlikely that environmental exposure plays a role in pathogenesis.

Clinical presentation, diagnosis, and staging

Primary sites

The most common sites of the primary lesion are the pelvis, long bones of the extremities, ribs, scapula, and vertebrae (Fig. 20.2). Compared with the skeletal distribution of osteogenic sarcoma, the flat bones of the trunk are more often affected. In long bones, the tumour originates from the diaphysis, either centrally or towards the ends, in contrast with the typical metaphyseal presentation of osteosarcoma.

Fig. 20.1 Age and gender. Data based on 1426 patients from the (EI)CESS studies.

Signs and symptoms

EFT patients most commonly present with pain. Additional symptoms depend on site and bulk of disease. The pain is often mistakenly related by the patient or health care professional to minor injuries incurred during sport or everyday activity, but it persists for an unusual length of time and is independent of activity (e.g. it occurs at night). A history of increasing persistent pain, followed by swelling of the affected area, can sometimes be misdiagnosed as infection, further delaying the diagnosis. Involvement of the spinal cord or peripheral nerves may produce neurologic symptoms. Bony metastases may be palpable on the skull, ribs, or any superficial bone. Slight to moderate fever is reported in about one-third of patients, and occurs more often in those with metastatic disease at diagnosis.

Laboratory investigations

There are no specific blood or urine tests which identify EFT. Blood tests usually show a moderately elevated erythrocyte sedimentation rate and may reveal some degree of anaemia and leucocytosis. An elevated serum level of lactate dehydrogenase (LDH) correlates with tumour burden and for this reason is of indirect prognostic significance. Urine catecholamine levels are normal.


Plain radiographs, MRI, and CT are required to assess the extent of disease at the primary site, for locoregional extension, and to search for metastases. It is important to define and document tumour size, including its soft tissue components, at diagnosis and to use these measurements to monitor response to chemotherapy, as this is of prognostic significance and helps to plan local therapy. The typical radiograph of a Ewing bone tumour shows a destructive osteolytic lesion of the diaphysis with destruction of the osseous cortex, elevation of the periosteum, and infiltration of surrounding soft tissue. MRI is superior to CT for measuring the extent of the tumour in bone marrow and soft tissue and in defining the relationship between the tumour and its adjacent structures (e.g. blood vessels or nerves). However, a CT scan can identify cortical and bony changes more precisely than MRI. Figure 20.3 shows MRI of a right iliac crest EFT. Chest CT is recommended for detection of pulmonary metastases, and radionuclide bone scanning with technetium-99m is recommended for detection of bone metastases. [18 Fluorine] fluorodeoxyglucose positron emission tomography (FDG-PET) has been shown to be of value as another potentially sensitive screening method for detection of metastases in EFT.

Fig. 20.2 Skeletal distribution of Ewing tumours. Percentages based on 1426 patients entered into the consecutive (EI)CESS trials.


Open biopsy or needle-core biopsy of the primary tumour is required to confirm the diagnosis. The biopsy track and adjacent tissues must be regarded as contaminated by tumour, and the approach should be planned so that the site can be subsequently included in the local treatment. For this reason, diagnostic procedures should be performed by specialists familiar with this rare disease. About 25 per cent of patients present with detectable metastases at diagnosis (Fig. 20.2), and if there is any doubt, these sites should also be confirmed by biopsy. The biopsy specimen must be sufficient to provide material for standard histology and immunocytochemistry, and fresh/frozen material for molecular biology studies.

Fig. 20.3 MRI of right iliac crest Ewing tumour.

Bone marrow involvement

Disseminated tumour cells may not be homogeneously distributed in the bone marrow and thus may escape detection. Therefore it is recommended that marrow samples are obtained from several sites to improve the accuracy of detection. In addition to standard pathologic assessment, micrometastatic disease in the bone marrow can be detected by molecular techniques using reverse transcriptase–polymerase chain reaction (RT–PCR) for tumour-specific EWS-fusion transcripts. The prognostic impact of the detection of minimal disease in bone marrow by such techniques is under investigation in clinical trials.


Staging at diagnosis is based on the the size and extent of the primary tumour and on the presence or absence of lung and/or bone metastases. These factors are of prognostic significance and influence the choice of treatment. A list of staging investigations at diagnosis is given in Tabl 20.1.

Differential diagnosis

The most important clinical differential diagnosis in EFT is osteomyelitis. The radiologic appearance may be very similar, and occasionally EFT of bone may be secondarily infected. On histologic examination, EFTmust be differentiated from other small round-cell tumours, in particular embryonal rhabdomyosarcoma, neuroblastoma (in children aged 5 years), smallcell osteosarcoma, and non-Hodgkin lymphoma.


Ewing tumours of bone and soft tissue have a similar appearance and are composed of firm grey-white soft tissue with a glistening moist appearance on sectioning. Macroscopically, the intraosseous component of the tumour is usually of firm consistency, while the extraosseous component tends to be of less firm texture with areas of haemorrhage and cystic degeneration secondary to tumour necrosis. EFT are always high-grade malignancies and when routinely stained with haematoxylin and eosin show microscopically monomorphic small blue round primitive cells, with round nuclei and scanty cytoplasm with variable amounts of glycogen deposition which stain periodic acid–Schiff (PAS) positive (Fig. 20.4). pPNET cells differ from other EFT on electron microscopy by their expression of distinct features of neural differentiation with prominent neurite-like cell processes containing neurosecretory granules and neurofilaments; rosettes and Homer–Wright pseudorosettes are also occasionally identified. However, EFT cannot be differentiated from other PAS-positive small blue round cell tumours on morphologic features alone and the diagnosis must be assisted by immunocytochemistry. The identification of monoclonal and polyclonal antibodies for various differentiation markers provides supportive evidence for the diagnosis of EFT. CD99 (Mic-2 antigen) expression is positive in >95 per cent cases but is not unique to EFT. EFTs show varying degrees of neural differentiation and can be subdivided into typical undifferentiated, atypical differentiated ES, and fully differentiated pPNET by their increasing expression of neural markers such as S-100 protein, neuron-specific enolase (NSE), and synaptophysin. Other useful markers include vimentin, desmin, smooth muscle actin, and CD45 (leucocyte common antigen) which help to differentiate between small round cell tumours of myogenic, fibrogenic, and haematopoietic origin.

Table 20.1. Staging investigations at diagnosis


Primary tumour site

Staging for metastases

Radiograph in two planes, whole bone with
adjacent joints


At suspicious sites

MRI and/or CT, affected bone(s) and adjacent joints


At suspicious sites

Biopsy, histology, molecular biology


At suspicious sites

Chest radiograph and CT



Bilateral bone marrow aspirates and trephine biopsies



Whole-body technetium bone scan




Indicated, if available

Indicated, if available

Fig. 20.4 Typical histologic appearance of Ewing sarcoma with round or oval nuclei and poor delineation of cytoplasma of cells.

Molecular biology

Genetic definition of the Ewing sarcoma family of tumours

Historically, ES and pPNET have been considered distinct tumour entities along a gradient of limited neuroglial differentiation. However, after the identification of a tumour-specific chromosomal translocation t(11;22)(q24;q12) and its molecular equivalent, the EWS–FLI-1 gene fusion in both ES and pPNET, the case was made for a genetically based union. The presence of this gene rearrangement enables unambiguous diagnosis in 85 per cent of ES and pPNET. In the majority of the remaining cases, variant translocations of the EWS gene can be identifed. Today ES, pPNET, and an increasing number of extraskeletal small round cell tumours carrying this genetic marker or one of its similarly structured variants are referred to as the Ewing family of tumours (EFT). This unification implies a common pathogenetic origin and accounts for similar morphology and, at least in part, immunohistochemistry (i.e. high CD99 expression). Patients with EFT are treated according to the same clinical protocols and, despite earlier reports of different clinical outcomes in ES and pPNET, similar treatment results are achieved for ES and pPNETwith current multimodal treatment strategies. However, the exact tissue of origin for EFT is still not known. The almost complete absence of unequivocal differentiation markers and the sporadic occurrence of biphenotypic tumours with an EWS–FLI-1 gene rearrangement imply a primitive pluripotent neuroectodermal or mesenchymal stem cell. At present, however, it cannot be concluded with certainty that ES, pPNET, and all other tumours carrying one of the EFT-specific gene rearrangements originate from the same tissue.

The EWS–ETS gene fusion

The molecular key to the disease is the rearrangement of the Ewing sarcoma gene EWS on chromosome 22q12 with the FLI-1 gene on chromosome 11q24.2 EWS encodes an ubiquitously expressed RNA binding protein that appears to be involved in transcription and processing of messenger RNA. FLI-1 codes for a protein with a carboxy terminal DNA binding domain that is characteristic of members of the ETS transcription factor family. As a consequence of the gene fusion, the EWS RNA binding domain is replaced by the FLI-1-derived ETS DNA binding domain, resulting in a novel transcription factor with a strong in vitro transactivation potential (Fig. 20.5). Alternative EWS gene fusions to other members of the ETS gene family (predominantly ERG) can be observed in about 10–15 per cent of EFTs (Table 20.2). While EWS, as well as its close relatives [TLS (FUS) and RPB56 (hTAFII68)], and ERG genes have been found to be involved in several chromosomal translocations in other solid tumours and acute leukemias, the specific combination of EWS with an ETS gene is restricted to, and therefore diagnostic of, EFT. The EWS and FLI-1 genes comprise 17 and 9 exons, respectively. Breakpoints occur between EWS exons 7 and 12 and FLI-1 exons 4 and 9. The most frequent gene fusions in EFT result in a type 1 chimeric RNA (fusion of EWS exon 7 to FLI-1 exon 6) in about 50 per cent of cases or a type 2 product (fusion of EWS exon 7 to FLI-1 exon 5) in about 27 per cent of cases. In a significant proportion of EFT the specific chromosomal translocation would lead to an out-of-frame gene fusion resulting in a truncated non-functional protein. However, in these cases, alternative splicing restores the correct reading frame for protein synthesis, indicating that there is selective pressure in EFT for the expression of a functional EWS–ETS protein.3 EWS–FLI-1 antagonists suppress EFT cell growth. Introduction of EWS–FLI-1 into immortalized murine fibroblasts leads to in vitro transformation, and EWS–FLI-1 transformed cells are tumorigenic in nude mice, with the tumour cells displaying a small round cell phenotype resembling that of EFT.4 In contrast, oncogenic fusions of EWS to non-ETStranscription factors induce a distinct morphology suggesting that the transcription factor moiety of the chimeric gene product determines the tumour phenotype. Most recent studies have demonstrated that EWS–FLI-1 is capable of suppressing differentiation in pluripotent mesenchymal cells, possibly explaining the largely undifferentiated phenotype of EFT cells. Together, these observations indicate that EWS–ETS gene fusion is essential for EFT pathogenesis.

Fig. 20.5 Chromosomal translocation t(11;22)(q24;q12) in EFT. The chromosomal translocation t(11;22)(q24;q12) leads to the generation of a DNA binding fusion protein in EFT. The gene EWS for the RNA binding protein is located on chromosome 22 (chr22) in region q12, and the gene FLI-1 for the DNA binding transcription factor is localized on chromosome 11 (chr11) at q24. The EWS–FLI1 chimeric protein is expressed from the abnormal chromosome 22 (abn22), while the reciprocal translocation product on chromosome 11 does not give rise to a gene product in EFT.

Table 20.2. EWS–ETS gene rearrangements in EFT

Gene fusion

Cytogenetic equivalent

Observed frequency (%)
















Paradoxically, introduction of EWS–FLI-1 is toxic to most primary cell types. Impairment of the p53 pathway by either mutation of p53 or loss of the INK4A gene, which play a key role in the regulation of apoptosis in response to cellular stress, including that conferred by several oncogenes, rescues EWS–FLI-1-expressing primary cells from programmed cell death. Interestingly, p53 mutations are rare (<10 per cent) in primary EFT and INK4A deletions occur in only 25 per cent of cases.5 However, if present, these alterations appear to be associated with an adverse prognosis. It remains to be established whether other types of secondary mutations rescue EFT cells from EWS–ETS-induced cell death or if the EFT progenitor cell is generally tolerant to the chimeric oncogene.

The transforming ability of the EFT-specific oncoprotein requires the presence of at least the EWS amino terminus and the ETS DNA binding domain, which together confer high transcriptional activity to the fusion gene product. Therefore it is assumed that the oncogenic function of EWS–ETS proteins is mediated by deregulation of ETS responsive genes. Almost equal numbers of upand downregulated genes have been observed in response to ectopic EWS–FLI-1 expression. Several candidate genes downstream of EWS–FLI-1 have been identified, encoding cell cycle regulatory genes (cyclins D1, p57KIP2, p21), growth factors (PDGF-C), receptors (TGFbRII, EGR), transcriptional regulatory proteins (c-Myc, Id2, c-Fos), molecules involved in intraand extracellular signalling (EAT2, MNFG), and proteins with a presumed role in the metastatic process (stromelysin, tenascin C). Recently, induction of the cell surface sialoglycoprotein CD99 (MIC2), the immunologic hallmark of EFT, has been found to be induced by transgenic EWS–FLI-1 in immortalized human fibroblasts.6 The pattern of EWS–FLI-1 associated gene expression appears to depend on the cellular context, and it is unclear whether most of the genes identified so far are directly or indirectly affected by EWS–FLI-1 and what role they might play in EFT pathogenesis. Furthermore, the different EWS–ETS gene fusions present in EFT differ with respect to the sets of genes induced/repressed in transgenic murine fibroblasts with only slight overlap. Even different EWS–FLI-1 fusion types (i.e. type 1 and type 2) differ in the strength of their activating function. How far this variability translates into different clinical behaviour remains a matter of debate. For the group of patients with localized disease, retrospective studies indicated a better event-free survival of patients with tumours expressing type 1 EWS–FLI-1 when compared with all other EWS–FLI-1 fusion types. This finding is currently under prospective evaluation as part of the Euro-EWING 99 study. In contrast, no difference in clinical behaviour has been observed between EWS–FLI-1 and EWS–ERG-containing EFT.

In addition to its transcriptional regulatory activity, EWS–FLI-1 is likely to be involved in distinct processes that do not depend on a functional DNA binding domain. Because of its interaction with several RNA processing proteins and its documented influence on splice site selection in an in vitro model, it may be speculated that alternative splicing might be involved in the oncogenic activity of EWS–FLI-1. So far, no targets for this hypothetical EWS–FLI-1 activity have been identified in EFT.

The strong interest in downstream targets of the primary genetic aberration in EFT derives from the hope of identifing tumour-specific targets for biologically tailored EFT therapy. CD99 (MIC2), which always parallels EWS–ETS expression in EFT, and for which engagement by some specific antibodies was found to induce cell death, may provide an example in this respect. However, as this antigen is expressed at variable levels in many tissues, its therapeutic potential may be limited.

Additional chromosomal aberrations

Although t(11;22) or one of its variants may be the only cytogenetically detectable chromosomal anomaly in some EFT, additional recurrent genetic aberrations are frequent. Among them, trisomy 8 and/or 12 are observed in 44 per cent and 29 per cent of cases, respectively. Structural changes commonly affect chromosomes 1 and 16, most frequently leading to a gain of 1q and a loss of 16q and the formation of a derivative chromosome der(1;16). The molecular consequences of these aberrations are not known. Deletion of the chromosomal region 9p21 (the INK4A gene), which is lost in 25 per cent of EFT, remains cytogenetically invisible in most cases.

Molecular diagnostics

Because of its consistent presence in the disease, EWS–ETS gene fusions serve as an unambiguous diagnostic marker for EFT cells. The classical cytogenetic demonstration of t(11;22)(q24;q12) has largely been replaced by fluorescence in situ hybridization (FISH) using probes flanking the EWS breakpoint region on chromosome 22. Even complex chromosomal rearrangements and rare EWS translocations with other ETS partner genes can be revealed using this method. However, the most frequently used method of detecting the fusion gene is RT–PCR. This diagnostic approach requires immediate snap-freezing of the biopsy since it relies on extraction of good-quality RNA. The strength of the method lies in its high sensitivity which enables identification of single tumour cells among 105—106 normal cells. Therefore it is used to evaluate the prognostic significance of minimally disseminated EFT cells in bone marrow, peripheral blood, and stem cell collections for autologous transplantation. Retrospective studies reveal bone marrow involvement in a majority of patients with metastatic disease and in about one-third of patients with localized disease. Preliminary data imply adverse prognosis when the bone marrow tests RT–PCR positive at the time of diagnosis,7 although these results await confirmation in prospective studies such as Euro-EWING 99.

RT–PCR relies on gene expression. However, nothing is known about the activity of the EWS–ETS fusion gene in resting EFT cells. Therefore, at present, it is possible that dormant EFT cells may not be detectable by RT–PCR. Furthermore, as tumour cells may not be evenly distributed in the bone marrow, they may also escape detection. Hence it is necessary to sample the bone marrow from several sites.

Extraskeletal Ewing sarcoma

Extraskeletal ES is uncommon. The major differential diagnosis includes ES of bone with extensive soft tissue extension and an inapparent intraosseous component. The predominant site of initial presentation is the trunk. As distinct from ES of bone, there is no predominance in boys, but the same rules apply for the investigation and systemic treatment of extraskeletal ES as for ES of bone. However, as there appears to be a higher risk of lymphatic spread, local therapy, especially radiation, must be planned according to the same principles as applied to soft tissue malignancies like rhabdomyosarcoma. It can be differentiated from rhabdomyosarcoma, neuroblastoma, and lymphoma by histocytochemistry, and in addition, like all EFT, extraskeletal ES shows the typical chromosomal translocation t(11;22).


Until the late 1960s, despite the well-known radioresponsiveness of these localized tumours, >90 per cent of patients died of metastases within 2 years of diagnosis. Since the introduction of multimodal treatment regimens including combination chemotherapy, surgery (where possible), and radiotherapy, cure rates of between 50–65 per cent have been achieved. The high incidence of systemic relapse in localized EFT, despite local and systemic treatment, suggests that disseminated occult tumour cells (micrometastases), possibly present at the time of diagnosis or which develop very early in the course of disease, escape detection and may not be eradicated by current treatment approaches. Improved treatment is the subject of all ongoing clinical trials and, whenever possible, patients with EFT should participate and be treated in a center participating in such studies.

Local therapy

EFT is a systemic disease; nevertheless there is no cure without local control. Local therapy following an induction phase with chemotherapy is now regarded as standard procedure. Radiotherapy is no longer the local treatment of first choice, as surgery, when feasible, has been shown to produce a survival advantage in several studies. Data from the European CESS and EICESS studies indicate that selection bias, owing to the better outcome for small peripheral lesions for which surgery was more feasible, cannot be the only explanation for the difference in local control.


Increasing awareness of the risk of local recurrence following radiotherapy has encouraged the use of surgery or surgery with radiotherapy, with chemotherapy as optimal therapy. EFT of bone is rarely limited to the bony compartment and the presence of a soft tissue component is common, classifying most tumours as highly malignant extracompartmental tumours. The biopsy track must always be included in the surgical resection. In order to allow comparison of surgical interventions, all surgical procedures should be classified, for example according to the Enneking criteria: intralesional, marginal, wide, or radical resection (Table 20.3). Response of the tumour to initial chemotherapy is classified histologically based on the percentage of viable tumour cells remaining in the resected tumour (Salzer–Kuntschik grades 1–6). In cases of marginal or intralesional resection, or where there is evidence of poor histologic response to initial chemotherapy (10 percent viable tumour cells), surgery should be combined with postoperative radiotherapy. According to experience from several clinical study groups [e.g. the (EI)CESS studies], surgical margins and response to initial therapy correlate with outcome in terms of local control rate and risk of distant metastases (Fig. 20.6). Hence good response to initial chemotherapy and tumour-free margins must be achieved whenever possible.

Table 20.3. Enneking classification of surgical intervention

Intralesional resection

The tumour is opened during surgery, the surgical field is contaminated, and there is microscopic or macroscopic residual disease

Marginal resection

The tumour is removed en bloc. However, the line of resection runs through the pseudocapsule of the tumour. Microscopic residual disease is likely

Wide resection

The tumour and its pseudocapsule are removed en bloc surrounded by healthy tissue within the tumour-bearing compartment

Radical resection

The whole tumour-bearing compartment is removed en bloc (e.g. above-knee amputation in tibial tumours)

Situations in which surgical resections are preferable to radiation therapy include lesion in expendable bone, pathologic fracture, distal extremity, bulky primary tumour, and poor response to initial chemotherapy. In every case, a team approach is essential for the evaluation of these patients before local therapy. The treatment plan must be individualized depending upon the location and size of the tumour and the anatomical structures in the vicinity of the tumour which might be affected by the type of treatment selected.

In recent years, surgical options have broadened to include the development of modern modular endoprosthesis systems, autograft techniques such as replacement of tibial bone by contralateral fibula, and modified amputation techniques such as rotation-plasty which have improved the mobility of the patient.


The radiosensitivity of ES was recognized by James Ewing in 1921 and radiotherapy has always played a major role in obtaining local control. Preoperative radiotherapy is indicated when there is tumour progression during chemotherapy, in an emergency such as spinal cord compression, or when incomplete surgical resectability is anticipated. Postoperative radiotherapy is indicated by incomplete resection of the tumour or where poor histologic response of the tumour to chemotherapy is determined.

Traditionally, whole-bone irradiation was advised because the tumour was thought to arise in the bone marrow, putting the whole marrow cavity at risk. The advent of improved imaging with MRI, which accurately demonstrates the extent of marrow involvement, and effective chemotherapy has questioned this approach. In a randomized study performed by POG and completed in 1989, it was shown that the results of radiotherapy to initial tumour volume with a 2-cm safety margin are as good as those obtained after whole-bone irradiation. The compartment dose for radiotherapy of inoperable tumours is 44.8 Gy with a tumour boost to at least 54.4 Gy, but it can vary depending on site of tumour and age of the patient. The standard target volume dose for preoperative radiotherapy is 54.4 Gy. The dose for postoperative radiotherapy depends on histologic assessment of the resected tumour margins and the tumour response to chemotherapy, and varies from 44.8 to 54.4 Gy. Irradiation of the whole anatomical compartment is not required if adequate safety margins around the initial volume of tumour (usually 2–5 cm) can be assured. Areas of scars after biopsy or tumour resection must be included in the radiation fields. To avoid constrictive fibrosis, an adequate strip of skin and subcutaneous tissue should be left when irradiating limb sites and the epiphyseal plates should be spared if possible, particularly in growing children.

Fig. 20.6 (a) Local therapy and site of failure. (b) Site of failure in surgically treated (with or without radiotherapy) Ewing sarcoma patients according to surgical margins and response.


EFT have been shown to be particularly responsive to alkylating agents such as ifosfamide and cyclophosphamide, to the anthracycline doxorubicin (Adriamycin), and to other agents such as vincristine, actinomycin D, and the topoisomerase II inhibitor etoposide. A combination of agents with complementary mechanisms of action haveimproved disease-free survival rates. The Intergroup Ewing Sarcoma (IESS) studies showed the superiority of a four-drug regimen with vincristine (V), actinomycin D (A), cyclophosphamide (C), and doxorubicin (D) over a threedrug regimen with vincristine, actinomycin D, and cyclophosphamide (VAC) in terms of disease-free survival (60 versus 24 per cent) and effectiveness of local control (96 versus 86 per cent). Rosen et al8 from the Memorial Sloan–Kettering Cancer Center (MSKCC) reported an advantage of using these drugs in combination rather than sequentially. It is important to recognize that a considerable number of patients in earlier series received cumulative doxorubicin doses of over 700 mg/m2 before its use was restricted to a maximum of 400 500 mg/m because of the risk of doxorubicin-related cardiomyopathy. The MSKCC and IESS experience led to the wide use of similar four-drug regimens in most therapeutic trials. Thereafter, the treatment of EFT was extended to incorporate ifosfamide (I) (CESS 86) and etoposide (E). In the European EICESS-92 study, patients with localized high-risk disease (tumour volume >200ml) receiving VIDA did not seem to benefit from the addition of etoposide, whereas in the randomized POG–CCG Ewing trial (VACD versus VACD-iE) patients treated with VACD-iE appeared to have a more favourable outcome. Results of selected reported trials823 in localized EFTare summarized in Table 20.4.

The incorporation of granulocyte colony-stimulating factor (GCSF) into treatment regimens allowed for dose intensification by either increasing the dose per cycle or shortening the interval of time between treatment. The IESS-II study compared high-dose intermittent chemotherapy with moderate-dose continuous chemotherapy, resulting in a significant benefit from the more intensive regimen [68 versus 48 per cent disease-free survival at 5 years (Table 20.4)]. The second POG–CCG randomized study explored dose intensification by delivering chemotherapy over either 30 or 48 weeks, although the results so far have shown no difference in outcome between the standard and the dose-intensified arms. The COG are currently exploring dose intensification by decreasing the intervals between cycles (interval compression) while maintaining the same dose per cycle with the use of GCSF.

In patients with metastatic disease, who have a very poor prognosis, treatment intensification has been applied with the use of high-dose chemotherapy and autologous stem cell support. Avariety of agents have been used, such as busulphan, melphalan, cyclophosphamide, thiotepa, etoposide, and carboplatin. The addition of total body irradiation does not seem to offer any further benefit, but significantly contributes to toxicity. Analyses from the European EBMT registry have indicated that combinations including busulphan are more effective than other conditioning regimens.

Currently, large phase II–III studies are being conducted in several groups (e.g. in the USA, Scandinavia, and Italy) and within a joint European study group [GPOH, UKCCSG, SFOP, SIAK, and EORTC–STBSG, forming the European Ewing Tumour Working Initiative of National Groups (Euro-EWING)]. The Euro-EWING 99 study was initiated in late 1999 and serves as an example of a current treatment regimen. All patients receive induction with six courses of vincristine, ifosfamide, doxorubicin, and etoposide (VIDE), followed by consolidation therapy stratified and randomized according to established prognostic factors. Three risk groups are based on initial tumour volume (<200 ml or ≥200 ml), presence and site of metastatic disease at diagnosis, and histologic response to induction therapy. For standardrisk cases with good histologic response to VIDE (<10 per cent viable tumour cells in the surgical specimen), vincristine, actinomycin, and ifosfamide (VAI) consolidation is randomized versus vincristine, actinomycin, and cyclophosphamide (VAC). In high-risk patients with poor histologic response to VIDE (≥10 per cent viable tumour cells in the surgical specimen), VAI is randomized against a high-dose regimen of busulphan and melphalan with autologous stem cell support. In high-risk patients with initial lung metastases, VAI plus lung irradiation is compared with high-dose busulphan and melphalan. The main endpoints of this study are to determine event-free survival and toxicity and to evaluate the role of high-dose therapy in the treatment of patients with high-risk disease. In addition, the predictive value of the detection, by RT–PCR, of minimal metastatic or residual disease in the bone marrow is being investigated for its prognostic significance. Figure 20.7 shows the Euro-EWING 99 treatment strategy (

Table 20.4. Treatment of localized Ewing tumours: results of selected clinical studies




No. of patients

5-year DFS



Intergroup Ewing Sarcoma Studies
IESS-I (1973–1978)





VAC vs VAC+WLI: 0.001
VAC vs VACD: 0.001

Value of D
Benefit of WLI?

IESS-II (1978–1982)






Value of aggressive cytoreduction







Value of combination IE

Second POG–CCG


VCD +IE (48 weeks)
VCD +IE (30 weeks)


75% (3 yr)
76% (3 yr)


No benefit of dose-time compression

Memorial Sloan–Kettering Cancer Center
T2 (1970–1978)


VACD (adjuvant)




After local therapy only, cumulative
dose of D up to 600 mg/m2

P6 (1990–1995)




77% (2 yr)


C dose escalation, 4.2 g/m2course

St Jude Children's Research Hospital




<8 cm 82% (3 yr)
≥8 cm 64% (3 yr)


Tumour size as prognostic factor



Therapeutic window with IE


Clinical responses
in 96%


IE combination effective



VCD/IE intensification


78% (3 yr)


Tumour size (<8 cm or ≥8 cm) loses
prognostic relevance with more
intensive treatment

Rizzoli Orthopaedic Institute, Bologna, Italy






Surgery in 78% of patients

SFOP, France






Histologic response better predictor
of outcome than tumour volume

UK Children's Cancer Study Group–Medical Research Council




Extr. 52%
Axial 38%
Pelvic 13%


Tumour site most important
prognostic factor





Extr. 73%
Axial 55%
Pelvic 41%


Importance of administration of
high-dose alkylating agents (I)





<100 ml 80% (3 yr)
≥100 ml 31% (3 yr)
Viable tumour
<10%, 79% (3 yr)
>10%, 31% (3 yr)


Tumour volume (<100 ml or ≥100 ml)
and histologic response are
prognostic factors



<100 ml (SR): VACD
≥100 ml (HR): VAID


52 % (10 yr)
51 % (10 yr)


Intensive treatment with I for high-risk
patients. Tumour volume (<200 ml or
≥200 ml) and histologic response as
prognostic factors






0. 5834
0. 2913

Tumour volume (<200 ml or ≥200 ml)
and histologic response as prognostic
factors. Addition of E no major benefit

DFS, disease-free survival; NA, not available; V, vincristine; A, actinomycin D; D, doxorubicin; E, etoposide, I, ifosfamide; WLI, whole-lung irradiation; HD, high dose; MD, moderate dose; SR,
standard risk; HR, high risk.

aP-values are given only for trials comparing randomized treatment arms.

Metastatic disease

Approximately 15 per cent to 35 per cent of EFT patients present at diagnosis with detectable metastases in the lung and/or in bone and/or in bone marrow. The presence of metastatic disease at diagnosis is a major adverse prognostic factor. The EICESS studies showed that patients with isolated lung metastases may have a better prognosis than those with extrapulmonary metastases; however, the survival of all has been disappointing (Fig. 20.8). If pulmonary disease is present, patients also receive bilateral pulmonary irradiation at a dose of 14–20 Gy. Patients with solitary or circumscribed bony lesions can receive irradiation to those sites at doses of 40 to 50 Gy, in addition to radiation to the primary site. However, survival rates for patients with multiple bony metastases and/or bone marrow disease are reported to be < 10 per cent at 2 years and the discouraging results of treatment for metastatic disease have led to the more aggressive approach with the use of high dose chemotherapy and autologous stem cell reinfusion.

Fig. 20.7 Outline of Euro-EWING 99 treatment protocol. Euro-EWING 99 is a cooperative trial involving GPOH, UKCCSG, SFOP, SIAK, and EORTC. R1, risk group 1 (localized, good histologic response); R2, risk group 2 (localized, poor histologic response, all patients with metastases limited to lungs/pleura); R3, risk group 3 (extrapulmonary metastases); VIDE, vincristine, ifosfamide, doxorubicin, and etoposide; VAI, vincristine, actinomycin D, and ifosfamide; VAC, vincristine, actinomycin D, and cyclophosphamide; Bu-Mel, high-dose busulfan–melaphalan therapy with stem cell rescue; HDT, Bu-Mel or other experimental high-dose therapy.

Fig. 20.8 Disease-free survival according to metastatic state.

The survival of patients who develop metastatic disease either on or off therapy is poor and second remissions are usually short lived.

Prognostic factors

The classical factors related to poor prognosis include presence of metastases, older age, large tumours, and primary site in the trunk and pelvis (Table 20.5). The specific prognostic factors which concurrently influence treatment regimens include initial volume of the primary tumour (>200 ml), the presence of pulmonary and/or extrapulmonary metastases at diagnosis, the presence of tumour cells in the surgical margins, and poor histologic response of the tumour to induction chemotherapy. The predictive value of other factors such as the degree of neural differentiation (ES versus pPNET), the amount of bone marrow disease detected by RT–PCR, and the type of fusion transcript remains under investigation. So far, studies have shown that EFTs with marked neuroectodermal differentiation such as pPNET do not have a worse prognosis than the typical undifferentiated ES. Among putative molecular determinants of prognosis, loss of the INK4A gene and mutation of p53 have been described as independently associated with poor outcome. The possible influence of the specific EWS–FLI-1 gene fusion type on the prognosis of patients with localized disease is currently under prospective evaluation in Euro-EWING 99. The most clear-cut treatment-related prognostic indicator is tumour response to chemotherapy. Poor histologic response is defined as the persistence of viable tumour cells within the surgical specimen after presurgical chemotherapy and is a very strong predictor of poor outcome. Finally, early relapse (within 2 years of initial diagnosis) was another predictor of adverse outcome in a recent analysis of two large European trials.

Table 20.5. Prognostic factors in Ewing tumours in order of relevance (poor < good prognosis)


Visible at diagnosis < undetectable

Bone/bone marow < lungs / pleura only

Histologic respose

Poor < good

Tumour size

Tumour extension above 8 cm < less than 8 cm

Tumour volume above 200 ml < above 100 ml < below 100 ml

Tumour site

Central < proximal < distal

Local therapy

Radiotherapy alone < surgery ± radiotherapy


Over 15 years < under 15 years


Elevated serum LDH < normal serum LDH

LDH, lactate dehydrogenase.

Future prospects

Since EFT are fundamentally systemic diseases, the addition of multi-agent chemotherapy to local control with radiation and/or surgery has improved the disease-free survival rate from only 10 per cent with local therapy alone to 50–70 per cent. Careful analyses of patterns of failure and the tailoring of the treatment approach to prognostic subgroups of patients appears to be the task for the future in an effort to overcome tumour resistance. Initial chemotherapy following biopsy-confirmed diagnosis prior to local therapy leads to impressive tumour shrinkage in >90 per cent of patients. The optimal combination of agents and the length of initial treatment is still under evaluation, as is the use of high-dose therapy regimens with stem cell rescue for high-risk patients and poor responders to conventional therapy.

High-throughput microarray-based gene expression profiling studies comparing series of different types of small blue round cell tumours have identified a unique gene expression pattern associated with EFT.24 Thus, in the future, it should be possible to assign a small blue round cell tumour to a specific diagnosis based on its overall gene expression profile. It can also be expected that gene expression studies on large EFT cohorts will lead to the discrimination of prognostically distinct groups to allow better treatment stratification.

In terms of biologically based therapy, hopes have recently focused on the use of tyrosine kinase receptor inhibitors [imatinib mesylate/STI-571 (Gleevec1)]. Among potential targets of this drug are the stem cell factor receptor c-kit and PDGF-Rb which are expressed by 30 per cent and 90 per cent of primary EFT, respectively. So far, however, in vitro studies and early clinical results suggest only limited therapeutic activity of imatinib mesylate on EFTs.25 Nevertheless, biologic modifiers with potential antitumour effect may emerge to play an important role in the future treatment of EFT.


The prognosis for EFT patients is determined primarily by tumour dissemination and tumour burden as well as by response to treatment. Tumour burden can only be limited by early diagnosis, but response is subject to the impact of better therapies: These include the design of more effective chemotherapy combinations, increased drug intensity with improved supportive care, and surgical removal of all tumour. In addition, more precise determination of prognostic factors should result in opportunities to improve stratification of treatment intensity, offering the opportunity to limit the risk of late effects from treatment in patients with favourable disease. Because of the rarity of these tumours, treatment should always be performed in centers experienced in EFT treatment and within the framework of controlled clinical trials. In particular, experimental approaches should be restricted to clinical trial settings.


This work was supported by Deutsche Krebshilfe and EU-Biomed grants. We thank Christine Jürgens for carefully reading the manuscript.


1. Ewing J (1921). Diffuse endothelioma of bone. Proc N Y Pathol Soc 21, 17–24

2. Delattre O, Zucman J, Plougastel B, et al. (1992). Gene fusion with an ETS DNA-binding domain caused by chromosome translocation in human tumours. Nature 359, 162–5.

3. Zucman-Rossi J, Legoix P, Victor JM, Lopez B, Thomas G (1998). Chromosome translocation based on illegitimate recombination in human tumors. Proc Natl Acad Sci USA 95, 11786–91.

4. Thompson AD, Teitell MA, Arvand A, Denny CT (1999). Divergent Ewing sarcoma EWS ETS fusions confer a common tumorigenic phenotype on NIH3T3 cells. Oncogene 18, 5506–13.

5. Kovar H, Jug G, Aryee DN, et al. (1997). Among genes involved in the RB dependent cell cycle regulatory cascade, the p16 tumor suppressor gene is frequently lost in the Ewing family of tumors. Oncogene 15, 2225–32.

6. Lessnick SL, Dacwag CS, Golub TR (2002). The Ewing sarcoma oncoprotein EWS/FLI induces a p53-dependent growth arrest in primary human fibroblasts. Cancer Cell 1, 393–401.

7. Schleiermacher G, Peter M, Oberlin O, et al. (2003). Increased risk of systemic relapses associated with bone marrow micrometastasis and circulating tumor cells in localised Ewing tumor. J Clin Oncol 21, 85–91.

8. Rosen G, Caparros B, Mosende C, et al. (1978). Curability of Ewing sarcoma and considerations for future therapeutic trials. Cancer 41, 888–899.

9. Nesbit ME Jr., Gehan EA, Burgert EO Jr., et al. (1990). Multimodal therapy for the management of primary, nonmetastatic Ewing sarcoma of bone: a long-term follow-up of the First Intergroup study. J Clin Oncol 8, 1664–74.

10. Burgert EO Jr., Nesbit ME, Garnsey LA, et al. (1990). Multimodal therapy for the management of nonpelvic, localized Ewing sarcoma of bone: intergroup study IESS-II. J Clin Oncol 8, 1514–24.

11. Grier HE, Krailo MD, Tarbell NJ, et al. (2003). Addition of ifosfamide and etoposide to standard chemotherapy for Ewing sarcoma and primitive neuroectodermal tumor of bone. N Engl J Med 348, 694–701.

12. Granowetter L, Womer R, Devidas M, et al. (2001). Comparison of dose intensified and standard dose chemotherapy for the treatment of non-metastatic Ewing sarcoma (ES) and primitive neuroectodermal tumor (PNET) of bone and soft tissue: A Pediatric Oncology Group–Children's Cancer Group phase III trial. Med Pediatr Oncol 37, 172 (abstr).

13. Kushner BH, Meyers PA, Gerald WL, et al. (1995). Very-high-dose short-term chemotherapy for poor-risk peripheral primitive neuroectodermal tumors, including Ewing sarcoma, in children and young adults. J Clin Oncol 13, 2796–804.

14. Hayes FA, Thompson EI, Meyer WH, et al. (1989). Therapy for localized Ewing sarcoma of bone. J Clin Oncol 7, 208–13.

15. Meyer WH, Kun L, Marina N, et al. (1992). Ifosfamide plus etoposide in newly diagnosed Ewing sarcoma of bone. J Clin Oncol 10, 1737–42.

16. Marina NM, Pappo AS, Parham DM, et al. (1999). Chemotherapy dose-intensification for pediatric patients with Ewing family of tumors and desmoplastic small round-cell tumors: a feasibility study at St Jude Children's Research Hospital. J Clin Oncol 17, 180–90.

17. Bacci G, Mercuri M, Longhi A, et al. (2002). Neoadjuvant chemotherapy for Ewing tumour of bone: recent experience at the Rizzoli Orthopaedic Institute. Eur J Cancer 38, 2243–51.

18. Oberlin O, Deley MC, Bui BN, et al. (2001). Prognostic factors in localized Ewing tumours and peripheral neuroectodermal tumours: the third study of the French Society of Paediatric Oncology (EW88 study). Br J Cancer 85,1646–54.

19. Craft AW, Cotterill SJ, Bullimore JA, Pearson D (1997). Long-term results from the first UKCCSG Ewing Tumour Study (ET-1). United Kingdom Children's Cancer Study Group (UKCCSG) and the Medical Research Council Bone Sarcoma Working Party. Eur J Cancer 33,1061–9.

20. Craft A, Cotterill S, Malcolm A, et al. (1998). Ifosfamide-containing chemotherapy in Ewing sarcoma: the Second United Kingdom Children's Cancer Study Group and the Medical Research Council Ewing Tumor Study. J Clin Oncol 16, 3628–33.

21. Jürgens H, Exner U Gadner H, et al. (1988). Multidisciplinary treatment of Ewing primary sarcoma of bone: a 6-year experience of a European Cooperative Trial. Cancer 61, 23–32.

22. Paulussen M, Ahrens S, Dunst J, et al. (2001). Localized Ewing tumor of bone: final results of the cooperative Ewing sarcoma study CESS 86. J Clin Oncol 19,1818–29.

23. Paulussen M, Craft AW, Lewis I, et al. (2002). Ewing tumor of bone—updated report of the European Intergroup Cooperative Ewing Sarcoma Study EICESS 92. Proc Am Soc Clin Oncol 21, 393a (abstr).

24. Khan J, Wei JS, Ringner M, et al. (2001). Classification and diagnostic prediction of cancers using gene expression profiling and artificial neural networks. Nat Med 7, 673–9.

25. Scotlandi K, Manara MC, Strammiello R, et al. (2003). ckit receptor expression in Ewing sarcoma: lack of prognostic value but therapeutic targeting opportunities in appropriate conditions. J Clin Oncol 21, 1952–60.

If you find an error or have any questions, please email us at Thank you!