Clinical presentation, staging, and prognostic factors of the Ewing sarcoma family of tumors Treatment of the Ewing sarcoma family of tumors Radiation therapy for Ewing sarcoma family of tumors Epidemiology, pathology, and molecular genetics of the Ewing sarcoma family of tumors Adjuvant and neoadjuvant chemotherapy for soft tissue sarcoma of the 52 Bone sarcomas: Preoperative evaluation, histologic classification, and principles of surgical management 29 42 extremities 15 62 Osteosarcoma: Epidemiology, pathogenesis, clinical presentation, diagnosis, and histology 81 Chemotherapy and radiation therapy in the management of osteosarcoma Treatment protocols for soft tissue and bone sarcoma 110 BİYOPSİ ÖZET 116 Chondrosarcoma 93 118 1 Clinical presentation, staging, and prognostic factors of the Ewing sarcoma family of tumors 2 3 4 5 6 7 Aug 2013. | This topic last updated: Oca 14, 2013. INTRODUCTION — Ewing sarcoma (ES) and peripheral primitive neuroectodermal tumor (PNET, previously called peripheral neuroepithelioma) were originally described in the early 1900s as distinct clinicopathologic entities. It became evident that these entities are actually part of a spectrum of neoplastic diseases known as the Ewing sarcoma family of tumors (EFT), which also includes extraosseous ES (EES), PNET, malignant small-cell tumor of the thoracopulmonary region (Askin's tumor), and atypical ES. Because of their similar histologic and immunohistochemical characteristics and shared nonrandom chromosomal translocations, these tumors are considered to be derived from a common cell of origin. Although the histogenetic origin has been debated over the years, increasing evidence from immunohistochemical, cytogenetic, and molecular genetic studies supports a mesenchymal progenitor cell origin for all EFT [1]. The EFT can develop in almost any bone or soft tissue, but is most common in the pelvis, axial skeleton, and femur; patients typically present with localized pain and 8 swelling. Although overt metastatic disease is found in fewer than 25 percent at the time of diagnosis, subclinical metastatic disease is presumed to be present in nearly all patients because of the 80 to 90 percent relapse rate in patients undergoing local therapy alone. As a result, systemic chemotherapy has evolved as an important component of treatment [2]. Advances in multidisciplinary management of EFT over the past 30 years have resulted in a marked improvement in survival and a greater likelihood of limb-sparing surgery rather than amputation [3-6]. In data derived from the Surveillance, Epidemiology and End Results (SEER) program of the National Cancer Institute, five year survival rates for patients with ES rose from 36 to 56 percent during the periods 1975 to 1984 and 1985 to 1994 [3]. With modern multidisciplinary treatment, long-term survival can be achieved in 70 to 80 percent of patients presenting with nonmetastatic disease [4,5,7]. Here we will discuss the clinical presentation, diagnosis, and staging of EFT. The epidemiology, pathology, molecular genetics, and treatment of these tumors, principles underlying the performance of a diagnostic bone biopsy, and central (supratentorial) PNET are discussed elsewhere. CLINICAL PRESENTATION Primary sites — EFT most often arises in the long bones of the extremities (predominantly the femur, but also the tibia, fibula and humerus), and the bones of the pelvis. The spine, hands, and feet are affected considerably less often [8,9]. In a compilation of data from 975 patients from the European Intergroup Cooperative Ewing Sarcoma Studies (EI-CESS) trials, the distribution of primary sites was as follows [9]: Axial skeleton – 54 percent (pelvis 25 percent, ribs 12 percent, spine 8 percent, scapula 3.8 percent, skull 3.8 percent, clavicle 1.2 percent) Appendicular skeleton – 42 percent (femur 16.4 percent, fibula 6.7 percent, tibia 7.6 percent, humerus 4.8 percent, foot 2.4 percent, radius 1.9 percent, and hand 1.2 percent) Other bones – 0.7 percent A minority of Ewing sarcomas arise in soft tissue. Compared to undifferentiated ES of bone, patients with extraosseous Ewing sarcomas (EES) are more frequently older, more likely to be female, and arise more often within the axial rather than the appendicular skeleton [8,10-12]. Signs and symptoms — Patients with EFT typically present with localized pain or swelling of a few weeks or months duration [13-15]. Trauma, often minor, may be the initiating event that calls attention to the lesion. The pain may be mild at first, but intensifies fairly rapidly; it may be aggravated by exercise and is often worse at night. A distinct soft tissue mass can sometimes be appreciated. When present, it is usually firmly attached to the bone and moderately to markedly tender to palpation [16,17]. Swelling of the affected limb with erythema over the mass is not uncommon. Patients with juxta-articular lesions may present with loss of joint motion, while lesions involving the ribs can be associated with direct pleural extension and large extraosseous masses [18]. When the spine or sacrum is involved, nerve root irritation or compression can result in back pain, radiculopathy, or symptoms of spinal cord compression (eg, weakness or loss of bowel and/or bladder control). Constitutional symptoms or signs, such as fever, fatigue, weight loss, or anemia, are present in about 10 to 20 percent of patients at presentation [13]. Fever is related to cytokines produced by the tumor cells, and along with other systemic symptoms, is associated with advanced disease. Approximately 80 percent of patients present with clinically localized disease, although as noted previously, subclinical metastatic disease is presumed to be present in nearly 9 all. Overt metastases may become evident within weeks to months in the absence of effective therapy. The significance of this lies in the frequent delay between the onset of symptoms and diagnosis, which in one report averaged over nine months [15,19]. Patients with primary pelvic tumors are significantly more likely to present with metastatic disease compared to other sites (25 versus 16 percent) [9]. Other factors that may be associated with clinically evident metastatic disease at presentation include high level of lactic dehydrogenase (LDH), the presence of fever, an interval between onset of symptoms and diagnosis less than three months, and age older than 12 years [20]. In one series, the rate of metastatic disease at presentation in the subset of patients with none of these risk factors was only 4 percent with two factors it was 23 percent, while it was almost double (44 percent) if three or four factors were present. Sites of metastatic disease at diagnosis are similar to those seen with recurrent disease; lung and bone/bone marrow metastases predominate, in roughly equal proportions. The spine is the most frequently involved bone [21,22]. Lung metastases represent the first site of distant spread in 70 to 80 percent of cases, and are the leading cause of death for patients with EFT. Lymph node, liver, and brain involvement are distinctly uncommon [14,23]. STAGING EVALUATION — The goals of the initial evaluation are to establish the diagnosis, evaluate local disease extent, and determine the presence and sites of metastatic spread. Clinical staging includes all of the data obtained prior to definitive therapy, including the results of imaging, laboratory studies, physical examination, and tissue biopsy. Radiographic studies — The diagnostic work-up is usually initiated with a plain radiograph of the affected area. ES involving bone typically presents as a poorly marginated destructive lesion, most often associated with a soft tissue mass. The tumors tend to be large, and in long bones are metaphyseal or diaphyseal in location (image 1 and figure 1). The radiographic appearance has been described as "permeative" or "moth-eaten", indicative of a series of finely destructive lesions that become confluent over time. The cortex at the site of the lesion is often expanded, and the periosteum displaced by the underlying tumor, resulting in the clinical sign of Codman's triangle. The characteristic periosteal reaction produces layers of reactive bone, deposited in an "onion peel" appearance (image 2). The soft tissue component of the tumor rarely shows any calcification or ossification. Sclerosis, if present, represents a secondary bone reaction rather than the primary bone formation that characterizes osteosarcoma. A pathologic fracture is present at diagnosis in 10 to 15 percent of cases [13,15]. Compared to plain radiographs, a CT scan of the primary site better delineates the extent of cortical destruction and soft tissue disease (image 3). A multi-institutional study of 387 patients that included both children and adults concluded that CT and MRI were equally accurate for local staging of bone and soft tissue tumors [24]. However, MRI is preferred in most cases because of its superior definition of tumor size, local intraosseous and extraosseous extent, and the relationship of the tumor to fascial planes, vessels, nerves, and organs (image 4). Imaging of the entire involved bone is necessary to exclude the presence of skip lesions (ie, medullary disease within the same bone, but not in direct contiguity with the primary lesion). Differential diagnosis — For EFT presenting as primary bone tumors, the differential diagnosis includes both benign and malignant conditions. The most common nonmalignant possibility is subacute osteomyelitis, which may present similarly (especially the presence of fever and an elevated sedimentation rate), and be associated with intense radiotracer uptake on bone scan and a soft tissue mass on other imaging studies [25]. Aspiration of a tumor may yield purulent-appearing material; however, it will be sterile on culture. (See "Hematogenous osteomyelitis in adults".) 10 "Benign" bone tumors that can present as lytic lesions include eosinophilic granuloma and giant cell tumor of bone. Destructive eosinophilic granulomas usually occur at a younger age and are not associated with a sizable soft tissue mass. Malignant tumors that should be considered in the differential include other common solid tumors of childhood, including osteosarcoma, primary lymphoma of bone, undifferentiated high-grade pleomorphic sarcoma of bone (previously termed malignant fibrous histiocytoma of bone or spindle cell sarcoma), acute leukemia, and metastasis from a non-bone tumor, particularly neuroblastoma (table 1). A primarily lytic osteosarcoma may be difficult to distinguish from Ewing sarcoma on imaging studies. Most often, however, osteosarcoma is located in the metaphysis and usually has a rim of bone formation, which is very uncommon in ES. Primary lymphoma of bone occurs in an older age group, and is generally associated with less bone destruction than ES. Chondrosarcoma is uncommon in this age group; furthermore, the soft tissue mass usually contains calcifications. EES and soft tissue PNETs must be distinguished from a variety of benign and malignant soft tissue tumors. Metastatic work-up — Guidelines for imaging studies in patients with EFT are available from the Children’s Oncology Group Bone Tumor Committee (table 2) [26]: The metastatic work-up should include a CT scan of the chest to evaluate the thorax for metastatic disease. Criteria to guide the evaluation of suspected pulmonary metastases are available from the Children’s Oncology Group that were adopted from the European Ewing Tumor Working Group Initiative of National Groups Ewing Tumour Studies 1999 (EURO-E.W.I.N.G 99) (table 3) [27,28]. Radionuclide bone scan is recommended to evaluate the entire skeleton for the presence of multiple lesions. The utility of PET or integrated PET-CT for initial staging is unclear [29-34]. At least three series note better sensitivity for PET over bone scan or other conventional imaging modalities for detection of bone metastases [32,34,35]. However, one potential problem is that because integrated PET-CT scanning usually scans from the neck through the femurs, the bones are not all visualized as they would with a bone scan. PET has not yet replaced radionuclide bone scan for initial staging at many institutions. In contrast to the situation with bone metastases, the sensitivity of PET for detection of lung metastases appears to be lower than that of thoracic CT [32,33], although integrated PET/CT is superior to CT alone in this regard [36]. PET may have greater utility for monitoring the response to chemotherapy and/or radiation therapy (particularly neoadjuvant chemotherapy), and in the postoperative evaluation for possible recurrence [37]. Nevertheless, PET or PET/CT is increasingly used for the initial staging of patients with Ewing sarcoma. Consensus-based guidelines from the National Comprehensive Cancer Network recommend a PET scan and/or bone scan for initial workup [38], and a baseline PET is recommended at presentation by the Children’s Oncology Group, if the primary bone tumor is negative on bone scintigraphy (table 2) [26]. Staging system — No commonly used staging systems exist for EFT as they do for other solid tumors. Although there are tumor, node, and metastasis (TNM) staging systems for primary tumors of both bone (table 4) [39,40] and soft tissue (table 5) [41] available from the Musculoskeletal Tumor Society and the American Joint Committee on Cancer (AJCC)/International Union Against Cancer (UICC), they are not in widespread use for EFT. An obvious deficiency of these staging systems is that they do not specify the primary site, which is one of the most important prognostic factors (see 'Prognostic factors' below). In addition, at least for soft tissue tumors, features of primary tumor classification in the 2002 edition (table 6) are not maintained in the 2010 version. Specifically, the inclusion 11 of both superficial and deep tumors within the same prognostic groups for tumors of the same size eliminates a criterion that previously stratified patient risk, without supporting data to suggest it is unnecessary. Laboratory studies — Initial laboratory studies should include a complete blood count, serum chemistries, and lactate dehydrogenase (LDH), which is a known prognostic factor in patients with EFT (see 'Prognostic factors' below) [9,42,43]. For cases in which neuroblastoma is in the differential diagnosis, urine catecholamine levels may be useful, since they are elevated in neuroblastoma but normal in EFT. Bone marrow aspirate and biopsy — Because of the predilection of EFT for spread to bone marrow [44], some clinicians advocate bone marrow biopsy (at least unilateral) in all patients to exclude widespread metastatic disease. Neither MRI nor bone scan are sufficient to evaluate the possibility of bone marrow metastases. Although MRI is sensitive to changes in the marrow, they are not specific. Any decrease in the marrow fat content may alter the signal, creating difficulty in differentiating tumor from active hematopoiesis. Tumor biopsy — The considerations for appropriate biopsy of a soft tissue or bone mass resemble those for other soft tissue and bone sarcomas. (See "Bone tumors: Diagnosis and biopsy techniques".) The surgeon should be consulted before any biopsy is carried out, and the procedure should be carefully planned to obtain adequate diagnostic tissue without compromising a later operation, particularly the opportunity for limb salvage. Biopsies should take place after the completion of imaging studies of the primary site, and the surgeon, radiation oncologist, medical or pediatric oncologist, and pathologist should review these studies in detail so that each member of the team is fully informed of the diagnostic considerations. Adequate amounts of tissue are necessary in order to provide sufficient diagnostic material. The extensive pathologic evaluation that is often required to ascertain the correct diagnosis within the group of "small round blue cell tumors" means that tissue is usually needed for special studies, and these samples require special handling. This topic is discussed in detail elsewhere. The diagnosis is most often established by CT guided core-needle biopsy (image 3). If only necrotic material is obtained on core biopsy, an open biopsy may be required. In either case, specimens should be obtained for microbiology to rule out osteomyelitis (see 'Differential diagnosis' above). Fine needle aspiration biopsy is not acceptable as the only diagnostic material, and should only be used to sample metastatic sites, or areas suspicious for recurrence when the histologic diagnosis is known. These issues are discussed in detail elsewhere. PROGNOSTIC FACTORS — Several clinical and biologic characteristics can assist in defining prognosis and directing the intensity of therapy [45]. These include the presence or absence of metastases, primary tumor location and size, age, the response to therapy, and the presence of certain chromosomal translocations. Disease extent — The key prognostic factor in ES is the presence or absence of metastases. Approximate five-year survival rates for patients with localized disease are 70 percent, while they average 33 percent for those who have overt metastases at diagnosis. The location of distant disease also impacts outcomes. In the European Intergroup experience, the five year relapse-free survival rates for patients with localized and metastatic disease at presentation were 55 and 21 percent, respectively [9] . Patients with bone and lung metastases fared significantly worse than those with bone metastases alone, who in turn, fared worse than those with isolated lung metastases. Approximately 30 percent of patients with metastases limited to the lungs will survive five years, as compared to only 10 percent of those with bone or bone marrow involvement. 12 Patients with limited pulmonary metastases who are candidates for resection may have a reasonable opportunity for cure. Tumor site and size — For patients presenting with localized disease, those with axial primary tumors (ie, pelvis, rib, spine, scapula, skull, clavicle, sternum) have a worse treatment outcome than those with extremity lesions [9,46]. In one series, the five year relapse-free survival rates were 40 versus 61 percent [9]. In addition, patients with small primary tumors (<100 mL) fare better than those with larger tumors [9,47]. Fever, anemia, and elevated serum LDH all correlate with a greater volume of disease and a poorer prognosis [9,20,42,43]. The poorer prognosis of large primary tumors, and those involving the pelvis and spine, is at least partly attributable to the difficulty in achieving wide negative resection margins, and higher rates of local failure after radiotherapy for large lesions. .) Extraosseous as compared to osseous origin does not have an adverse influence on survival. Origin has a significant adverse influence on outcome [11,48,49]. In fact, EFT that arise in skin or subcutaneous sites have a generally favorable prognosis [49-51]. Response to therapy — Patients with apparently localized disease have only a 10 to 20 percent likelihood of cure if treated with surgery or radiotherapy alone; this is improved dramatically when chemotherapy is added to treatment. (See "Treatment of the Ewing sarcoma family of tumors".) Both the completeness of surgical resection and the response to induction therapy are important prognostic factors. Patients who are left with significant amounts of viable tumor in the resected specimen following neoadjuvant chemotherapy do worse than those with minimal or no residual tumor [9,52-55]. These topics are discussed in detail elsewhere. .) Histology — In most but not all studies, the presence of neural differentiation (eg, as in PNETs) does not have an adverse influence on survival [10,56-58]. Age — Older age has been linked to a poor prognosis in some reports [59-61], but not others [62,63]. Younger children seem to have a better prognosis than older ones. As an example, in one report, the five-year relapse-free survival was significantly better for children younger than 10 compared to older children (86 versus 55 percent, respectively) [60]. The situation for adults is less clear. Although many series have reported a less favorable outcome in adults (particularly older adults [59]) as compared to children, a greater tumor bulk in adults or use of lower doses of alkylating agents [64] may explain some of these observations. Others have noted that adults fare as well as children, particularly when full dose chemotherapy protocols are utilized [65]. Molecular findings — The EFT are characterized by distinct nonrandom chromosomal translocations, which all involve the Ewing sarcoma (EWS) gene on chromosome 22. These translocations result in the fusion of distinct genes on different chromosomes, and these fused genes then encode hybrid proteins, which are thought to be involved in tumorigenesis. ( At least 18 different structural possibilities for gene fusions have been reported in these tumors. There are two sources of variability: the EWS fusion partner (eg, FLI1, ERG, ETV1, E1A, or FEV) and the breakpoint locations within the genes. This molecular heterogeneity may have some influence on the prognosis of EFT [66-69]. As an example, a better outcome has been reported for patients with localized tumors expressing the most common chimeric transcript (the so-called type I transcript in which EWS exon 7 is fused to FLI1 exon 6, which is present in about 60 percent of cases) compared to other fusion types [67,68]. However, more recent treatment protocols appear to have eliminated prognostic differences based upon fusion type [70]. Studies evaluating the prognostic significance of other cytogenetic and molecular alterations in EFT are limited. However, deletion of the short arm of chromosome 1p, 13 homozygous deletions of CDKN2A and p16/p14ARF, and p53 mutations have all been associated with poor response to chemotherapy and a worse prognosis [66,71,72]. The association of genetic variants of EFT with clinically significant features will require further study. Use of prognostic factors for treatment stratification — As noted previously, there is no widely accepted formal staging system for EFT. One of the most important uses of a staging system is to stratify patients for treatment purposes based upon expected outcome. Because of the very powerful prognostic power of the presence of metastatic disease, most studies in EFT have stratified patients by the presence or absence of metastatic disease. Although the primary site (ie, extremity versus pelvis or spine) also has prognostic significance, only a few studies have used site as the basis for determining the aggressiveness of therapy. Some investigators consider patients with localized disease, but who show features of poor prognosis, as "advanced" or "high-risk", grouping them for the purposes of treatment with others who have overt metastatic disease. The selection of treatment based upon individual prognostic factors is referred to as risk-adapted therapy [52]. The data indicating prognostic differences among patients with varying fusion genes raise the possibility that heterogeneity in the structure of chimeric transcripts may also be used to define clinically distinct risk groups. As an example, a direct comparison between treatment response and EWS fusion type may reveal a role of certain chimeras in therapy resistance. However, prospective clinical studies are just beginning to be performed to address this hypothesis. Inclusion of EWS fusion-type determination into future clinical trials is appropriate. SUMMARY — Ewing sarcoma (ES) and peripheral primitive neuroectodermal tumors (PNET) comprise the same spectrum of neoplastic diseases known as the Ewing sarcoma family of tumors (EFT), which also includes malignant small-cell tumor of the chest wall (Askin tumor). Because of their similar histologic and immunohistochemical characteristics and shared nonrandom chromosomal translocations, these tumors are considered to be derived from a common cell of origin, although its histogenic origin is debated. (See 'Introduction' above.) EFT most often arises in the long bones of the extremities (predominantly the femur, but also the tibia, fibula and humerus), and the bones of the pelvis. The spine, hands, and feet are affected considerably less often. (See 'Primary sites' above.) Patients with EFT typically present with localized pain or swelling of a few weeks or months duration. Trauma, often minor, may be the initiating event that calls attention to the lesion. Constitutional symptoms or signs, such as fever, fatigue, weight loss, or anemia, are present in about 10 to 20 percent of patients at presentation. (See 'Signs and symptoms' above.) The goals of the initial evaluation are to establish the diagnosis, evaluate local disease extent, and determine the presence and sites of metastatic spread. (See 'Staging evaluation' above.) The diagnostic work-up is usually initiated with a plain radiograph of the affected area. EFT involving bone typically presents as a poorly marginated destructive lesion with a permeative or “moth-eaten” appearance, most often associated with a soft tissue mass. While CT scan better delineates the extent of cortical destruction and soft tissue disease, definition of tumor size, local intraosseous and extraosseous extent, and the relationship of the tumor to fascial planes, vessels, nerves, and organs is best achieved by magnetic resonance imaging (MRI). (See 'Radiographic studies' above.) The metastatic work-up includes a CT scan of the chest to evaluate the thorax for metastatic disease, and a radionuclide bone scan to evaluate the entire skeleton for the presence of multiple lesions. PET or PET/CT is increasingly used for the initial staging 14 and response assessment of EFT. Some clinicians advocate bone marrow biopsy (at least unilateral) in all patients to exclude widespread metastatic disease. A biopsy should be carefully planned to obtain adequate diagnostic tissue without compromising a later operation, particularly the opportunity for limb salvage. Several clinical and biologic characteristics can assist in defining prognosis and directing the intensity of therapy. These include the presence or absence of metastases, primary tumor location and size, age, the response to therapy, and the presence of certain chromosomal translocations. Treatment of the Ewing sarcoma family of tumors 15 16 17 18 19 INTRODUCTION — Ewing sarcoma (ES) is a rare malignancy that most often presents as an undifferentiated primary bone tumor; less commonly, it arises in soft tissue (extraosseous Ewing sarcoma, EES). Both are part of a spectrum of neoplastic diseases known as the Ewing sarcoma family of tumors (EFT), which also includes the more differentiated peripheral primitive neuroectodermal tumor (PNET, previously called peripheral neuroepithelioma, adult neuroblastoma, and Askin's tumor of the chest wall) [1]. PNET can also present either in bone or soft tissue. Because these tumors share similar histological and immunohistochemical characteristics and unique nonrandom chromosomal translocations, they are considered to have a common origin [2-5]. In addition to their immunohistochemical and cytogenetic similarities, the EFT share important clinical features. These include a peak incidence between the age of 10 and 20 (70 percent of affected patients are under the age of 20), a tendency towards rapid spread to lungs, bone, and bone marrow, and responsiveness to the same treatments that include chemotherapy and radiotherapy. As with osteosarcoma (the other major sarcoma affecting bone), advances in multidisciplinary management over the past 30 20 years have resulted in a marked improvement in long-term survival. In data derived from the Surveillance, Epidemiology and End Results (SEER) program of the National Cancer Institute, five-year survival rates for patients with Ewing sarcoma rose from 36 to 56 percent during the periods 1975 to 1984 and 1985 to 1994 [6]. .) Here we will discuss the management of the EFT. The epidemiology, pathology, molecular genetics, clinical presentation, and diagnosis of these tumors, surgical principles, indications for limb sparing surgery, indications for radiation therapy (RT), and treatment of central (supratentorial) PNET tumors are discussed elsewhere. (See appropriate topic reviews). GENERAL TREATMENT PRINCIPLES — Despite the fact that fewer than 25 percent of patients have overt metastases at the time of diagnosis, EFT is a systemic disease. Because of the high relapse rate (80 to 90 percent) in patients undergoing local therapy alone, it is surmised that the majority of patients have subclinical metastatic disease at the time of diagnosis, even in the absence of overt metastases. Chemotherapy can successfully eradicate these deposits, and modern treatment plans all include chemotherapy, usually administered prior to and following local treatment. For patients with localized disease, the addition of several months of intensive multiagent chemotherapy to local therapy has had a dramatic impact on survival, and reported fiveand 10-year survival rates are now approximately 70 and 50 percent, respectively [716]. Even patients with advanced disease may be cured by multimodality therapy, although the long-term survival rates are clearly lower than for localized disease [17]. Approaching a child who has metastatic EFT with noncurative intent is rarely, if ever, appropriate, since it is not possible to predict a priori whose disease will be cured. Because of these issues, clinicians experienced in the treatment of Ewing sarcoma must direct the surgery and RT, and coordination with the medical oncologist is essential [11,18,19]. Adult patients — Fewer than 5 percent of cases of EFT arise in adults over the age of 40. There are no clinical trials that address treatment in adults; the vast majority of published studies have specifically excluded older individuals. The available information on treatment and prognosis is limited to six small single institution series of patients over the age of 15 who were treated for EFT (table 1) [20-25]. It is difficult to conclude from these data whether treatment outcomes are different in adults as compared to children. Early studies suggested worse outcomes for patients >15 years of age as compared to those 11 to 15 or under the age of 10 at the time of diagnosis (five-year relapse-free survival 43, 50, and 66 percent, respectively) [9]. However, only nine of the 331 individuals in this retrospective report were over the age of 20 at the time of diagnosis. The interpretation of studies addressing outcomes in adults is further complicated by the fact that adults more often have pelvic or extraskeletal primaries, which are associated with a poorer prognosis [20,22,23,25]. Despite this fact, adults who are treated with modern adjuvant and neoadjuvant chemotherapy for localized EFT may do as well as children [20]. Treatment of adults with EFT should be guided by the same general principles as are used for younger individuals. TREATMENT FOR LOCALIZED DISEASE Chemotherapy — Most modern treatment plans utilize initial (induction or neoadjuvant) chemotherapy followed by local treatment and additional chemotherapy. Reduction of local tumor volume is accomplished in the majority of patients, and this can facilitate resection. This is particularly important with regard to limb-sparing procedures for extremity lesions. Since most treatment failures are attributable to systemic metastatic 21 disease, local therapy considerations should never compromise the administration of effective systemic therapy. Although patients with EES have been treated in the past on protocols for rhabdomyosarcoma and have a similar response to multimodality therapy [13], both EES and PNET respond to the same chemotherapy regimens as osseous ES, and these EFT variants should be treated similarly [26-29]. Adjuvant treatment has evolved, largely due to the efforts of several cooperative groups: In the first Intergroup Ewing's Sarcoma Study (IESS-I), the combination of vincristine, doxorubicin, cyclophosphamide and actinomycin D (VDCA or VACA) was associated with a significantly better five-year relapse-free survival than vincristine, actinomycin D, and cyclophosphamide (VAC) alone or VAC plus adjuvant bilateral pulmonary irradiation (60 versus 24 versus 44 percent, respectively) [9]. Increasing the doxorubicin dose intensity during the early months of therapy further improved response, and in the second intergroup study (IESS-II), the fiveyear relapse-free survival rates using intermittent high dose four drug therapy improved to 73 percent for non-pelvic lesions [30]. Because of concerns about limiting the dose intensity of doxorubicin in regimens containing actinomycin D [31], actinomycin D was omitted from most trials thereafter, with no adverse impact on long-term outcome. Adding alternating cycles of ifosfamide (I) and etoposide (E) to a VDC backbone provides further benefit [14,32-34]. In the randomized IESS-III study, the addition of I/E to VDCA was associated with significantly better five-year relapsefree survival compared to VDCA alone (69 versus 54 percent, respectively) in patients with nonmetastatic, but not metastatic EFT or PNET [32]. A similar outcome was shown with a slightly different IE/VDC regimen at Memorial Sloan Kettering (four-year event-free survival 82 percent) [33] and with ifosfamide plus VDCA by the Orthopedic Institute from Rizzoli (five-year event-free survival 79 percent) [14]. Earlier trials from the Rizzoli group had suggested no additional benefit (five-year event-free survival rates 54 and 50 percent for VDCA plus I/E, and for VDCA alone in two consecutive phase II trials) [35]. As a result of these data, current standard chemotherapy for EFT in the United States includes vincristine, doxorubicin, and cyclophosphamide alternating with ifosfamide and etoposide (VDC/IE) (table 2) [32]. Typically, four to six cycles of chemotherapy are given before local therapy. So long as there is any sign of response to preoperative chemotherapy, additional cycles of the same treatments are given post-operatively, and the total duration of therapy is approximately 48 weeks. Relief of pain, decrease in tumor size, fall in LDH level, radiologic improvement, or evidence of necrosis in the resected specimen all argue for continued chemotherapy for those who can tolerate its sometimes considerable side effects. Dose intense chemotherapy — The sensitivity of EFT to alkylating agents, which have a steep dose-response curve, has prompted the evaluation of dose intense regimens in patients with poor risk disease. The majority of patients in these studies have relapsed or metastatic disease. The benefit of dose-intensive therapy for patients with poor risk localized disease is unclear. Dose escalation did not improve outcomes in a COG trial [36]. Furthermore, concerns for an increased risk of secondary malignancies in patients receiving dose-intense therapy have tempered enthusiasm for this approach. Another approach is “dose-dense” therapy as originally reported by Womer in 2000 [37]. A randomized trial by the Children’s Oncology Group (COG) randomized patients with localized Ewing sarcoma family of tumors to receive VDC/IE every 21 or 14 days, 22 respectively. This trial demonstrated a significant event-free survival (EFS) benefit for interval-compressed VDC/IE given every 14 days as compared to the same chemotherapy given at 21-day intervals (five-year EFS 73 versus 65 percent) [38]. The toxicity of both regimens was similar. The results of this trial suggest that interval compressed therapy is preferred for front-line treatment of children with localized Ewing sarcoma family of tumors. Whether this approach is preferred in adults is unclear. In a preliminary report of this COG randomized trial presented at the 2007 meeting of the Connective Tissue Oncology Society, the benefit of interval compressed therapy seemed limited to patients aged 17 or younger [39]. These data have not yet been reported in a peer-reviewed publication. Local treatment — Local control for EFT can be achieved by surgery, radiation, or both. The choice of RT or surgery usually represents a tradeoff between functional result, and the risk of secondary radiation-induced malignancy. Patients who lack a functionpreserving surgical option because of tumor location or extent may be recommended RT. However, surgery is preferred for potentially resectable lesions and for those arising in dispensable bones (eg, fibula, rib, small lesions of the hands or feet) for the following reasons: It avoids the risk of secondary radiation-induced sarcomas (see "Radiation therapy for Ewing sarcoma family of tumors"). An analysis of the degree of necrosis in the excised tumor can permit refinements in the estimate of prognosis. In the skeletally immature child, resection may be associated with less morbidity than RT, which can retard bone growth and cause deformity. Although there are no randomized trials comparing surgery with RT for local control, multiple retrospective series suggest superior local control and survival for surgery compared to RT alone. However, selection bias may account for at least some of these results (ie, smaller, more favorably situated peripheral tumors are more likely to be resected while larger, axial lesions are radiated). Radiation dose and proper field planning are also important factors in local control. The role of RT in the local management of Ewing sarcoma is discussed separately. The surgical principles that apply to resection of the primary tumor and reconstruction are similar to those in patients with osteosarcoma and are discussed elsewhere. However, there are three main differences between EFT and osteosarcoma: EFT are radiosensitive while osteosarcoma is much less so. EFT tend to occur in a younger population, where skeletal immaturity and concern for radiation-induced growth inhibition must be considered. EFT tend to arise in different areas of the long bones than osteosarcoma (the diaphysis compared to epiphysis, (figure 1 and figure 2)) [40]. Adjuvant RT — RT is an essential component of therapy for patients undergoing resection if the surgical margins are inadequate, although effective chemotherapy can also reduce the risk of local failure in such patients [9,15]. RT is usually avoided in patients without residual disease, to avoid exposing them to the risk of a radiationinduced malignancy. TREATMENT FOR ADVANCED DISEASE — Patients with overt metastatic disease at presentation have a significantly less favorable outcome than those with localized disease. However, aggressive multimodality therapy can relieve pain, prolong the progression-free interval, and cure some patients of their disease. In a review of 12 different series in which patients with metastatic EFT were predominantly treated with chemotherapy, five year event-free and overall survival rates averaged 25 (range 9 to 23 55) and 33 (range 14 to 61) percent, respectively [7,11,41-50]. The small numbers of patients in each series and the heterogeneity in location and extent of metastatic disease probably accounts for these wide variations in outcome. In particular, the site of metastatic disease is an important variable. For patients with isolated lung and pleural metastases, cure rates up to 40 percent are reported with multimodality therapy; for metastases involving bone or bone marrow, cure rates fall to 20 to 25 percent and for combined sites, to less than 15 percent [49]. Because it can be difficult to predict which patients with metastatic disease will be long-term relapse-free survivors [42], treatment should be administered with curative intent. Several issues are pertinent to patients with metastatic EFT: What is the optimal initial chemotherapy regimen? Does dose escalation provide benefit in the initial treatment phase? Is there a role for dose-intense therapy as consolidation after initial therapy? Is RT to sites of metastatic involvement of any benefit? How should control of the primary site be approached? Chemotherapy — Patients with disseminated disease at diagnosis often respond well to the same type of systemic chemotherapy as that used for localized disease. Standard treatment options include conventional doses of doxorubicin plus cyclophosphamide, vincristine, doxorubicin, and cyclophosphamide with or without actinomycin D (VDCA or VDC), alternating courses of VDC and ifosfamide plus etoposide (I/E), regimens incorporating higher doses of alkylating agents, and more intensive multiagent combinations that generally require hematopoietic support. Unfortunately, there are few conclusions that can be garnered from the published literature as to the superiority of any of these regimens, because there are no randomized studies that directly compare different regimens. Conventional dose Doxorubicin plus cyclophosphamide – The best results for doxorubicin and cyclophosphamide were obtained by investigators at St. Jude using both agents sequentially [43]. Ten of 18 patients (55 percent) receiving oral cyclophosphamide (150 mg/m2 daily for seven days) followed by doxorubicin (35 mg/m2) who delayed surgery and delayed lower dose limited field radiation were alive and event-free at a median 47 months of follow-up. However, the same regimen administered every three weeks resulted in a five year EFS of only 21 and 23 percent in two subsequent trials by the Pediatric Oncology Group (POG) and the French Society of Pediatric Oncology [11,51]. VDCA – Results from two larger Intergroup Ewing's Sarcoma studies for patients with metastatic disease (IESS-MD-I and II) using VDCA seem superior to these results [50]. In the first trial (IESS-MD-I), conducted from 1975 to 1977, 53 eligible patients received VDCA with concomitant RT to areas of gross disease, while in IESS-MD-II, conducted from 1980 to 1983, 69 eligible patients received 5-fluorouracil (5-FU) in addition to VDCA, and XRT was delayed until week 10 [50]. In both studies, the best response rate (complete and partial remissions combined) was similar (73 and 70 percent in IESS-MD-I and II, respectively), as was the response duration (approximately 30 percent of the patients on either study survived five years or more), leading the authors to conclude that 5-FU was not beneficial. Contrasting results were noted by others, using the same VDCA regimen and RT to the primary site with or without surgery [46]. Only 9 percent of patients with metastatic disease were relapse-free at five years; however, RT was not 24 administered to sites of metastatic disease (pulmonary in 10 of 22 patients). (See 'Pulmonary metastases' below.) VDC plus standard dose I/E – In contrast to data for nonmetastatic disease, specific benefit for the addition of I/E to the VDC backbone has not been shown for patients with metastatic disease at diagnosis [32,34,44,45,52,53]. Higher doses administered with conventional scheduling (ifosfamide 6 to 9 g/m2 and doxorubicin 60 mg/m2), or interval compression of standard dose VDC/IE chemotherapy using every two as compared to every three week cycling in conjunction with hematopoietic growth factor support (as has been used for high-risk localized disease) has resulted in possibly better results, particularly in patients with isolated pulmonary metastases. Up to 40 percent of such patients are still alive and event-free at five years; results are less favorable in patients with nonpulmonary metastases [37,41,45,49]. Lung irradiation was used in the majority of patients in these trials. (See 'Dose intense chemotherapy' above.) High-dose chemotherapy — Because of the failure to cure the majority of patients with metastatic disease at diagnosis and the steep dose response curve that exists for alkylating agents, more intensive therapies incorporating high-dose chemotherapy with hematopoietic cell support (HDT) have been studied. Induction phase – Initial treatment with very high doses of chemotherapy with or without total body irradiation and autologous hematopoietic stem cell support has been reported to improve survival in some small nonrandomized series [5459] but not others [60-65]. Variability in the definition of high risk patients and in the regimens used makes it difficult to compare results and arrive at any firm conclusions regarding the benefit of this approach. Allogeneic stem-cell transplantation offers no advantage over autologous transplantation [60] and has a higher complication rate. (See 'Secondary malignancies' below.) Consolidation phase – Another approach is to administer HDT as consolidation following standard dose chemotherapy and optimal local treatment. A number of small trials which include subsets of patients with EFT suggest benefit [45,56,58,66-69], but this is not universal. The following represents the range of findings: Some of the most favorable results were reported in a prospective study of 97 unselected patients with newly diagnosed metastatic EFT/PNET, who underwent conventional chemotherapy; those with a complete or partial response (n = 75) went on to HDT/autologous transplant [70]. The five-year event-free survival rate after HDT was 47 percent; for those with isolated pulmonary metastases it was 52 percent. On the other hand, disappointing results were shown for patients with metastatic EFT in a large prospective trial conducted by Children's Cancer Group, in which 32 patients with bone or marrow metastases received induction therapy (three cycles of VDC alternated with IE) and surgery and/or RT for local control (and RT to sites of metastatic disease), followed by TBI over three days, high-dose melphalan and etoposide, and peripheral blood stem cell reinfusion [71]. The two-year event-free survival rate was only 16 percent, suggesting that this approach did not represent a therapeutic advance. Similar disappointing results (five-year survival 21 percent) were noted among patients with metastatic EFT reported by the European Bone Marrow Transplant Solid Tumor Registry [72]. Thus, given the mixed results seen to date in patients who present with advanced disease, high-dose chemotherapy with autologous stem cell transplantation remains an investigational approach and should be considered only in the context of a clinical trial. The ongoing EURO-EWING 99 trial is comparing high-dose chemotherapy versus 25 standard approaches in a variety of clinical settings (including patients with metastatic disease at diagnosis and those with localized tumors and a poor initial histologic response to induction chemotherapy) [73]. It is hoped that the results will provide some insight as to which patients benefit from this approach. Secondary malignancies — Although the majority of EFT survivors have some late effects of therapy [74], an additional detrimental impact of high dose chemotherapy regimens is the risk of secondary myelodysplasia (MDS) and leukemia [60,65,75,76]. This was illustrated in a report from the Children's Oncology Group of 578 children with EFT/PNET who were treated with three different regimens over a six-year period [75]. Overall, 11 children developed secondary MDS/AML, and the cumulative risk was significantly higher among children treated with a regimen incorporating higher doses of doxorubicin, cyclophosphamide, and ifosfamide as compared to those receiving standard dose VDCA with or without IE (11, 0.9, and 0.4 percent at five years, respectively). Surgery and radiation therapy — Outcomes are best when chemotherapy is combined with radiation and sometimes resection of sites of gross metastatic disease [50,61,77,78]. This multimodality approach has been most successful for patients with limited pulmonary metastases. Pulmonary metastases — Patients with a limited number of lung metastases do not share the same dismal prognosis of metastatic disease at other sites (ie, bone or bone marrow), because of the ability to accomplish wide resections both easily and repeatedly. In such patients, chemotherapy is a necessary component of therapy, and between 20 and 40 percent five-year survival can be achieved [49,79,80]. For patients with more advanced pulmonary disease at diagnosis, supplemental radiation may provide benefit, even after a complete response to chemotherapy. Most of the available data supporting radiation in this setting are from nonrandomized series, and patient selection factors confound interpretation of the data. Nevertheless, low dose bilateral lung irradiation (15 to 20 Gy, in daily 1.5 to 2.0 Gy fractions) can control gross metastatic disease in the lungs without significant pulmonary toxicity and is usually recommended after chemotherapy, despite the paucity of data. This topic is discussed in detail elsewhere. Bone and soft tissue metastases — Patients with solitary or circumscribed bone or soft tissue lesions can be irradiated at those sites, usually to doses of 45 to 56 Gy, in addition to irradiation of the primary tumor. However, the likelihood of long-term survival is considerably lower than for patients with isolated pulmonary metastases. Local control — Local control can pose major issues, even for patients with metastatic disease. The complete response rate to chemotherapy can be increased with subsequent radiation therapy or selected excision for all sites of evident disease; however, long-term relapse-free survival is still only achieved in the minority of patients [42]. Diffuse metastatic disease can make it difficult to justify a large resection. In selected patients, resection may be reconsidered if chemotherapy results in significant volume reduction, particularly if areas of small volume metastatic disease are also amenable to surgical resection. On the other hand, radiation therapy, which is more often considered in patients with metastatic disease, will usually provide adequate local control with acceptable morbidity. If substantial amounts of bone marrow will need to be included in the radiation treatment volume, then radiating the primary tumor and/or metastases may be delayed until the end of systemic therapy to avoid interfering with chemotherapy. POSTTREATMENT SURVEILLANCE — For all EFT of bone or soft tissue, there are no prospective data that address the appropriate schedule or selection of tests for surveillance after initial treatment for localized disease. Consensus-based guidelines from the National Comprehensive Cancer Network (NCCN) and from the Children's Oncology Group recommend physical examination, a complete blood count, chest 26 imaging, and local imaging of the primary site every three months for two years, every four months for year three, every six months for years 4 and 5, and annually thereafter (table 3) [81,82]. The appropriate duration of follow-up is unknown; however, the vast majority of recurrences are observed within 10 years. Physicians performing posttreatment surveillance must be cognizant of concerns for radiation exposure and the risk for secondary malignancies, particularly in younger individuals. RECURRENT DISEASE — The majority of relapses occur within five years, but late relapse is not uncommon [83-85]. In a report from the Childhood Cancer Survivor Study, the 20-year cumulative incidence of a late recurrence among five-year survivors of Ewing sarcoma was 13 percent [85]. For this reason it is advisable that patients be followed for the potential of late relapse indefinitely. There are no definite protocols for this, but the authors suggest that an examination and chest radiograph be performed every two to five years for the life of the patient. Symptoms at the primary site or elsewhere should raise concern and be appropriately investigated. In general, survival after an early relapse is poor, with few survivors among those who relapse within two years of therapy. In contrast, up to 15 to 20 percent of those who relapse later may survive long-term [86,87]. Other prognostic factors for death in patients with recurrent Ewing sarcoma include recurrence at combined local and distant sites, and an elevated lactate dehydrogenase (LDH) at initial diagnosis [86,87]. Evaluation — Patients with a suspected recurrence should undergo evaluation of both the primary site and for the presence of metastatic disease before a treatment plan is formulated. The majority of patients with a local recurrence have either gross or microscopic metastatic disease. The documentation of a local recurrence can be difficult. The interpretation of irradiated areas on imaging studies can be challenging, and soft tissue masses may represent residual fibrosis rather than recurrent tumor. The evaluation of intraosseous sites is even more difficult, since the response to prior therapy and variability in reossification complicates the interpretation of radiographic studies. Progressive cortical destruction or increasing lytic areas suggest local recurrence, as do bone scans that demonstrate increased uptake. Bone biopsies are associated with local morbidity (ie, wound complications and bone fracture) and are subject to a high rate of sampling error. Treatment — Although the prognosis for patients with recurrent disease is poor, some patients can be successfully salvaged [88,89]. The sites of recurrence, prior treatment, and relapse-free interval affect remaining treatment choices. Management of a local recurrence usually includes surgery (and possibly an amputation if the local recurrence involves an irradiated extremity), radiotherapy, or both. Radiation therapy to bone lesions usually provides pain relief, while surgery can cure some cases with limited isolated lung metastases [79]. The likelihood of response to chemotherapy retreatment increases with longer duration of disease control. Late relapsers do better than those who progress during initial chemotherapy (primary refractory disease). In one report, 26 of 34 patients relapsing off treatment attained a second remission compared to none of 18 who had primary refractory disease [90]. Twelve of the 26 responders remained in second remission for periods up to 19 years. The role of intensive myeloablative regimens is unclear, although a few patients may be salvaged [59]. The choice of chemotherapy regimen depends upon initial treatment and the duration of the relapse-free interval. Patients relapsing after a disease-free interval off chemotherapy may respond again to the same agents. Patients who have not previously received ifosfamide and etoposide may respond to this regimen [91]. At our center, the regimen used by the third Intergroup Ewing's Sarcoma Study is typically chosen [32], but doses are escalated as tolerated if there is not a prompt response. Most patients 27 with relapsed disease will be treated at centers that have their own individualized approach. Clinical trials investigating new agents (eg, monoclonal antibodies and small molecule inhibitors targeting the insulin-like growth factor-I (IGF 1)-receptor [92-95]) and new combinations (eg, irinotecan plus temozolomide [96]) warrant consideration, since completion of these trials is crucial to improving outcomes. Although most of the available phase I or I/II studies do not specifically target advanced EFT and most phase III studies target early rather than advanced stage or recurrent disease, opportunities remain [97]. The National Cancer Institute clinical trials website provides information about ongoing clinical trials Future therapies will probably emerge from the exploitation of the molecular targets accompanying the unique genetic alterations that are held responsible for these malignancies [17,98]. COMPLICATIONS IN LONG-TERM SURVIVORS — Although the survival of patients with Ewing sarcoma has improved steadily since the 1970, long-term survivors have considerable future health risks [16,99-102]. These include second malignancy, pathologic fractures, other radiation-associated complications (wound complications, pulmonary fibrosis, neuropathy, limb leg discrepancy, femoral head necrosis), and chemotherapy-related complications (renal insufficiency, bowel toxicity, neuropathy, and cardiomyopathy) [99]. The health status of long-term (≥5 years) survivors was addressed in a cohort study of 568 individuals who were diagnosed with Ewing sarcoma before age 21 from 1970 to 1986, including a subset of 403 patients who were participating in the Childhood Cancer Survivor study [102]. Cumulative mortality among all survivors was 25 percent at 25 years after diagnosis, and the cumulative incidence of secondary malignancy was 9 percent. Disease progression/recurrence accounted for 60 percent of all deaths, while other causes included secondary neoplasms, cardiac disease, other medical causes, and pulmonary disease. The cumulative mortality attributed to subsequent malignant neoplasms and cardiopulmonary disease potentially attributed to treatment was 8.3 percent at 25 years. In addition, compared with their siblings, survivors had significantly higher rates of severe, disabling, or chronic health conditions, significantly lower fertility rates, and higher rates of self-reported moderate to extreme adverse health status. It is anticipated that the adoption of tailored radiation ports in the last 15 years (as opposed to whole bone ports used in the period 1960 to 1980) and use of lower and risk-adopted radiation therapy doses as well as an appreciation of the rise in radiationinduced second malignancy risk with doses above 60 Gy will reduce the risk of late effects. Nevertheless, the relatively high complication rates seen with many of these earlier treatment approaches, the delayed nature of many of the complications, and the possibility that trends in chemotherapy intensification may alter the pattern of secondary malignancies [100] underscore the need for long-term follow-up. SUMMARY Patients with Ewing sarcoma or another of the Ewing sarcoma family of tumors (EFT) require referral to centers that have multidisciplinary teams of sarcoma specialists. With rare exception, systemic combination chemotherapy and definitive local therapy is required in all patients, and care should be coordinated among the medical oncologist, surgeon, and radiation therapist. In most cases, treatment will begin with chemotherapy. The primary tumor is then most often treated with surgical resection. Radiation, or a combination of radiation 28 and surgery, is reserved for cases in which negative margins cannot be achieved while still preserving function. The choice between surgery and RT is dictated by the age of the patient, the location and size of the primary, and functional as well as long-term consequences of therapy (ie, radiation-induced growth inhibition or secondary malignancy). Following local treatment, chemotherapy is usually continued, typically for several months. Thus, the total duration of therapy ranges from 10 to 12 months. Patients with clinically detectable metastatic disease at presentation, and those who relapse after initial therapy also require multimodality therapy. All patients with advanced disease should be approached with potentially curative treatment. Up to 40 percent of patients with limited pulmonary metastatic disease who undergo intensive chemotherapy and pulmonary resection with or without radiation therapy may be long-term survivors. The prognosis for other subsets of patients with advanced disease is less favorable. There is no conclusive evidence that high-dose therapy with or without hematopoietic stem cell infusion at any point during treatment is beneficial for patients with poor-risk localized and metastatic EFT. Most patients with advanced or recurrent disease need new approaches to improve outcomes, and participation in clinical trials should be encouraged. Long-term follow-up is needed following therapy because disease relapse, treatment-related complications, and second malignancies all occur beyond five years after treatment is initiated. Radiation therapy for Ewing sarcoma family of tumors 29 INTRODUCTION — Ewing sarcoma (ES) is a rare malignancy that most often presents as an undifferentiated primary bone tumor; less commonly, it arises in soft tissue (extraosseous Ewing sarcoma, EES). Both are part of a spectrum of neoplastic diseases known as the Ewing sarcoma family of tumors (EFT), which also includes the more differentiated peripheral primitive neuroectodermal tumor (PNET, previously called peripheral neuroepithelioma, adult neuroblastoma, and Askin's tumor of the chest wall) [1]. PNET can also present either in bone or soft tissue. Because these tumors share similar histological and immunohistochemical characteristics and unique nonrandom chromosomal translocations, they are considered to have a common origin. EFT also share important clinical features, including a peak incidence between the age of 10 and 20, a tendency towards rapid spread to lungs, bone, and bone marrow, and responsiveness to the same chemotherapeutic regimens and radiotherapy (RT). Because relapse rates are high in patients undergoing local therapy alone (80 to 90 percent), it is surmised that the majority have subclinical metastatic disease at the time of diagnosis, even in the absence of overt metastases. The routine administration of intensive multiagent chemotherapy, which can eradicate these deposits, has had a dramatic impact on outcomes. Five-year survival rates for patients with Ewing sarcoma in the United States rose from 36 to 56 percent during the periods 1975 to 1984 and 1985 to 1994 [2]. Local control at the primary tumor site can be accomplished by surgery, RT, or both. The choice of modality usually represents a tradeoff between functional result and treatmentrelated morbidity, particularly the risk of a secondary radiation-induced malignancy. Although modern treatment protocols emphasize surgery for optimal local control, 30 patients who lack a function-preserving surgical option because of tumor location or extent, and those who have clearly unresectable primary tumors following induction chemotherapy are appropriate candidates for RT. Since more than 90 percent of patients with EFT have either detectable or subclinical metastases at diagnosis, local therapy, if delivered correctly, is probably not the critical event in determining survival. However, if local therapy is delivered poorly or given in such a way that it compromises the delivery of adequate chemotherapy, survival can be greatly compromised. Moreover, local failure is associated with a very poor survival outcome [3]. Here we will discuss the role of radiation therapy in the local management of the EFT. Epidemiology, pathology, molecular genetics, clinical presentation, and diagnosis of these tumors, as well as surgical principles and the use of chemotherapy are presented elsewhere. Central (supratentorial) PNET tumors are also discussed elsewhere. RT FOR LOCAL CONTROL OF THE PRIMARY General principles — The radiosensitivity of Ewing sarcoma was first noted in the original description by James Ewing [4]. Historically, RT was preferred over surgery because of less acute morbidity, acceptable rates of local control (50 to 77 percent), and poor longterm prognosis, largely due to distant dissemination in the absence of effective chemotherapy [5]. The introduction of multiagent chemotherapy into the management of EFT in the 1970s led to an increase in cure rates as well as higher local control rates after RT (72 to 90 percent) [6-11]. As more children survived their tumors, balancing long-term local disease control with long-term functional results assumed greater importance. The recognition of late effects of RT (eg, fracture, second malignant neoplasms) as well as improvements in surgical techniques resulted in a shift towards the use of surgery rather than RT. Current protocols tailor local treatment to the individual patient with the goal of maximizing local tumor control while minimizing treatment-related morbidity. In general, patients are selected for local therapy in such a way that they are treated with surgery or RT but not both, since the combined use of surgery and RT places patients at risk for morbidity from both modalities [12]. However, there is a role for combined therapy in some circumstances, particularly for large tumors (particularly involving the pelvis), and for cases in which resection margins are positive or close. (See 'Adjuvant RT' below.) RT versus surgery — Both surgery and RT represent effective local treatment for Ewing sarcoma. There are no randomized trials which directly compare both modalities, and their relative roles continue to be debated. Contemporary treatment guidelines emphasize surgical resection as the local control modality of choice if it is believed that the lesion can be resected with negative margins, without excessively morbidity, and with the expectation of a reasonable functional result. Surgery is preferred for potentially resectable lesions, and for those arising in dispensable bones (eg, fibula, rib, small lesions of the hands or feet) for the following reasons: It avoids the risk of secondary radiation-induced sarcomas. An analysis of the degree of necrosis in the excised tumor can permit refinements in the estimate of prognosis. In the skeletally immature child, resection may be associated with less morbidity than radiation, which can retard bone growth and cause deformity. Thus, in most cases, the decision to use RT for treatment of the primary site is made only after review of the potential surgical options for a given lesion. Patients who lack a function-preserving surgical option because of tumor location or extent, and those who 31 have clearly unresectable primary tumors following induction chemotherapy are appropriate candidates for RT. Influence of tumor site — When considering the local control strategy for an individual patient, the need to attain complete tumor eradication must be weighed against the twin goals of maximizing function and minimizing long-term morbidity. Although treatment decisions must be individualized according to the expected deficits from surgical resection at any site, the following generalizations can be made regarding the influence of tumor site on the choice of local treatment (see "Bone sarcomas: Preoperative evaluation, histologic classification, and principles of surgical management"): It is generally agreed that surgery is preferred for lesions arising in dispensable bones (eg, fibula or rib). Tumors affecting the long bones of the leg, distal humerus, or ulna can usually be resected and reconstructed using intercalary techniques (allografts, autografts, or metallic prostheses) or joint replacement, depending on tumor location. On the other hand, primary tumors involving the proximal humerus and upper scapula may be best treated with RT, since limb reconstruction is difficult and shoulder morbidity may be substantial. Patients with lesions of the skull, facial bones, or vertebrae are often candidates for nonsurgical treatment because of the difficulty in achieving negative margins without substantial functional deficit [13]. A number of series report better results with surgery over RT alone for pelvic bone tumors [14-18], but this is a controversial area. Surgical treatment for pelvic sarcomas, which are often bulky, is challenging. Lesions of the iliac wing, ischium, or pubis can usually be resected after a good response to chemotherapy without substantial functional morbidity. For tumors of the periacetabular region and those that cross the sacroiliac joint, there is less enthusiasm for surgical resection given the significant resulting functional deficit. Outcomes — Because no randomized trial has directly compared both modalities, only a relative comparison can be made from retrospective reports and the prospective trials that have mainly tested different multiagent chemotherapy regimens. In many retrospective series, rates of local control and survival are superior after surgery compared to RT alone [9,10,16,19-30]. However, larger cooperative group studies have failed to reflect this advantage, and selection bias likely accounts for at least some of these results. Smaller, more favorably situated peripheral tumors are more often resected while larger, axial lesions (which have a higher rate of local failure in most series and a poorer prognosis overall [9,20,21,31]) are referred for RT. Furthermore, RT delivery techniques may also have contributed to poorer outcomes in early series. As an example, in the Cooperative Ewing's Sarcoma Study (CESS)-81 trial, relapse-free survival (RFS) for patients undergoing irradiation of the primary site in the initial phase of the trial was only 55 percent, significantly lower than for surgically treated patients [22]. On review, the high rate of local failure after RT alone (50 percent) was attributed to geographic "miss". After a quality assurance program was initiated, local control rates improved significantly, and RFS increased to 80 percent. In a subsequent study, CESS-86, which included central quality assurance, the five-year RFS rates following chemotherapy plus either RT or surgery were 67 and 65 percent, respectively [6]. The importance of RT dose and field planning to local control is discussed below. Several studies support the adequacy of RT alone for local control: One retrospective report included 39 patients who were treated between 1975 and 1991 at two Boston hospitals with multiagent chemotherapy and varying 32 combinations of surgery and RT [32]. Twenty had RT alone, 16 both surgery plus RT, and 3 surgery alone. Despite the use of combination chemotherapy, five-year disease-free survival (DFS) was only approximately 30 percent, emphasizing the high-risk nature of these mainly axial lesions. However, local control rates were similar among patients treated with RT alone or surgery (85 and 84 percent, respectively). Another case series included 60 patients treated for EFT at Memorial Sloan Kettering between 1990 and 2004; 31 underwent RT as the sole modality for local control, the remainder in conjunction with surgery [33]. The relatively poor prognosis of this group is reflected by the fact that most were centrally located tumors (spine, pelvis, proximal extremities), 52 percent were ≥8 cm in size, and one-third were metastatic at diagnosis. Nevertheless, with a median follow-up of 41 months, DFS and overall survival rates for patients with nonmetastatic disease at presentation were 70 and 86 percent respectively. There were only nine local recurrences, five of which developed in patients who presented with overt metastases. The three-year actuarial local control rates for the entire group and those undergoing definitive RT alone were 77 and 78 percent, respectively. The presence of overt metastases emerged as the only significant adverse prognostic factor for local control, a finding that has been confirmed by others [9,34]. Long-term treatment-related complications were minimal, and there were no secondary malignancies. (See 'Late effects' below.) The relationship between local control modality (surgery, RT, or both) and the subsequent risk for local failure was examined in a subgroup of 75 patients treated for nonmetastatic pelvic EFT in US Intergroup study 0091 (IESS III), a randomized comparison of VACA (vincristine, doxorubicin, cyclophosphamide, and dactinomycin) with or without alternating IE (ifosfamide plus etoposide) [35]. Choice of local control modality was left to the treating clinicians; 12 underwent surgery, 44 received RT, and 19 received both. The rates of five-year, event-free survival (EFS) and cumulative incidence of local failure (as any component of first failure) in the entire group were 49 and 21 percent, respectively. There was no significant difference in either endpoint according to tumor size (<8 versus ≥8 cm) or choice of local control modality. However, there was a trend toward improved local control in patients receiving VACA/IE as compared to VACA alone (cumulative incidence of local failure 11 versus 30 percent; p = .06), which was present regardless of the local control modality. Although not a universal finding [6,33,36], many studies suggest that smaller rather than larger tumors are more likely to be locally controlled after RT alone [9,15,19,3741]. As an example, in a report of 100 patients with localized Ewing sarcoma of bone, local control rates were significantly higher after RT for the 14 lesions that were ≤100 cm3 in volume compared to larger lesions (93 versus 60 percent, respectively) [38]. Adjuvant RT — RT is usually not recommended for patients who undergo a complete resection in order to avoid exposing them to the risk of a secondary malignancy. Although indications for postoperative RT are not firmly established, the following are common scenarios in which RT is recommended in conjunction with surgery for EFT [42]: For bulky tumors in difficult sites such as the pelvis, combined surgery and RT might allow for a more limited surgical procedure, better functional outcome, and enhanced local control as compared to single modality therapy [9,15,33,43]. RT can be given either preoperatively or postoperatively, based upon institutional protocols and experience. 33 Patients who are left with macroscopic amounts of viable tumor in the resected specimen following neoadjuvant chemotherapy have worse survival than those with minimal or no residual tumor [10,20,37,41,44-47]. There are no data evaluating the effectiveness of adjuvant RT in this situation; however, it can be considered if there is concern about the adequacy of the margins. The role of adjuvant thoracic irradiation is unsettled. The United States Intergroup Ewing's Sarcoma Study (IESS)-I trial showed that bilateral whole lung irradiation was an effective adjuvant treatment for patients with localized EFT, but prolonged follow-up favored four-drug multiagent chemotherapy [8]. Prophylactic whole lung irradiation has not been studied in any subsequent trial, and no studies have used it in addition to modern combination chemotherapy regimens [48], except in patients with lung metastases at diagnosis. On the other hand, adjuvant hemithorax irradiation improves outcomes in patients with high-risk chest wall primary tumors (close or involved margins, initial pleural effusion, pleural infiltration, and intraoperative contamination of the pleural space) [49,50]. Postoperative RT is usually recommended if there is residual microscopic or gross disease after surgery, or inadequate surgical margins. Inadequate resection margins (ie, a marginal or intralesional resection (table 1) are associated with a worse outcome as compared to radical or wide resection) [27,51]. The impact of surgical margins on outcome was described in a series of 244 patients who were enrolled in CESS studies for localized EFT and underwent surgery as a component of local therapy [27]. Ninety-four had definitive surgery alone, 131 received postoperative RT, and 19 had preoperative RT. Surgical margins were radical, wide, marginal, and intralesional in 29,148, 39, and 28 patients, respectively. The local and combined (local plus systemic) relapse rate was significantly lower after a complete resection (radical or wide) with or without RT than after an incomplete resection (marginal or intralesional) with or without RT (5 versus 12 percent, p = 0.045). However, the relapse rate was not significantly lower in patients who received RT after incomplete surgery compared to those who did not (12 versus 14 percent). Ten-year survival rates for the cohorts undergoing radical, wide, marginal, and intralesional resection were 58, 65, 61, and 71 percent, respectively; the differences were not statistically significant. However, the data to support the benefit of RT in this setting are surprisingly scant, and conflicting: The efficacy of combined surgery and RT was evaluated in a retrospective series of 39 patients treated for localized EFT at the St Jude Hospital between 1978 and 2001 [52]. With combined local therapy, local control rates were excellent, even for patients with positive surgical margins (eight-year local failure rates for patients with positive and negative surgical margins were 17 and 5 percent, respectively), and the corresponding overall survival rates were 71 and 94 percent, respectively. On the other hand, a lack of benefit for adjuvant RT was suggested in a large singleinstitutional retrospective series, in which the addition of moderate dose adjuvant RT (45 Gy) to patients with inadequate margins did not seem to improve local control or overall disease-free survival. The authors concluded that patients for whom inadequate margins were anticipated at the time of preoperative evaluation might be better treated by full dose radiotherapy alone [53]. Summary — Although most clinical protocols for EFT emphasize surgery for treatment of the primary lesion, RT is an effective option for local control in patients who lack a function-preserving surgical option because of tumor location or extent, and those who have clearly unresectable primary tumors despite induction chemotherapy. If a tumor 34 appears to be categorically unresectable following induction therapy, debulking surgery should be avoided and the patient referred for definitive RT. Indications for adjuvant RT in patients who have undergone resection of a EFT include: Bulky tumors in difficult sites (eg, the pelvis); in this setting, RT can be given either preoperatively or postoperatively, based upon institutional protocols and experience. Our own preference, when positive margins are considered likely, is to use preoperative irradiation. If there is residual microscopic or gross disease after surgery, or inadequate surgical margins Patients who are left with significant amounts of viable tumor in the resected specimen and less than wide margins following neoadjuvant chemotherapy Adjuvant hemithorax irradiation is indicated in patients with high-risk chest wall primary tumors (close or involved margins, initial pleural effusion, pleural infiltration, and intraoperative contamination of the pleural space) RT FOR METASTATIC DISEASE — Patients with overt metastatic disease at presentation have a significantly less favorable outcome than do those with localized disease. However, aggressive multimodality therapy can relieve pain [54], prolong the progression-free interval, and provide long-term relapse-free survival (and possible cure) in some patients. Management of the primary site — Local control can pose major issues, even for patients with overt metastatic disease. In such cases, it may be difficult to justify a large resection of the primary site because of the poor long-term prognosis [15,55]. However, resection may be reconsidered in selected patients if chemotherapy results in significant volume reduction, particularly if areas of small volume metastatic disease are also amenable to surgical resection. On the other hand, RT can provide adequate local control with acceptable morbidity [6,21,22,56]. If substantial amounts of bone marrow will need to be included in the field, RT may be delayed until the end of systemic therapy to avoid interfering with chemotherapy. Pulmonary metastases — Patients with a limited number of lung metastases do not share the same dismal prognosis of metastatic disease at other sites (ie, bone or bone marrow), because of the ability to accomplish wide resections both easily and repeatedly. Chemotherapy is a necessary component of therapy. In highly selected patients, between 20 and 40 percent five-year survival can be achieved [57-60]. Retrospective reports from large cooperative groups suggest that the addition of lowdose bilateral lung irradiation (15 to 20 Gy) benefits patients with EFT presenting with pulmonary metastases, even if all lesions are resected or completely respond to chemotherapy [8,22,48,57,59-63]. In the CESS and EICESS trials, the rate of pulmonary relapse was reduced by 50 percent over that for patients who did not undergo lung RT, and this was accompanied by improvements in event free survival (from 19 to 40 percent in one report [60]) as well [59,60]. However, most of the available data supporting benefit from RT in this setting are from nonrandomized series, and patient selection factors confound interpretation of the data. There are no randomized trials that confirm the benefit of this approach. Despite the lack of controlled trials, low-dose bilateral lung irradiation (15 to 20 Gy, in daily 1.5 to 2.0 Gy fractions) with a focal boost dose to a total of 40 to 50 Gy to large deposits is commonly recommended for patients with pulmonary metastases who have had a good response to chemotherapy. Long-term toxicity appears to be acceptable. This was shown in a retrospective analysis of data from the prospectively performed European Intergroup Cooperative Ewing's Sarcoma Study, in which pulmonary function tests were available for 37 patients who were treated with whole lung irradiation (with or 35 without a boost) [64]. At a median follow-up 25 months, none, mild, moderate or severe pulmonary complications were seen in 43, 29, 21 and 7 percent of patients treated with whole lung irradiation without further boost. Slightly higher complication rates were reported in patients who had an additional radiation boost or surgery to the thorax in addition to RT. Bone and soft tissue metastases — Patients with solitary or circumscribed bone or soft tissue lesions fare less well in the long run than those with isolated pulmonary metastases. Nevertheless, with aggressive multimodality therapy, perhaps 10 percent will be long-term survivors. For patients with solitary or limited bone metastases, RT is delivered to metastatic lesions (doses of 40 to 50 Gy) in addition to irradiation of the primary tumor. Total body irradiation — Total body irradiation (TBI) or sequential hemibody irradiation as a component of systemic treatment for patients with high-risk features or metastatic disease at presentation has been investigated in a limited fashion at a few institutions [36,65-67]. Unfortunately, the results of these trials have been disappointing; TBI has not contributed significantly to the control of metastatic disease. This approach should not be considered at present outside of the context of a clinical trial. The use of TBI as a conditioning regimen prior to high-dose chemotherapy and hematopoietic stem cell transplantation for patients with poor-risk disease is discussed elsewhere. RADIATION TREATMENT PLANNING — Because of the relative rarity of EFT, the potential for cure, and the importance of minimizing late side effects from therapy (see below), radiation therapy (RT) must be administered by radiation oncologists who are familiar with the optimal treatment techniques. RT is currently planned using CT simulation to allow for three dimensional (3-D) conformal treatment planning. Biopsy sites and surgical scars should be marked with radioopaque markers to allow inclusion into the target volume when deemed appropriate. Treatment volume — Historically, Ewing sarcoma was thought to be a tumor of the bone marrow. Consequently, RT was administered to the entire marrow cavity of the involved bone, and the site of gross disease boosted to a higher dose. However, an analysis of the RT fields for the Intergroup Ewing's Sarcoma Study Group (IESS) trial I suggested that most relapses were at the site of initially bulky tumor [7,8,68]. Subsequent efforts were geared toward reducing the irradiated field and targeting higher doses to the site of the initial primary tumor [6,19]. In 1983, the Pediatric Oncology Group attempted a randomized trial of whole bone versus tailored-field RT after 12 weeks of induction chemotherapy [21]. After preliminary analysis showed that tailored field was as effective as whole-bone RT, the randomized study was stopped and subsequent patients were treated with limited fields. At three years, the rates of local control and event-free survival were 76 and 54 percent, respectively. IESS trial III, which opened in 1988, was the first cooperative group trial to include tailored RT ports, and the first to be carried out with modern MRI imaging of the primary site and CT-based treatment planning [69]. The main randomization was between standard versus more intensive chemotherapy (VAC with or without IE). The addition of IE significantly improved five-year survival (72 versus 61 percent) and event-free survival rates (69 versus 54 percent), particularly for patients with pelvic primary tumors. Moreover, the five-cumulative rate of local recurrence either alone or with systemic relapse was 20 percent in the VAC group compared to only 7 percent in the VAC/IE group. These data underscore the contribution of intensive chemotherapy to both local control and survival for patients with nonmetastatic EFT at diagnosis. Current recommendations for RT define an initial clinical target volume to include the original bone and soft tissue tumor with a 2.0 cm margin. As noted above, for patients 36 with chest wall tumors and malignant effusions, the field includes the entire ipsilateral hemithorax to a dose of 15 Gy prior to cone down to the chest wall tumor. (See 'Adjuvant RT' above.) Attention to potential RT effects on normal tissue is critical in radiation planning to minimize late effects, particularly in children: Even partial treatment of uninvolved epiphyseal growth plates is avoided in order to minimize treatment-induced limb shortening [70]. Circumferential irradiation of a limb is avoided to reduce the risk of limb edema and fibrosis. Gonadal avoidance or additional shielding (for the testes) is important to retain fertility. Excellent results can be achieved from RT of lesions of the hand or foot [71]. However, walking places repetitive stress on the soft tissues of the sole of the feet, and dose sparing for the sole of the foot can improve the functional outcome of treatment. Irradiation of the Achilles tendon is usually avoided. Lesions of the calcaneus can be treated; although some atrophy of the heel pad has been reported, the functional impairment is usually minor [72]. Nail beds should be excluded from the radiated field when possible. For pelvic tumors, distention of the bladder prior to each day's treatment can reduce the amount of small bowel in the radiated field, but care must be taken to avoid radiation to significant portions of the bladder in patients receiving ifosfamide to reduce the risk of hemorrhagic cystitis. It is anticipated that improvements in RT technique over the last 20 years (ie, the adoption of tailored radiation ports as opposed to whole bone ports used in the period 1960 to 1980, three-dimensional (3D) conformal RT, intensity modulated RT (IMRT), and proton therapy (see below) and use of lower and risk-adapted RT doses [56,73]) will reduce the risk of late effects. Nevertheless, the relatively high complication rate seen with earlier treatment approaches and the delayed nature of many of these complications underscore the need for long-term posttreatment follow-up. (See 'Sequelae of treatment' below.) Proton beam therapy — One way to reduce the volume of normal tissue irradiated is with charged particle irradiation using protons. The advantage of protons over conventional photons is in their dose distribution. The physical characteristics of the proton beam result in the majority of the energy being deposited at the end of a linear track, called a Bragg peak, with the dose falling rapidly to zero beyond the Bragg peak. Compared to photon beam irradiation, proton beam therapy permits the delivery of high doses of RT to the target volume while reducing the radiation dose received by normal tissues distal to the target. Depending on the tumor target configuration and the portal selection, this can result in up to 60 percent reduction in the dose received by uninvolved normal tissue [74]. This approach may be particularly beneficial for Ewing sarcomas arising in the bones of the paranasal sinus (image 1A-B), the pelvis, and the spine: Because of the proximity of the spinal cord, the dose to vertebral body primaries using conventional photon irradiation has been often limited to 45 Gy. Proton beam therapy permits the delivery of higher doses while respecting spinal cord constraints (image 2) [75]. When used for pelvic lesions, proton beam therapy is associated with better sparing of the intestine, rectum, bladder, pelvic bone marrow, and femoral head as compared to photon irradiation [76]. 37 In addition to less acute morbidity, one would also anticipate a reduction in late, RTinduced tumors in patients treated with protons because of the lower volume of normal tissue in the irradiated field (see below) [77]. Proton beam irradiation is approved for use in Children's Oncology Group protocols for EFT [78]. Our group recently reported experience with proton beam irradiation in 30 children undergoing chemoradiation with or without surgery for Ewing sarcoma [79]. At a median follow-up of 38.4 months, the three-year actuarial rates of event-free survival, local control, and overall survival were 60, 86, and 89 percent, respectively. RT was acutely well tolerated, with mostly mild-to-moderate skin reactions. The only serious late effects were four hematologic malignancies (three acute myelogenous leukemia [AML] and one myelodysplastic syndrome [MDS]), which are known risks of topoisomerase and anthracycline exposure. (See 'Second malignancies' below.) Concern has been raised that neutron scatter radiation associated with passively scattered proton beam lines may also result in secondary malignancies [80]. The magnitude of this risk is uncertain and may be low in the beam lines in modern facilities [81]; it is expected to be significantly reduced by the use of scanned proton beams [82]. These issues will only be addressed by careful further follow-up of patients treated with protons. Intensity modulated radiation therapy — Intensity modulated RT (IMRT) requires the same careful three dimensional radiation treatment planning used for 3D-CRT. However, IMRT utilizes variable, computer-controlled intensities within each RT beam, in contrast to the uniform doses within each 3D-CRT beam. Compared to most other treatment techniques, IMRT can achieve a higher degree of accuracy in conforming the radiation to the planned target while sparing normal tissue and reducing the incidence of some late complications. The advantages of IMRT are particularly evident when the target volumes have complex shapes or concave regions. Initial experience with IMRT for Ewing sarcoma has been promising [33]. However, critics argue that the risk of a second malignancy might be higher because the multiple portals used with IMRT to achieve conformality of the high-dose region around the target result in delivery of low-moderate doses of radiation to a larger volume of surrounding tissue than with 3D conformal RT techniques. In addition, for IMRT delivery, the linear accelerator is delivering more monitor units and, therefore, a larger total-body dose because of leakage of radiation [80]. IMRT supporters point out that since the risk of radiation-induced tumors is dose dependent, this risk may not in fact be increased. These questions will only be answered by careful further follow-up of patients treated with IMRT. Dose — RT dose is an important factor in local control, particularly for large tumors [34,56]. For patients treated with chemotherapy and RT, recent cooperative studies in the United States have employed 45 Gy in 25 fractions to the initial clinical target volume as defined above, followed by a 10.8 Gy boost in six fractions to the site of original bony disease and any residual soft tissue disease that remains following chemotherapy. Because of the proximity of the spinal cord, the dose to vertebral body primaries has been often limited to 45 Gy, although techniques such as protons and IMRT can permit the safe delivery of higher doses. ( For patients undergoing adjuvant postoperative RT, doses of 45 to 50.4 Gy are recommended for microscopic, and 55.8 Gy for gross residual disease. If preoperative RT is given for bulky tumors, doses in the range of 45 Gy are used. Lower doses (eg, 30 to 36 Gy) have been studied at St Jude and Sloan Kettering [56,83]. In one early report of patients undergoing definitive RT, local control rates were 90 percent using RT doses of 35 Gy for smaller lesions (≤8 cm) with a favorable response to chemotherapy, but they were inferior (52 percent) for larger lesions [56]. Others suggest that doses as low as 30 Gy may be sufficient in the adjuvant setting 38 [83]. However, further studies are needed before lower doses of RT can be accepted as standard practice. Radiation schedule — Conventional RT schedules usually consist of once daily RT doses of 1.8 to 2.0 Gy per fraction. Accelerated fractionation RT does not seem to improve rates of local control or survival [6,19,22,33,84]. However, data from the University of Florida suggest that hyperfractionated RT (1.2 Gy twice daily with a six hour interfraction interval) may be associated with less long-term toxicity [19]. The series consisted of 75 patients with localized Ewing tumor of the extremity or pelvis (lower extremity, 30; upper extremity, 19; pelvis, 26) who were treated with definitive RT at the University of Florida between 1970 and 2006 . RT was performed on a oncedaily (40 percent, median dose was 55.2 Gy in 1.8-Gy daily fractions) or twice-daily (60 percent, median dose 55.0 Gy in 1.2 Gy twice-daily fractions) basis. Functional outcome was assessed using the Toronto Extremity Salvage Score [85]. Larger tumors had significantly worse cause-specific survival (81 versus 39 percent for tumors <8 cm versus ≥8 cm), but there were no patient characteristics or treatment variables that were predictive of local failure. No fractures occurred in patients treated with hyperfractionation or with tumors of the distal extremities. Severe late complications were more frequently associated with use of <8 MV photons and fields encompassing the entire bone or hemipelvis. A significantly better Toronto Extremity Salvage Score was associated with a late-effect biologically effective dose of <91.7 Gy [68]. The authors concluded that limited field sizes with hyperfractionated high-energy RT could minimize long-term complications and provide superior functional outcomes. IORT — A benefit for intraoperative RT (IORT) has been suggested in retrospective series involving a small number of patients [86,87]. However, peripheral nerves are dose-limiting tissue structures for IORT, so the risk of severe neuropathy and soft tissue necrosis must be considered if this approach is used. SEQUELAE OF TREATMENT — Potentially curative treatment of EFT is necessarily aggressive. The toxicity of RT given in conjunction with intensive chemotherapy can be separated into short-term and long-term complications. The severity of both can be limited by careful attention to technique. Acute effects — Acute reactions are those that occur during or shortly after the completion of RT. The most prominent affect tissues within the radiated field that contain rapidly dividing cells, and include desquamation of the skin, myelosuppression, mucositis, diarrhea, nausea, and cystitis. All except myelosuppression are site-specific. Severity depends upon the amount of normal tissue in the radiated field, the radiation fraction size, and the timing of chemotherapy in relation to the RT. Many acute effects, particularly myelosuppression, are exacerbated or potentiated by the use of intensive systemic chemotherapy. As examples: Both cyclophosphamide and ifosfamide can cause hemorrhagic cystitis; the amount of bladder in the treatment field should be limited to avoid compounding this problem. (See "Cystitis in patients with cancer".) Dactinomycin and doxorubicin act as radiation sensitizers and can enhance acute radiation toxicity. If these drugs are used as a component of systemic chemotherapy, breaks should be instituted during the RT or administration of these agents should be delayed until acute RT reactions subside. Current practice in the Children's Oncology Group (COG) studies is to administer RT concurrently with the ifosfamide/etoposide cycles of chemotherapy. (See "Treatment of the Ewing sarcoma family of tumors".) Acute reactions are usually self-limited and subside within 10 to 14 days of RT completion. Dry desquamation can be managed topically with emollient creams. The 39 development of wet desquamation may require a temporary break in therapy in addition to thrice daily cleansing with warm water and mild soap, petrolatum-based ointments, nonadherent dressings, and, if patients are neutropenic, a topical antibiotic ointment or systemic antibiotics depending on the clinical scenario. Mucositis and diarrhea are managed with supportive measures; if severe, a treatment break may be required. Patients receiving proton beam irradiation for treatment of vertebral lesions tend to have less nausea and diarrhea than those undergoing photon beam irradiation because of the lack of an exit dose to bowel anterior to the spine. Patients receiving whole lung irradiation are at risk for radiation pneumonitis. Late effects — Late reactions occur months to years after completing a course of RT. Severity is not always predicted by the severity of acute effects. Late changes in normal tissues caused by RT are related to field size, the volume of normal tissues in the field, and the dose received by normal tissue. Other important factors include patient age at the time of treatment, skeletal maturity, adherence to a rehabilitation program, and whether a pathologic fracture is present at diagnosis. As examples: Younger, prepubertal children are at greatest risk for radiation-induced arrest of bone growth [88]. Sparing of uninvolved epiphyseal plates minimizes limb shortening after RT of extremity lesions. RT doses above 60 Gy are associated with markedly increased rates of soft tissue induration and fibrosis [89,90]. High-dose circumferential irradiation of an extremity is associated with edema, fibrosis, and compromised limb function [91]. This can be avoided by sparing of an adequate strip of tissue. Weight-bearing bones are at risk for pathologic fractures [90]. The highest risk is within the first 18 months of RT completion Refinements in diagnostic imaging, RT planning, and techniques over time (tailored field size, hyperfractionated treatment schedules, IMRT, proton beam irradiation) have resulted in better limb function among long-term survivors, and more recent series suggest that excellent functional results can be obtained in the majority of patients following RT for EFT [19,72,89]. A posttreatment rehabilitation program, including active range of motion of affected joints, is also important in improving and maintaining limb function. Second malignancies — There is an increased risk of secondary neoplasia after treatment for EFT [73,92-95]. RT-induced osteosarcomas and therapy-related (alkylating agents, epidophyllotoxins) leukemias predominate. Most of the radiation-related bone tumors are osteosarcomas, although other tumors are reported. The magnitude of risk is variable [46,73,92-94,96-98]. Early data suggested that the cumulative risk of a secondary sarcoma after treatment for EFT was approximately 35 percent at 10 years [97]. The Late Effects Study Group, examining the late effects of treatment for childhood cancer, reported an estimated cumulative incidence of secondary sarcomas after treatment of EFT that approached 22 percent at 20 years [93]. These and other investigators identified RT dose as the major predisposing factor for development of a radiation-induced sarcoma, particularly above 60 Gy. Subsequent experience with protocols utilizing lower doses of RT and tailored RT fields suggest that the magnitude of the risk is somewhat lower [46,73,94,98]. The two largest series are described in detail: The largest series with longest follow-up includes 266 survivors of EFT treated at St Jude, the National Cancer Institute, and the University of Florida, and followed for a median 9.5 years (range 3 to 30) [73]. The cumulative RT doses were none, 21 to 48, 48 to 60, and >60 Gy in 8, 23, 45, and 24 percent, respectively. 40 Overall, 16 children developed second malignancies, which included 10 sarcomas. The estimated cumulative risk at 20 years was 9.2 percent for any malignancy, and 6.5 percent for a secondary sarcoma. The median time to the diagnosis of the second malignancy after completion of therapy was 7.6 years. All the secondary sarcomas occurred near or at the primary site of the Ewing sarcoma and within the primary irradiated field. As has been noted by others, the cumulative incidence rate of secondary sarcoma was radiation dose-dependent. No secondary sarcomas developed among patients who had received less than 48 Gy, while the absolute risk of secondary sarcoma was 130 cases per 10,000 person-years of observation among patients who had received ≥60 Gy. Dunst et al reported on second malignancies among 674 patients enrolled in the Cooperative Ewing's Sarcoma Study Group (CESS)-81 and 86 studies, and followed for an average of 5.1 years [94]. Only eight developed a second malignancy, four were treatment-related leukemias/myelodysplastic syndrome, and there were three sarcomas. The cumulative risk of a second malignancy at 10 and 15 years was 2.9 and 4.7 percent, respectively. Long-term follow-up guidelines after treatment of childhood malignancy have been published by the Children's Oncology Group and are available online [99]. RT and other risk factors for secondary sarcomas after treatment of childhood cancer are discussed further elsewhere. SUMMARY AND RECOMMENDATIONS — Ewing sarcoma is a rare malignancy that most often presents as an undifferentiated primary bone tumor; less commonly, it arises in soft tissue. Both are part of a spectrum of neoplastic diseases known as the Ewing sarcoma family of tumors (EFT), which also includes the more differentiated peripheral primitive neuroectodermal tumor (PNET). These entities share important clinical features, including a tendency towards rapid spread to lungs, bone, and bone marrow, and responsiveness to the same chemotherapeutic regimens and radiotherapy (RT). Because relapse rates are high in patients undergoing local therapy alone (80 to 90 percent), it is surmised that the majority have subclinical metastatic disease at the time of diagnosis, even in the absence of overt metastases. Intensive multiagent chemotherapy can eradicate these deposits, and is a critical component of therapy for both localized and advanced disease. (See 'Introduction' above.) Localized disease — For patients with localized EFT, local and systemic therapy are both necessary to achieve cure. Either surgery or RT can provide effective local control. Contemporary treatment guidelines emphasize surgical resection as the modality of choice if it is believed that the lesion can be resected with negative margins, without excessively morbidity, and with the expectation of a reasonable functional result. A major advantage of surgery is the lack of association with treatment-related sarcomas. For patients who lack a functionpreserving surgical option because of tumor location or extent, and those who have clearly unresectable primary tumors following induction chemotherapy, we recommend RT (Grade 1A). (See 'RT versus surgery' above.) Although the indications for adjuvant RT are not firmly established in patients undergoing surgery for localized EFT, we suggest adjuvant RT in the following circumstances (Grade 2C) (see 'Adjuvant RT' above): 41 Bulky tumors in difficult sites (eg, the pelvis); in this setting, RT can be given either preoperatively or postoperatively, based upon institutional protocols and experience If there is residual microscopic or gross disease after surgery, or inadequate surgical margins (ie, a marginal or intralesional resection as compared to a wide or radical excision) Patients who are left with significant amounts of viable tumor in the resected specimen following neoadjuvant chemotherapy who have less than wide surgical margins Adjuvant hemithorax irradiation is indicated in patients with high-risk chest wall primary tumors (close or involved margins, initial pleural effusion, pleural infiltration, and intraoperative contamination of the pleural space) Advanced disease — Although prognosis is clearly worse than for localized disease, a subset of patients who present with overt metastatic disease can be cured, particularly if they have limited pulmonary metastases. For patients with metastatic EFT, we suggest RT rather than surgery for treatment of the primary site in most patients (Grade 2B). Surgery could be reconsidered in selected patients if chemotherapy results in significant volume reduction, particularly if areas of small volume metastatic disease are also amenable to surgical resection. (See 'RT for metastatic disease' above.) The role of whole lung irradiation after resection of pulmonary metastases is unclear. Nevertheless, because of the potential for reduced pulmonary relapse and improved event-free survival, and the low rate of pulmonary toxicity, we suggest bilateral lowdose lung irradiation (15 to 18 Gy) even if all lesions are resected or completely respond to chemotherapy (Grade 2C). (See 'Pulmonary metastases' above.) Epidemiology, pathology, and molecular genetics of the Ewing sarcoma family of tumors Literature review current through: Aug 2013. | This topic last updated: May 23, 2013. 42 43 44 INTRODUCTION — Ewing sarcoma (ES) and peripheral primitive neuroectodermal tumor (PNET) were originally described as distinct clinicopathologic entities: In 1918, Stout described a tumor of the ulnar nerve with the gross features of a sarcoma, but composed of small round cells focally arranged as rosettes; this entity was subsequently designated neuroepithelioma, and then PNET [1]. ES was described by James Ewing in 1921 as an undifferentiated tumor involving the diaphysis of long bones that, in contrast to osteosarcoma, was radiationsensitive. Although most often a primary bone tumor, ES was also reported to arise in soft tissue (extraosseous Ewing sarcoma [EES]) [2]. 45 However, over the last several decades, it has become clear that these entities comprise the same spectrum of neoplastic diseases known as the Ewing sarcoma family of tumors (EFT), which also includes malignant small-cell tumor of the chest wall (Askin tumor) [3], and atypical ES [4,5]. Because of their similar histologic and immunohistochemical characteristics and shared nonrandom chromosomal translocations, these tumors are considered to be derived from a common cell of origin, although the histogenic origin of this cell is debated [6-13]. The EFT can develop in almost any bone or soft tissue, but the most common site is in a flat or long bone, and patients typically present with localized pain and swelling. Although overt metastatic disease is found in fewer than 25 percent at the time of diagnosis, subclinical metastatic disease is assumed to be present in nearly all patients because of the 80 to 90 percent relapse rate in patients undergoing local therapy alone. As a result, systemic chemotherapy has evolved as an important component of treatment. Advances in multidisciplinary management of EFT over the past 30 years have resulted in a marked improvement in survival and a greater likelihood of limb-sparing surgery rather than amputation. In data derived from the Surveillance, Epidemiology and End Results (SEER) program of the National Cancer Institute, five-year survival rates for patients with ES rose between 1975 and 2000 in all age groups, although the improvements were most marked in younger individuals [14]. Five-year survival rates in those treated between 1993 and 2000 were >65 percent, whereas for those over the age of 15, they were below 50 percent. Among young patients presenting with nonmetastatic disease who receive modern multidisciplinary treatment, long-term survival can be achieved in 70 to 80 percent [15,16]. Here we will discuss the epidemiology, pathology, and molecular genetics of EFT/peripheral PNET. The clinical features, diagnosis, and treatment of these tumors, principles underlying the performance of a diagnostic bone biopsy, and central PNET tumors are discussed elsewhere. EPIDEMIOLOGY AND RISK FACTORS — Primary bone tumors are responsible for 6 percent of all childhood cancers [17]. Although rare, the EFT represent the second most common primary bone malignancy (table 1) affecting children and adolescents, after osteosarcoma [14,18]. Despite this, EFT is responsible for only 3.5 percent of cancers in American children 10 to 14 years old, and 2.3 percent of those arising among 15 to 19 year olds. A similar incidence (2.8 percent of all tumors in children aged 15 to 19 years old) is described in adolescents in England [19]. In the United States, 650 to 700 children and adolescents younger than 20 years old are diagnosed with bone tumors every year, of which only 200 are EFT and the remainder osteosarcomas [20]. The peak incidence is between 10 to 15 years of age. However, 30 percent of cases arise in children under the age of 10, and another 30 percent are in adults over the age of 20 [21,22]. As with many pediatric tumors, there is a slight male predominance. Racial and ethnic factors are of epidemiologic importance. For unclear reasons, the EFT affect mainly Caucasians and are extremely uncommon among blacks (both in the United States and Africa) and Asians (figure 1) [22-25]. The reason for this striking ethnic distribution is not known. However, interethnic differences exist for certain alleles of the Ewing sarcoma gene (the EWSR1 gene, MIM#133450), which is consistently disrupted in these tumors [26] (see 'EWSR1 translocations in EFT' below). Furthermore, genome-wide association studies have raised the possibility that genetic susceptibility factors may contribute to the observed geographical/ethnic differences in Ewing sarcoma incidence [27]. Risk factors — EFT have not been consistently associated with any familial or congenital syndromes [28,29], although at least one series reports an excess of congenital 46 mesenchymal defects in affected patients [30], and another suggests an increase in neuroectodermal tumors and stomach cancer in the families of patients with ES [31]. Specific environmental exposures have not been identified as risk factors [29,32]. EFT develop rarely after treatment of a primary cancer during childhood, but most cases do not appear to be related to radiation therapy [33,34]. An association has been suggested between a parental occupation of farming (particularly if the mother farmed) and the development of Ewing sarcoma [35]. Ewing sarcoma may be more common in children who have hernias [35-39]. A metaanalysis of three case-control studies (totaling 357 patients with Ewing sarcoma and 745 controls [37-39]) showed that affected children were more likely to have had a hernia in childhood (odds ratio 3.2, 95% CI 1.9 to 5.7) [35]. The association was strongest for umbilical hernias. The mechanism underlying a possible association between hernias and Ewing sarcoma is unclear. HISTOGENESIS AND HISTOLOGY — As noted above, the EFT represent a spectrum of tumors that range from the undifferentiated ES to atypical poorly differentiated ES and the differentiated PNET. Histogenesis — The histogenic origin of EFTs has been debated over the years. A neuroectodermal origin had been proposed based upon variable expression of neuronal immunohistochemical markers, cytogenetic, and ultrastructural features, as well as the ability of ES cell lines to differentiate along neural pathways in vitro [10-12,40,41]. Because of the resemblance of the PNET to classic neuroblastoma, the neural crest was regarded as the most likely progenitor. Although ES is typically described as an undifferentiated tumor whose primitive cells lack neural differentiation, some tumors contain Homer-Wright rosettes, which are indicative of neural differentiation. Furthermore, ES can express neural antigens (eg, gastrin releasing peptide, a protein which is normally expressed by the brain and neuroendocrine cells) on their surface [40], and both ES and PNET synthesize choline acetyltransferase, a marker suggesting a common neural origin [41]. (See 'Histologic features' below.) Based upon these features, and the lack of precursor synthesis for the adrenergic pathway, some investigators believe that EFT are derived from postganglionic parasympathetic primordial cells, which are located throughout the parasympathetic nervous system, accounting for their widespread distribution [10,12,42]. In contrast, neuroblastomas synthesize neurotransmitters that are associated with the sympathetic autonomic nervous system [10,42]. In contrast to this "neural crest hypothesis", an alternative theory is emerging that suggests that EFT arise from mesenchymal progenitor or mesenchymal stem cells (MSC). MSCs isolated from bone marrow can differentiate into osteogenic, chondrogenic, or adipogenic lineages. Expression of the ES-specific oncoprotein EWS-FLI (see below) blocks MSC differentiation [43], while removal of EWS-FLI from patient-derived ES cell lines allows the cells to gain the ability to differentiate into multiple lineages and adapt a gene expression profile that resembles that of MSCs [13]. Additional data will be required to draw a final conclusion as to the cell of origin in ES. Histologic features — The morphologic appearance of classic ES is that of a primitive, undifferentiated neoplasm. On hematoxylin and eosin-stained sections, there are monotonous sheets of small round blue cells with hyperchromatic nuclei and scant cytoplasm that is clear because of the presence of abundant glycogen (picture 1) [44]. There is usually extensive necrosis with preservation of viable tumor around blood vessels. Nuclear atypia, palisading, and formation of rosettes (where the tumor cells are arranged in a circle about a central fibrillary space, indicative of neural differentiation) or pseudorosettes are typically absent. Invasion of blood vessels can be seen, although the 47 tumors are not characteristically highly vascular. Mitotic figures are rare, and aneuploidy is uncommon on flow cytometry studies [45]. Atypical ES are usually described as poorly differentiated rather than undifferentiated tumors. They contain larger cells, a greater degree of cellular pleomorphism, and have a higher mitotic rate (generally more than two mitoses per high power field). Spindle cells may be present at the periphery, and these tumor cells invade the adjacent tissue in an infiltrative manner. The cells can be arranged in sheets, or there may be an organoid, lobular, or alveolar architectural pattern. As in the typical ES, evidence of neural differentiation is usually absent. The histopathologic features of a peripheral PNET may resemble the typical undifferentiated ES or atypical ES; however, a neural immunophenotype or evidence of neural differentiation on light microscopic (ie, Flexner-type rosettes or Homer-Wrighttype pseudorosettes) or ultrastructural examination is present in about one-half of cases. However, mature neural elements (ie, ganglion cells, nerve fascicles, or a neuropil background) are absent. The neoplastic cells are typically arranged into an organoid, alveolar, or lobular pattern. Differential diagnosis — Morphologically, the appearance of EFT and PNET tumors is similar to that of other small round blue cell tumors involving bone and soft tissue, including lymphoma, small cell osteosarcoma, mesenchymal chondrosarcoma, medulloblastoma, dedifferentiated synovial sarcoma, desmoplastic small round cell tumors, and rhabdomyosarcoma. As a group, these tumors often pose difficult diagnostic problems when examined by light microscopy alone. There is little in the routine histologic appearance of an EFT to relate it to a component of normal bone or other tissue from which it might have arisen. Even the determination of bone versus soft tissue origin can be challenging, because EFT involving bone often have an extensive soft tissue component, while extraosseous tumors may invade bone secondarily. Although ES can exhibit a variable degree of neural differentiation, this is usually subtle and often only detected by immunohistochemical staining for markers of neural differentiation (table 2) or by ultrastructural examination. Although no routinely used histochemical or immunohistochemical stain can positively distinguish EFT from other undifferentiated tumors of childhood, the vast majority of EFT express high levels of a cell surface glycoprotein (designated CD99, MIC2 surface antigen or p30/32MIC2) that is encoded by the CD99 (MIC2X) gene [6,46]. The finding of membrane-localized MIC2 expression is a sensitive diagnostic marker for the EFT; however, it lacks specificity since other tumors (eg, rhabdomyosarcoma, which is often in the differential diagnosis) and normal tissues are also immunoreactive with anti-MIC2 antibodies [47]. Another new potential immunohistochemical marker for EFT is NKX2.2, the protein product of the NKX2-2 gene [48]. Cytogenetic or molecular genetic studies looking for particular chromosomal translocations and/or their fusion transcripts are usually required to secure the diagnosis. (See 'Molecular diagnostics' below.) MOLECULAR GENETICS — Remarkable progress has been made in defining and classifying bone and soft tissue tumors on the basis of characteristic chromosomal abnormalities (table 3). Because of their unique shared chromosomal translocations, the EFT have become a model system for nonmorphologic approaches to diagnosis and subclassification using both cytogenetic and molecular techniques. These translocations interrupt specific genes and recombine them to create novel fusion genes, whose products are tumor-specific and present in virtually all cases of an individual tumor category. Thus, their characterization not only yields profound insights into tumor biology, but also holds great promise for diagnostic and therapeutic approaches. .) EWSR1 translocations in EFT — The unique shared pattern of nonrandom chromosomal translocations in the EFT provides one of the most compelling arguments for a common 48 cellular origin. Virtually all cases express one of several different reciprocal translocations, most of which involve breakpoints that are clustered within a single gene locus designated the EWSR1 gene on chromosome 22q12 (table 3) [7,8,44,49-52]. The EWSR1 gene normally encodes a widely expressed protein, the EWS protein, whose amino terminal domain shares some homology with eukaryotic RNA polymerase II and whose carboxy terminus (which is replaced by tumor-specific translocations) contains an RNA recognition motif. The EWS protein is a member of a family of highly conserved RNA-binding proteins that are believed to mediate interaction with RNA or singlestranded DNA [53,54]. In 85 to 90 percent of cases of EFT, a recurrent chromosomal translocation, t(11;22)(q24;q12), fuses the 5' portion of the EWSR1 gene on chromosome 22 to the 3' portion of the FLI1 (friend leukemia integration locus-1) gene on chromosome 11 [49,50]. This can be detected using fluorescence in situ hybridization (FISH, (image 1)). FLI1 encodes the FLI protein, a member of the ETS family of transcription factors, and is involved in the control of cellular proliferation, development, and tumorigenesis [55]. Members of the ETS family are defined by the presence of a highly conserved 85-amino acid domain termed the erythroblastosis virus-transforming sequence (ETS) domain, which mediates specific binding to purine-rich DNA sequences [56]. As a result of the t(11;22), the EWS-FLI fusion protein is expressed. EWS-FLI binds DNA via its FLIderived ETS domain, and regulates gene expression through the EWS portion of the fusion. In EFT that lack the EWSR1-FLI1 translocation, analogous translocations fuse the EWSR1 gene to other genes of the ETS family (ie, ERG, ETV1, ETV4, or FEV) that have structural homology to FLI1, forming t(21;22)(q22;q12), t(7;22)(p22;q12), t(17;22)(q12;q12), or t(2;22)(q35;q12) translocations, respectively [57-60]. The EWSR1-ERG translocation [t(21;22) (q22;q12)] is present in 5 to 10 percent of EFT, while the others are less common [42,60-62]. In addition to EWSR1-based translocations, rare cases of EFT are associated with translocations involving a related gene, FUS (also called TLS) [63,64]. Tumors exhibiting t(16;21)(p11;q24) or t(2;16)(q35;p11) translocations express FUS-ERG or FUS-FEV fusions, respectively. Because EWS and FUS proteins are highly similar, it is believed that FUS-based fusion proteins function similarly to EWS-based fusions. It is important to note that tumors with these translocations can be a source of diagnostic confusion. (See 'Molecular diagnostics' below.) Ultimately, specific translocations are important diagnostic features of the EFT, and the protein products of the fusion genes are believed to play an important role in tumor development and biology. A growing body of evidence suggests that these proteins function as transcriptional regulators, although their transcriptional targets are just beginning to be identified. (See 'Molecular pathogenesis' below.) EWSR1 translocations in other soft tissue tumors — Although translocations involving the EWSR1 and ETS families of genes are specific for EFT, translocations that involve fusion of the EWSR1 gene to other genes have been observed in many other tumors (eg, clear cell sarcoma/malignant melanoma of soft parts, myxoid liposarcoma [65-67]). The structure of these resultant fusion genes is analogous to that of EWSR1-FLI1 (ie, all result in the fusion of the 5' portions of the EWSR1 gene or an EWSR1-like gene [eg, the FUS gene] to 3' portions of genes encoding transcription factors). The resulting chimeric proteins presumably have a similar function to the EWS-FLI fusion protein, and it is thought that disruption of transcriptional control contributes to their transforming potential [44]. (See 'Molecular pathogenesis' below.) Despite these similarities, with few exceptions, a clinicopathologically distinct tumor entity has been consistently associated with each specific translocation involving a unique class of transcription factors. This tumor specificity is most likely related to cell49 specific influences on DNA recombination, chimeric gene expression, and intrinsic proteins that are necessary to complement the action of the chimeric protein. Other genetic changes in EFT — In addition to the characteristic reciprocal translocations described above, other numerical and structural chromosomal aberrations are occasionally found in EFT. These include a gain of chromosomes 1q, 2, 8 and 12,losses of 9p and 16q, and the non-reciprocal translocation t(1;16)(q12;q11.2) [68-70]. In addition, alterations in known tumor suppressor genes have been identified in some cases. As an example, homozygous deletion of the CDKN2A gene has been described in 18 to 30 percent of cases, and TP53 mutations are detected in 5 to 20 percent [44,7173]. The biologic implications of these findings are incompletely understood. Molecular pathogenesis — EFT-associated translocations result in the production of chimeric proteins that contain the amino-terminal domain of the normal EWS protein fused to the nucleic acid-binding domain of the transcription factor translocation partner. Because of their ubiquitous presence in EFT, they are thought to be intimately connected to the biology of these tumors. The available data suggest two possible mechanisms by which these unique chimeric proteins may contribute to neoplastic transformation and/or cell growth: Chimeric proteins such as EWS-FLI may influence transcription of some of the same genes that are normally regulated by native FLI, but because the chimeric molecule is expressed in either a different temporal or quantitative manner, the downstream targets are deregulated, leading to uncontrolled cellular growth. EWS-FLI may affect target genes that are different from those regulated by the native FLI or EWS proteins. This would explain why the chimeric gene product (EWS-FLI) has transforming properties that are not shared by either the EWS or the FLI wild-type counterparts [74]. Some chimeric fusion proteins can act as transcriptional activators, deregulating many genes associated with cell signaling, proliferation, apoptosis, tissue invasion and metastasis [44,74-80]. It is likely that this ability to regulate transcriptional activity is associated with their oncogenic potential. EWS-FLI and other chimeric fusion proteins are all capable of transforming cell lines equally well. As an example, expression of either the EWS-FLI or EWS-ERG cDNAs can transform mouse NIH3T3 fibroblasts so that they acquire tumor-like properties (eg, growth in soft agar or in immunodeficient mice) [74,81]. Notably, EWS-FLI can only transform cells in which insulin like growth factor signaling is intact, and thus there is a potential role for IGF1R inhibitors in the treatment of Ewing sarcoma [82-84]. The protein products of the fused genes are also important for maintaining the growth characteristics of EFT cell lines, and emerging data suggest the potential for molecularly targeted therapy [85]: Introduction of EWS-FLI RNA antisense molecules or short-interfering or short hairpin RNAs (such as RNAi that mediate RNA-interference) into EFT cells decreases the expression of the EWS-FLI fusion protein and results in growth inhibition and/or loss of oncogenic potential, both in vitro and in vivo [86-90]. Likewise, inhibition of the interaction between EWS-FLI and its cellular binding partner RNA helicase A (RHA) by a small peptide or small molecule inhibitor reduces the growth of EFT orthotopic xenografts [91]. Truncated ETS domain-binding molecules can act as competitive inhibitors and have a dominant negative effect on cell growth [92]. p21(WAF1/CIP1), which is important in cell cycle regulation, might be one of the direct targets of EWS-FLI, suggesting that this molecule could serve as a target for a molecularly based therapy for EFT [85]. 50 Because EWS-FLI and related fusions are thought to function primarily as transcriptional regulators, a number of studies have sought to define the genes that are regulated by the fusion proteins [88-90,93,94]. Some of the EWS-FLI regulated genes are involved in the oncogenic phenotype of Ewing's sarcoma model systems (eg, IGFBP3, CAV1, NKX2.2, NR0B1, GSTM4, and others) [76,88-90,95,96]. While many of the details are still being worked out, some of these gene targets and associated pathways may prove to be future targets for molecularly-targeted therapy. Another mechanism whereby EWS-FLI may exert a tumorigenic effect is via deregulation of programmed cell death (apoptosis). Cancer is believed, in part, to represent an imbalance between cell growth and cell death. Expression of EWS-FLI1 or EWS-ERG in NIH3T3 cells inhibits the apoptosis that would normally occur with serum deprivation or calcium ionophore treatment [97], while antisense inhibition of EWS-FLI increases susceptibility to chemotherapy-induced apoptosis in ES cell lines. Similarly, reduced expression of EWS-FLI using RNAi-based approaches also resulted in increased apoptosis [89]. EWS-ETS fusion proteins also activate human telomerase activity in Ewing tumors through upregulation of the telomerase reverse transcriptase gene expression, probably by acting as a transcriptional coactivator [98]. Telomerase is a ribonucleoprotein enzyme that compensates for telomere shortening during cell division by synthesizing telomeric DNA, thereby maintaining telomere length. In normal somatic cells, telomerase activity is usually undetectable, with the exception of some cell types such as hematopoietic cells, hair follicles, intestinal crypt cells, and basal cells of the epidermis. Upregulated telomerase activity in cancer cells correlates with the stabilization of telomere length and cellular immortalization (ie, the ability of the cells to divide indefinitely). This raises the possibility of therapeutic approaches using telomerase inhibitors. Molecular diagnostics — The translocations that characterize EFT are exquisitely sensitive and specific tumor markers that have rapidly become the standard for confirming the diagnosis, particularly for cases in which morphology is inconclusive [7]. Standard cytogenetic analysis can reveal a wide variety of chromosomal alterations, but technical constraints and variant translocations may complicate the interpretation, and results are often obtained too late to influence therapy. Molecular assays are more attractive because of the small amount of tissue required, rapid results, and higher sensitivity than traditional cytogenetics. Both the fusion genes and their hybrid transcripts can be identified in tumor cells using molecular genetic approaches. Two techniques are in use, fluorescent in situ hybridization (FISH) and reverse transcriptase polymerase chain reaction (RT-PCR), and both are more rapid than cytogenetic analysis [99-102]. In particular, RT-PCR to detect the presence of fusion transcripts in tumor cells has become a mainstay in molecular diagnosis of the EFT. It is extremely sensitive and specific and allows detection of very low levels of tumor cells even among large numbers of normal cells (eg, peripheral blood and bone marrow), providing an extremely robust method for molecular staging and monitoring treatment response (see below). In addition to providing diagnostic support for challenging cases, the identification of specific types of transcripts was at one time thought to have prognostic significance [68,103,104]. As an example, in one report, the presence of type 1 EWS-FLI transcripts (EWSR1 exon 7 fused with FLI1 exon 6) was associated with a threefold reduction in the risk of developing metastatic disease as compared to non-type 1 transcripts [103]. However, more recent independent studies from both the United States and Europe do not support the prognostic value of translocation type [105,106]. It appears that current intensive treatment protocols may have eliminated this apparent difference in prognosis. While molecular tests for EFT-associated translocations are important and often useful adjuncts in the diagnostic work-up, in most cases, molecular pathology laboratories do not test for all of the "alternate" translocations (such as those that encode EWS-ERG, 51 EWS-ETV1, EWS-ETV4, EWS-FEV, FUS-ERG, and FUS-FEV). Thus, a "negative" FISH or RT-PCR result does not necessarily rule out the diagnosis of EFT. Physicians are strongly advised to consult with their molecular diagnostic laboratory to understand exactly what tests are performed, and the sensitivity and specificity of those tests for EFT. Ultimately, the diagnosis of EFT is based upon a combination of histopathology, special stains, and molecular diagnostics that are interpreted in the context of the patient's clinical presentation. Molecular staging — Disease stage as determined by standard imaging modalities constitutes the most powerful predictor of prognosis for patients with EFT. However, some patients considered to have localized disease have an unfavorable outcome, even with multimodality therapy, and this may be related to the persistence of minimal metastatic disease that is not detected by traditional methods. Minimal disease (ie, circulating ES cells) can be detected in peripheral blood and bone marrow using PCR-based methods in up to 30 percent of patients with apparently localized disease and in approximately 50 percent of those with advanced or relapsed disease [107-111]. However, the prognostic significance of this finding, particularly in patients with localized disease, is controversial [108,110,111], and the ultimate contribution of this finding to patient management is unclear [110,111]. Further prospective studies using molecular methods to detect the presence of minimal residual disease are needed in order to assess its impact on outcome. SUMMARY — Ewing sarcoma and peripheral primitive neuroectodermal tumors (PNET) comprise the same spectrum of neoplastic diseases known as the Ewing sarcoma family of tumors (EFT), which also includes malignant small-cell tumor of the chest wall (Askin tumor). Because of their similar histologic and immunohistochemical characteristics and shared nonrandom chromosomal translocations, these tumors are considered to be derived from a common cell of origin, although its histogenic origin is debated. (See 'Introduction' above.) Although rare, the EFT represent the second most common primary bone malignancy (table 1) affecting children and adolescents. EFT have not been consistently associated with any familial or congenital syndromes, and specific environmental exposures have not been identified as risk factors. Morphologically, the appearance of EFT is similar to that of other small round blue cell tumors involving bone and soft tissue, including lymphoma, small cell osteosarcoma, mesenchymal chondrosarcoma, medulloblastoma, dedifferentiated synovial sarcoma, desmoplastic small round cell tumors, and rhabdomyosarcoma. As a group, these tumors often pose difficult diagnostic problems when examined by light microscopy alone. Cytogenetic or molecular genetic studies looking for particular chromosomal translocations and/or their fusion transcripts are usually required to secure the diagnosis. (See 'Differential diagnosis' above and 'Molecular diagnostics' above.) Virtually all cases express one of several different reciprocal translocations, most of which involve breakpoints that are clustered within a single gene locus designated the EWSR1 gene on chromosome 22q12 EFT-associated translocations result in the production of chimeric proteins that contain the amino-terminal domain of the normal EWS protein fused to the nucleic acid-binding domain of the transcription factor translocation partner. Because of their ubiquitous presence in EFT, these chimeric proteins are thought to be intimately connected to the biology of these tumors. (See 'EWSR1 translocations in EFT' above and 'Molecular pathogenesis' above.) Adjuvant and neoadjuvant chemotherapy for soft tissue sarcoma of the extremities 52 53 Literature review current through: Aug 2013. | This topic last updated: Tem 12, 2013. INTRODUCTION — Soft tissue sarcomas (STS) are uncommon malignant tumors that arise from extraskeletal connective tissues, including the peripheral nervous system. They can arise at any body site, but are most common in the extremities, particularly the lower limbs. In treating STS of the extremities, the major therapeutic goals are long-term survival, avoidance of a local recurrence, maximizing function, and minimizing morbidity. Surgical resection is the cornerstone of potentially curative treatment. For nearly all patients with extremity sarcomas >5 cm, the addition of radiation therapy (RT) improves local control, 54 and it has also had a significant impact on limb salvage. There are advantages to preoperative (neoadjuvant) as compared to postoperative (adjuvant) administration of RT, and for neoadjuvant therapy utilizing combinations of RT and chemotherapy, particularly for recurrent and large, high-grade primary tumors. These topics are discussed in detail elsewhere. Systemic chemotherapy is a routine component of treatment for several sarcomas that occur predominantly in children (eg, rhabdomyosarcoma, Ewing sarcoma, and osteogenic sarcoma). However, the value of adjuvant chemotherapy in patients undergoing resection of the adult-type localized extremity STS (eg, leiomyosarcoma, liposarcoma, synovial sarcoma) remains controversial due to the complexity of the group of diagnoses involved. This topic review will discuss the use of adjuvant and neoadjuvant chemotherapy in the treatment of adult-type extremity STS. The role of chemotherapy in the treatment of retroperitoneal STS, rhabdomyosarcoma and the Ewing sarcoma family of tumors, and neoadjuvant combined modality approaches for patients with large, high-grade or recurrent extremity STS are discussed in detail elsewhere. ADJUVANT CHEMOTHERAPY Pediatric-type sarcomas — The addition of systemic chemotherapy to local therapy significantly improves outcomes for the common pediatric types of sarcoma (rhabdomyosarcoma, osteogenic sarcoma, and the Ewing sarcoma family of tumors of both soft tissue and bone). Most modern treatment plans utilize initial (induction or neoadjuvant) chemotherapy followed by local treatment and additional (adjuvant) chemotherapy. Rhabdomyosarcoma — The vast majority of patients with rhabdomyosarcoma are children. The routine use of multiagent chemotherapy (typically vincristine, dactinomycin, and cyclophosphamide, VActinoC, or regimens used for Ewing sarcoma), in addition to surgery and RT, has contributed significantly toward increasing cure rates among those with localized disease. The rare cases of rhabdomyosarcoma that arise in adults are managed similarly. (See "Rhabdomyosarcoma and undifferentiated sarcoma in childhood and adolescence: Treatment".) Ewing sarcoma — Among children, Ewing sarcoma is much more common in bone than in soft tissue, while in adults, it more often presents in soft tissue. Regardless of whether it arises in bone or soft tissue, the tumor is treated in the same multidisciplinary manner in adults as in children. A combination of neoadjuvant and adjuvant chemotherapy using vincristine, doxorubicin, and cyclophosphamide (VAC), with alternating cycles of ifosfamide and etoposide, is a US standard of care in addition to local therapy; in Europe a somewhat different combination of many of these agents is recommended (vincristine, ifosfamide, doxorubicin, etoposide [VIDE]). (See "Treatment of the Ewing sarcoma family of tumors", section on 'Treatment for localized disease'.) Extraosseous osteogenic sarcomas — Osteosarcomas rarely arise in the soft tissue rather than bone (image 1 and image 2). In contrast to Ewing sarcoma, when osteogenic sarcomas arise in soft tissue rather than bone, management has to date followed the principles established for soft-tissue tumors rather than primary bone tumors. While adjuvant chemotherapy is a standard component of treatment for primary bone osteosarcomas in all age groups, it has been thought to be much less effective for tumors of extraosseous origin [1]. However, at least one study suggests that outcomes may be better when children with extraosseous osteosarcoma are treated with the same combination chemotherapy protocols as used for bone-primary osteogenic sarcoma [2]. Sarcomas more commonly seen in adults — The remainder of this discussion will focus on the use of adjuvant chemotherapy for the more common adult-type STS, such as liposarcomas, leiomyosarcomas, and synovial sarcomas. 55 Over 20 randomized trials and two meta-analyses have addressed the potential benefit of adjuvant chemotherapy for resected extremity STS in adults. Unfortunately, these have yielded conflicting data, and as a result, the benefit of adjuvant chemotherapy remains uncertain. Early randomized trials — The majority of early trials used doxorubicin alone or with dacarbazine but did not employ ifosfamide, a compound only developed in the mid1980s. Among the first 14 published randomized trials of adjuvant doxorubicin-based therapy versus surgery alone, two reported a significant survival advantage for combination chemotherapy, three found higher survival in the observation arm, and the remainder showed no difference in outcome in the treated group [3]. SMAC meta-analysis — Individual patient data from these trials, which involved 1568 adults with localized resectable STS (only some of which were localized to an extremity), were analyzed by the Sarcoma Meta-Analysis Collaboration (SMAC) and published in 1997 [4,5]. All evaluated studies that randomly assigned patients postoperatively to receive or not receive adjuvant doxorubicin-containing chemotherapy; fewer than 5 percent of patients received a chemotherapy regimen that included ifosfamide. The following benefits were noted in the chemotherapy group: Significantly longer local recurrence-free interval — hazard ratio [HR] for local recurrence 0.73 (95% CI 0.56-0.94). Significantly longer distant recurrence-free interval — HR 0.70 (95% CI 0.570.85). Significantly higher overall recurrence-free survival — HR for any recurrence 0.75 (95% CI 0.64-0.87). This corresponded to an absolute 6 to 10 percent improvement in recurrence free survival at 10 years. There was a trend towards better overall survival that favored chemotherapy, but it was not statistically significant (HR for death 0.89, 95% CI 0.76-1.03). There was no consistent evidence of a difference in any endpoint according to age, sex, stage, site, grade, histology (although there was no central pathology review), extent of resection, tumor size, or exposure to RT. However, the strongest evidence of a beneficial effect on survival was shown in the subset of patients with extremity and truncal sarcomas. Among these patients who received adjuvant doxorubicin-containing chemotherapy, there was a modest but statistically significant benefit for chemotherapy (HR for death 0.80, p = 0.029), which translated into a 7 percent absolute benefit in overall survival at 10 years. The updated meta-analysis from this group, which includes many later trials, is presented below. (See "Clinical features, evaluation, and treatment of retroperitoneal soft tissue sarcoma", section on 'Adjuvant chemotherapy' and 'Updated meta-analyses' below.) Later randomized trials — Given the suggestion of a survival benefit for extremity and truncal STS in the SMAC meta-analysis, later randomized trials largely focused on these sites. There are few randomized trials addressing the benefit of adjuvant chemotherapy for visceral or head and neck soft tissue sarcomas. Four additional randomized trials explored the benefit of anthracycline and ifosfamidebased combination adjuvant chemotherapy in extremity STS [6-10], two of which suggest a possible survival benefit for adjuvant chemotherapy [6-8]: In an Italian trial in which 46 percent of the enrolled patients had either synovial sarcoma or liposarcoma (two relatively chemosensitive histologies), 104 patients with high grade large (≥5 cm) or recurrent spindle cell sarcomas involving the extremities or girdles were randomly assigned to no postoperative therapy or to five cycles of chemotherapy [6,7]. Chemotherapy consisted of a dose intensive 56 epirubicin/ifosfamide combination (epirubicin 60 mg/m2 on days 1 and 2 plus ifosfamide 1.8 g/m2 on days 1 to 5) with mesna and granulocyte colonystimulating factor support. Accrual was prematurely discontinued at two years, when a significant difference in the cumulative incidence of distant metastasis was found (45 versus 28 percent), favoring the chemotherapy group. Four-year overall survival was significantly greater in favor of chemotherapy (69 versus 50 percent), but the difference that favored the chemotherapy group lost statistical significance with median follow-up of over seven years, a finding that was attributed to the limited number of enrolled patients. Curiously, overall relapse rates (local and distant) were similar in the two groups (44 and 45 percent). It is difficult to interpret these results, since the main expected benefit of adjuvant systemic chemotherapy is to reduce the rate of distant relapse. However, this may have reflected a preponderance of patients with local recurrence later amenable to local surgical salvage approaches. A follow-up Italian trial randomly assigned 88 patients with high-risk extremity sarcoma to surgery with or without RT (n = 43) or to surgery plus chemotherapy (n = 45, 26 with epirubicin alone, and 19 to epirubicin plus ifosfamide) with or without RT [8]. The five-year survival rate of patients treated with chemotherapy was significantly higher than that of patients who did not receive chemotherapy (72 versus 47 percent). However, the large number of treatment variables and the small number of studied patients makes interpretation of this result difficult. In contrast to these results, a survival benefit from adjuvant doxorubicin and ifosfamidecontaining chemotherapy could not be shown in two other trials [9,10]: The EORTC randomly assigned 351 patients with completely resected STS (67 percent extremity tumors, 60 percent high-grade, 40 percent ≥10 cm) to observation versus five cycles of adjuvant chemotherapy (doxorubicin 75 mg/m2 and ifosfamide 5 g/m2 per cycle) [10]. The estimated five-year relapse-free survival was similar in both arms as was overall survival (67 versus 68 percent, HR 0.94, 95% CI 0.68-1.31). Interpretation of these results is limited by the inclusion of patients with nonextremity, small, and low/intermediate-grade primaries, as well as the relatively low ifosfamide dose. An Austrian trial of 59 patients assigned to perioperative chemotherapy or surgery alone was also negative, but the small number of patients makes it likely that the study was underpowered to detect a small difference between the arms, if one were present [9]. Updated meta-analyses — An updated 2008 meta-analysis was conducted on published data from 18 randomized trials of 1953 patients with localized and potentially resectable soft tissue sarcoma that were reported between 1973 and 2002, including the Austrian and both Italian trials discussed above, but not the most recent large EORTC trial [11]. Five of the trials used doxorubicin plus ifosfamide, while the others used doxorubicin alone or in combination with other agents. The odds ratio (OR) for local recurrence was 0.73 (95% CI 0.56 to 0.94) in favor of chemotherapy; the corresponding value for distant and overall recurrence was 0.67 (95% CI 0.56 to 0.82), again favoring chemotherapy. These values are nearly identical to those found in the earlier meta-analysis (see 'SMAC meta-analysis' above). However, in contrast to the earlier analysis, the use of doxorubicin with ifosfamide was associated with a statistically significant overall survival benefit (OR for death 0.56, 95% CI 0.36 to 0.85). The absolute risk reduction for doxorubicin in combination with 57 ifosfamide was 11 percent (30 versus 41 percent risk of death). Benefit could not be shown for doxorubicin alone (OR 0.84, 95% CI 0.68 to 1.03), implying the fundamental importance of ifosfamide in the adjuvant treatment of sarcomas overall. Pooled analysis of the EORTC trials — In contrast to this result, a pooled analysis of individual patient data from the two largest adjuvant trials of doxorubicin and ifosfamide-based chemotherapy (both performed by the EORTC [10,12] and totaling 819 patients) presented at the 2008 ASCO meeting was negative [13]. Compared to surgery alone, the use of postoperative adjuvant chemotherapy was not associated with a significant survival benefit, except in the subset of patients undergoing incomplete (R1) resection. In multivariate analysis, tumor size, histologic subtype, and grade were not associated with any progression-free or overall survival benefit from adjuvant chemotherapy. Impact of histology — Adjuvant clinical trials in adult sarcomas have, out of necessity, included patients with multiple histologies, in contrast to pediatric studies, which focus on one specific sarcoma subtype. It is well recognized that myxoid/round cell liposarcomas and synovial sarcomas are relatively chemosensitive subtypes of STS, at least in the setting of metastatic disease. (See "Systemic treatment of metastatic soft tissue sarcoma", section on 'Histologic subtype and response to chemotherapy'.) Although it has been proposed that the benefit of adjuvant chemotherapy may be preferentially seen when patients are selected based upon tumor histology, grade [14] or tumor size, this hypothesis has never been validated in a prospective clinical trial in adult sarcoma patients, nor in any of the pooled analyses of randomized trial data [11,13]. Furthermore, the results from retrospective reports evaluating adjuvant chemotherapy in patients with the more chemotherapy-sensitive histologic subtypes are conflicting [1519]. While three contemporary retrospective series suggest a potential survival benefit for adjuvant chemotherapy in patients with the liposarcoma and synovial sarcoma subtypes of extremity STS [15,16,18], two others do not [17,19]. As examples: Benefit for adjuvant chemotherapy was suggested in a single-center Italian report of 251 patients (aged 5 to 87 years) with a localized synovial sarcoma [16]. Adjuvant chemotherapy was administered to 61 of 215 patients who had a macroscopically complete resection (28 percent), while the remainder received no adjuvant systemic therapy. Five-year metastasis-free survival was greater for those treated with chemotherapy (60 versus 48 percent), and benefit appeared to be greatest for patients aged 17 or older who had tumors measuring >5 cm (fiveyear metastasis-free survival 47 versus 27 percent, respectively). On the other hand, a lack of long-term benefit from adjuvant chemotherapy was suggested in a report of the combined experience of two major cancer cancers (MSKCC and M. D. Anderson) that included 674 consecutive adults undergoing resection of a high-grade, ≥5 cm extremity STS between 1984 and 1999 [17]. Adjuvant doxorubicin-based chemotherapy was administered to 336 (50 percent), while the remainder received local therapy only. Although not a randomized trial, there were no statistically significant differences between the chemotherapy and local therapy alone groups with respect to tumor size, anatomic site, histopathologic subtype, or resection margin status. With a median follow-up of 6.1 years, the effect of chemotherapy appeared to vary over time. During the first year, the HR for disease-specific survival for chemotherapy versus no chemotherapy was 0.37 (95% CI 0.20-0.69); thereafter, the HR was 1.36 (95% CI 1.02-1.81) and did not vary according to histology or tumor size. 58 Interpretation of retrospective data such as these is hampered because several biases are operative that favor the control arm: Chemotherapy was likely recommended for those patients whose tumors were thought to have the highest risk of recurrence, while those thought to have more favorable outcomes were not offered chemotherapy. Although none of the differences were statistically significant, there were differences in tumor histology and size between the two groups. In the combined report from MSKCC and M. D. Anderson, 50 percent of the patients who did not receive chemotherapy had 5 to 10 cm primary tumors, while only 42 percent of those receiving chemotherapy had this relatively favorable tumor size. It is well recognized that the risk of metastasis increases for every centimeter increase in the size of a primary sarcoma. In addition, 21 percent of the control patients had liposarcoma histology compared to 14 percent in the treatment group. These data can be interpreted positively or negatively; the negative view is that overall survival was not improved with the use of chemotherapy. The positive view is that patients with inferior prognosis had their survival improved to that of lower risk patients with the use of systemic chemotherapy in the adjuvant setting. Summary — The role of chemotherapy for patients with a resected extremity STS remains uncertain and controversial [20,21]. The updated meta-analysis from the Sarcoma Meta-Analysis Collaboration [11] suggests that use of an optimal, adequately dosed anthracycline/ifosfamide-containing regimen significantly prolongs survival in patients with resected extremity STS [11] with the ifosfamide component as the more important of the two drugs. However, the analysis did not include data from the single largest negative trial, which so far has been reported only in abstract form [12]. A preliminary report of a pooled analysis of this trial and another European large trial, both of which tested the value of an anthracycline and ifosfamide-containing regimen, indicated no benefit from adjuvant anthracycline-ifosfamide chemotherapy [13]. Taken together, despite the most recent positive meta-analysis, it is difficult to recommend adjuvant chemotherapy as a standard practice for all patients with extremity STS. If there is a survival benefit for doxorubicin-based adjuvant chemotherapy, it appears to be small, no more than 5 percent absolute increase in survival at 5 to 10 years [11]. Conversely, the positive meta-analysis from 2008 is arguably the strongest evidence in support of the use of adjuvant chemotherapy. The challenge wrought by treating many different sarcoma subtypes with one particular therapeutic plan has not worked in other cancers, and should be not be expected to apply to what are 50 or more sarcoma subtypes. In keeping with the consensus-based guidelines of the National Comprehensive Cancer Network (NCCN) and the European Society of Medical Oncology [22], which encompass adjuvant chemotherapy as an option for high-risk patients, our present approach is to individualize therapy, taking into consideration the patient's performance status, comorbid factors (including age), site of disease, and histologic subtype (eg, younger patients with synovial sarcoma or the round cell version of myxoid liposarcoma). The potential for benefit from adjuvant chemotherapy must be discussed in the context of expected treatment-related toxicities, including potential sterility in younger people, cardiomyopathy, renal damage, second cancers, and overall impairment of quality of life. Even for those patients in whom the decision has been made to administer adjuvant chemotherapy, the optimal regimen is undefined. We prefer five to six cycles of doxorubicin (usually 75 mg/m2 per cycle in split bolus doses or continuous infusion over three days), ifosfamide (9 to 10 g/m2 in split doses over three hours per day for three to four days), with mesna [AIM (table 1 and table 2)] rather than MAID (mesna, doxorubicin, ifosfamide, and dacarbazine [23]) since this permits the administration of 59 maximal doses of the two most active drugs for sarcoma (doxorubicin and ifosfamide), rather than adding myelotoxicity with dacarbazine. Three cycles of chemotherapy may suffice, although the data supporting this statement are limited to a single trial of three versus five cycles of chemotherapy in 328 patients with high-risk (deeply seated, high grade, or large [≥5 cm] primary, or locally recurrent) extremity sarcomas [24]. Given the limitations of the available data, we still consider that five or six cycles of chemotherapy represents the preferred approach. (See "Treatment protocols for soft tissue and bone sarcoma".) Given that the benefit of chemotherapy appears to be dependent upon the use of ifosfamide, at least from the most recent meta-analysis, and the known age-dependent toxicities of ifosfamide, it makes a decision to use adjuvant chemotherapy in older patients even more difficult. In view of the uncertainty of long-term benefit for patients with most sarcoma subtypes, extreme caution is indicated in treating elderly patients with adjuvant chemotherapy. NEOADJUVANT CHEMOTHERAPY — Theoretical advantages to neoadjuvant approaches to therapy include tumor cytoreduction, immediate treatment of micrometastases, and an early indication as to the effectiveness of chemotherapy/radiotherapy. Cytoreduction may allow less radical surgical resection to be performed, and this approach is often considered in patients with large extremity sarcomas, particularly if the patient is a borderline candidate for limb salvage surgery. Most often, when induction therapy is considered for a patient with a large or recurrent extremity sarcoma, particularly if limb salvage is an issue, radiotherapy is chosen with or without chemotherapy. Adjuvant radiotherapy with and without chemotherapy is discussed in detail elsewhere. The benefit of induction chemotherapy alone in this setting is uncertain. Adequately powered, randomized phase III trials are not available, and the results of uncontrolled studies are conflicting [25,26]. A randomized phase II EORTC study in which 150 patients were randomly assigned to three cycles of neoadjuvant doxorubicin (50 mg/m2 per cycle) plus ifosfamide (5 g/m2 per cycle) versus surgery alone failed to show better survival in the chemotherapy arm, and the trial was stopped before expansion into a phase III study [27]. The low chemotherapy intensity used in this study may have contributed to the negative result. At least some data supports the use of neoadjuvant trabectedin, where available, in patients with locally advanced myxoid liposarcoma. In an uncontrolled study of 23 patients, seven had an objective partial response, and at surgery, three had a pathologic complete response after three to six cycles of trabectedin (1.5 mg/m2 over 24 hours, once every 21 days) [28]. Trabectedin is available in Europe and in some other countries, but not in the US. (See "Systemic treatment of metastatic soft tissue sarcoma", section on 'Trabectedin'.) Given the lack of data regarding the impact of adjuvant or neoadjuvant trabectedin on long-term outcomes, its use in these settings is still considered experimental. Chemotherapy with regional hyperthermia — A European randomized trial reported benefit from the addition of regional hyperthermia to neoadjuvant chemotherapy compared to neoadjuvant chemotherapy alone among patients with large high grade tumors, or initially unresectable disease. This approach, which is not widely used outside of Europe, is discussed in detail elsewhere. (See "Treatment of locally recurrent and unresectable, locally advanced soft tissue sarcoma of the extremities", section on 'Chemotherapy with regional hyperthermia'.) Summary — The role of neoadjuvant chemotherapy in the management of STS remains undefined. Nevertheless, NCCN guidelines suggest that preoperative chemotherapy (followed by postoperative RT) is an acceptable alternative to preoperative RT or preoperative chemoradiotherapy for patients with large (>10 cm) potentially resectable 60 STS or if there is concern for adverse functional outcomes from initial surgery. For most of these patients, we prefer radiation therapy with or without chemotherapy rather than chemotherapy alone. These patients should be managed in a center with multidisciplinary expertise in the management of STS, and ideally, treatment should be administered in the context of a clinical trial. If such a trial is not available or if patients are ineligible, we suggest sequential rather than concurrent chemotherapy and RT. We administer chemotherapy first (typically AIM rather than MAID) followed by RT and then surgery. Others utilize concomitant chemoradiotherapy using low radiosensitizing doses of chemotherapy during RT. The specific approach used remains a function of an institution's experience and expertise. However, it should be emphasized that there are no data that simultaneous administration of chemotherapy and RT improves local control over sequential use; it is well recognized from larger randomized studies that adjuvant chemotherapy improves local relapse-free survival rates. (See "Local treatment for primary soft tissue sarcoma of the extremities and chest wall", section on 'Initial chemoradiotherapy for large high-grade STS'.) SUMMARY AND RECOMMENDATIONS Surgical resection is the cornerstone of treatment for virtually all patients with an extremity soft tissue sarcoma. The combination of surgery and radiation therapy (RT) achieves better outcomes than either treatment alone for nearly all soft tissue sarcomas more than 5 cm in greatest dimension. (See 'Introduction' above.) Systemic chemotherapy is a routine component of treatment for several soft tissue sarcomas that occur predominantly in children (ie, rhabdomyosarcoma, Ewing sarcoma). (See 'Pediatric-type sarcomas' above.) However, despite many randomized trials, the role of adjuvant chemotherapy for the more common adult subtypes of soft tissue sarcoma (such as liposarcoma, synovial sarcoma, and leiomyosarcoma) remains uncertain: The most recent analysis from the Sarcoma Meta-Analysis Collaboration (SMAC) suggests a significant 11 percent improvement in survival for doxorubicin and ifosfamide-based adjuvant chemotherapy compared to resection alone [11]. However, a pooled analysis of individual patient data from the two largest adjuvant trials of doxorubicin and ifosfamide-based chemotherapy (both performed by the EORTC [10,12], only one of which was included in the SMAC meta-analysis) presented at the 2008 ASCO meeting was completely negative [13]. (See 'Pooled analysis of the EORTC trials' above.) There is no evidence that adjuvant chemotherapy is relatively more beneficial for the chemotherapy-sensitive subtypes of STS. (See 'Impact of histology' above.) Thus, in our view, this approach cannot be adopted as a standard treatment for all extremity soft tissue sarcomas, regardless of histology. In keeping with the guidelines of the National Comprehensive Cancer Network (NCCN) and the European Society for Medical Oncology [22], we suggest that the appropriateness of adjuvant chemotherapy be addressed on a case by case basis, taking into consideration the patient's performance status, comorbid factors (including age), disease location, tumor size, and histologic subtype. The potential for benefit must be discussed in the context of expected treatment-related 61 toxicities including sterility in younger individuals, cardiomyopathy, renal damage, second cancers, and overall impairment of quality of life. For patients who decide to undergo adjuvant chemotherapy despite the uncertainty of benefit and the potential for treatment-related toxicity, we recommend using a regimen that contains both ifosfamide and doxorubicin (Grade 1A). We prefer doxorubicin plus ifosfamide and mesna [AIM (table 2 and table 1)] rather than the MAID regimen. (See 'Summary' above and "Treatment protocols for soft tissue and bone sarcoma".) Neoadjuvant therapy is most often considered in the setting of a large or recurrent high-grade tumor, particularly if limb salvage is an issue. In these situations, RT is most commonly selected with or without chemotherapy. The optimal neoadjuvant regimen, and how best to integrate radiation therapy, chemotherapy, and surgery are unknown. The benefit of induction chemotherapy alone in this setting is uncertain. However, NCCN guidelines indicate that preoperative RT alone, preoperative chemotherapy (with postoperative RT), or preoperative chemoradiotherapy are all acceptable options. Where available, another option is neoadjuvant chemotherapy combined with regional hyperthermia. There are no data to support one approach over the other, and the specific neoadjuvant approach chosen typically depends upon institutional expertise and experience. If possible, we prefer that these patients be treated in the context of a clinical trial. Bone sarcomas: Preoperative evaluation, histologic classification, and principles of surgical management 62 63 64 65 INTRODUCTION — Osteosarcoma is the most common primary malignant tumor of bone. Osteosarcomas are characterized by the production of osteoid tissue or immature bone by the malignant cells [1-3]. Osteosarcomas are uncommon tumors compared to carcinomas, with approximately 900 cases diagnosed each year in the United States, mainly in children and adolescents [4]. Among 15 to 29 year olds, bone tumors account for 3 percent of all tumors, and osteosarcoma accounts for about one-half of these cases [5]. Most osteosarcomas present as high-grade tumors and most are located around the anatomic regions of high growth rate. The survival of patients with malignant bone sarcomas has improved dramatically over the past 30 years, largely as a result of chemotherapeutic advances. Before the era of 66 effective chemotherapy, 80 to 90 percent of patients with osteosarcoma developed metastatic disease despite achieving local control from surgery and died of their disease. It was surmised (and subsequently demonstrated [6]) that the majority of these patients had subclinical metastatic disease that was present at the time of diagnosis, even in the absence of overt metastases. In osteosarcoma, chemotherapy can successfully eradicate microscopic deposits in the majority of cases if initiated at a time when disease burden is low (ie, following resection of the primary tumor). As a result, all patients with intermediate- or high-grade osteosarcoma receive chemotherapy, although the optimal timing (ie, preoperative or postoperative) is controversial (see 'Adjuvant therapy' below) [7]. Low-grade osteosarcomas, such as parosteal osteosarcomas, are not treated routinely with chemotherapy because the risk of metastatic spread is low. With modern therapy, approximately two-thirds of patients with non-metastatic extremity osteosarcoma will be long-term survivors, up to 50 percent of those with limited pulmonary metastases may be cured of their disease, and long-term relapse-free survival can be expected in about 25 percent of those who present with metastatic disease overall [8-12]. Surgical management has evolved in parallel with the emergence of effective chemotherapy. Although complete extirpation of the tumor remains the primary objective, the nature and scope of the approach taken to accomplish this goal has changed, with an emphasis on more conservative surgery in order to maintain function. Functional outcome depends not only on the extent of resection and the amount of muscle that is removed, but also the quality of the reconstruction and its associated complications. Limb-sparing surgery rather than amputation is now possible in the majority of patients, particularly when preoperative (neoadjuvant) chemotherapy is used. Here we will discuss the principles of surgical management for primary bone sarcomas. Although the focus of this topic review will be on osteosarcoma, most of the same surgical principles apply to other bone sarcomas as well. Intermediate or high-grade fibrosarcoma of bone (previously referred to as malignant fibrous histiocytoma) is treated in a similar manner as osteosarcoma. Other bone sarcomas, such as leiomyosarcoma and chondrosarcoma, may be managed with a slightly different approach because of differences in their responsiveness to chemotherapy and radiation therapy. As an example, neoadjuvant chemotherapy is usually not considered routine for chondrosarcomas, because they are relatively chemoresistant. Leiomyosarcoma of bone has typically been treated in a similar manner to that of the leiomyosarcoma in soft tissues. In contrast, Ewing sarcoma is usually approached initially with systemic chemotherapy due to its chemosensitivity. Radiation and/or surgery may be used to aid in local control for Ewing sarcoma. Surgery can be a component of management of metastatic disease both at initial presentation, and at the time of recurrence, is addressed elsewhere. PREOPERATIVE EVALUATION — The goal of the preoperative evaluation is to establish the tissue diagnosis, evaluate disease extent, and assess the feasibility of a limb-sparing approach. Clinical staging includes all of the data obtained prior to definitive therapy, including the results of imaging, physical examination, laboratory studies, and tissue biopsy. A thorough history and physical examination that evaluates all systems is necessary prior to definitive treatment. However, physical examination should focus particularly on the involved bone or joint. The regional lymph nodes (although they are a rare site for metastases), other bones (sometimes synchronous but may be site for distant disease), and lungs (the most common metastatic site) are sites for disease dissemination. 67 Radiographic imaging — Plain radiographs are nearly always a good way to begin to evaluate a bone lesion, as they provide data that often cannot be replicated even with MRI or CT. Nonetheless, although plain radiographs can often predict the probable histology of a potentially malignant bone lesion, the definition of tumor size and local intraosseous and extraosseous extent is most accurately achieved by magnetic resonance imaging (MRI). The entire involved bone should be imaged to avoid missing skip metastases (ie, medullary disease within the same bone, but not in direct contiguity with the primary lesion) (image 1). CT scans are best suited to evaluate the thorax for metastatic disease, which is essential because approximately 80 percent of metastatic lesions in osteosarcoma occur in the lungs [13,14]. The most common metastatic site for all bone sarcomas is pulmonary. Thin-section imaging of the chest using high-resolution helical CT is the preferred modality, detecting approximately 20 to 25 percent more nodules than conventional CT, and the reliable detection of nodules as small as 2 to 3 mm. Radionuclide bone scanning with technetium is the preferred method for evaluating the entire skeleton for the presence of multiple lesions and help in identifying skip metastases [15]. Although positron emission tomography (PET) scans may have greater utility for assessing the response of the primary tumor to neoadjuvant chemotherapy, at least one study suggests it is inferior to radionuclide bone scanning for the detection of osseous metastases from osteosarcoma [16] and to spiral (helical) CT for detecting pulmonary metastases [17]. Nevertheless, guidelines from the National Comprehensive Cancer Network (NCCN) suggest a PET scan and/or bone scan in the workup of a suspected osteosarcoma, and imaging guidelines from the Children’s Oncology Group Bone Tumor Committee for both osteosarcoma and Ewing sarcoma recommend radionuclide bone scan and/or PET scan for whole body staging. This subject is addressed in detail elsewhere. Tissue biopsy — Biopsy of the tumor completes the staging process. As with soft tissue sarcomas, the biopsy must be carefully planned to ensure that adequate tissue is obtained for diagnosis without compromising the opportunity for limb salvage. Biopsies should take place after the completion of the staging studies, and the surgeon, radiologist, and pathologist should review these studies in detail so that each member of the team is fully appraised of the diagnostic considerations. Either a core needle or open biopsy may be performed as long as adequate tissue is obtained. For CT-guided core biopsies, multiple cores of tissue should be obtained for frozen section (to ensure that adequate tissue has been obtained), permanent hematoxylin and eosin (H&E) stained sections, microbial culture (in case the diagnosis should turn out to be osteomyelitis), and one core is reserved for special stains or cytogenetic studies. Although fine needle aspiration biopsy under radiologic guidance may obviate the need for open biopsy, the risk of a nondiagnostic or nonrepresentative sample must be considered [18]. In one report of 359 patients with musculoskeletal lesions, the accuracy rates of CT-guided biopsies and fine needle aspirates were 74 and 63 percent, respectively [19]. If an open biopsy is performed, the incision should be placed in accordance with the planned surgical resection; the primary tumor and the entire biopsy tract should be resected en bloc. Meticulous hemostasis and the judicious use of a drain are important to avoid the spread of hematoma-containing tumor cells. If a soft tissue mass is not present, or material is nondiagnostic, a bone defect may be required to obtain tissue. If so, it should be a small, oval defect, and a polymethylmethacrylate plug may be used to close the hole in order to minimize hematoma. Tumor staging — The staging system used primarily for bone sarcomas was developed by Enneking et al at the University of Florida and based upon a retrospective review of 68 cases of primary malignant tumors of bone treated by primary surgical resection (table 1) [20,21]. This system characterizes nonmetastatic malignant bone tumors by grade (low grade [stage I] versus high grade [stage II]), and further subdivides these stages according to the local anatomic extent. The compartmental status is determined by whether the tumor extends through the cortex of the involved bone. Patients with distant metastases are categorized as stage III. The American Joint Committee on Cancer (AJCC) adopted a similar staging system in its 1997 fifth edition [3], but this was modified in 2002, with additional minor refinements in the latest 2010 edition (table 2) [22]. The TNM classification is not widely used for primary bone tumors, particularly for Ewing sarcoma, since it does not take into account several important clinicopathologic features such as primary tumor site, patient age, and histology. The staging classification used for Ewing sarcoma is complex and discussed in detail elsewhere. HISTOLOGIC CLASSIFICATION — Primary bone sarcomas are classified according to their cytologic features and cellular products (eg, osteoid) (table 3). The different types of osteosarcomas are briefly described here. For other sarcomas of bone (eg, chondrosarcoma), subtypes exist as well, based upon the histology. .) The histologic appearance of the Ewing sarcoma family of tumors is described elsewhere. The diagnosis of a primary bone malignancy should be rendered only in the setting of radiologic studies, including plain films and MRI or CT of the primary lesion. While radiologic or histologic findings may not be specific by themselves, a histologic diagnosis in the context of consistent radiographic findings secures the diagnosis with much greater certainty. The histologic diagnosis of an osteosarcoma depends upon the presence of a malignant sarcomatous stroma associated with the production of tumor osteoid or tumor bone. Osteosarcomas are thought to arise from a mesenchymal stem cell that is capable of differentiating towards fibrous tissue, cartilage, or bone. As a result, they share many features with chondrosarcomas and fibrosarcomas of bone. However, chondrosarcomas and fibrosarcomas of bone are distinguished by their lack of osteoid substance, which is required for the diagnosis of osteosarcoma (table 4). Of note, the degree of osteoid production in osteosarcomas may be limited; because of this and variability in tumor morphology, electron microscopy may be required for histologic confirmation. The largest group of osteosarcomas are conventional (intramedullary high grade) osteosarcomas, which account for approximately 90 percent of all osteosarcomas [23]. These tumors usually involve the metaphysis of long bones in adolescents and young adults (figure 1). Conventional osteosarcomas can be classified as osteoblastic, chondroblastic, or fibroblastic, depending upon the predominant cellular component (table 3); all are managed similarly if they are of the same grade: Osteoblastic osteosarcoma, which accounts for 50 percent of cases, is characterized by abundant osteoid production, which forms a fine or coarse lacelike pattern around the tumor cells; massive amounts may result in distortion of the malignant stromal cells. Some tumors contain thick trabeculae of osteoid that form an irregular anastomosing network. The amount of mineralization is variable. Fibroblastic osteosarcomas are predominantly composed of a high-grade spindle cell stroma that contains only focal osteoid production. More pleomorphic tumors may resemble malignant fibrous histiocytoma, but the distinction can be made by the identification of osteoid. In chondroblastic osteosarcomas, cartilaginous matrix production may be evident throughout most of the tumor or only in certain locations, underscoring the heterogeneity of these tumors. While some chondroblastic osteosarcomas tend to 69 be lower grade, the chondroid areas may contain cytologically atypical cells that are characteristic of higher grade tumors. These chondroblastic foci are admixed with malignant spindle cells that produce osteoid trabecula. In addition to the three categories of conventional osteosarcoma, there are several variants (ie, telangiectatic, small cell, multifocal, and the malignant fibrous histiocytoma [fibrosarcoma] subtypes) that were originally thought to carry a worse prognosis. However, with modern aggressive therapy, they behave in general like conventional osteosarcomas and are treated similarly: Small cell osteosarcoma is noteworthy for the confusion that may arise in distinguishing it from other "small round blue cell tumors" such as Ewing sarcoma by conventional light microscopy of H&E stained sections [24]. Immunohistochemical staining, cytogenetics, and molecular genetic studies may be required to establish the diagnosis. Telangiectatic osteosarcoma has a purely lytic appearance on plain radiographs, and there is a high rate of pathologic fracture [25]. Grossly, these tumors appear like a "multicystic bag of blood"; a solid mass of tumor is usually absent [23]. Telangiectatic osteosarcomas have overlapping features with aneurysmal bone cysts and giant cell tumors. However, the cells are highly pleomorphic, unlike the benign appearance of the aneurysmal bone cyst. Because they often contain minimal osteoid and numerous multinucleated giant cells, they may be mistaken for a giant cell tumor. Although once thought to carry a poorer prognosis that conventional osteosarcomas, more recent series indicate similar survival with multimodality therapy [25,26]. Fibrosarcoma of bone (previously referred to as malignant fibrous histiocytoma of bone) has an appearance similar to osteosarcoma, but without osteoid production. Although these tumors tend to have a lower rate of tumor necrosis following induction chemotherapy, long-term survival rates for intermediate and high-grade fibrosarcomas of bone are similar to conventional osteosarcomas [27], with the possible exception of spine primaries [28]. With very careful examination, occasionally osteoid will be found in rare sections of tumor, suggesting that this subtype of primary bone tumor may represent an undifferentiated version of a more classic osteosarcoma. In contrast to these variants, surface or juxtacortical osteosarcomas differ as a group with respect to prognosis and therapy. These osteosarcomas include [29-31]: Parosteal osteosarcoma is a low-grade osteosarcoma Periosteal osteosarcomas are usually chondroblastic intermediate grade tumors (image 2) There are rare high-grade surface osteosarcomas Among the surface osteosarcomas, surgery alone may be curative, particularly for tumors that are low-grade and do not enter the marrow cavity [32,33]. However, at least some data support the benefit of adjuvant chemotherapy in periosteal and highgrade surface osteosarcomas [31]. Given the rarity of this diagnosis, it is worthwhile obtaining a second opinion regarding the pathology and radiology for such tumors at expert centers. Because of the rarity of osteosarcoma in adults, other cancers should be considered in the differential diagnosis, such as primary lymphoma of bone and metastatic carcinoma, in addition to the primary bone tumors discussed above. (See "Primary lymphoma of bone".) 70 Besides osteosarcomas, fibrosarcomas, and chondrosarcomas, other less common primary bone tumors include primitive neuroectodermal tumor/Ewing sarcoma, chordomas, angiosarcomas (a subtype with a particularly high rate of metastatic disease at presentation [34]), and a variety of other rare tumor types (table 3). Histologic characteristics of the Ewing sarcoma family of tumors are discussed elsewhere. SURGICAL MANAGEMENT OF THE PRIMARY TUMOR — Despite the favorable response of osteosarcomas to chemotherapy (see 'Neoadjuvant chemotherapy' below), surgery is a necessary component of curative therapy [35]. The specific surgical procedure is dictated by the location and extent of the primary tumor [36]. Axial tumors tend to do worse probably because they are larger when discovered and more difficult to resect. Although all patients with extremity sarcomas are candidates for amputation, emphasis on functional outcome has focused efforts on limb-sparing procedures. However, not all patients are candidates for more conservative surgery (see 'Patient selection' below). In order to avoid sacrificing oncologic outcome, tumor control must be the primary therapeutic concern, and functional outcome a secondary goal. One of the most important tasks of the preoperative evaluation is to assess the feasibility of performing a limb-salvage procedure based upon the clinical presentation and disease extent (see 'Preoperative evaluation' above). Types of resections — Bone resections fall into one of three categories, depending upon the anatomic site and the extent of the involved bone that needs to be excised. Because most bone sarcomas arise in the metaphysis of the long bones near a joint (figure 2), the majority of resections, for tumors in the lower extremity, include both the segment of tumor-bearing bone and the adjacent joint (osteoarticular resection). Most often, the incision is performed through the joint (intraarticular resection); however, when the tumor extends along the joint capsule or ligamentous structures or invades the joint, the entire joint can be resected (extraarticular resection) to avoid cutting through tumor. Less frequently, tumor arises within the diaphysis or shaft region of a long bone, and the bone alone is resected (an intercalary resection). Uncommonly, extensive involvement along the length of the bone precludes adequate resection and reconstruction without sacrificing the entire bone, and a whole bone resection, including both proximal and distal joints, is required. Resection margins — An important consideration in selecting the type of operation is the ability to attain a negative margin of resection with the planned surgical procedure. The quality of the tissue forming the margin is as important as the distance between tumor and uninvolved tissue. The Musculoskeletal Tumor Society recognizes a wide local excision either by amputation or a limb-sparing procedure as the recommended surgical approach to bone sarcomas [37]. The wide local excision removes the primary tumor en bloc along with its reactive zone and a cuff of normal tissue in all planes. However, in many cases, particularly for tumors involving the spine, wide local excision cannot be accomplished easily. An "intralesional" surgical resection margin is more frequently obtained during a procedure that removes the tumor in a piecemeal fashion or by curettage, while a "marginal" resection margin refers to a situation where the pseudocapsule and reactive zone surrounding the tumor form the surgical margin. These operations may result in residual tumor cells being left behind. Extremity lesions: limb-sparing procedures — For lesions involving either the upper or lower extremity, limb salvage can improve functional outcome without sacrificing local disease control as long as complete tumor resection is anatomically possible and adjuvant chemotherapy is used [12,38,39]. Although the functional results are generally better, the available data do not support the position that quality of life or long-term psychosocial outcome is substantially superior after limb salvage than after amputation [40-44]. There is, however, a higher complication rate in patients who undergo limb salvage as compared to amputation [40,45]. 71 Despite the increasing numbers of limb salvage procedures, there are no randomized prospective studies that prove its oncologic safety compared to amputation. However, retrospective series with long-term follow-up of patients with osteosarcoma who undergo amputation or limb salvage do not show a difference in overall or disease-free survival [46]. Nevertheless, appropriate patient selection is critical (see below). If there is any doubt that a wide local excision can be accomplished, amputation, the oncologic "gold standard", is the indicated procedure in order to avoid local recurrence, which is invariably followed by metastatic tumor spread and diminished survival [47-50]. Compared to amputation, limb salvage procedures are usually associated with narrower surgical margins, which can increase the likelihood of a local failure [48,51]. The degree of chemotherapy-induced tumor necrosis and surgical margin status are the important prognostic factors for local control [49,52]. For patients undergoing neoadjuvant chemotherapy in order to increase the feasibility of a limb sparing procedure, it is unclear whether the reduction in surgical margins afforded by a favorable antitumor response increases the likelihood of a local recurrence. The available data are contradictory. Several multicenter series suggest an increased risk of local recurrence, even in patients with a good response to chemotherapy [49,51,53], while other single institution studies indicate similar local recurrence rates for limb-sparing and amputation procedures [48,52]. Patient selection — Tumor location and extent are the most important determinants of the feasibility of limb-sparing surgery. Among the contraindications for limb-sparing surgery are nerve and/or vessel encasement by tumor, the presence of a large biopsyrelated hematoma, and possibly the presence of a pathologic fracture. A pathologic fracture is present at diagnosis or occurs during the course of treatment in 5 to 10 percent of patients with osteosarcoma (image 3). Historically a pathologic fracture was considered an indication for immediate amputation because of the theoretical presence of tumor cells within the hematoma, and a greater risk of both local recurrence and shorter survival with limb-sparing procedures [54-58]. However, more recent studies have suggested that limb salvage is possible with an acceptable rate of local control: A multi-institutional series from the Musculoskeletal Tumor Society compared outcomes for 52 patients with osteosarcoma and a pathologic fracture versus those in 55 concurrently treated patients, matched for age and tumor location, who had an osteosarcoma without a pathologic fracture [54]. Overall, both fiveyear overall (55 versus 77 percent) and local recurrence-free survival rates (75 versus 96 percent) were significantly poorer for those presenting with a pathologic fracture. However, there were no significant differences in outcome when limb salvage was compared to amputation in patients with a pathologic fracture: 11 of 30 patients managed with limb salvage (37 percent) versus 10 of 22 who underwent amputation (45 percent, p = 0.05) died of their disease. A single-institution retrospective review compared outcomes of 31 patients with high-grade osteosarcoma who presented with a pathologic fracture versus 201 patients without a pathologic fracture [59]. In the pathologic fracture group, 19 patients had limb salvage surgery (61 percent), while 12 underwent an amputation; limb salvage surgery was performed in 86 percent of the non-fracture group. Five-year overall survival was significantly worse for patients presenting with a pathologic fracture (41 versus 60 percent), but there was no difference in local recurrence rate after limb-sparing surgery in the pathologic fracture versus no-fracture group (10 versus 8 percent). Other reports indicate that healing of pathological fractures during neoadjuvant chemotherapy may permit limb-salvage surgery to be undertaken safely, as long as wide surgical margins can be achieved. In many of these series, the majority of patients were 72 treated with limb-sparing approaches, and the presence of a pathologic fracture did not constitute a negative prognostic factor for either local recurrence or survival [55,60]. Despite these data, many investigators still consider a pathologic fracture to be a relative rather than an absolute contraindication to limb-sparing procedures. Indications for amputation — In certain circumstances, amputation or rotationplasty (see below) may be preferred over limb-sparing surgery for extremity sarcomas. Prosthetic improvements have enhanced the amputee's functional results. Furthermore, some types of amputation do not require the patient to wear a prosthesis, and they are associated with lesser cosmetic and functional deformity as compared with those individuals who have undergone an amputation and need a prosthesis to return to the normal activities of daily living. Although amputation does not preclude a local recurrence, the risk is less than 5 percent. Stump recurrence has been attributed to extensive intramedullary tumor spread and the existence of skip lesions. It is important to assess the joint that is adjacent to the primary extremity sarcoma to determine whether an intraarticular or extraarticular resection will be necessary and to exclude the presence of skip metastases within the involved bone (image 1). Among the factors that lessen the likelihood of local recurrence are a wide surgical margin and a good histologic response to neoadjuvant therapy (see below) [49]. Rotationplasty and the skeletally immature patient — Growth considerations following resection and reconstruction are not a major concern for patients with lower extremity lesions who are at or near skeletal maturity or for lesions involving the upper extremity. However, for skeletally immature children with lower extremity lesions, vigorous functional demands, and limb length inequality following a limb-sparing procedure is a major problem. In this setting, surgical options include rotationplasty, amputation, or reconstruction with an expandable metal endoprosthesis. (See 'Allografts and endoprostheses' below.) The Van Ness rotationplasty is a surgical technique for lesions involving the femur that was originally described in 1930. It involves resection of the tumor and surrounding tissues as an intercalary amputation, saving only the sciatic nerve, while either twisting and curling the femoral vessels or segmentally resecting a portion of the vessels with a subsequent vascular anastomosis [61-63]. Typically, the distal femur and its adjacent musculature are excised, and the tibia and foot preserved by maintaining an intact neurovascular bundle. The tibia is then fixed with plates and screws to the proximal femur, and the distal extremity rotated 180º, permitting the foot and ankle to function like a knee (picture 1). The vessels adjacent to the tumor can be resected and reanastomosed if necessary. The retained foot acts as the amputation stump for a transtibial amputation, and this prosthetic arrangement requires less energy expenditure than that required for a transfemoral amputation. Although the appearance of the reconstructed limb is cosmetically displeasing, the major advantage of rotationplasty in the skeletally immature patient is the ability to tailor the limb length at maturity. Functional results are excellent, with patients able to achieve recreational, sporting, and career goals [64]. A series of 25 children treated with rotationplasty focused on risk factors for failure and postoperative complications [65]. The majority of cases (22 of 25) produced excellent functional outcome. The few failures were a result of vascular impairment by the tumor, necessitating amputation. Patients with larger tumors, unresponsiveness to chemotherapy, or a preoperative pathologic fracture appeared to be at higher risk of failure. In view of these favorable functional results, rotationplasty is often the reconstructive procedure of choice for children under the age of eight who have proximal lower extremity sarcomas. However, some patients and their families refuse rotationplasty and 73 amputation, and allografts are frequently used to reconstruct the defect in the skeletally immature patient, with standard limb equalizing procedures carried out at a later date. An alternative to allografts is the use of expandable or modular metal prostheses which can be lengthened as the child grows. All of these approaches have limitations, and their advantages and disadvantages in very young patients are discussed below (see 'Reconstruction techniques' below). Upper extermity lesions involving the shoulder — Tumors of the scapula present an unusual challenge. Although subtotal or complete scapulectomy can be considered for some lesions [66,67], limb reconstruction is difficult and shoulder morbidity may be substantial. For massive tumors of the shoulder girdle (including the proximal humerus), the Tikhoff-Linberg (interscapulothoracic) resection is an alternative to forequarter amputation as long as resection of the brachial plexus is not necessary [68]. The proximal humerus and scapula are resected without the need for bony replacement, and the function of distal arm, elbow, and hand is preserved. Because the resulting shoulder region is unstable, with telescoping of the intervening structures when stresses are applied to the hand or forearm, stabilization is necessary using dacron tape. This is usually sufficient to prevent telescoping and stress to the remaining neurovascular structures. When the scapula can be preserved, an arthrodesis with or without allograft can be attempted even if the surrounding deltoid and rotator cuff musculature requires resection. Non-extremity lesions — The principles derived from extremity sarcoma resections have now been applied to the axial structures. Osteosarcomas involving the axial skeleton have a worse prognosis than those involving the extremities, in part due to the difficulty in achieving complete surgical resection in these difficult locations [69-71]. In particular, primary malignant non-extremity tumors are more difficult to resect secondary to their large size, surrounding neurologic and vascular structures, reconstructive challenges, and frequent comorbidities. Pelvic and spine tumors are associated with unique challenges. In the spine, en bloc resections may be intralesional, marginal, or wide. Radical resections or removing the entire epidural space is not feasible. The dura and epidural space offer a unique region for tumor management. Pelvic tumors — Pelvic tumors represent a unique challenge. A tumor arising in the pelvis can involve a segment of nonarticular bone, the acetabulum, or both. En bloc excision of the hemipelvis with preservation of the extremity (internal hemipelvectomy) produces equivalent oncologic results when compared to external formal hemipelvectomy (hindquarter amputation) and better functional outcome [72,73]. It is preferred over external hemipelvectomy, usually because of patient preference. External hemipelvectomy may be necessary in up to one-third of patients, and despite this, local recurrence rates are high because of difficulty obtaining adequate margins [74-77]. In a large series of 67 patients with pelvic osteosarcoma, 50 underwent surgical excision, 12 of whom required external hemipelvectomy; the remainder had limb-sparing procedures [76]. The incidence of local recurrence overall was 70 percent, and it was 62 percent for those undergoing definitive surgery. Despite the use of multiagent systemic chemotherapy, the five-year overall and progression-free survival rates were only 27 and 19 percent, respectively. Postoperative radiation therapy improved the outcome for patients with an intralesional or no surgical excision. Hemipelvectomy has been combined with intraoperative radiation therapy (IORT) in an attempt to improve local control [78]. Initial results are promising, but further clinical experience is needed. (See "Radiation therapy for Ewing sarcoma family of tumors", section on 'Radiation schedule'.) The extent of resection determines the functional deficit and influences the decision regarding the type of reconstruction, if any, to optimize function. An internal 74 hemipelvectomy may not require allograft or a metallic endoprosthesis reconstruction. Resection without reconstruction is a reasonable option, particularly in view of the high complication rate with reconstruction of pelvic defects using allografts (approximately 50 percent) [79]. The femur and remaining pelvis can be left flail and the gap is bridged by scar tissue, or postoperative traction can be applied to initiate the development of scar tissue. Although this allows for stable weight bearing, it also results in a significant limb length inequality, and most patients require crutches for ambulation. Metallic endoprosthesis reconstruction (saddle prosthesis) has occasionally been successful, but fixation is extremely difficult and prosthesis loosening can be a problem long-term [73,80,81]. Another method of obtaining structural integrity is to fuse the proximal femur to the ilium or sacrum [82]. The choice of procedure is dependent on the clinical situation and the surgeon and patient preference, but they in general offer reasonable alternatives to external hemipelvectomy, with superior functional and psychological results. Spine tumors — The spine presents a particularly difficult problem with regard to surgical margins. Local recurrence rates are closely tied to surgical margin status [69]. In one report of 30 patients with primary sarcomas of the mobile spine, the resection was classified as wide, marginal or intralesional in 23, 10, and 67 percent respectively [83]. Resection margins were histologically positive in 60 percent and associated with a fivefold increased risk of a local recurrence. At least one series suggests an advantage for either wide or marginal surgery compared to intralesional or no surgery for osteosarcomas of the spine [69]. They recommended that these patients be treated with a combination of chemotherapy and at least marginal surgery. En bloc resection (vertebrectomy) followed by stabilization and fusion is possible with wide margins in some cases as long as dural invasion is absent. If the tumor abuts or invades the dura (image 4), a segment can be excised. However, this procedure may be accompanied by seeding of tumor cells into the cerebrospinal fluid. Postoperative (adjuvant) radiation therapy can be employed to manage a microscopically positive dural margin. Largely because of the potential for dural involvement, local recurrence rates are higher for tumors located in the vertebral body and posterior elements (particularly those with extra-osseous extension into the canal and paraspinal region) as compared to the anterior vertebral column. Patients with sacral tumors do especially poorly. In one report that included 22 patients with primary high-grade osteosarcoma of the spine or sacrum, all of whom received neoadjuvant chemotherapy, only two of the 12 who underwent surgery had wide excision to negative margins [69]. Although total sacrectomy may improve local control, neurologic and sexual dysfunction is inevitable [84]. Local recurrence may still develop despite maximal surgical resection and intraoperative or postoperative radiation therapy [70,78]. Postoperative radiotherapy may be beneficial [69]. Newer methods of adjuvant radiation therapy, including combined proton beam and photon irradiation using three dimensional conformal techniques, may permit a higher dose of radiation to be administered while sparing adjacent normal tissues [85]. Allograft reconstruction is an effective means of filling the massive osseous defects following spine resection. Combined anterior and posterior instrumentation provides the best stability following tumor vertebral body resection. In the anterior spine, femoral and humeral shafts used to substitute for resected vertebral bodies have provided excellent stability. Metal cages provide an alternative to allografts in the anterior spine, and they do not require biologic healing of the graft to the host bone [86]. New posterior instrumentation, bone graft products, and pedicle screws provide improved fixation and better rates of fusion. In setting where radiation therapy has been used to improve local control, complications such as fractures and non-unions may occur more frequently. The 75 bridging of a large osseous defect may also be augmented with vascularized grafts. (See 'Vascularized grafts' below.) RECONSTRUCTION TECHNIQUES — The majority of patients with osteosarcoma will require reconstructive procedures in order to restore structural integrity. However, some locations and circumstances do not always require reconstruction to deal with the surgical defect. As an example, pelvic tumors may not require postresection reconstruction following an internal hemipelvectomy (see 'Non-extremity lesions' above). Expendable bones that do not require replacement include the ulna, patella, fibula, scapula, and ribs. Resection techniques for tumors in these locations are well described, and the resultant functional disability without bony reconstruction is frequently minimal with the exception of complete scapulectomy, which may result in significant morbidity, especially if muscle loss is great. For patients who do require reconstruction, available methods include metal endoprostheses/hardware, allografts, arthrodeses, vascularized fibular transfers, rotationplasty, and various unusual methods that are specific for certain anatomic locations. The choice of reconstructive method should be individualized. The following factors influence the selection of a reconstructive method [87]: Anatomic location and integrity of the surrounding structures The stage of disease and extent of resection The tissue diagnosis Preference of the surgeon The likelihood and nature of complications related to a particular type of reconstruction The patient's age, expectations, and anticipated functional demands The availability of the materials for the reconstructive procedure Because the majority of patients receive preoperative (neoadjuvant) chemotherapy, planning of the surgical procedure is essential in order to coordinate chemotherapy schedules and anticipated bone marrow recovery with the time needed to manufacture a prosthesis or procure the necessary reconstructive materials. An understanding of the capabilities and limitations of available reconstructive techniques is necessary so that the appropriate method is chosen for the surgical reconstruction. A brief description of some of these techniques follows. Rotationplasty is discussed above. (See 'Rotationplasty and the skeletally immature patient' above.) Allografts and endoprostheses — Allografts and metal endoprostheses are common means of reconstructing bone defects that result from sarcoma surgery. Other methods listed below may be preferred to fill the osseous defect depending on the clinical situation and the availability of the product. Allografts — While autologous bone grafting is of limited use in patients undergoing resection of bone tumors because of the large size of the defect, allografts have been successfully used for many years. Allografts provide the potential for long-lasting reconstruction of large bony defects by providing a structural lattice for the ingrowth of the patient's own bone elements (image 5) [88-94]. The host normal tissue slowly invades the allograft by creeping substitution of normal bone and vascular elements at the osteosynthesis site and periosteum. Large segments of allografted bone probably do not completely fill with autogenous bone, and this may lead to allograft fracture over time (this occurs in about 18 percent of the cases). The articular cartilage is slowly replaced by an inflammatory pannus-like tissue, and some patients may ultimately require joint resurfacing. Allografts are available from tissue banks and need to be matched to the size of the resected bone. Although bone is a relative nonantigenic structure, matching for the class II major histocompatibility antigens results in better clinical outcomes [95,96]. Freezing 76 further decreases the likelihood of immune rejection, and the addition of glycerol or dimethyl sulfoxide during freezing preserves the articular cartilage somewhat. Soft tissues are left attached to the allograft to increase function by providing a site of attachment for host soft tissues. This is particularly important for reconstructions around the knee. The major advantages of allografts over endoprosthetic reconstruction are restoration of bone stock, sparing of the uninvolved portion of adjacent joints, and providing a site of attachment for host soft tissues. Intercalary allograft reconstructions tend to perform better than osteoarticular grafts, and clearly better than allograft arthrodeses, which have the highest rate of complications [97]. There are some disadvantages to allografts. Compared to metal endoprostheses, allografts must be fixed to the host bone and allowed to heal. Thus, they must be protected from weight bearing for prolonged periods. Early complication rates (15 to 20 percent) are higher than those seen in patients undergoing placement of metal prostheses. While implant loosening does not occur as it does for metal prostheses, nonunion and fractures may result in allograft failure [98]. As with endoprosthetic reconstructions, infection in the reconstructed site is devastating and may necessitate amputation of the extremity. Late complications include degenerative arthritis and joint instability. Despite these limitations, satisfactory functional results are achieved in 70 percent of patients undergoing allografting [89,93,99,100]. Metal endoprostheses — Initially, the metallic implants that were used to reconstruct bone defects following tumor resection were custom made and manufactured for specific patient requirements. More recently, modular metal prostheses have become available that permit reconstruction without having to resort to a custom implant. The components are readily available and can be assembled at the time of surgery to match the extent of the defect (image 6). Modular metal prosthetic reconstructions are performed most frequently about the shoulder, hip, and knee [101,102]. Total femoral reconstructions have been performed in certain circumstances. Most bone tumors are located adjacent to joints, and metal endoprostheses are well suited to joint replacement. The implants used for reconstruction after sarcoma surgery are similar to those that are used in patients with arthritis. (See "Total joint replacement for severe rheumatoid arthritis".) As compared to allografts, endoprostheses allow for immediate joint stability and early weight bearing. However, they are not free of complications, which include fatigue fracture, loosening, and infection. The fatigue fracture potential of the metal is a design and stress problem that can be potentially improved by manufacturing changes. Loosening at the bone-cement interface is a function of repetitive stresses during activities of daily living [87,103]. Loosening can be minimized by using a meticulous cementing technique and stress-reducing total joint mechanics such as a rotating-hinge knee design. Nevertheless, most, if not all, implants will need to be replaced at some point in the long-term survivor's lifetime. The survivorship of the prosthetic arthroplasty reconstruction varies, and ranges from 60 to 90 percent estimated survival at five years to 40 to 80 percent at ten years [102,104-106]. These data have generated the impetus for development of cementless or press-fit implants. Implants placed without cement may have a better long term survival, but this is a controversial area. Because the prostheses act as a large foreign body, infections, which occur in 2 to 5 percent of cases, are difficult to treat [87,102,104,105,107]. Deep infections usually require removal of the implant and sometimes an amputation. Expandable prostheses — Expandable prostheses have been developed for reconstruction in skeletally immature children with malignant bone tumors [108-110]. Different types of prostheses are available; one type has a telescoping unit that can be expanded using a gear device [110,111], and another employs the principle of a loaded 77 spring which is gradually released via external exposure to an electromagnetic field [112]. Because these devices permit an expansion of the overall length of the prosthesis, they can be implanted in skeletally immature patients, increasing the numbers of those eligible for limb-salvage surgery. Estimation of growth potential in the younger child is important in determining whether this means of reconstruction is appropriate for the situation. The Lewis Expandable Adjustable Prosthesis (LEAP) was initially successful in young patients (five to eight years). However, newer expandable prostheses that can be lengthened without invasive surgery are popular in this age group. One of these, the Phoenix, uses a magnetic field to provide the energy to lengthen the prosthesis. The modular prostheses are better for large preadolescent and adolescent children. Most children require several lengthenings in those prostheses, and it is frequently necessary to excise the pseudocapsule around the implant to gain length; as a result, the lengthening procedures are not small operations. Expandable prostheses frequently require replacement after they have been fully extended because they lose some structural strength. Difficulties in using this prosthesis occur primarily in the rehabilitation phase, but loosening, fatigue failure of the implant, and inability to gain equal limb lengths can also occur. Close follow-up is crucial to optimize the functional outcome and properly time periodic lengthenings. Allograft-prosthetic composites — For extremity reconstructions, allografts can be combined with metal prostheses (the prosthesis is cemented to the allograft) to form what is called an allograft-prosthetic composite, or alloprosthesis (image 7). This has certain advantages over allografts or metal prostheses alone. Since the articular surface is formed by the metal portion of the construct, it does not fracture or collapse as may happen with osteoarticular allografts. The allograft portion of the composite restores bone stock and provides soft tissue attachment points for insertion of muscle tendons. Even though a great deal of research has been done to improve attachments of muscle units to prostheses, this is still an unsolved problem with this type of reconstruction. Alloprosthetic reconstructions usually result in good or excellent functional results, but the operation is more complex and time consuming than either osteoarticular allografts or prostheses alone [113]. Arthrodesis — Arthrodesis (joint fusion) is important for tumors involving the spine; otherwise, the hardware fails as a result of nonunion, and pain may be a persistent complaint. In the spine almost all fusions require implantation of bone graft material. Transplantation of structured or morcelized autologous corticocancellous bone obtained from the iliac crest is the most frequently used technique. Arthrodesis may also be considered for selected patients with extremity lesions. Although a nonmobile joint is less desirable than a mobile one to the majority of patients, it may provide the best mode of reconstruction in certain cases and is one method by which durable function can be achieved. Arthrodesis of various joints following extraarticular resection of malignant bone tumors may be the best reconstructive option if resection includes most of the surrounding joint soft tissues and joint arthroplasty cannot be accomplished [114,115]. The most common joints that are considered for arthrodeses following resection of bone sarcomas resection are the hip, knee, and shoulder. Hip and shoulder arthrodeses do not produce as much functional disability as knee fusions because of the capacity for excellent compensatory motion at other joints in the affected extremity. Despite functional limitations, knee arthrodesis can be achieved in the majority of patients, even when a long segment of bone has been excised. However, this procedure is complicated by a high incidence of nonunion, fatigue fracture, and infection. Although the operation successfully achieves tumor-free margins, the revascularization and rehabilitation 78 processes are prolonged and complicated. Patients must adjust their lifestyles to accommodate the arthrodesis. Vascularized grafts — Vascularized bone grafts are also used to reconstruct osseous defects and augment other reconstructive techniques, such as intercalary allograft segments [116,117]. They are of particular benefit in the treatment of allograft nonunions and fractures and in surgical beds which have poor vascularity and perfusion following radiation. If the vascular graft remains viable, healing occurs more rapidly than with nonvascularized grafts and is a more durable construct. The surgical procedure for implantation of a vascularized graft is lengthy and the vascular anastomosis is technically demanding. One of the most commonly used grafts for this purpose is the fibula; fortunately, donor site complications are rare. Vascularized fibular grafts have also been used to augment healing of intercalary grafts following intraepiphyseal resections of the distal femur and proximal tibia, where only a short segment of epiphyseal bone remains for fixation. ADJUVANT THERAPY — As previously noted, more than 80 percent of patients with osteosarcoma or Ewing sarcoma treated with surgery alone develop metastatic disease, despite having adequate local tumor control. It is surmised that subclinical metastatic disease is present at diagnosis, even in the absence of overt metastases. Chemotherapy can eradicate these deposits if initiated at a time when disease burden is low. As a result, adjuvant chemotherapy is considered a standard component of management for these primary bone tumors. For osteosarcoma, postoperative chemotherapy was used initially, and five year overall survival rates rose from less than 20 percent to between 40 and 60 percent in the 1970s [118]. Two subsequent randomized studies conducted in the 1980s demonstrated a significant relapse-free and overall survival benefit for adjuvant chemotherapy, although the trials were limited in size, and the survival benefits were modest. (See "Chemotherapy and radiation therapy in the management of osteosarcoma".) Neoadjuvant chemotherapy — For patients with osteosarcoma, presurgical or neoadjuvant chemotherapy was originally considered because of the time needed for fabrication of custom metallic implants; chemotherapy was given while awaiting definitive surgery. Due to its success in shrinking tumors, neoadjuvant chemotherapy evolved to a method of increasing the proportion of patients who were suitable candidates for limb-salvage surgery and to facilitate limb-sparing procedures by diminishing tumor burden. A major concern in this regard is the possibility that inexperienced surgeons may expand selection criteria to accommodate inappropriate candidates who might be better served by amputation. Adjuvant chemotherapy is never a substitute for sound surgical principles. One of the most compelling rationales for neoadjuvant chemotherapy is its ability to function as an in vivo drug trial to determine the drug sensitivity of an individual tumor and to customize postoperative therapy. A grading system for assessing the effect of preoperative chemotherapy in osteosarcoma was developed at Memorial Sloan Kettering and is in widespread use (table 5) [118]. The response to neoadjuvant chemotherapy is a major prognostic factor. Patients with a near-complete absence of viable tumor cells in the resection specimen after neoadjuvant therapy do well when the same therapy is continued after surgery. In contrast, if the tumor contains 10 percent or more residual viable cells after neoadjuvant chemotherapy, a change in the chemotherapeutic regimen might be beneficial, although this is a controversial area. This topic is discussed in detail elsewhere. The majority of limb-sparing surgical procedures for osteosarcoma are performed at institutions using presurgical chemotherapy. On the other hand, immediate resection is an acceptable option in situations where the surgical procedure would not be changed by 79 a good response to neoadjuvant chemotherapy. In such cases, adjuvant chemotherapy should be administered postoperatively. Issues surrounding the use of neoadjuvant and adjuvant chemotherapy for Ewing sarcoma are also discussed elsewhere. Radiation therapy — Unlike Ewing sarcoma, conventional osteosarcoma is relatively resistant to radiation therapy. Because of this, primary radiation therapy is not usually adequate to achieve local disease control, particularly for bulky tumors. Issues surrounding the use of radiation therapy for Ewing sarcoma are discussed elsewhere. (See "Radiation therapy for Ewing sarcoma family of tumors".) There has been renewed interest in a potential role for radiation therapy for patients with osteosarcoma whose tumors respond to chemotherapy and in whom surgery would be debilitating. A recent report described a five-year local control rate of 56 percent among 31 patients with nonmetastatic extremity osteosarcoma who refused surgery and were instead treated with radiotherapy (median dose 60 Gy) [119]. No local failures developed in 11 patients who responded well to chemotherapy with both a radiographic and biochemical response (normalization of alkaline phosphatase), and the five-year metastasis-free survival rate was 91 percent. Among patients who achieved local control, 86 percent had "excellent" limb function. Although local adjuvant radiation therapy and prophylactic whole lung radiation have been used in an attempt to improve outcomes following surgery, neither is effective in the absence of systemic chemotherapy [13,120,121]. Furthermore, in patients treated with effective surgery and chemotherapy, adjuvant radiation does not improve survival and increases the risk for secondary tumors [122]. Adjuvant radiation should be considered only in the setting of an unresectable or incompletely resected primary tumor or in patients with the small cell variant of osteosarcoma, which may be more radiosensitive [123]. The role of adjuvant radiation therapy in patients with Ewing sarcoma is discussed elsewhere. (See "Radiation therapy for Ewing sarcoma family of tumors", section on 'Adjuvant RT'.) SUMMARY Osteosarcoma is the most common primary malignant tumor of bone; other less common types of bone sarcoma include peripheral primitive neuroectodermal tumor (PNET)/Ewing sarcoma, malignant fibrous histiocytoma, fibrosarcoma, chondrosarcoma, and chordoma. (See 'Histologic classification' above.) The goal of the preoperative evaluation for a primary bone sarcoma is to establish the tissue diagnosis, evaluate disease extent, and assess the feasibility of a limbsparing approach. Clinical staging includes all of the data obtained prior to definitive therapy, including the results of imaging, physical examination, laboratory studies, and tissue biopsy. The biopsy must be carefully planned to ensure that adequate tissue is obtained for diagnosis without compromising the opportunity for limb salvage. Biopsies should take place after the completion of the staging studies. (See 'Preoperative evaluation' above.) Surgery and systemic chemotherapy are the mainstays of treatment for patients with osteosarcomas and other primary bone tumors such as fibrosarcoma of bone. Chondrosarcoma and chordoma have not been routinely or historically treated with chemotherapy or radiation and therefore surgery is the mainstay of management. (See "Chondrosarcoma" and "Chordoma and chondrosarcoma of the skull base" and "Spinal cord tumors", section on 'Chordomas'.) 80 Although there is no specific survival benefit to preoperative as compared to postoperative chemotherapy in osteosarcoma patients, the neoadjuvant approach may permit a greater number of patients to undergo limb-sparing procedures. However, chemotherapy is no substitute for sound surgical judgment when assessing the need for amputation versus limb-sparing surgery. The optimal chemotherapy regimen has not been established. In general, patients with a nearly complete response to neoadjuvant chemotherapy do better than those with a lesser response. Even if the patient has a chemosensitive tumor, it may be reasonable to proceed to immediate surgery followed by adjuvant chemotherapy if the nature of the resection would not necessarily be influenced by a good response to chemotherapy. (See "Chemotherapy and radiation therapy in the management of osteosarcoma", section on 'Neoadjuvant chemotherapy' and 'Neoadjuvant chemotherapy' above.) Reconstructive options vary depending upon the factors listed above. In addition, there is surgeon preference and no absolute consensus in many circumstances as to the optimal reconstructive method. In some circumstances, if the tumor is located in an expendable bone, reconstruction is not even necessary. (See 'Reconstruction techniques' above.) There is no role for adjuvant radiation therapy except in patients with unresectable or incompletely resected sarcomas and possibly in the rare patient with a small cell osteosarcoma. Radiation therapy for local control can be used in patients who decline surgery or for whom there is no effective surgical option. (See 'Radiation therapy' above.) In contrast to osteosarcoma and other primary bone sarcomas, Ewing sarcoma is more radiosensitive, and radiation may be considered an effective option for local control. Although modern treatment protocols emphasize surgery for optimal local control, patients who lack a function-preserving surgical option because of tumor location or extent, and those who have clearly unresectable primary tumors following induction chemotherapy are appropriate candidates for radiation therapy. Furthermore, adjuvant radiation may be considered for bulky tumors in difficult sites (eg, the pelvis), if there is residual microscopic or gross disease after surgery, and for patients with high-risk chest wall primary tumors (close or involved margins, initial pleural effusion, pleural infiltration, and intraoperative contamination of the pleural space). (See "Radiation therapy for Ewing sarcoma family of tumors", section on 'Adjuvant RT'.) Osteosarcoma: Epidemiology, pathogenesis, clinical presentation, diagnosis, and histology Literature review current through: Aug 2013. | This topic last updated: May 30, 2013. 81 82 83 INTRODUCTION — Osteosarcomas are primary malignant tumors of bone that are characterized by the production of osteoid or immature bone by the malignant cells [13]. Osteosarcomas are uncommon tumors. Approximately 750 to 900 new cases are diagnosed each year in the United States, of which 400 arise in children and adolescents younger than 20 years of age [4,5]. Despite their rarity, osteosarcomas are the most common primary malignancy of bone in children and adolescents (figure 1), and the fifth most common malignancy among adolescents and young adults aged 15 to 19 [6,7]. The survival of patients with malignant bone sarcomas has improved dramatically with effective chemotherapy. Prior to the use of chemotherapy, 80 to 90 percent of patients with osteosarcoma developed metastatic disease despite achieving local tumor control and died of their disease. It was surmised (and subsequently demonstrated) that the majority of patients had subclinical metastatic disease that was present at the time of diagnosis, even in the absence of overt clinical metastases [8,9]. Chemotherapy can successfully eradicate these deposits if initiated at a time when disease burden is low. As a result, all patients with osteosarcoma are treated with 84 adjuvant chemotherapy, and most receive this treatment modality in the preoperative period. With multimodality therapy, at least two-thirds of patients with non-metastatic extremity osteosarcomas will be long-term survivors, up to 50 percent of those with limited pulmonary metastases may be cured of their disease, and long-term relapse-free survival can be expected in about 25 percent of all patients who present with more extensive metastatic disease. This topic review will provide an overview of the epidemiology, clinical presentation, diagnosis, staging, and histopathology of patients with osteosarcoma. Diagnostic evaluation and biopsy techniques for primary bone tumors, an overview of treatment and outcomes, principles guiding surgical management of bone sarcomas, and chemotherapy in the treatment of osteosarcoma are discussed in detail elsewhere. (See "Bone tumors: Diagnosis and biopsy techniques" and "Bone sarcomas: Preoperative evaluation, histologic classification, and principles of surgical management" and "Chemotherapy and radiation therapy in the management of osteosarcoma".) EPIDEMIOLOGY — As noted previously, osteosarcoma is an uncommon tumor; it accounts for only 1 percent of all cancers diagnosed annually in the US. In contrast to Ewing sarcoma, which is extremely rare in older adults, there is a bimodal age distribution of osteosarcoma incidence, with peaks in early adolescence and in adults over the age of 65 [10]. There are differences in tumor site and survival according to age at presentation. Children — Bone sarcomas account for about 6 percent of all childhood cancers, and osteosarcomas account for approximately 3 percent of childhood cancers overall [11]. However, osteosarcoma is the most common primary bone tumor affecting children and young adults. Osteosarcomas comprise 56 percent of all bone cancers in individuals under the age of 20, while Ewing sarcoma accounts for 34 to 36 percent, and chondrosarcomas are responsible for less than 10 percent [10]. (See "Epidemiology, pathology, and molecular genetics of the Ewing sarcoma family of tumors".) In children, the peak incidence is between 13 and 16 years of age (figure 1), a time that appears to coincide with the adolescent growth spurt. For unclear reasons, osteosarcomas are more common in boys than in girls, and in blacks and other races as compared to Caucasians [4,10,12]. The most common sites of osteosarcoma in children are the metaphyses of long bones, especially the distal femur (75 percent of cases in one large population-based series [10]), proximal tibia, and proximal humerus [13,14]. Adults — Osteosarcomas in adults are often considered secondary neoplasms, attributed to sarcomatous transformation of Paget disease of bone or some other benign bone lesion [10]. In the US, over one-half of all osteosarcomas arising in patients over the age of 60 arise are secondary [15]. In contrast, in Asia, Paget's disease is less frequent, and a higher percentage of osteosarcomas in patients over the age of 60 arise primarily [16]. Osteosarcomas arising in the setting of Paget's disease (Pagetic sarcomas) have a worse prognosis overall. (See 'Risk factors' below.) Compared to cases diagnosed in children, osteosarcomas arising in adults more commonly occur in axial locations (although as in children, the lower leg bones are the single most common site [10]), and in areas that have been previously irradiated or that have underlying bone abnormalities. As is seen in children, among older adults, males are affected more often than females [10]. However, in contrast to children, osteosarcoma is more common in whites than in blacks or individuals of other races [10]. RISK FACTORS AND PATHOGENESIS — In children, the majority of osteosarcomas are sporadic, while inherited predisposition accounts for a minority of cases. In older adults, about one-third of cases arise in the setting of Paget disease of bone or as a second or later cancer [10]. 85 Risk factors — Several predisposing factors have been identified [17]. Prior irradiation or chemotherapy — Osteosarcoma is the most frequent second primary cancer occurring during the first 20 years following radiation therapy for a solid cancer in childhood. Early estimates suggested that approximately 3 percent of osteosarcomas could be attributed to prior irradiation. However, a higher incidence is likely to be revealed, as more patients survive long enough after primary irradiation to develop this complication. The interval between irradiation and the appearance of a secondary osteosarcoma averages 12 to 16 years; it is shorter in childhood cancer survivors. (See "Radiation-associated sarcomas", section on 'Epidemiology and histologic distribution' and "Radiation-associated sarcomas", section on 'Radiation dose and age of exposure'.) Prior exposure to chemotherapy, particularly alkylating agents, is also associated with secondary osteosarcomas in childhood cancer survivors and may potentiate the effect of previous radiation. (See "Radiation-associated sarcomas", section on 'Chemotherapy agents'.) Paget disease and other benign bone lesions — Cases of osteosarcoma in patients older than 40 years of age are frequently associated with Paget disease, a focal skeletal disorder characterized by an accelerated rate of bone turnover [18]. Although the incidence of bone tumors is markedly increased in patients with Paget disease, they only occur in 0.7 to 1 percent of cases [19]. Sarcomatous transformation is most often seen in long-standing Paget disease, but is not necessarily related to the extent of skeletal involvement. Although histologically indistinguishable from other osteosarcomas, multiple bone involvement is common, and the prognosis is poor overall. (See "Clinical manifestations and diagnosis of Paget disease of bone".) The etiology of Paget disease is unclear, but genetic factors are thought to play a pathogenetic role. Both Paget disease and pagetic osteosarcomas are associated with loss of heterozygosity of chromosome 18, possibly involving the same site of a postulated tumor suppressor gene [20-22]. Somatic mutations in the SQSTM1 gene on chromosome 5q35 have been identified in both sporadic Paget disease of bone as well as pagetic osteosarcomas [23]. In addition to Paget disease, other benign bone lesions are associated with an increased risk of malignant degeneration to a primary bone tumor. These include chronic osteomyelitis, multiple hereditary exostoses, fibrous dysplasia, sites of bone infarcts, and sites of metallic implants for benign conditions [24,25]. Inherited conditions — Genetic conditions with a known predisposition to osteosarcoma include hereditary retinoblastoma, Li-Fraumeni syndrome, Rothmund-Thomson syndrome, and the related Bloom and Werner syndromes. Because of the association of these genetic conditions with osteosarcomas, particularly in the setting of multiple primary malignancies [26], careful detailing of family history is important for patients with newly diagnosed osteosarcoma. The genetic abnormality associated with heritable forms of retinoblastoma (ie, germline mutations of the retinoblastoma gene) are associated with an increased risk of developing second primary tumors, 60 percent of which are soft tissue sarcomas and osteosarcomas [27-29]. The risk for later development of osteogenic sarcoma is not only within irradiated fields, but also in long bones distant from radiation ports. Patients with the sporadic type of retinoblastoma are at a much lower risk. As an example, in one study of 1604 patients with retinoblastoma, the cumulative incidence of a second cancer at 50 years after diagnosis was 51 percent for hereditary cases, compared to only 5 percent for nonhereditary (sporadic) disease [27]. (See "Overview of retinoblastoma", section on 'Epidemiology' and "Radiation-associated sarcomas", section on 'Genetic predisposition'.) Li-Fraumeni syndrome is a familial cancer syndrome in which affected family members display a spectrum of cancers, including breast, soft tissue, adrenocortical, and brain 86 tumors, leukemias, and osteosarcomas [30]. Many of these patients carry germline inactivating mutations in the p53 tumor suppressor gene, which is involved in cell cycle regulation and maintaining the integrity of the genome [31,32]. (See "Li-Fraumeni syndrome".) Despite this important association, the number of osteosarcomas that are attributable to Li-Fraumeni syndrome is small [33,34]. In one series of 235 unselected children with osteosarcoma, only 3 percent carried constitutional germline mutations in p53 [33]. Rothmund-Thomson syndrome (RTS, also called poikiloderma congenitale) is an autosomal recessive condition characterized by distinctive skin findings (atrophy, telangiectasias, pigmentation), sparse hair, cataracts, small stature, skeletal anomalies, and a significantly increased risk for osteosarcoma [35,36]. In one cohort of 41 patients with RTS, 13 (32 percent) developed osteosarcoma [35]. Clinically, these tumors tend to develop at a younger age than seen in the general population. A specific loss of function mutation in the RECQL4 gene has been identified in approximately two-thirds of patients with RTS and is closely associated with the risk for osteosarcoma. In one series of 33 patients with RTS, none of the 10 patients without a truncating mutation in this gene developed osteosarcoma, while among the 23 patients with truncating mutation, the incidence of osteosarcoma was five cases per year with 230 person-years of observation [37]. Other members of the RecQ gene family are mutated in Bloom syndrome and Werner syndrome, two diseases with overlapping clinical features, including a predisposition to develop osteosarcoma [37]. Molecular pathogenesis — Although the etiology of osteosarcoma is unclear, a relationship between rapid bone growth and the development of osteosarcoma is suggested by the following: The peak incidence of osteosarcoma occurs during the adolescent growth spurt. The tumor appears most frequently at sites where the greatest increase in bone length and size occurs (the metaphyseal portions of the distal femur, proximal tibia, and proximal humerus) (figure 2). Osteosarcomas occur at an earlier age in girls, corresponding to their more advanced skeletal age and earlier adolescent growth spurt. These data have led to speculation that bone tumors arise from an aberration of the normal process of bone growth and remodeling at a time when rapidly proliferating cells are particularly susceptible to oncogenic agents, mitotic errors, or other events leading to neoplastic transformation [38]. However, studies examining the relationship between factors related to growth and development and the risk of bone sarcomas have revealed no consistent pattern [39,40]. The specific nature of this aberration or aberrations that leads to tumorigenesis remains elusive and is the subject of intense investigation [17,41,42]. In contrast to other sarcomas, there are no characteristic translocations or other molecular genetic abnormalities in osteosarcomas. (See "Pathogenetic factors in soft tissue and bone sarcomas", section on 'Genetics and molecular pathogenesis'.) Most osteosarcomas have a complex unbalanced karyotype. The highest frequency of loss of heterozygosity (which implies the loss of a putative tumor suppressor gene) in osteosarcomas is reported for chromosomes 3q, 13q (the location of the retinoblastoma gene), 17p (the location of the p53 gene), and 18q, the chromosomal region that has been linked to osteosarcomas arising in the setting of Paget disease (see 'Paget disease and other benign bone lesions' above) [20,21,43-45]. Combined inactivation of the retinoblastoma and p53 tumor suppressor pathways is common in osteosarcomas [17,46,47]. The potential contribution of p53 pathway abnormalities to tumorigenesis is particularly intriguing, given the increased incidence of 87 osteosarcoma in families with the Li-Fraumeni syndrome, an inherited condition in which p53 is mutationally inactivated. (See 'Inherited conditions' above.) Normal or "wild type" p53 appears to play a role in the normal development and physiology of bone [48], since p53-null mice display failure of skull growth and delayed longitudinal bone growth in utero [49]. Furthermore, bone cell lysates from p53-null mice also fail to activate normal apoptotic pathways [46]. The contribution of p53 and other molecular pathways, such as the Wnt, Notch, IGF and mTOR signaling pathways, to the pathogenesis of osteosarcoma is beyond the scope of this discussion; excellent reviews are available [41,50]. As will be discussed below, the majority of patients are presumed to have metastatic disease at presentation, with the majority being subclinical. Gene expression profiling through the use of DNA microarray analysis as well as whole genome sequencing efforts are beginning to uncover the molecular events that dictate metastatic potential, findings that may pave the way for future molecularly targeted therapies [41]. CLINICAL PRESENTATION — The majority of patients with osteosarcoma present with localized pain, typically of several months' duration. Pain frequently begins after an injury, and may wax and wane over time. Systemic symptoms such as fever, weight loss, and malaise are generally absent. The most important finding on physical examination is a soft tissue mass, which is frequently large and tender to palpation. Osteosarcomas have a predilection for the metaphyseal region of the long bones (figure 3). The most common sites of involvement, in descending order, are: distal femur, proximal tibia, proximal humerus, middle and proximal femur, and other bones [51]. Laboratory evaluation is usually normal, except for elevations in alkaline phosphatase (in approximately 40 percent) [52], lactate dehydrogenase (LDH, in approximately 30 percent) [53], and erythrocyte sedimentation rate. Laboratory abnormalities do not correlate with disease extent, although a very high LDH level is associated with a poor clinical outcome [54]. At the time of presentation, between 10 and 20 percent of patients have demonstrable macrometastatic disease and are classified as stage III according to the staging system used by the Musculoskeletal Tumor Society (see 'Staging system' below). Distant metastases most commonly involve the lungs, but can also involve bone [55]. Occult micrometastases are presumed to be present in the majority of those who appear to have clinically localized disease, since before the era of adjuvant chemotherapy, over 80 percent of patients with osteosarcoma developed metastatic disease despite achieving local tumor control. It was postulated that these patients had subclinical metastases that were present at the time of diagnosis [56]. With routine use of systemic adjuvant chemotherapy, at least two-thirds of children and adolescents with nonmetastatic osteosarcoma will be long-term survivors, implying the success of chemotherapy in eradication of micrometastases. Prognosis is worse in adults with osteosarcoma, particularly those over the age of 65 [57]. DIAGNOSIS AND STAGING EVALUATION — The first diagnostic test to arouse suspicion for a primary bone tumor is generally a plain radiograph of the affected area [58]. Characteristic features of conventional osteosarcomas (which account for the majority of cases, see below) include destruction of the normal trabecular bone pattern, indistinct margins, and no endosteal bone response. The affected bone is characterized by a mixture of radiodense and radiolucent areas, with periosteal new bone formation, lifting of the cortex, and formation of Codman's triangle (image 1). The associated soft tissue mass is variably ossified in a radial or "sunburst" pattern. (See "Bone tumors: Diagnosis and biopsy techniques".) Differential diagnosis — The correct histologic diagnosis of osteosarcoma may be predicted in up to two-thirds of patients who have a characteristic radiographic appearance, clinical features, and tumor location [59]. However, no radiographic finding 88 is pathognomonic, and biopsy is required for definitive diagnosis. The differential diagnosis includes other malignant bone tumors (ie, Ewing sarcoma, lymphoma, and metastases), benign bone tumors (eg, chondroblastoma, osteoblastoma), and nonneoplastic conditions, such as osteomyelitis, eosinophilic granuloma, and aneurysmal bone cysts. Occasionally, no abnormalities will be evident on plain radiographs. In such cases, magnetic resonance imaging (MRI) should be obtained if clinical suspicion for a bone tumor is high. Even for patients with characteristic plain radiographic findings, MRI is indicated for surgical planning (see 'Staging system' below). Staging work-up — Patients with overt metastatic disease at presentation have a significantly worse outcome than those with localized disease. Because a significant proportion of patients with metastases (including up to one-half of those with limited pulmonary involvement) may be amenable to cure, a thorough staging workup is imperative to facilitate surgical planning. Imaging studies — The staging work-up should include the following (table 1) [60]: MRI of the entire length of the involved long bone. A multi-institutional study of 387 patients that included both children and adults concluded that CT and MRI were equally accurate for local staging of bone and soft tissue tumors [61]. However, MRI is preferred in most cases because of its superior definition of soft tissue extension, particularly to the neurovascular bundle, joint and marrow involvement, and the presence of skip lesions (ie, medullary disease within the same bone, but not in direct contiguity with the primary lesion) (image 2) [62]. CT scans are best suited to evaluate the thorax for metastatic disease, which is essential because approximately 80 percent of osteosarcoma metastases involve the lungs [56,63]. Because of the possibility of false-positive results, histologic confirmation is indicated for suspected sites of metastatic disease, particularly if a given lesion was not detected on plain films. However, CT may underestimate the extent of pulmonary involvement by metastatic tumor [64,65]. In one study, metastases would have been missed in more than one-third of cases by any method other than manual palpation of the lung during open thoracotomy [64]. These data raise doubt as to the advisability of minimal access procedures (eg, thoracoscopic metastasectomy) when the goal is resection of all pulmonary metastases [66-68]. (See "Surgical resection of pulmonary metastases: Benefits; indications; preoperative evaluation and techniques".) Distinguishing metastatic lesions from benign nodules can be difficult, particularly in adults who have a high prevalence of granulomatous disease and in children living in areas with endemic fungal disease, especially histoplasmosis. While calcification can be a sign of benign disease, it may also be seen in metastases from osteosarcoma [69]. Criteria to guide the evaluation of suspected pulmonary metastases are available from the European and American Osteosarcoma Study Group, as are being used in the EURAMOS 1 (AOST0331) trial (table 2) [70]. Radionuclide bone scanning with technetium is the preferred method for evaluating the entire skeleton for the presence of multiple lesions. Although a positron emission tomography (PET) scan may have greater utility for assessing the response to preoperative chemotherapy, at least one study suggests it is inferior to radionuclide bone scanning for the detection of osseous metastases from osteosarcoma [71] and to spiral CT for detecting pulmonary metastases [72]. 89 The role of PET and integrated PET/CT imaging in patients with osteosarcoma is incompletely characterized, and there is no consensus on their use [72,73]. Guidelines from the National Comprehensive Cancer Network (NCCN) suggest a PET scan and/or bone scan in the workup of a suspected osteosarcoma [74]. Imaging guidelines from the Children’s Oncology Group Bone Tumor Committee recommend radionuclide bone scan and/or PET scan for whole body staging (table 1) [60]. Regardless of which is chosen, the same imaging modality should be used throughout treatment and during posttreatment surveillance. Biopsy — Once the diagnosis of a primary bone tumor is suspected, referral should be made to a facility with expertise in pediatric oncology for further management, including a diagnostic biopsy. The biopsy should be carried out by an orthopedic surgeon who is experienced in the management of osteosarcoma and ideally by the same surgeon who will perform the definitive surgery. If a core needle biopsy is planned by an interventional radiologist, the orthopaedic surgeon and radiologist should communicate about the placement of the biopsy tract. Proper planning of the biopsy with careful consideration of the future definitive surgery is important so as not to jeopardize the subsequent treatment, particularly a limb salvage procedure [75]. This topic is discussed in detail elsewhere. (See "Bone tumors: Diagnosis and biopsy techniques".) Staging system — The Musculoskeletal Tumor Society (MSTS) staging system is most often used for bone sarcomas and was developed by Enneking at the University of Florida [76,77]. This is a surgical staging system and is not used to decide medical (chemotherapy) treatment for patients with osteosarcoma. The MSTS staging system characterizes nonmetastatic malignant bone tumors by grade (low-grade [stage I] versus high-grade [stage II]) and further subdivides these stages according to the local anatomic extent (intracompartmental [A] versus extracompartmental [B]). For bone tumors, the compartmental status is determined by whether the tumor extends through the cortex of the involved bone; the majority of high grade osteosarcoma are extracompartmental. Patients with distant metastases are categorized as stage III. HISTOLOGIC CLASSIFICATION — The histologic diagnosis of an osteosarcoma is based on the presence of a malignant sarcomatous stroma, associated with the production of tumor osteoid and bone. Osteosarcomas are thought to arise from a mesenchymal stem cell that is capable of differentiating towards fibrous tissue, cartilage, or bone. As a result, they share many features with chondrosarcomas and fibrosarcomas, tumors of the same family of bone sarcomas (table 3), with which osteosarcoma can be easily confused. However, chondrosarcomas and fibrosarcomas are distinguished by their lack of woven bone matrix, which is required for the diagnosis of osteosarcoma. Because some osteosarcomas have a limited degree of osteoid production and variable histomorphology, immunohistochemistry may be required for confirmation of the diagnosis. In contrast to Ewing sarcoma and many soft tissue sarcomas, osteosarcomas are not associated with any characteristic chromosomal translocations [78]. Conventional osteosarcomas — The largest group are the conventional (intramedullary high-grade) osteosarcomas, which account for approximately 90 percent of all osteosarcomas [79]. These tumors typically involve the metaphysis of long bones (figure 2) and are most common in adolescents and young adults. Depending upon the predominant cellular component, conventional osteosarcomas are subclassified as osteoblastic (accounting for 50 percent of conventional osteosarcomas), chondroblastic (25 percent), or fibroblastic (25 percent) (table 3) [58]. Despite histologic differences, their clinical behavior and management are similar. Osteoblastic osteosarcoma is characterized by abundant osteoid production that forms a fine or coarse lacelike pattern around the tumor cells; massive amounts 90 may result in distortion of the malignant stromal cells (picture 1). Some tumors contain thick trabeculae of osteoid that form an irregular anastomosing network. The degree of mineralization is variable. Fibroblastic osteosarcomas are predominantly composed of high-grade spindle cell stroma that contains only focal osteoid production (picture 2). More pleomorphic tumors may resemble malignant fibrous histiocytoma, but the distinction can be made by the identification of osteoid. In chondroblastic osteosarcomas, cartilaginous matrix production is evident throughout most of the tumor. While the majority of the tumor tends to be of lower grade, the chondroid areas may contain cytologically atypical cells that are characteristic of higher-grade tumors. These chondroblastic foci are admixed with malignant spindle cells that produce osteoid trabecula (picture 3). Histologic variants — In addition to the three subcategories of conventional osteosarcoma, there are several variants: Small cell Telangiectatic Multifocal Malignant fibrous histiocytoma [MFH] subtype These variants were originally thought to carry a worse prognosis. However, with modern aggressive therapy, they appear to behave similarly. Other subtypes (the juxtacortical parosteal and periosteal osteosarcomas) are associated with a more indolent biologic behavior. Parosteal osteosarcomas are more common in older patients and periosteal osteosarcoma has a similar age distribution to classic osteosarcoma. Small cell osteosarcoma — Small cell osteosarcoma is noteworthy for the confusion that may arise in its distinction from other "small round blue cell tumors" such as Ewing sarcoma by conventional light microscopy of hematoxylin and eosin (H&E) stained sections [80]. Immunohistochemical staining, cytogenetics, and molecular genetic studies may be required to establish the diagnosis [81,82]. (See "Treatment of the Ewing sarcoma family of tumors".) Telangiectatic osteosarcomas — Telangiectatic osteosarcomas are high-grade, vascular tumors which contain little osteoid. Because of their purely lytic appearance on plain radiographs, they can be confused with aneurysmal bone cysts or giant cell tumors of bone (GCTB). Grossly they appear as a "multicystic bag of blood", and a solid mass of tumor is usually absent [79]. As a result, it may be difficult to obtain diagnostic tissue on biopsy. Histologically, the minimal osteoid formation and numerous multinucleated giant cells (picture 4) are reminiscent of a benign GCTB. However, the cells are highly pleomorphic. (See "Giant cell tumor of bone".) The age distribution and treatment are identical to classic high-grade osteosarcoma. The response to chemotherapy and survival are similar to conventional osteosarcomas [83]. Multifocal osteosarcoma — Rarely, patients present with multiple synchronous sites at diagnosis, all resembling the primary tumor. It is difficult to determine whether these represent synchronous multiple primary lesions or considered metastases. Regardless of their designation, the prognosis is usually dismal. Multicentric osteosarcoma may also be metachronous in that other bony lesions occur years after treatment of the first. Malignant fibrous histiocytoma — MFH of bone appears similar to osteosarcoma, but without osteoid production. Although they tend to have a lower rate of tumor necrosis following induction chemotherapy, long-term survival rates are similar to conventional osteosarcomas [84]. Surface (juxtacortical) osteosarcomas — In contrast to these intramedullary variants, the surface osteosarcomas differ with respect to prognosis and therapy. These 91 osteosarcomas include the parosteal low-grade type, the intermediate-grade periosteal type, and the high-grade surface osteosarcomas [84,85]. Surgery alone may be curative for the low-grade and intermediate-grade varieties, as long as they do not enter the marrow cavity or have a de-differentiated component [86,87]. Parosteal osteosarcoma — The most common of these entities, typical parosteal osteosarcoma, is a surface lesion composed of low-grade fibroblastic cells that produce woven or lamellar bone. It occurs in an older age group than conventional intramedullary osteosarcoma, usually between the ages of 20 and 40 years. The posterior aspect of the distal femur is the most commonly involved site, but other long bones may be affected. The tumor arises from the cortex as a broadly based lesion. With time, however, this lesion may invade the cortex and enter the endosteal cavity. Treatment for conventional parosteal osteosarcomas is surgical resection alone with expected survival rates of approximately 90 percent [88,89]; however, de-differentiated parosteal osteosarcomas may require adjuvant chemotherapy. Periosteal osteosarcoma — Periosteal osteosarcoma is a moderate-grade, chondroblastic surface osteosarcoma frequently located in the proximal tibia, which has the same age distribution as conventional intramedullary osteosarcoma (image 2). The likelihood of metastases is greater than that for the low-grade parosteal tumors, but lower than the classic intramedullary osteosarcoma. The role of adjuvant chemotherapy for periosteal osteosarcomas is controversial. Adjuvant chemotherapy is recommended in many centers because of the estimated 20 percent metastatic rate. However, retrospective reports indicate more favorable outcomes than seen with classic osteosarcomas (84 percent tenyear survival rate in one study [90]), a lack of benefit when patients who received adjuvant chemotherapy were compared to those treated by surgery alone [90,91], and a concerning number of second malignant neoplasms in patients treated with adjuvant chemotherapy [90]. Randomized trials have not been conducted, which limits the conclusions that can be drawn from the available literature. High-grade surface osteosarcoma — Conventional high-grade osteosarcomas may also develop on the surface of the bone, where they may be confused with parosteal or periosteal osteosarcoma. They are treated similar to conventional intramedullary osteosarcomas. Osteosarcoma of the jaw — Another distinct variant is osteosarcoma of the jaw, which tends to occur in older patients, has an indolent course, and is more often associated with local recurrences than with distant metastases. Extraosseous osteosarcoma — Extraosseous osteosarcoma is a malignant tumor arising in the soft tissue, without involving the bone or periosteum, that produces osteoid, bone, or chondroid material (image 3 and image 4) [92-97]. Most arise in the setting of prior radiation exposure. In contrast to osseous osteosarcomas, extraskeletal osteosarcomas present in older age groups, they have a different anatomic predilection (the most common site is in the thigh) and show relative chemoresistance to doxorubicin-based chemotherapy [93-95]. They are generally treated as a soft tissue sarcoma with aggressive behavior. SUMMARY Osteosarcomas are uncommon primary malignant tumors of bone that are characterized by the production of osteoid or immature bone by the malignant cells. (See 'Introduction' above.) 92 The age distribution of osteosarcoma incidence is bimodal, with peaks in early adolescence and in adults over the age of 65. Osteosarcoma is the most common primary bone tumor affecting children and young adults; the peak age is between 13 and 16. (See 'Children' above.) In children, the majority of osteosarcomas are sporadic. A minority of cases are associated with inherited predisposition syndromes such as hereditary retinoblastoma, Li-Fraumeni syndrome, Rothmund-Thomson syndrome, and the related Bloom and Werner syndromes. (See 'Risk factors' above.) Osteosarcomas in adults are often considered secondary neoplasms, attributed to sarcomatous transformation of Paget disease of bone or some other benign bone lesion. In the US, over one-half of all osteosarcomas arising in patients over the age of 60 are secondary, while in, Asia, Paget disease is less frequent and a higher percentage arise primarily. (See 'Adults' above and 'Paget disease and other benign bone lesions' above.) The majority of patients with osteosarcoma present with localized pain, typically of several months' duration. The most important finding on physical examination is a soft tissue mass, which is frequently large and tender to palpation. At presentation, between 10 and 20 percent have demonstrable metastatic disease, most often involving the lung. (See 'Clinical presentation' above.) Biopsy is required for definitive diagnosis. Proper planning of the biopsy with careful consideration of future definitive surgery is important so as not to jeopardize subsequent treatment, particularly the opportunity for limb salvage. (See 'Biopsy' above.) The histologic diagnosis of an osteosarcoma is based on the presence of a malignant sarcomatous stroma, associated with the production of tumor osteoid and bone. Conventional (intramedullary high-grade) osteosarcomas account for approximately 90 percent of all osteosarcomas. Other less common histologic variants are small cell, telangiectatic, multifocal, surface (juxtacortical), and extraosseous osteosarcomas, malignant fibrous histiocytoma, and osteosarcoma of the jaw. Jaw osteosarcoma is more common in older patients. (See 'Histologic classification' above.) Once the diagnosis of an osteosarcoma is established, the staging evaluation should include MRI of the entire length of the involved bone, CT of the thorax, radionuclide bone scanning with technetium, and/or a PET scan. (See 'Staging work-up' above.) Chemotherapy and radiation therapy in the management of osteosarcoma Literature review current through: Aug 2013. | This topic last updated: Tem 8, 2013. 93 94 95 INTRODUCTION — Osteosarcomas are primary malignant tumors of bone that are characterized by the production of osteoid or immature bone by the malignant cells. Osteosarcomas are uncommon; only about 750 to 900 cases are diagnosed each year in the US, of which 400 are in children and adolescents under the age of 20 [1,2]. This topic review will cover the use of adjuvant and neoadjuvant chemotherapy and radiation therapy (RT) in the management of osteosarcoma. The same principles apply to other primary bone tumors such as fibrosarcoma and undifferentiated high-grade pleomorphic sarcoma (previously referred to as malignant fibrous histiocytoma of bone), and they are treated similarly [3,4]. On the other hand, primary bone angiosarcomas do 96 not behave clinically like other primary bone tumors, and these patients are treated according to the principles of soft tissue sarcoma rather than osteosarcoma. The surgical management of patients with primary bone tumors, the clinical features, epidemiology, diagnosis, pathology and pathogenesis of osteosarcoma, management of osteosarcomas arising in the head and neck, management of chondrosarcomas and of chordomas and chondrosarcomas arising in the skull base, and the treatment of Ewing sarcoma are addressed separately. Rarely osteosarcomas can occur in soft tissue. There are conflicting data on their management as soft tissue sarcomas or osteosarcoma of bone [5,6]. These tumors are treated as soft tissue sarcomas at some institutions with surgery and radiation alone, or surgery, radiation, and soft tissue-based chemotherapy. At other institutions, they are treated as osteogenic sarcomas, with surgery, radiation and doxorubicin plus cisplatin with or without methotrexate. The choice of chemotherapy is best determined by the availability of clinical trials, and otherwise is decided on a patient-by-patient basis, balancing the risks and benefits of chemotherapy for this very high risk diagnosis. PROGNOSIS AND EVOLUTION OF TREATMENT — The survival of patients with malignant bone sarcomas has improved dramatically over the past 30 years, largely as a result of the use of effective chemotherapy. Previously 80 to 90 percent of patients with bone sarcomas developed metastases despite achieving local tumor control, and died of their disease. It was surmised (and subsequently demonstrated [7]) that subclinical metastatic disease was present at the time of diagnosis in the majority of patients and that chemotherapy can successfully eradicate these deposits if initiated at a time when disease burden is low. The benefits of chemotherapy are best illustrated by a systematic review of the literature, which showed that long-term survival after local tumor control without chemotherapy was only 16 percent (95% CI 9 to 23 percent) [8]. In contrast, the addition of systemic chemotherapy with three or more drugs provided a five-year overall survival rate of 70 percent. Chemotherapy is now considered a standard component of osteosarcoma treatment, both in children and in adults. The choice of regimen and optimal timing (ie, preoperative versus postoperative) are controversial; however, many centers preferentially utilize preoperative chemotherapy, particularly if a limb-sparing procedure is being contemplated for an extremity osteosarcoma. With modern therapy, at least two-thirds of children, adolescents, and adults under the age of 40 with nonmetastatic extremity osteosarcomas will be long-term survivors and presumably cured of their disease. In addition, up to 35 to 40 percent of those with limited pulmonary metastases may be cured with multimodality therapy. In contrast, long-term survival can be expected in less than 20 percent of all other patients who present with or develop overt metastatic disease. In most (but not all [9]) series, prognosis is worse for adults than for children [10-13]. PRIMARY MANAGEMENT Adjuvant chemotherapy — More than 80 percent of patients with osteosarcoma treated with surgery alone develop metastatic disease, despite achieving local control. It is presumed that subclinical metastases are present at diagnosis in the majority of patients. Chemotherapy can eradicate these deposits if initiated at a time when disease burden is low. Initially, postoperative chemotherapy was used, and five-year survival rates rose from less than 20 percent to between 40 and 60 percent in the 1970s [14]. Two subsequent randomized studies conducted in the 1980s demonstrated a significant relapse-free and overall survival benefit for adjuvant chemotherapy that persisted over time [15-18], although the trials were limited in size and the survival benefits modest. The chemotherapy regimens used in these studies included high-dose methotrexate 97 (HDMTX) plus doxorubicin, bleomycin, cyclophosphamide and dactinomycin and either vincristine [15], or cisplatin [16,17]. Neoadjuvant chemotherapy — The concept of induction or neoadjuvant chemotherapy arose in concert with the evolving use of limb-sparing surgery. Because of the time needed for fabrication of custom metallic endoprostheses, chemotherapy was often given while awaiting definitive surgery [19]. Accumulating experience with this approach led to the suggestion that induction chemotherapy might also be improving survival. Neoadjuvant versus adjuvant chemotherapy — These observations ultimately led to a randomized clinical study conducted between 1986 and 1993 by the Pediatric Oncology Group (POG trial 8651) that compared immediate surgery and postoperative chemotherapy versus 10 weeks of the same chemotherapy regimen followed by surgery in 100 patients under the age of 30 with nonmetastatic high-grade osteosarcoma [20]. Chemotherapy consisted of alternating courses of HDMTX with leucovorin rescue, cisplatin plus doxorubicin, and bleomycin, cyclophosphamide, and dactinomycin (BCD). The five-year relapse-free survival rates were similar between the two groups (65 versus 61 percent for adjuvant and neoadjuvant therapy, respectively) as was the limb salvage rate (55 and 50 percent for immediate and delayed surgery, respectively). The study was criticized for the relatively low rate of limb-sparing surgery in both groups (by modern standards) and the inclusion of BCD as a component of the regimen. The contribution of BCD to the therapeutic efficacy of this regimen is unclear, while it can clearly contribute to long-term bleomycin-related pulmonary toxicity. (See "Bleomycininduced lung injury".) Nevertheless, this trial established that survival was similarly improved by either presurgical or postsurgical chemotherapy and established a benchmark outcome for future studies (five-year event-free survival 65 percent). Due to its success in killing cancer cells (although actual tumor shrinkage during treatment is not common, particularly with chondroblastic osteosarcomas), neoadjuvant chemotherapy has evolved to a method of increasing the proportion of patients who are suitable candidates for limb-salvage surgery. The majority of limb-sparing surgical procedures for extremity osteosarcomas are now performed at institutions using presurgical chemotherapy. While it is clear that the number of patients with osteosarcoma who are deemed suitable candidates for limb-sparing surgery has increased in parallel with the increasing use of presurgical chemotherapy, this finding may reflect improvements in reconstructive techniques and the increased experience and confidence of tumor surgeons rather than a benefit attributable to the use of induction chemotherapy. A major concern in this regard is the possibility that inexperienced surgeons may expand selection criteria to accommodate inappropriate candidates who might be better served by amputation. Neoadjuvant chemotherapy is never a substitute for sound surgical principles. Histology and response to chemotherapy — Response to neoadjuvant chemotherapy is histology dependent (table 1) [21,22]. In the larger series, in which 1058 patients received presurgical chemotherapy for osteosarcoma over a 20-year period, the likelihood of a "good" response (>90 percent necrosis) was significantly higher for fibroblastic and telangiectatic osteosarcomas (83 and 80 percent, respectively) compared to chondroblastic (43 percent) or osteoblastic osteosarcomas (58 percent) [21]. Five-year survival rates paralleled the quality of the response to induction chemotherapy (83, 75, 62, and 60 percent for fibroblastic, telangiectatic, osteoblastic, and chondroblastic osteosarcomas, respectively). For the two principal subtypes (osteoblastic and chondroblastic), the difference in “good” response rate has not translated into any survival difference. Thus, the lack of a “good” response to induction chemotherapy does not necessarily predict a poor outcome. 98 Tailored postoperative chemotherapy based upon initial response — One of the most compelling rationales for neoadjuvant chemotherapy is its ability to function as an in vivo drug trial to determine the drug sensitivity of an individual tumor and to customize postoperative therapy. A grading system for assessing the effect of preoperative chemotherapy on the tumor was developed at Memorial Sloan Kettering and is in widespread use (table 2) [14]. Responsiveness of an osteosarcoma to neoadjuvant chemotherapy is a major determinant of clinical outcome [21-28]. Five-year survival rates for patients with an extremity sarcoma and a "good" response to chemotherapy (as defined by 90 percent or more necrosis in the surgical specimen) are significantly higher than for those with a lesser response (71 to 80 versus 45 to 60 percent, respectively) [21,22,27,29]. Response to chemotherapy is usually defined by histologic appearance in the resected specimen (picture 1). However, at least some data suggests that three-phase thallium scintigraphy is a useful method for noninvasively evaluating tumor response to neoadjuvant chemotherapy [30-32]. This approach may be most useful in a patient with an equivocal clinical response for whom alternative chemotherapy might be attempted prior to resection. Patients with a near-complete absence of viable tumor cells in the resection specimen after neoadjuvant therapy tend to do well when the same therapy is continued after surgery [14,23-25,33]. However, when the tumor contains 10 percent or more residual viable cells after neoadjuvant chemotherapy, a change in the chemotherapeutic regimen might be beneficial. This strategy of altering the postoperative chemotherapy regimen based upon the response to the neoadjuvant regimen was pioneered in the T10 protocol at the Memorial Sloan-Kettering Cancer Center (MSKCC) [14,33] and confirmed by at least one other group [34,35]. However, this strategy came under scrutiny for the following reasons: With more mature follow-up, the initial beneficial results from the T10 protocol have not been sustained at MSKCC [36]. Major cooperative groups (eg, Children's Cancer Group, German Society for Pediatric Oncology, Scandinavian Sarcoma Group) have been unable to confirm that altering the postoperative chemotherapy regimen in patients with a less than complete response to presurgical chemotherapy improves their overall outcome [37-41]. As noted above, POG trial 8651 failed to show a survival benefit for neoadjuvant as compared to adjuvant chemotherapy [20]. It seems reasonable to conclude from these data that the administration of presurgical chemotherapy (with or without treatment individualization based upon initial tumor response) does not appear to have improved cure rates. On the other hand, experience with ifosfamide plus etoposide (I/E) has once again brought the issue of modifying postsurgical treatment based upon response to induction therapy to the forefront [34,35,42,43]. The high level of antitumor activity of I/E in patients with measurable metastatic osteosarcoma (66 percent in one trial [42]) has prompted investigators to consider the use of I/E in patients with resectable disease who have a poor initial histologic response to standard chemotherapy. Because of the potential for additional toxicity (gonadal and renal) from the addition of these agents to standard chemotherapy, the Children's Oncology Group (COG) performed a pilot trial that demonstrated that the addition of I/E to doxorubicin, cisplatin and HDMTX was feasible and safe [43]. Subsequently, an international multigroup trial (COG AOST 0331, the EURAMOS1 trial [44]) was developed to test prospectively the benefit of altering postoperative chemotherapy based upon the response to initial chemotherapy. In this trial, patients who had a poor histological response to standard 99 induction chemotherapy were randomly assigned to HDMTX, cisplatin, and doxorubicin with or without I/E after resection, while those with a good histologic response after induction chemotherapy were randomly assigned to continued HDMTX, cisplatin, and doxorubicin with or without the addition of pegylated interferon alpha 2b for two years. The only result that is available is an analysis of the 715 patients with a confirmed “good response” who were randomized to maintenance IFN or no IFN, a preliminary report of which was presented at the 2013 annual meeting of the American Society of Clinical Oncology [45]. At a median follow-up of 3.1 years, EFS for MAP plus interferon is not statistically superior to MAP alone (77 versus 74 percent, HR 0.82, 95% CI 0.61-1.11). The ability to detect a benefit from maintenance IFN may have been limited by the fact that one-fourth of patients did not even start IFN, and of those that did, 45 percent terminated treatment prematurely. Nevertheless, maintenance IFN cannot be considered a standard approach for patients with a good histologic response to upfront MAP. Until results of the “poor histologic response” arm from this trial are available, in the absence of data to indicate that a change in therapy improves cure rate, we suggest not pursuing a change in the chemotherapeutic regimen after surgery in patients who have less than a complete response to neoadjuvant chemotherapy. While the results with chemotherapy as initiated are not as good in a patient with a poor response to chemotherapy than with a good response to chemotherapy, outcomes are overall better with chemotherapy than without it, arguing for completion of the regimen that had been started. Better methods are needed to prospectively identify chemoresistant tumors at diagnosis are needed. Regardless of the regimen chosen, based on available evidence, in patients with localized disease of an extremity, we suggest resuming chemotherapy within 21 days after definitive surgery, when feasible [46]. Choice of regimen — There is no worldwide consensus on a standard chemotherapy approach for osteosarcoma. The development of adjuvant chemotherapy has been largely empiric, with the majority of regimens incorporating doxorubicin and cisplatin, with or without high-dose methotrexate (HDMTX, 6 to 12 g/m2 with leucovorin rescue) (table 3) [16,17,35,39,47-50]. (See "Therapeutic use and toxicity of high-dose methotrexate".) Role of methotrexate — The role of HDMTX has been questioned, particularly in adults: Nearly all of the trials purporting to show benefit for chemotherapy regimens that include HDMTX are phase II trials that have been conducted predominantly in children, although many have enrolled patients up to age 40 [8]. Neither of the two randomized studies comparing HDMTX plus doxorubicin and cisplatin versus doxorubicin/cisplatin alone have shown an advantage to threedrug therapy [40,49]. However, one of these studies was criticized because the outcome in the group receiving HDMTX, cisplatin, and doxorubicin was particularly poor by modern standards (41 percent five-year disease-free survival compared to 57 percent with doxorubicin/cisplatin alone) [40]. A single randomized trial comparing higher as compared to intermediate doses of methotrexate did not show a survival advantage for high-dose therapy [51]. However, a benefit for HDMTX is supported by at least one phase II trial demonstrates a superior outcome with high-dose as compared to intermediatedose methotrexate in the context of a multiagent chemotherapy regimen [26]. Furthermore, many studies have shown a correlation between peak serum levels of methotrexate, tumor response, and outcome [38,52-54]. Thus, it is possible that determining a benefit for HDMTX has been compromised by the use of insufficient doses [55] or administration schedules. Additional information comes from a literature-based systematic review of chemotherapy trials for localized high-grade osteosarcoma [8]. In an analysis of 100 18 trials, almost all conducted in patients under the age of 40, treatment with three or more drugs (including methotrexate, doxorubicin, and cisplatin) was associated with a significant improvement in both event-free survival (hazard ratio [HR] 0.701, 95% CI 0.615-0.799) and overall survival (HR 0.792, 95% CI 0.6770.926) compared to a two-drug regimen. However, only two of the trials included in this analysis were randomized trials comparing a three-drug regimen containing methotrexate doxorubicin and cisplatin versus a two-drug regimen. Investigators at St. Jude’s Hospital have demonstrated good outcomes (five-year event free and overall survival rates 66 and 75 percent) with a non-methotrexatecontaining chemotherapy regimen consisting of carboplatin plus ifosfamide and doxorubicin [56]. The role of HDMTX in chemotherapy for osteosarcoma requires further study [57]. Until further information is available, we and others consider that a methotrexate-containing regimen represents a standard approach for children and adults age 40 or younger. The optimal methotrexate-containing regimen is not established. The most recent study from COG, AOST 0331, has completed accrual [44]. We consider the control arm of the protocol (MAP: high dose methotrexate plus doxorubicin and cisplatin) (table 4) to represent a reasonable and appropriate standard treatment [44]. At least some data suggest that substitution of carboplatin for cisplatin is associated with inferior long-term outcomes, at least in patients with metastatic disease at diagnosis [58]. Older adults are more commonly offered the combination of doxorubicin and cisplatin, although control arm of the AOST 0331 protocol (MAP chemotherapy) (table 4) is also a reasonable standard of care in this population. (See 'Chemotherapy for adults' below.) Practical tips on administration of HDMTX, including the rationale for leucovorin rescue and prevention and management of prolonged high plasma MTX levels, is discussed in detail elsewhere. Benefit of adding a fourth drug — The upfront addition of ifosfamide with or without etoposide to HDMTX, doxorubicin, and cisplatin improves initial tumor response rates, but it also increases toxicity, and the influence on overall and event-free survival is unclear [59-63]. A meta-analysis of 14 trials in localized osteosarcoma (the majority studying the addition of ifosfamide to methotrexate plus doxorubicin and cisplatin) concluded that treatment effect was not significantly better with four as compared to three drugs (HR for event-free survival 0.956, 95% CI 0.779-1.177, HR for overall survival 1.043, 95% CI 0.851-1.280) [8]. Thus, the routine addition of ifosfamide (with or without etoposide) to adjuvant chemotherapy for osteosarcoma is difficult to recommend outside of a clinical trial. It is not known if adding or changing to ifosfamide for poor responders will help increase overall survival. Benefit of mifamurtide — The benefit of ifosfamide in conjunction with the liposomal formulation of the immune stimulant muramyl tripeptide phosphatidylethanolamine (MTP-PE, mifamurtide, Junovan®) was evaluated in a single large phase III study involving 677 patients (age up to 30 years) with nonmetastatic osteosarcoma [59]. All patients received doxorubicin, cisplatin, and methotrexate, and were randomized in a 2 x 2 scheme to receive or not receive ifosfamide, and then to receive or not receive liposome encapsulated mifamurtide (see 'Newer and investigational approaches' below). The addition of mifamurtide significantly improved the overall survival rate (78 versus 70 percent at six years), a finding which led to European regulatory approval of this agent for patients with osteosarcoma [64]. However, the results of this trial are complex for several reasons. Importantly, the data were initially analyzed, then reanalyzed when more complete survival data could be assembled, leading to two separate publications of the same data set with somewhat different results. The initial published report stated that there was an interaction between ifosfamide and mifamurtide that prevented the results from being analyzed as originally intended [65]. In addition, the improvement in 101 event free survival (67 versus 61 percent) with mifamurtide did not reach statistical significance, which is very unusual in an oncology study in which the intervention leads to a statistically significant improvement in overall survival. Such paradoxical results have been observed in other diseases treated with immunotherapy (eg, sipuleucel-T and prostate cancer). As a result of these data, there is interest in further studying the efficacy of this agent in osteosarcoma prior to including mifamurtide as a standard component of chemotherapy for osteosarcoma [59,66,67]. Although the drug has received regulatory approval in Europe and some other countries, further development of this agent (and drug approval in the United States) may prove difficult given a "not approvable" letter for use of mifamurtide in the adjuvant setting from the US Food and Drug Administration (FDA), resulting from the presentation of the initial data, not the follow-up dataset, to the FDA. Chemotherapy for adults — Intensive treatment with chemotherapy and resection is warranted in adults, since osteosarcoma is a potentially curable tumor [11,60]. Similar to children, the lack of a near-complete response to neoadjuvant chemotherapy predicts a poor prognosis [60]. In many (but not all [68]) series, adults, especially older adults, have a worse prognosis than do children with osteosarcoma. This was shown in a population-based series from the Surveillance, Epidemiology, and End Results (SEER) database of the National Cancer Institute [12]. Of the 3482 cases of osteosarcoma reported between 1973 and 2004, there were 1855 cases in the 0 to 24 age group, 974 cases in adults 25 to 59 years of age, and 653 in adults 60 to ≥85 years of age. Five-year survival rates in the three age groups were 62, 59, and 24 percent, respectively. When broken down by decade of age, five-year survival rates for adults in their 50s, middle to late 60s, and 80 to 84 years were 50, 17, and 11 percent, respectively. The optimal chemotherapy regimen for adults (at least those over the age of 40) is not established; almost all of the trials included in the systematic review discussed above were limited to individuals age 40 or younger [8]. Only one trial enrolled older adults up to the age of 65 (range 14 to 62; median 42 years) and failed to show a benefit for the addition of HDMTX to cisplatin and doxorubicin [49]. Older adults are most often offered doxorubicin plus cisplatin (AP, doxorubicin 25 mg/m2 per day days 1 to 3, cisplatin 100 mg/m2 day 1, every three weeks for six cycles) [3,40,49]. However, the dose and tolerability of high-dose cisplatin and the role of HDMTX remain unanswered questions. A methotrexate-containing regimen is a reasonable standard of care in this population, if patients can tolerate it. Options include a five-week cycle of cisplatin (100 mg/m2 day 1) and doxorubicin (25 mg/m2 days 1 to 3), followed by two weekly doses of HDMTX (6 to 12 g/m2 with leucovorin rescue), with three cycles administered preoperatively and three postoperatively [59], or the control arm of the AOST 0331 protocol (MAP chemotherapy) (table 4) [44], recognizing that dose reductions may be necessary in older adults. Monitoring renal function before and during treatment is paramount in such patients to minimize the risk of irreversible renal failure. Radiation therapy — In contrast to Ewing sarcoma, conventional osteosarcoma is believed to be relatively resistant to radiation therapy (RT), although the small cell variant may be more radiosensitive [69]. Primary RT is usually inadequate to achieve local control, particularly for bulky tumors; surgery is preferred if possible. Prophylactic whole lung radiation has been used in an attempt to improve outcomes following surgery for nonmetastatic localized disease; however, it is not effective in the absence of systemic chemotherapy [70-72]. Furthermore, in patients treated with effective surgery and chemotherapy, adjuvant radiation does not improve survival and increases the risk for secondary tumors; it should be considered only in the setting of an unresectable or incompletely resected primary tumor [73]. 102 With the improvements in RT techniques over time, there has been renewed interest in the use of RT for patients whose tumors respond to chemotherapy and in whom surgery would be debilitating. One report described a five-year local control rate of 56 percent among 31 patients with nonmetastatic extremity osteosarcoma who refused surgery and were instead treated with radiotherapy (median dose 60 Gy) [74]. The five-year metastasis-free survival rate was 91 percent, and there were no local failures among the 11 patients who responded well to chemotherapy (ie, had both a radiographic and biochemical response with normalization of serum alkaline phosphatase). Among patients who achieved local control, 86 percent had "excellent" limb function. However, this single study does not represent sufficient data to recommend the use of RT as a replacement for surgery in patients with resectable tumors. RT should not be considered a substitute for inadequate surgery. The benefit of RT for local control may need to be readdressed in view of the availability of newer radiation techniques such as intensity modulated RT (IMRT), proton beam irradiation [75], and carbon ion radiotherapy [76]. Summary — Surgery and systemic chemotherapy are the mainstays of treatment for patients with nonmetastatic osteosarcomas. Preoperative as compared to postoperative administration may permit a greater number of patients with extremity tumors to undergo limb-sparing procedures. However, chemotherapy is not a substitute for sound surgical judgment when assessing the need for amputation. It is reasonable to proceed to immediate surgery followed by adjuvant chemotherapy if the nature of the resection would not necessarily be influenced by a good response to chemotherapy. Response to neoadjuvant chemotherapy is a major prognostic factor. It is unclear whether outcomes can be improved in poor responders by changes in the postoperative chemotherapy regimen. The optimal regimen has not been established; however, the available evidence supports the benefit of a three-drug as compared to a two-drug regimen in particular for children and younger adults [8]. For children and adolescents, we recommend the MAP regimen as was used in the control arm of the AOST 0331 protocol (table 4) [44]. If preoperative therapy is chosen, we administer 10 weeks of chemotherapy prior to surgery and continue postoperative chemotherapy for 29 weeks, beginning three weeks postoperatively. Where available, mifamurtide (MTP-PE) can be used in addition to chemotherapy. For adult patients under the age of 40, we use the same regimen as in children and adolescents. For older patients, in whom the biology of the tumor may be somewhat different, we generally employ doxorubicin and cisplatin only, although a methotrexatecontaining regimen is also a reasonable approach. There is no role for adjuvant RT except in patients with incompletely resected osteosarcomas and possibly in the rare patient with a small cell osteosarcoma. RT for local control can be used in patients who decline surgery or for whom there is no effective surgical option. TREATMENT OF PATIENTS WITH METASTATIC DISEASE AT DIAGNOSIS — Patients who present with overtly metastatic osteosarcoma have a poor prognosis; long-term survival rates with standard chemotherapy and surgery range from 10 to 50 percent [9,28,61,77,78]. This is in contrast to patients with apparently localized disease at presentation, two-thirds of whom will achieve long-term survival with appropriate therapy. Analogous to the situation with primary osteosarcomas, the ability to control all foci of macroscopic disease is an essential element for successful treatment [27,79,80]. The minority of patients with overt metastatic disease who achieve long-term survival and are presumably cured have usually been treated with a combination of surgery, chemotherapy, and sometimes RT [9,78,81-83]. 103 The location of the metastases is of prognostic importance. Patients with only pulmonary disease appear to have a chance of long-term event-free survival on the order of 20 to 30 percent [82,84]. In contrast, most series report a dismal prognosis for patients with bone metastases [58,85,86], with the exception of a single study that added etoposide and high-dose ifosfamide to standard chemotherapy (see 'Choice of chemotherapy' below) [42]. Although it might be argued that pulmonary lesions are more susceptible to chemotherapy than are bone metastases, there is also a tendency to surgically pursue pulmonary lesions more aggressively. In some studies, long-term survival correlates with the number of pulmonary nodules at diagnosis and the ability to successfully resect those that persist after chemotherapy [9,42,82,83,87-89]. As noted previously, patients with bone metastases do less well. There are some longterm survivors among patients in whom bone disease has been treated aggressively (usually with chemotherapy in addition to complete resection, sometimes with irradiation) [42,77,86,90]. Choice of chemotherapy — Optimal management for patients who present with metastatic osteosarcoma has not been defined by randomized clinical trials, and thus, there is no single standard approach. The most active drugs in patients with measurable disease (HDMTX, doxorubicin, cisplatin, ifosfamide) have single-agent response rates between 20 and 40 percent [58,85,87,91-93]. Response rates are higher with multiagent regimens but a lower proportion of patients treated for metastatic disease show a good histological response to neoadjuvant chemotherapy as compared to those with apparently localized disease [52,81,94]. This suggests an underlying difference in the biological behavior. In an effort to improve outcomes, the Children’s Oncology Group (COG) and others have utilized a strategy of applying novel agents to patients with newly diagnosed metastatic disease prior to standard therapy (termed the "therapeutic window" approach) [42,58,83,87]. Using this approach, the Pediatric Oncology Group (one of the predecessors of COG) identified the combination of ifosfamide/etoposide as effective induction therapy, particularly for those with metastatic bone disease. In one report, 43 patients under the age of 30 with measurable metastatic osteosarcoma at diagnosis (28 lung only, 12 with metastatic bone involvement, with or without lung metastases), received two threeweek courses of etoposide and high-dose ifosfamide (17.5 g/m2 per cycle) with hematopoietic growth factor support, followed by surgery and an additional 34 weeks of "continuation therapy" [42]. Postoperative therapy consisted of ten courses of HDMTX with leucovorin rescue, four courses of doxorubicin/cisplatin, one course of doxorubicin alone, and three additional courses of etoposide plus lower-dose ifosfamide [42]. Treatment-related toxicity was prominent, and included 83 percent grade 4 neutropenia, 24 percent sepsis, 29 percent grade 4 thrombocytopenia, and five patients with Fanconi's syndrome. However, the overall response rate was 59 percent, and it was 80 percent for patients with synchronous bone metastases. The projected two-year progression-free survival rates were 34 and 58 percent for patients with lung-only and bone involvement, respectively. These results are superior to any other reports of patients with metastatic disease at diagnosis, particularly bone metastases. Nevertheless, few patients with metastatic osteosarcoma are cured, and new therapeutic approaches are needed. Dose-intensive chemotherapy with peripheral blood stem cell support is ineffective in metastatic osteosarcoma [95-97]. For patients who present with overt metastatic disease, participation in experimental trials should be encouraged. Such a trial, AOST06P1, tested the feasibility of adding zoledronic acid to multiagent chemotherapy 104 (MAP chemotherapy plus ifosfamide and etoposide); results from this study are pending [98]. (See 'Bisphosphonates' below.) POSTTREATMENT SURVEILLANCE — For all bone sarcomas, there are no prospective data that address the appropriate schedule or selection of tests for surveillance after initial treatment for localized disease. As with other cancers, there are no data indicating that a particular follow-up schedule improves survival. Consensus-based guidelines from the National Comprehensive Cancer Network (NCCN) and from the Children's Oncology Group recommend physical examination, a complete blood count, chest imaging, and local imaging of the primary site every three months for two years, every four months for year three, every six months for years 4 and 5, and annually thereafter (table 5) [99]. The appropriate duration of follow-up is unknown; however, the vast majority of recurrences are observed within 10 years. At diagnosis, an initial technetium bone scan or PET scan is performed to explore for bone metastases. There are no comparative data of bone scan versus PET scan in this context. COG guidelines suggest either as appropriate (table 5). A restaging bone scan (or PET, depending on which study was originally done, and whether or not the disease was FDG-avid on the baseline PET) may be performed in some institutions at the completion of therapy (and is included in imaging guidelines from the bone tumor committee of the Children’s Oncology Group (table 5) [99]), but it is rarely positive. At most institutions, bone scans and PET scans are utilized primarily for evaluation of new symptoms following therapy. In the uncommon subset of patients with unresectable tumors (eg, in the proximal sacrum), PET scan may be preferable to bone scan for following the patient after radiation therapy since bone scans can show persistent activity reflective of ongoing bone remodeling in this setting. Following primary therapy, long term follow-up of patients should continue indefinitely, because of treatment-related toxicity (including secondary malignancy) and the possibility of a late relapse, which may occur as late as 20 years after therapy [100]. Recommendations for long-term follow-up are available from the Children’s Oncology Group Survivorship Guidelines. For individuals treated with anthracycline-containing chemotherapy, the recommendation includes an echocardiogram or equilibrium multigated blood pool imaging (MUGA scan) at the first long-term follow-up visit, then periodically reassessment of cardiac function, with the frequency of testing dependent on age at initial treatment, the total anthracycline dose, and whether chest irradiation was also administered. TREATMENT OF RECURRENT DISEASE — Patients with a disease recurrence after resection alone can often be salvaged with additional surgery and chemotherapy, although their long-term survival is inferior to that of patients who received conventional multiagent chemotherapy in conjunction with surgery upfront [101,102]. Treatment of relapse in patients who have already received adjuvant and/or neoadjuvant chemotherapy is a more difficult situation. Such patients usually have received most of the effective drugs, and presumably their tumors are more chemotherapy-resistant than those that have never been exposed to antineoplastic agents [103]. Nevertheless, salvage is still possible and is more likely in patients with a longer relapsefree interval. In a large database of 565 osteosarcoma patients who relapsed after being treated on one of three different neoadjuvant chemotherapy protocols within the European Osteosarcoma Intergroup, five year survival post-relapse in those whose disease recurred after two years versus within two years of randomization was 35 versus 14 percent, respectively [104]. Other favorable prognostic factors in recurrent osteosarcoma include no more than one or two pulmonary nodules, the presence of unilateral pulmonary involvement, lack of pleural disruption, and achieving a second surgical remission [29,101,105-108]. 105 The predominant site of disease relapse is in the lung, but the increasing use of adjuvant/neoadjuvant chemotherapy for localized osteosarcoma may have changed the pattern of relapse. Patients whose disease recurs following adjuvant chemotherapy tend to relapse later and with fewer pulmonary lesions than those treated with surgery alone; they also have a higher chance of relapsing at distant bony sites [103,109-111]. However, others have observed no change in relapse pattern with the increasing use of upfront intensive chemotherapy [112]. Although rare, the possibility of metachronous, sometimes multiple osteosarcomas must be considered in a patient who has what appears to be isolated bone relapse without pulmonary involvement [113]. Therapeutic approach — The therapeutic approach to recurrent disease and the likelihood of prolonged survival are related to the location, extent, and timing of relapse [101,106,114]. Attempted resection is probably the most appropriate initial treatment for patients who relapse late (ie, beyond one year after initial therapy) with a small number of pulmonary nodules that do not invade the pleura and that can be completely resected [80,115-122]. Surgery — Complete resection of all metastatic sites is a prerequisite for long-term survival. The importance of resection can be illustrated by a series of 249 consecutive patients who developed a second or subsequent recurrence after modern combined modality treatment of osteosarcoma [114]. While only one of 205 patients with recurrence survived past five years without surgical remission, the five-year survival and event-free survival rates were 32 and 18 percent for 119 second, 26 and 0 percent for 45 third, 28 and 13 percent for 20 fourth, and 53 and 0 percent for five fifth recurrences, respectively, in which a surgical remission was achieved. Some patients with isolated pulmonary metastases that develop more than one year after original treatment will be long-term survivors with just surgical removal of their lung nodules [80,115]. Multiple procedures may be necessary, although the odds of long-term survival decrease with each subsequent relapse. Many investigators advocate the use of postthoracotomy chemotherapy (and even lung irradiation) to destroy presumed residual microscopic deposits after surgical treatment of overt metastatic disease [123]. The contribution of such therapy to long-term outcomes has not been examined in a controlled study and remains to be defined. However, it is probably reasonable to administer such treatment to patients who undergo resection of more than three lesions appearing within six months to one year of surgery. Patients who relapse early (ie, within six months of surgery) seem to fare poorly even if their disease is amenable to complete resection [80,105,116,124]. This is most likely because of drug resistance that develops during chemotherapy. Although aggressive use of surgery, chemotherapy, and at times RT aimed at eradicating all sites of disease can lead to prolonged survival in some of these patients [103,105,117,123,125-127], it is difficult to be overly optimistic about the chances of cure after early relapse, especially relapse involving nonpulmonary sites or unresectable pulmonary disease [23,105,116,118]. Unresectable disease — Patients who relapse with unresectable metastatic disease are most often incurable and should be considered for palliative therapy (eg, RT) [128]. For selected patients with a limited volume of unresectable disease, particularly those who have no prior exposure to systemic chemotherapy, consideration should be given to administering chemotherapy before resection. While chemotherapy rarely produces a complete response at metastatic sites, some patients with inoperable metastases (including the rare patient with bone metastases [129]) may respond sufficiently to permit complete resection at a later date. While it has not been shown that chemotherapy given before or after resection of metastatic lung disease improves patient survival in comparison to patients who have resection of metastases alone [88], there are no randomized trials that directly address this issue. 106 Choice of chemotherapy regimen — Most of these patients will already have received standard chemotherapy with HDMTX, doxorubicin, and cisplatin. Combinations of etoposide and ifosfamide are more active than either agent alone [83,130] and commonly used, without or with carboplatin (ICE) [131-133]. In general, response rates are lower than in the setting of newly diagnosed disease. However, although ifosfamide clearly shows a higher response rate in newly diagnosed patients, objective responses have been reported in patients whose disease relapses after they have received ifosfamide-containing adjuvant therapy [83,131,134]. Combinations of cyclophosphamide with etoposide are also active [135]. In an early report 14 of 42 children with recurrent osteosarcoma (33 percent) responded to etoposide (100 mg/m2 daily for three days) and lower dose ifosfamide (2 g/m2 daily for three days) [134]. With the expectation that adding G-CSF would permit dose intensification, a phase I trial of etoposide and escalating doses of ifosfamide in patients with recurrent osteosarcoma was completed; in a preliminary report, six of 13 patients exhibited a partial response [136]. Other options include cyclophosphamide plus etoposide [135,137,138] or gemcitabine plus docetaxel [139]. Clinical trials, when available, are appropriate as well. Samarium — Samarium-153 lexidronam is a bone-seeking radiopharmaceutical that may provide pain palliation for patients with an unresectable local recurrence or skeletal metastases [140,141]. However, as a radioactive agent that is taken up by bone, Samarium-153 can be toxic to bone marrow and can leave patients transfusion dependent. Thus, marrow reserve should be considered before its use. (See "Radiation therapy for the management of painful bone metastases", section on 'Radiopharmaceuticals'.) NEWER AND INVESTIGATIONAL APPROACHES — A study of the mTOR inhibitor ridaforolimus in patients with metastatic sarcoma suggested potential activity for this class of compounds in patients with osteosarcoma [142], raising the possibility of using agents of this drug class in patients with metastatic disease; since the RECIST response rate in this study was very low, mTOR inhibitors remain investigational; perhaps it will become possible to identify patients who do benefit from these agents. Among other interesting agents that may have clinical utility are inhibitors of insulin-like growth factor I receptor (IGF IR), since IGF signaling is critical for bone formation during development. Early studies with a variety of monoclonal antibodies and small molecule inhibitors of the IGF IR have been completed, with largely disappointing results even for the combination of an mTOR inhibitor and an IGF IR inhibitor [143]. A study evaluating the feasibility of adding trastuzumab to standard chemotherapy for patients with HER2-positive osteosarcoma was just completed by the Children's Oncology Group (COG) [144]. The results of this study are not yet available. Bisphosphonates — Numerous in vitro and xenograft studies support the concept that bisphosphonates have activity against osteosarcoma alone or in combination with chemotherapy (reviewed in [145]); the safety of combining pamidronate with chemotherapy in patients with newly diagnosed osteosarcoma has been shown in a small phase II trial [145]. However, until further data are available, this approach should not be considered outside of the context of a clinical trial. The COG is evaluating the feasibility of adding zoledronic acid to standard chemotherapy for patients with metastatic osteosarcoma. Immunotherapy — Immune responses may influence the survival of patients with osteosarcoma. Cytotoxic lymphocytes are present in such patients [146,147], and in at least one study, the degree of lymphocytic infiltration correlated with survival [147]. These findings have prompted investigators to explore a variety of immunotherapeutic approaches for patients with advanced osteosarcoma. 107 The addition of Bacille Calmette-Guerin (BCG) and interferon did not improve survival when added to multiagent chemotherapy [148,149]. However, as noted above, liposomal muramyl tripeptide-phosphatidyl-ethanolamine (mifamurtide), an agent derived from BCG that activates macrophages and increases circulating cytokine levels [150,151], was studied in a randomized study, described above, in which patients were assigned, using a 2 x 2 factorial design, to standard chemotherapy with or without ifosfamide and then to receive or not receive mifamurtide [59]. The addition of mifamurtide to standard chemotherapy significantly improved overall survival but there was only a trend toward improved event-free survival. (See 'Benefit of mifamurtide' above.). However, when the analysis was restricted to the 91 patients with metastatic disease at diagnosis, there was only a nonstatistically significant trend toward improved five-year event free survival (42 versus 26 percent) and overall survival (53 versus 40 percent) that favored mifamurtide [152]. Thus, the role of mifamurtide in patients with metastatic osteosarcoma remains uncertain and a further randomized trial seems warranted. Another immunotherapeutic approach that is being pursued for pulmonary metastatic disease is inhalation of aerosolized granulocyte macrophage colony-stimulating factor (GM-CSF). GM-CSF stimulates the proliferation and differentiation of hematopoietic progenitor cells and augments the functional activity of neutrophils, monocytes, macrophages, and dendritic cells. Early data using aerosolized GM-CSF in a variety of cancers with pulmonary metastases suggested potential efficacy [153,154]. However, benefit could not be shown in a trial of inhaled GM-CSF in 43 patients with pulmonary relapse from osteosarcoma in the American Osteosarcoma Study Group [AOST] protocol 0221 [155]. There was no detectable immunostimulatory effect on the pulmonary metastases or suggestion of improved outcomes post relapse. Aerosolized GM-CSF is not a standard approach and should only be considered in the context of a clinical trial. SUMMARY AND RECOMMENDATIONS Localized disease The survival of patients with malignant bone sarcomas has improved dramatically over the past 30 years, largely owing to chemotherapeutic advances. Adjuvant chemotherapy significantly improves survival compared to surgery alone and is considered a standard component of therapy for both children and adults. (See 'Prognosis and evolution of treatment' above.) The optimal timing of chemotherapy (ie, preoperative versus postoperative) has not been established. There is no survival benefit for neoadjuvant as compared to adjuvant chemotherapy, but many centers preferentially utilize preoperative chemotherapy, particularly if a limb-sparing procedure is being contemplated for an extremity osteosarcoma. For children, adolescents, and young adults up to age 40, we suggest a methotrexate-containing chemotherapy regimen over a two-drug nonmethotrexate containing regimen (Grade 2B). (See 'Role of methotrexate' above.) The optimal methotrexate-containing regimen has not been established. For children, adolescents, and young adults up to age 40 with localized osteosarcoma, the control arm of the recently closed protocol AOST0331 (MAP chemotherapy, HDMTX plus doxorubicin and cisplatin, (table 4)) is a reasonable and appropriate choice for standard therapy. If preoperative therapy is chosen, we administer 10 weeks of chemotherapy prior to surgery and continue postoperative chemotherapy for 29 weeks, starting one week postoperatively. For most patients, we suggest against a four-drug as compared to a three-drug regimen (Grade 2B). However, in countries where mifamurtide is available, mifamurtide in conjunction with ifosfamide, methotrexate, doxorubicin, and cisplatin chemotherapy may be used. (See 'Benefit of mifamurtide' above.) 108 For older adults, the benefit of HDMTX is unproven, and we suggest cisplatin and doxorubicin alone (Grade 2B). However, a HDMTX-containing regimen is a reasonable alternative in patients with adequate renal function. (See 'Chemotherapy for adults' above.) Practical tips on administration of HDMTX, including therapeutic drug monitoring, the rationale for leucovorin rescue and prevention and management of prolonged high plasma MTX levels, are discussed in detail elsewhere. For patients who demonstrate intolerance of HDMTX and for institutions that cannot provide pharmacokinetic monitoring for HDMTX, the combination of carboplatin, ifosfamide, and doxorubicin appears to be a reasonable alternative. However, the increased doses of alkylating agents may possibly increase the risk of second malignancies (leukemia). (See 'Role of methotrexate' above.) For patients treated off protocol, we suggest not pursuing a change in the chemotherapeutic regimen after surgery in patients who have less than a complete response to the initial neoadjuvant chemotherapy (Grade 2B). (See 'Tailored postoperative chemotherapy based upon initial response' above.) There is no role for adjuvant RT in patients with completely resected osteosarcoma, and for most patients, we recommend not pursuing this approach because of the risk for late toxicity (Grade 1B). We restrict the use of adjuvant RT to those patients with incompletely resected sarcomas and for the rare patient with a small cell osteosarcoma. RT should be considered as the primary modality for local control only for patients who decline surgery or for whom there is no effective surgical option. (See 'Radiation therapy' above.) Metastatic disease at diagnosis Patients who present with overtly metastatic osteosarcoma have a poor prognosis, although long-term survival is possible in up to 50 percent of those with isolated pulmonary metastases who are treated with combined systemic and surgical therapy. For children and adolescents with metastatic osteosarcoma at the time of diagnosis who do not have a surgical option, there is no single standard approach to treatment, and these patients should be encouraged to enroll on clinical trials testing new therapies. If protocol enrollment is not available or if patients are ineligible, treatment with the control arm of the AOST0331 protocol (MAP chemotherapy alone, (table 4)) or as per AOST06P1 without zoledronic acid (MAP plus ifosfamide and etoposide) are reasonable and appropriate choices for standard therapy. Posttreatment surveillance — We follow guidelines for posttreatment surveillance from the Children’s Oncology Group, which are outlined in the table (table 5). (See 'Posttreatment surveillance' above.) Recurrent disease The predominant site of disease relapse following initially successful treatment is lung. Attempted resection is probably the most appropriate initial treatment for patients who relapse beyond one year after initial therapy with a small number of pulmonary nodules that do not invade the pleura and that can be completely resected. Although some of these patients may be cured with resection alone, most receive a combination of surgery plus chemotherapy. Patients who relapse with unresectable metastatic disease are incurable and should be considered for palliative therapy (eg, RT, chemotherapy). However, in selected patients with a limited volume of unresectable disease, particularly those who have no prior exposure to systemic chemotherapy, consideration should be 109 given to administering chemotherapy before resection. While chemotherapy rarely produces a complete response at metastatic sites, some patients with inoperable metastases may respond sufficiently to permit complete resection at a later date. (See 'Therapeutic approach' above.) The choice of chemotherapy depends upon prior treatment. Most patients will have already received HDMTX, doxorubicin, and cisplatin. For these patients, we suggest treatment on a research study, when feasible. Off study, we typically use etoposide plus ifosfamide, without or with carboplatin (ICE). If chemotherapy has not been previously administered, we suggest the same regimens as are used for primary management. (See 'Choice of chemotherapy regimen' above.) Treatment protocols for soft tissue and bone sarcoma 110 111 112 113 114 115 116 BİYOPSİ SUMMARY Primary malignant bone tumors are uncommon malignancies, but they are an important cause of cancer morbidity and mortality, especially among teenagers and young adults. (See 'Introduction' above.) Primary malignant bone tumors are classified according to their cytologic features and cellular products (table 1). The goals of the preoperative evaluation are to establish the tissue diagnosis, evaluate disease extent, and assess the feasibility of a limb-sparing approach. Although plain radiographs can often predict the probable histology of a 117 potentially malignant bone lesion, the definition of tumor size and local intraosseous and extraosseous extent is most accurately achieved by magnetic resonance imaging (MRI). CT is the best method to evaluate the thorax for metastatic disease. PET scans are increasingly being used in place of radionuclide bone scans to evaluate the rest of the skeleton as well as other distant metastatic sites. (See 'Diagnostic and staging work-up' above.) The radiographic differential diagnosis depends upon recognition of the tissue type and the degree of aggressiveness (as determined by lesion size and relationship with the surrounding tissues [invasiveness]). Recognition of fat within a tumor is generally a sign that it is benign. (See 'Differential diagnosis' above.) Bone biopsy is indicated in the following circumstances (see 'Indication for biopsy' above): Whenever there is significant doubt as to the diagnosis of a benign or malignant lesion When the histologic distinction among possible diagnoses could alter the planned course of treatment When definitive confirmation of the diagnosis is required before undertaking a hazardous, costly, or potentially disfiguring treatment The biopsy of a suspected primary bone tumor must be carefully planned to avoid compromising the oncologic outcome. Although a fine needle aspiration biopsy may be adequate for the diagnosis of a metastatic or recurrent bone lesion, core needle biopsy, or an open biopsy is usually required for most primary bone tumors. Open biopsies are being done less frequently since most core needle biopsies can deliver sufficient tissue on multiple passes for all diagnostic studies. Given the heterogeneity of most sarcomas, the ability to guide the core needle to regions of more aggressive tumor is very important to accurately diagnosis and grade the sarcoma. (See 'Biopsy techniques' above.) The extensive pathologic evaluation that is often required to ascertain the correct diagnosis may require special handling for all or part of the specimen. This process can last one to two weeks depending on the number of tests necessary. Among patients undergoing an open biopsy, frozen section analysis is important in determining whether sufficient tissue has been acquired. Chondrosarcoma Literature review current through: Aug 2013. | This topic last updated: Tem 26, 2013. 118 119 120 121 122 INTRODUCTION — Chondrosarcomas are a heterogeneous group of malignant bone tumors that share in common the production of chondroid (cartilaginous) matrix [1]. Chondrosarcomas are the third most common primary malignancy of bone after myeloma and osteosarcoma [2]. They account for 20 to 27 percent of primary malignant osseous neoplasms [3]. Clinical behavior is variable. Ninety percent are conventional chondrosarcomas, 90 percent of which are low- to intermediate-grade tumors [4]. These tumors are slow growing with a low metastatic potential. They are considered relatively refractory to chemotherapy and radiation therapy. In contrast, high-grade chondrosarcomas, which include 5 to 10 percent of conventional chondrosarcomas as well as some rare variants, have a high metastatic potential and a poor prognosis following resection alone [4]. Some of the rare subtypes are more responsive to chemotherapy and radiation. 123 This topic review will provide an overview of the classification, clinical characteristics, and therapeutic options for chondrosarcoma. The rare chondrosarcomas involving the head and neck and skull base, as well as diagnosis and biopsy techniques for bone sarcomas in general, are discussed elsewhere. HISTOLOGIC GRADING AND PROGNOSIS — Histologic grade is one of the most important indicators of clinical behavior and prognosis [5-8]. Chondrosarcomas are graded on a scale from 1 to 3, based upon nuclear size, staining pattern (hyperchromasia), mitotic activity, and degree of cellularity (picture 1). Grade 1 chondrosarcomas were reclassified in the updated World Health Organization (WHO) 2013 classification system as "atypical cartilaginous tumours" (ACT/CS1) [1]. They are moderately cellular, with an abundant hyaline cartilage matrix. The chondrocytes have small, round nuclei and are occasionally binucleate. Mitoses are absent. ACT/CS1 almost never metastasize (1 percent risk in one series of patients [4]) and are therefore now considered a locally aggressive neoplasm rather than a malignant sarcoma. Ten-year survival is 83 to 95 percent [4,5,9]. Grade 2 chondrosarcomas are more cellular with less chondroid matrix than ACT/CS1 tumors. Mitoses are present, but widely scattered. The chondrocyte nuclei are enlarged and can be either vesicular or hyperchromatic. The metastatic potential is intermediate between low-grade and high-grade chondrosarcomas (approximately 10 to 15 percent). Ten-year survival is approximately 64 to 86 percent [4,5,9]. The vast majority of conventional (primary and secondary) chondrosarcomas are ACT/CS1 or chondrosarcoma grade 2 [4,9,10]. Grade 3 chondrosarcomas are highly cellular, with nuclear pleomorphism and easily detected mitoses. Chondroid matrix is sparse or absent. High-grade chondrosarcomas have a high metastatic potential (approximately 32 to 70 percent) and a poor prognosis with surgical resection alone [4,5]. The 10-year survival rate is about 29 to 55 percent [4,9]. In most cases, the histologic grade of differentiation of a recurrent chondrosarcoma is the same as the primary lesion; however, up to 13 percent of recurrences exhibit a higher grade of malignancy when compared with the original neoplasm [5,9,11]. This suggests that chondrosarcomas can progress biologically. Histologic grading is subject to interobserver variability [12,13], which can be problematic since surgical therapy for ACT/CS1 and grade 2 chondrosarcomas is often different. Because of this, there is an urgent need for molecular markers that can be used to predict clinical behavior, guide therapeutic decision making, and provide novel targets for molecularly targeted therapy [14]. CLASSIFICATION, HISTOLOGY, AND CLINICAL FEATURES Precursor lesions — Two benign cartilaginous lesions that can precede chondrosarcoma are described: Osteochondroma — An osteochondroma (osteocartilaginous exostosis) is a cartilagecapped bony projection arising on the external surface of a bone (picture 2 and image 1); it contains a marrow cavity that is continuous with that of the underlying bone. The majority are located in the long bones, predominantly around the knee. The inherited condition multiple osteochondromas (hereditary multiple exostoses) is characterized by the development of two or more osteochondromas in the appendicular and axial skeleton. This syndrome is inherited in an autosomal dominant fashion. The 124 prevalence in the general population is 1:50,000, and males are affected slightly more often than females. Almost 90 percent of cases of multiple osteochondromas are caused by inheritance of a germline mutation in one of the tumor suppressor genes EXT1 or EXT2. Although most are asymptomatic, osteochondromas can cause pain, functional problems, and deformity; they also carry a risk for fracture. Malignant transformation is estimated to occur in 5 percent of patients with a solitary or multiple osteochondromas [15-20]. In one series, the average time between initial diagnosis and malignant transformation was 9.8 years [19]. All chondrosarcomas arising in the setting of an osteochondroma are secondary peripheral tumors. A change in the size of an osteochondroma or new onset of symptoms warrants investigation, as each may herald malignant transformation [21,22]. Osteochondromas located at the pelvis, hips and shoulder girdle are reported to be particularly prone to malignant transformation [21,22]. Among patients with multiple osteochondromas, malignant transformation appears to be unrelated to the presence or absence of an EXT mutation, sex, severity of disease, or the number of lesions [20]. Enchondroma — Enchondromas are common benign cartilaginous tumors that develop in the medulla (marrow cavity) of bone (image 2). When multiple enchondromas are present, the condition is called enchondromatosis (image 3), of which the most common form is Ollier disease (estimated prevalence 1 in 100,000) [23]. When multiple enchondromas are associated with soft tissue hemangiomas, the designation is Maffucci syndrome (image 4). Both are congenital but not inherited. Although the vast majority are asymptomatic, clinical problems caused by enchondromas include skeletal deformity, limb-length discrepancy, and a risk for malignant transformation. Malignant transformation in a solitary enchondroma is presumed to be extremely rare (<1 percent) but it has been described (image 5) [19]. The risk of chondrosarcoma in patients with Ollier disease or Maffucci syndrome is as high as 50 percent [23-28]. The risk is highest with enchondromas located in the pelvis [27]. Malignant transformation usually presents after skeletal maturity and may be heralded by the development of pain [19]. The histologic and radiographic distinction between an enchondroma and a low-grade chondrosarcoma may be difficult, even in experienced hands (see 'Histologic appearance' below). Conventional chondrosarcomas Central chondrosarcoma — Central chondrosarcomas of bone arise within the medullary cavity and constitute approximately 75 percent of all chondrosarcomas (table 1). The majority are thought to arise primarily (ie, without a benign precursor lesion). However, the finding of remnants of a preexisting enchondroma in 40 percent of central chondrosarcomas and the fact that most enchondromas remain asymptomatic and clinically silent have led some to hypothesize that as many as 40 percent of central chondrosarcomas could be secondary to a preexisting enchondroma [29]. The majority of patients are over the age of 50. There is a slight male predominance. The most commonly involved skeletal sites are the proximal femur, bones of the pelvis (particularly the ilium), and proximal humerus (together accounting for about 75 percent of cases), followed by distal femur, ribs, tibia, and metacarpal and metatarsal bones [3,4,30,31]. Other less frequently involved sites include the spine, the skull base, and the craniofacial bones. (See "Chordoma and chondrosarcoma of the skull base" and "Head and neck sarcomas".) Local swelling and pain are the most common presenting symptoms. Pain is typically insidious, progressive, worse at night, and often present for months to years before presentation. A pathologic fracture is present at diagnosis in 3 to 17 percent of patients 125 [3]. Primary spinal chondrosarcomas are rare but can cause compression of the spinal cord. Peripheral chondrosarcoma — By definition, all peripheral chondrosarcomas arise within the cartilage cap of a preexisting osteochondroma (table 1). Patients with a peripheral chondrosarcoma are generally younger than those with a central chondrosarcoma (table 1) [21]. The clinical presentation is similar to that of central chondrosarcoma, usually pain and local swelling. The most commonly involved bones are the pelvis and bones of the shoulder girdle, although in some series, the long bones predominate [19]. The distinction between osteochondroma and secondary peripheral atypical cartilaginous tumour/chondrosarcoma grade I arising in osteochondroma can be difficult and should be made in a multidisciplinary team. The size of the cartilaginous cap is the most important parameter [32]. Periosteal chondrosarcoma — Fewer than one percent of conventional chondrosarcomas arise on the surface of a bone and are designated periosteal (previously termed juxtacortical) chondrosarcomas. They most frequently affect adults in their 20s and 30s, and have a slight male predilection. The metaphyses of long bones are most frequently involved, especially of the distal femur (figure 1). Patients typically present with a palpable, painless, slowly growing mass [3]. Periosteal chondrosarcomas usually have a good prognosis after adequate local surgery despite histologic features of a high-grade lesion [7,33-36]. Histological grading is therefore not used at this location. Histologic appearance — Despite their different origin and location, primary and secondary chondrosarcomas appear histologically similar (picture 1). At low magnification, there is abundant cartilage matrix production, and the irregularly shaped lobules of cartilage, often separated by fibrous bands, may permeate the bony trabeculae [1]. Necrosis or mitoses may be seen in high-grade lesions. The histology of periosteal chondrosarcomas is similar except that the appearance is often more worrisome with increased cellularity and nuclear atypia. The histologic (and radiographic) distinction between a benign cartilage lesion and an atypical cartilaginous tumour/chondrosarcoma grade 1 (ACT/CS1) can be extremely difficult [12,37,38]. ACT/CS1 is hypercellular when compared to benign cartilage lesions. The chondrocytes appear mildly to moderately atypical and contain enlarged hyperchromatic nucleoli. Permeation of preexisting host bone and mucomyxoid matrix changes are important characteristics that can be used to separate ACT/CS1 from an enchondroma [13]. Periosteal chondrosarcoma is distinguished from periosteal chondroma based upon size (≥5 cm) and the presence of cortical invasion. For phalangeal enchondromas, more worrisome histologic features are tolerated, and the diagnosis of chondrosarcoma at this site is based upon the presence of cortical destruction, soft tissue invasion, and mitoses [39]. Fortunately, the distinction between enchondroma and ACT/CS1 is not always essential for clinical decision making, since treatment (curettage and adjuvant phenol application or cryosurgery) is often similar for enchondromas as well as central ACT/CS1 (see 'Surgical treatment' below). Rare chondrosarcoma subtypes — In addition to conventional central, peripheral, and periosteal chondrosarcomas, several rare subtypes are described, together constituting less than 10 percent of all chondrosarcomas. Dedifferentiated chondrosarcoma — Dedifferentiated chondrosarcomas contain two juxtaposed components: a well-differentiated cartilage tumor (which can be either an enchondroma or a low-grade chondrosarcoma), and a high-grade non-cartilaginous sarcoma which most frequently is an osteosarcoma, fibrosarcoma, or an undifferentiated high-grade pleomorphic sarcoma (previously termed malignant fibrous histiocytoma) (picture 3) [40]. 126 Both components appear to share some genetic aberrations [41,42] with additional genetic changes in the high-grade component [41-44]. This suggests a common precursor cell with early divergence of the two components. The average age at presentation is older (between 50 and 60 years) than other chondrosarcoma subtypes (table 1). The majority occur centrally in medullary bone; the most common sites of involvement are the pelvis, femur, and humerus. The typical presentation is with pain, although swelling, paresthesias, and pathologic fractures are also common [45]. The majority of patients have an associated soft tissue mass [46]. Dedifferentiated chondrosarcomas are biologically aggressive and they have a poor prognosis [11,46,47]. In a multicenter review of 337 patients, 71 (21 percent) had metastases at the time of diagnosis; they had a 10 percent chance of survival at two years [11]. Even for patients without metastases at diagnosis, survival was only 28 percent at 10 years. Poor prognostic factors include pathologic fracture, pelvic location, and older age. Mesenchymal chondrosarcoma — Mesenchymal chondrosarcomas are highly malignant tumors that are characterized by differentiated cartilage admixed with solid highly cellular areas that are composed of undifferentiated small round cells (picture 3) [48]. The average age is 25, younger than that of other types of chondrosarcoma (table 1). There is a high proportion of extraskeletal primary tumors, which is not seen with other chondrosarcoma subtypes [49]. Of the approximately one-third of cases that affect the extraskeletal soft tissues, the meninges are one of the most common sites [50]. Also in contrast to conventional chondrosarcomas, mesenchymal tumors most commonly involve the axial skeleton, including the craniofacial bones (especially the jaw) [51], ribs, ilium, and vertebra, and there may be involvement of multiple bones. About 20 percent have metastatic disease at diagnosis [52]. The main symptoms are pain and swelling, and it is not uncommon for symptoms to have been present for many months. Mesenchymal chondrosarcomas have a tendency toward both local and distant recurrences, which may arise as long as 20 years following the initial diagnosis [53]. The prognosis is markedly worse than for conventional primary chondrosarcomas. Reported 10-year survival rates are 10 to 20 percent [49,52-54]. Clear cell chondrosarcoma — Clear cell chondrosarcoma is a rare low-grade variant of chondrosarcoma which is characterized by the presence of lobulated groups of blandappearing tumor cells with large, centrally-located nuclei and clear, empty cytoplasm in addition to hyaline cartilage (picture 3). Mitotic figures are rare. Many tumors contain zones of conventional chondrosarcoma with hyaline cartilage and minimally atypical nuclei. Although these tumors can arise at any age, most patients are between the ages of 25 and 50 (table 1). Men are three times more likely as women to develop this particular subtype [55]. Approximately two-thirds of tumors arise in the epiphyseal ends of the humerus or femur (figure 1). Pain, which may have been present for longer than one year, is the most common complaint. Serum alkaline phosphatase levels are often elevated at diagnosis and may provide a useful tumor marker [56]. Despite their low-grade nature, marginal excision or curettage is associated with a 70 percent or higher recurrence rate and should be avoided [56,57]. In incompletely excised cases, metastases may develop, usually to the lungs and other skeletal sites, and the overall mortality rate is up to 15 percent [55]. In contrast, en bloc wide local excision is usually curative. Disease recurrence may occur up to 24 years after initial diagnosis [56,57]. Long-term follow-up is mandatory. 127 Myxoid chondrosarcoma — It is debated whether myxoid chondrosarcoma of bone is a true separate entity or if it represents a high-grade conventional chondrosarcoma with prominent myxoid change. The light microscopic features of myxoid chondrosarcoma of bone are similar to those of the more commonly described extraskeletal myxoid chondrosarcoma (EMC), a soft tissue sarcoma which most commonly arises in the lower extremities [58,59]. However, these two entities are unrelated [60], and the term "chondrosarcoma" to describe EMC is a misnomer. Well-formed hyaline cartilage is found only in a minority of EMCs [61,62], while S100 expression (which is present in all or most chondrosarcomas) is often very focal or absent. Expression of collagen II and aggrecan (two other markers of cartilaginous differentiation) are absent in 86 percent of EMCs [62]. Furthermore, the translocation t(9:22) that is specific for EMC is generally absent in socalled myxoid chondrosarcoma of bone, and its ultrastructure is different [61,63]. The reported cases of myxoid chondrosarcoma of bone that contain a proven translocation of t(9:22) have a large soft tissue component which makes distinction from EMC with secondary bone destruction extremely difficult [64], although some convincing cases have been reported [65]. Thus, EMC and so-called myxoid chondrosarcoma of bone appear to represent two different entities. The 2013 WHO classification classifies the entity EMC in the "tumors of uncertain differentiation" category [1]. Myxoid chondrosarcomas of bone are not designated as a unique entity, and these tumors should be regarded as a myxoid variant of intermediate- or high-grade conventional chondrosarcoma. Molecular pathogenesis — Cartilaginous tumors are nearly always found in bones that arise from enchondral ossification. Research has uncovered some parallels between chondrocyte growth and differentiation in the normal growth plate and both benign and malignant cartilaginous tumors [66]. Within the normal growth plate, resting zone chondrocytes proliferate and differentiate, becoming hypertrophic. These cells undergo apoptosis, allowing the invasion of vessels and osteoblasts that start to form bone and lead to longitudinal bone growth. This physiologic process is tightly regulated by components of the Indian hedgehog (IHH)/parathyroid hormone related (PTHRP) protein signaling pathway. Patients with multiple osteochondromas (previously called hereditary multiple exostoses) have germline mutations in the exostosin (EXT1 or EXT2) genes [67-69], with loss of the remaining wild type allele in the cartilage cap of the osteochondroma [70]. The end result is decreased EXT expression. Loss of expression of the EXT genes through homozygous deletion of EXT1 is also seen in solitary osteochondromas that are unassociated with the hereditary syndrome [71,72]. The EXT gene products are involved in the biosynthesis of heparan sulfate proteoglycans (HSPGs), which are essential for cell signaling through IHH/PTHLH and other pathways [73]. In osteochondromas where EXT is inactivated, the HSPGs seem to accumulate in the cytoplasm and Golgi apparatus instead of being transported to the cell surface [72]. This hampers multiple growth signaling pathways (including the Indian hedgehog (IHH)/parathyroid hormone related (PTHRP) protein pathways), which, as noted above, are important for normal chondrocyte proliferation and differentiation within the normal human growth plate. In secondary peripheral chondrosarcomas arising in osteochondromas EXT is usually wild type, suggesting that the wild type cells in osteochondroma are prone to malignant transformation through EXT independent mechanisms [74]. The specific trigger underlying malignant transformation of osteochondromas is unknown. A role for IHH signaling has been suggested, although the data are not entirely consistent [75-79]: 128 PTHRP signaling, which is downstream of IHH and is involved in chondrocyte proliferation, is absent in osteochondromas, but upregulated with malignant transformation towards secondary peripheral chondrosarcoma, especially in highgrade lesions [76,77,80-83]. There is decreased expression of downstream targets in the IHH signaling cascade during tumor progression in peripheral chondrosarcomas, while they are still active in central chondrosarcomas [84]. Data from in vitro and in vivo models show that treatment of central chondrosarcoma cells with recombinant Hedgehog increases proliferation, whereas treatment with Hedgehog signaling inhibitors inhibits tumor proliferation and growth in a small subset of tumors and chondrosarcoma cell cultures [79,84,85]. A multistep genetic model for the development of secondary (peripheral) chondrosarcomas has been proposed (figure 2) [14]. Until recently, less was known about the molecular genetics of enchondromas and the far more common primary (central) chondrosarcomas. However, recently, point mutations in isocitrate dehydrogenase-1 and isocitrate dehydrogenase 2 genes IDH1 and IDH2 have been identified in 40 to 56 percent of enchondromas and chondrosarcomas and seem to be an early event [86,87]. Also, Ollier disease and Maffucci syndrome are caused by somatic mosaic mutations in IDH1 and IDH2 [86,88]. The identification of IDH1 and IDH2 mutations in four chondrosarcoma cell lines provides an in vitro model to study the role of these mutations in tumorigenesis [86]. In addition, although EXT is not involved, involvement of the IHH/PTHLH signaling pathway is suggested by the observations that parathyroid hormone-related protein (PTHRP) signaling is active in enchondromas [80,83], and hedgehog signaling is active in central chondrosarcomas [84]. Moreover, a mutation in the gene encoding the receptor for PTHRP (PTH-1 receptor or PTH1R) has been identified in enchondromatosis that is claimed to lead to constitutive activation of IHH signaling [78,89]. Three new heterozygous missense mutations have been described in the PTH1R gene in patients with Ollier disease, which result in reduced receptor function [90]. Mutations in PTH1R have not been found in sporadic chondrosarcomas, nor in Maffucci syndrome [86,88,91]; this gene may contribute to pathogenesis in only a very small subset (<5 percent) of patients with Ollier disease. Moreover, using whole exome sequencing, mutations were found in different genes involved in hedgehog signaling [92]. While enchondromas and low-grade chondrosarcomas are near-diploid and carry few karyotypic abnormalities, high grade chondrosarcomas are aneuploid and have complex karyotypes [39,93]. Some of the few consistent genetic aberrations include 12q13-15 and 9p21 rearrangements [39,93-96]. Chondrosarcoma progression has been linked to the CDKN2A (p16) tumor suppressor gene, located at 9p21 [97,98] and by alterations in p53 [99]. Mutations in COL2A1 are found in a subset of chondrosarcomas, the meaning of which is as yet unknown [92]. Activation and/or overexpression of platelet-derived growth factor receptor-alpha (PDGFRA) and beta (PDGFRB) has been described in conventional primary chondrosarcomas, although activating mutations have not been found [100,101]. The therapeutic implications of this finding are discussed below. (See 'Novel therapies' below.): A multistep genetic model for development of primary chondrosarcomas has been proposed (figure 3) [14]. Dedifferentiated chondrosarcomas also contain IDH1 or IDH2 mutations in approximately 50 percent of the cases [86,87,102]. 129 The majority of mesenchymal chondrosarcomas was shown to harbor a specific HEY1NCOA2 fusion product caused by an intrachromosomal rearrangement of chromosome arm 8q [103]. Alternatively, a IRF2BP2-CDX1 fusion gene brought about by translocation t(1;5)(q42;q32) was described [104]. For clear cell chondrosarcoma, no specific recurrent alterations have been found so far [102]. DIAGNOSTIC AND STAGING WORK-UP — The goals of the preoperative evaluation are to establish the tissue diagnosis and evaluate disease extent, in order to select the appropriate therapeutic approach. One fundamental principle applying to diagnosis of both tumors of bone and cartilage is that both the histology and radiography of bone tumors are not specific. Integration of the clinical history, radiography, and pathology is necessary to render a specific diagnosis. Radiographic imaging — The initial imaging study in a patient with a painful musculoskeletal swelling is often a plain radiograph. The location and radiographic appearance of the different chondrosarcoma subtypes are often characteristic [1]. However, although plain radiographs can provide a clue as to the probable histology of a potentially malignant bone lesion, evaluation of tumor size and local extent is most accurately achieved by magnetic resonance imaging (MRI) and/or CT [105]. CT is optimal to detect matrix mineralization, particularly when it is subtle or the lesions are located in an anatomically complex area. Because of their high water content, most chondrosarcomas are of low attenuation on CT. MRI is better for delineating the extent of marrow and soft tissue involvement. The high water content of chondrosarcomas is manifest as very high signal intensity on T2-weighted images [3]. In the long bones, central chondrosarcomas produce a fusiform expansion in the metaphysis or diaphysis (figure 4). The tumor has a mixed radiolucent and sclerotic appearance with the mineralized chondroid matrix appearing as a punctate or ring-andarc pattern of calcifications that may coalesce to form a more radiopaque flocculent pattern of calcification (the so-called chondroid type of calcification (image 6)). Higher grade chondrosarcomas often contain relatively less extensive areas of mineralization (image 7). The cortex is often thickened but a periosteal reaction is scant or absent. There may be features of endosteal scalloping and soft tissue extension. Evidence of a large, soft tissue mass, particularly if unmineralized, that is associated with a lesion whose radiologic features otherwise suggest a chondrosarcoma should raise the level of suspicion for a high-grade tumor (image 7 and image 8). Conventional radiographs are not reliable to distinguish between an enchondroma and central ACT/CS1 [12,37,106,107], although localization in the axial as opposed to the appendicular skeleton and size greater than 5 cm favor chondrosarcoma [37,108]. The presence of a soft tissue mass excludes the diagnosis of an enchondroma. The thickness and staining characteristics of the cartilaginous cap on dynamic contrast-enhanced MRI provides a fairly reliable assessment of the likelihood of malignancy in an osteochondroma [109,110]. However, an absolute distinction between benign and malignant cannot be made on radiologic grounds alone [37,111,112]. Osteochondromas appear as a sessile or broadly-based smoothly calcified lesion at the surface of bone, with the cortex of the bone typically extending into the stalk of the osteochondroma and normal trabeculation centrally. The cartilaginous cap is best assessed on T2-weighted MRI and should not exceed 1.5 to 2 cm. The thickness and appearance of the cap on dynamic contrast-enhanced MRI provide a fairly reliable assessment of the likelihood of malignancy. A thickened cap and irregular distribution of vague calcifications suggest the development of a secondary chondrosarcoma. 130 Periosteal chondrosarcomas appear as a round to oval soft tissue mass on the surface of bone, containing typical chondroid matrix mineralization. They cause variable amounts of cortical bone erosion and appear to be covered by an elevated periosteum (Codman triangles). The medullary canal is typically not involved. The radiographic differentiation between a periosteal chondrosarcoma, periosteal osteosarcoma, and parosteal osteosarcoma can be difficult. Clear cell chondrosarcomas have a predilection for the epiphyseal ends of the femur and humerus (figure 1). Radiographs reveal a well-defined, predominantly lytic lesion, sometimes with a sclerotic rim. Matrix mineralization is not as frequently apparent as with conventional chondrosarcoma. As noted above, aggressive chondrosarcomas, such as the mesenchymal and dedifferentiated subtypes, often contain areas of matrix mineralization that suggest a low-grade chondroid neoplasm; however these areas are relatively less extensive compared to conventional chondrosarcoma and usually ill-defined. On CT and MRI imaging, dedifferentiated chondrosarcomas may be seen to contain two distinct areas with differing radiographic characteristics: the low-grade conventional chondrosarcomatous component has low attenuation on CT and high signal intensity on T2-weighted MRI images, while the high-grade noncartilaginous component may have soft tissue attenuation on CT (isointense to muscle) and variable signal intensity on MRI T2-weighted images. There may be intraosseous lytic areas and an aggressive pattern of bone destruction with a moth-eaten or permeative pattern. These aggressive tumors are often associated with perforation of the cortex and a large soft tissue mass. Imaging features of extraskeletal mesenchymal chondrosarcomas are nonspecific, with chondroid-type calcification and foci of lost signal intensity within enhancing lobules [113]. Role of PET — The imaging methods described above provide limited information as to the biologic activity of a suspected chondrosarcoma. PET scanning with fluorodeoxyglucose (FDG) has been proposed as a noninvasive method to assess tumor grade, to distinguish benign from malignant chondroid lesions, to identify otherwise occult metastatic disease, and differentiate recurrent tumor from postoperative change [114-117]. However, the overall place of PET in the diagnostic and staging evaluation remains uncertain: Although grade 2 and 3 chondrosarcomas have a higher glucose metabolism (and therefore, a higher standardized uptake value [SUV]), PET cannot differentiate between benign cartilage tumors and ACT/CS1 [117]. The value of PET scanning to screen for metastatic or recurrent disease is also uncertain. Biopsy — For suspicious lesions, a diagnostic biopsy is frequently undertaken to establish the diagnosis and plan the surgical approach. The initial percutaneous biopsy may not accurately reflect the true histologic grade of the lesion because of lesion heterogeneity and the possibility of sampling error [118]. If it is undertaken, biopsy should always be directed at the more aggressive-appearing areas as seen on the radiographic studies (ie, the soft tissue components, or the more diffusely enhancing regions with limited or no matrix mineralization). Specific issues surrounding the diagnostic biopsy for suspected primary bone tumors are discussed elsewhere. Staging system — The staging system used for bone sarcomas was developed by Enneking et al at the University of Florida and based upon a retrospective review of cases of primary malignant tumors of bone treated by primary surgical resection (table 2) [119,120]. This system characterizes nonmetastatic malignant bone tumors by grade (low grade [stage I] versus high grade [stage II]), and further subdivides these stages 131 according to the local anatomic extent. The compartmental status is determined by whether the tumor extends through the cortex of the involved bone. Patients with distant metastases are categorized as stage III. The American Joint Committee on Cancer (AJCC) had a similar staging system in its 1997 fifth edition [121], but this was modified in 2002, with additional minor refinements in the latest 2010 edition (table 3) [122]. The TNM classification is not widely used for primary bone tumors. Completing the staging workup — As with other sarcomas, the lungs are the main site of metastatic disease; much less commonly, the regional nodes and liver are involved. Given the low rate of metastases in patients with ACT/CS1 (less than 10 percent), imaging of the lungs is generally not necessary. However, patients with intermediate and high-grade chondrosarcomas have a higher rate of metastatic disease (10 to 50 percent for grade 2 lesions and 50 to 70 percent for grade 3 lesions) [3,5]. In these patients, the staging evaluation should at least include a thoracic CT to rule out the presence of pulmonary metastases. As noted above, the place of PET scanning (which in other oncologic settings is generally more sensitive but less specific than CT for detection of metastatic disease) is uncertain. Although the use of PET can reveal sites of metastatic disease among patients with grade 2 or 3 chondrosarcomas, it is clear that sensitivity is lower than that of conventional CT for small lung metastases [117]. The utility of integrated PET/CT has not been addressed. SURGICAL TREATMENT — For all grades and subtypes of nonmetastatic chondrosarcoma, surgical treatment offers the only chance for cure. The optimal type of surgical management depends upon histologic grade, location, and tumor extent. Intermediate and high-grade tumors — Wide en bloc local excision is the preferred surgical treatment for all nonmetastatic intermediate- and high-grade chondrosarcomas [8]. Depending on the location of the primary, wide excision can lead to considerable morbidity and may require a demanding reconstruction. Low-grade central tumors — The goal of minimizing functional disability provides the rationale for pursuing less extensive surgery for ACT/CS1 that are confined to the bone. In appropriately selected cases, extensive intralesional curettage, followed by local adjuvant chemical treatment (phenolization) or cryotherapy, and cementation or bone grafting of the cavity produces satisfactory long-term local control while minimizing the need for extensive reconstruction [123-128]. The best outcomes are with small extremity conventional ACT/CS1. For patients with large tumor size, intraarticular or soft tissue involvement, and the periosteal, clear cell, mesenchymal and dedifferentiated subtypes [36], intralesional excision represents an inadequate form of local treatment, with high rates of local recurrence [3,129]. Wide local resection is preferred [129]. For large tumors, curettage can be technically difficult, and sampling error may result in a focus of higher-grade disease being missed. The tumor size cutoff beyond which a wide resection is preferred has not been studied and is highly dependent on location. Most authors also consider wide local excision to be the preferred treatment for a lowgrade chondrosarcoma involving the axial skeleton and pelvis. Several (but not all [8,56,130]) reports note higher local recurrence rates with curettage or marginal excision of tumors at these sites, with a higher tendency to metastasize [129-133]. The decision to perform a curettage rather than wide local excision on the basis of a diagnostic biopsy is complicated by tumor heterogeneity and variation in histopathologic interpretation. A failure to recognize higher-grade areas of differentiation in a predominantly low-grade chondrosarcoma is possible when a needle or limited open biopsy has been performed. Thus, a presumed ACT/CS1 treated by curettage and local 132 adjuvant treatment that is found to contain foci of intermediate- or high-grade differentiation on the final histologic sections might require additional surgery. In order to minimize this risk, the diagnostic biopsy should always be directed at the more aggressive-appearing areas on the radiographic studies (ie, the soft tissue components or the more diffusely enhancing regions with limited or no matrix mineralization). Peripheral chondrosarcomas — For patients with a preexisting osteochondroma, complete surgical removal of the cartilage cap with the pseudocapsule provides excellent long-term clinical and local results. In one series of 107 patients with a tumor arising in solitary or multiple osteochondromas, five- and 10-year local recurrence rates after surgery were 16 and 18 percent, and the 10-year mortality rate was only 5 percent [21]. Of the 63 patients who had their primary treatment at the author's institution, 26 had wide excision (none of whom recurred), 36 had a marginal excision (ten of whom recurred locally), and the one patient who had an intralesional excision also developed a local recurrence. Of the 45 patients who received treatment for a local recurrence, 15 died of their disease. When these tumors arise in the pelvis, the large cartilage cap can be difficult to excise, but outcomes are better if the excision is complete [129,134]. As an example, in a series of 61 patients with ACT/CS1 or grade 2 secondary peripheral chondrosarcoma of the pelvis, local recurrence rates after wide local or incomplete excision were 3 versus 23 percent, respectively [134]. Management of recurrent disease — Local recurrence of a ACT/CS1 in the long bones compromises survival, and aggressive management is warranted [135]. If the local recurrence is solitary, without progression in grade and located in the long bones, repeat intralesional resection with local adjuvant therapy is reasonable. Local recurrence of an intermediate- or high-grade chondrosarcoma located in the long bones or recurrence of any grade histology in the flat bones is an indication for a wide excision [130], although it is often challenging to reach adequate wide resection margins in these patients. RADIOTHERAPY — As most chondrosarcomas are slow growing, with a relatively low fraction of dividing cells and radiation-related cytotoxicity is dependent upon cell division, chondrogenic tumors are considered relatively (but not absolutely [136,137]) radioresistant. Nevertheless, radiation therapy (RT) may be of benefit in two situations: after an incomplete resection of a high-grade conventional, dedifferentiated, or mesenchymal chondrosarcoma to maximize the likelihood of local control (potentially curative intent) and in situations where resection is not feasible or would cause unacceptable morbidity (palliative intent). When RT is given with curative intent, doses in excess of 60 Gy are required to achieve local control. However, application of this dose with conventional high energy photons is often impossible in the vicinity of critical structures, especially in chondrosarcomas arising in the skull base and axial skeleton. Unfortunately, it is in this exact situation that postoperative RT is often indicated, as these tumors are less accessible for radical resection as compared to lesions in the appendicular skeleton. The benefits of RT can be illustrated by a series of 21 patients with primary chondrosarcoma of the spine who underwent 28 surgical procedures that included seven complete and 21 subtotal resections [138]. The median survival for the entire group was six years, and the addition of RT to resection prolonged the median disease-free interval from 16 to 44 months. Others have shown a high rate of local control (90 percent) with the addition of neoadjuvant or adjuvant RT to surgical resection in a group of 60 patients with high-risk extracranial chondrosarcoma, of whom 50 percent had either an R1 (microscopically positive) or R2 (grossly positive margins) resection [139]. Palliative RT is also a reasonable option for local treatment of a primary or locally recurrent chondrosarcoma if resection is not feasible or would cause unacceptable 133 morbidity. This is particularly true for mesenchymal chondrosarcomas, which, in our experience, are more radiosensitive than are other subtypes. The benefit of RT in this setting can be illustrated by a retrospective review of 15 patients with mesenchymal chondrosarcoma (all but one nonmetastatic, most extraosseous) treated in several protocols of the German Society of Pediatric Oncology and Hematology [49]. All patients had surgical resection, which was complete in eight; 13 received chemotherapy and six were irradiated. At a median follow-up of 9.6 years, four of seven incompletely resected patients were still alive, three of whom had been irradiated. Conventional RT can sometimes provide local control and symptom relief for other histologies, as long as sufficient doses are administered [140,141]. In an early report from M. D. Anderson of 20 patients with chondrosarcoma who were treated for cure, 5 of 11 patients who received RT as monotherapy achieved local control with doses from 40 to 70 Gy [140]. Charged particle irradiation — Given the limitations of conventional photon irradiation, alternative radiation modalities have been tested. Unlike photons, which lack mass and charge, particle beams interact more densely with tissue, causing greater levels of ionization per unit length, and therefore, an increased radiobiologic effect (RBE). The most data are available for protons, but there is limited experience with carbon ions as well. Proton beam irradiation — The theoretical advantage of charged particle irradiation using protons is in its dose distribution. The physical characteristics of the proton beam result in the majority of the energy being deposited at the end of a linear track, in what is called a Bragg peak. The radiation dose falls rapidly to zero beyond the Bragg peak. Proton beam therapy permits the delivery of high doses of RT to the target volume while limiting the "scatter" dose received by surrounding tissues. Proton beam RT has been studied most for incompletely resected chondrogenic tumors of the skull base and axial skeleton. Local control rates of 78 to 100 percent with mixed photon-proton or proton only protocols (doses up to 79 centigray-equivalents [CGE]) are reported by several authors [142-146], with limited severe late effects (<10 percent RTOG grade 3 toxicity). (See "Chordoma and chondrosarcoma of the skull base".) Carbon ions — Carbon ions represent another attractive radiation modality, which combines the physical advantages of protons with a higher radiobiological activity. The available data are in patients treated for skull based chondrosarcomas, which are discussed elsewhere. Although promising, these techniques are not widely available, in contrast to photon irradiation. Particle beam irradiation requires adaptation of particle accelerators designed for other purposes or specialized dedicated equipment. SYSTEMIC TREATMENT — Chemotherapy is generally considered ineffective in chondrosarcomas, especially for the most frequently observed conventional type and the rare (low grade) clear cell variant (table 1). Chemoresistance in these tumors may be attributable to several factors: Most chondrosarcomas are slow-growing, with a relatively low fraction of dividing cells; most conventional chemotherapeutic agents act on actively dividing cells. Expression of the multidrug-resistance-1 gene, P-glycoprotein, by chondrosarcoma cells may also play a role [147,148]. 134 Access of anticancer agents to the tumor may be hampered because of the large amount of extracellular matrix and the poor vascularity. High activity of anti-apoptotic and pro-survival pathways (eg, high expression of Bcl-2 family members) [149]. The benefit of chemotherapy for higher-grade (grade 2 or 3) chondrosarcomas is difficult to assess. Due to the rarity of chondrosarcomas in general and especially of intermediate and high-grade tumors, there are few prospective studies, and no randomized trials. The reported series are few in number, retrospective, and consist mainly of a small number of patients. There is an urgent need for inclusion of these patients in prospective trials. Adjuvant chemotherapy — While there is no role for adjuvant chemotherapy in patients who undergo surgical treatment for a ACT/CS1, its benefit for dedifferentiated and mesenchymal chondrosarcomas is unclear. Dedifferentiated chondrosarcoma — At least two studies (both retrospective) suggest that patients with dedifferentiated chondrosarcoma who are managed with surgery and chemotherapy may have a better outcome than those managed with surgery alone [150,151]. However, benefit from chemotherapy in patients with nonmetastatic disease is not a universal observation [11,46,47,152]. We suggest that eligible patients consider enrolling in clinical trials testing the value of adjuvant chemotherapy. As an example, patients with dedifferentiated chondrosarcomas are allowed in the EUROBOSS adjuvant chemotherapy trial of the Scandinavian Sarcoma Group [153]. Mesenchymal chondrosarcoma — Limited experience with adjuvant (or neoadjuvant) chemotherapy in patients with mesenchymal chondrosarcoma suggests a potential benefit for chemotherapy: An early study suggested chemotherapy responsiveness in patients with a prominent small round cell content [54]. In a retrospective series of nine patients with mesenchymal chondrosarcoma treated with preoperative chemotherapy, six received a multiagent doxorubicin-containing Rosen T-10 or T-11 osteosarcoma protocol, and there were three complete and three partial responses. The potential benefit of neoadjuvant chemotherapy in high-grade spindle cell sarcomas of bone (other than osteosarcoma or malignant fibrous histiocytoma) was addressed in a prospective study conducted by the European Osteosarcoma Intergroup [152]. Of the 21 patients with potentially resectable localized tumors who underwent attempted resection after three cycles of doxorubicin and cisplatin-based chemotherapy, one of two patients with a mesenchymal chondrosarcoma had a good pathological response in the primary tumor [152]. Experience with 15 cases of mesenchymal chondrosarcoma (all but one nonmetastatic, most extraosseous) was reported from the soft tissue and osteosarcoma study groups of the German Society of Pediatric Oncology and Hematology [49]. All patients had surgical resection, which was complete in eight; 13 received chemotherapy and six were irradiated. The treatment regimens consisted of a variety of chemotherapy drugs given in various combinations in several trials. Response to induction chemotherapy could be assessed in seven patients, four of whom had a 50 percent reduction in tumor volume or ≥50 percent "histologic devitalization". At a median follow-up of 9.6 years, characteristic late recurrences were not observed, and the actuarial event-free and overall survival rates at 10 years were 53 and 67 percent. The authors concluded that these outcomes were better than expected, and attributed the improved outcomes to combined modality treatment. 135 A retrospective series of 26 patients with mesenchymal chondrosarcoma treated at the Rizzoli Institute included 24 surgically treated patients, 12 of whom received chemotherapy [52]. After a median follow-up of 48 months, 10 remained alive. The 10-year rates of disease-free survival were markedly higher among those who received chemotherapy (31 versus 19 percent) as was overall survival (76 versus 17 percent). Questions will remain as to the worth of adjuvant chemotherapy until randomized trials are carried out. However, even in the absence of prospective trials, there is evidence that mesenchymal chondrosarcomas are sensitive to doxorubicin-based combination chemotherapy as used in other bone tumors. The poor prognosis seen with surgery alone provides a strong argument to consider the use of adjuvant chemotherapy in patients who are medically fit and able to tolerate the therapy. Initial induction chemotherapy is a reasonable option if the disease is locally advanced, and a response to therapy might increase the likelihood of function-preserving surgery. The best regimen is unknown, but it seems reasonable to use a doxorubicin/cisplatinbased chemotherapy regimen [152], as is used for osteosarcoma. Advanced disease — Although the majority of patients with recurrent or metastatic sarcoma do not respond to the usual chemotherapy regimens for advanced sarcoma, there have been isolated reports of successful treatment with ifosfamide alone or with doxorubicin, or single agent methotrexate [49,54,154-156]. High-grade chondrosarcomas seem to preferentially benefit, but this is unpredictable. In the small prospective European trial in high-grade spindle cell sarcomas of bone described above, two of six patients treated for an advanced dedifferentiated chondrosarcoma had a complete response to doxorubicin plus cisplatin, while the best response among five patients with metastatic mesenchymal chondrosarcoma was stable disease in two [152]. Novel therapies — The relative lack of efficacy of conventional chemotherapy and the discovery of novel signaling pathways in several histologic subtypes of chondrosarcoma has prompted interest in molecularly-targeted therapies, particularly for chemotherapy refractory nonoperable or metastatic chondrosarcomas [66]. As examples: Receptor tyrosine kinases are commonly activated in chondrosarcomas, the most active of which are Akt1/GSK3beta, the src pathway and PDGFR [157]. Activation and/or overexpression of platelet-derived growth factor receptor-alpha (PDGFRA) and PDGFR-beta (PDGFRB) has been described in conventional chondrosarcomas, although activating mutations have not been found [100,101]. Investigators have shown that SU6668, an inhibitor of several tyrosine kinase receptors, including PDGFRB, represses the growth of chondrosarcoma xenografts [158]. Although the role of these receptors in molecular pathogenesis or malignant transformation remains uncertain, these data suggest a therapeutic potential for inhibitors of these receptor tyrosine kinases. Unfortunately clinical trials have been disappointing to date: A phase II trial of imatinib, a PDGFR/C-KIT tyrosine kinase inhibitor, failed to demonstrate meaningful clinical activity in 26 patients with metastatic nonresectable chondrosarcoma [159]. The tyrosine kinase inhibitor dasatinib was explored in a SARC trial which included a cohort of 32 patients with chondrosarcoma within a larger group of indolent sarcoma subtypes [160]. The chondrosarcoma cohort did not meet its primary endpoint (>50 percent six month PFS) and the drug was considered inactive in this group. 136 Estrogen is important in the regulation of longitudinal skeletal growth. Estrogen receptors and aromatase activity have been identified in chondrosarcomas [161163]. Interference with estrogen signaling may have therapeutic value in this disease. However, in one study, dose-response assays showed no effect of any of the aromatase inhibitors on proliferation of conventional chondrosarcoma in vitro, and the median progression-free survival of patients treated with aromatase inhibitors did not significantly deviate from untreated patients [163]. There are other promising areas of investigation. Antitumor effects of histone deacetylase (HDAC) and angiogenesis inhibitors have been described in chondrosarcoma cell lines and in vivo models [158,164]. Several phase I studies report incidental responses of chondrosarcomas to newer targeted agents, such as vascular endothelial growth factor antisense molecules [165], recombinant human Apo2L/TRAIL [166], and a fully human IgG1 monoclonal antibody that triggers the extrinsic apoptosis pathway through death receptor 5 [167]. In the absence of effective systemic therapy, eligible patients should be encouraged to enroll in phase I or multi tumor-type trials testing novel strategies (www.clinicaltrials.gov). POSTTREATMENT SURVEILLANCE — As with other bone sarcomas, there are no prospective data that address the appropriate schedule or selection of tests for surveillance after initial treatment for localized disease. Consensus-based guidelines from the National Comprehensive Cancer Network (NCCN) recommend physical examination, complete blood count, and chest as well as local imaging every three months for the first two years, every four months during year three, every six months for years four and five, then annually [168]. Routine posttreatment surveillance should be extended to ten years, as late recurrences can occur [7]. Our practice is to perform MRI of the lesion one, two, and five years after initial surgery. SUMMARY AND RECOMMENDATIONS — Chondrosarcomas are a very heterogeneous group of malignant bone tumors that share in common the production of chondroid (cartilaginous) matrix. Clinical behavior is variable and predicted by the histologic grade. Ninety percent are conventional chondrosarcomas, the majority of which are low-grade tumors which are slow growing with a low metastatic potential. High-grade chondrosarcomas, which include 5 to 10 percent of conventional chondrosarcomas as well as the dedifferentiated and mesenchymal subtypes, have a high metastatic potential and a poor prognosis following resection alone. (See 'Introduction' above.) Experienced assessment of radiographic imaging studies (conventional X-ray, MRI, and CT) and histopathologic grade as assessed on a diagnostic biopsy are used to guide treatment decisions. (See 'Diagnostic and staging work-up' above.) Surgical resection — For all grades and subtypes of nonmetastatic chondrosarcoma, surgical treatment offers the only chance for cure. The optimal type of surgical management depends upon histologic grade, location, and tumor extent. The goal of surgery is complete excision while minimizing functional disability. (See 'Surgical treatment' above.) For small, central ACT/CS1 involving an extremity that are confined to the bone, we suggest intralesional curettage with local adjuvant therapy (phenol application or cryosurgery followed by cementation or bone graft of the cavity) rather than wide local excision (Grade 2C). For all other patients with intermediate or high-grade histology (including the mesenchymal and dedifferentiated subtypes), large tumor size, intraarticular or soft tissue involvement, periosteal or clear cell subtypes, a low-grade central 137 chondrosarcoma in the pelvis or axial skeleton, or a peripheral chondrosarcoma, we recommend wide local resection (Grade 1B). Locally recurrent chondrosarcomas should be managed aggressively, as they may be of higher histological grade than the original tumor, increasing the risk of metastases and fatal outcome. For nonmetastatic recurrence of a ACT/CS1, we suggest repeat intralesional resection with local adjuvant therapy if the local recurrence is solitary, without progression in grade and located in the long bones (Grade 2C). Otherwise, wide local excision is preferred. (See 'Management of recurrent disease' above.) Role of radiotherapy — While most low-grade chondrosarcomas are considered relatively radioresistant, radiation therapy (RT) may be of benefit in two situations (see 'Radiotherapy' above): We generally suggest adjuvant radiotherapy for incompletely excised high-grade conventional, dedifferentiated, or mesenchymal chondrosarcomas (Grade 2C). Doses of more than 60 Gy are needed for maximal local control after incomplete resection, and depending on the tumor site, this may not be feasible with conventional photon beam irradiation. In such cases, referral for treatment using newer techniques, such as proton beam irradiation, is appropriate, where available. Palliative RT is a reasonable option for local treatment of patients with a primary or locally recurrent chondrosarcoma if resection is not feasible or would cause unacceptable morbidity, as well as for those with symptomatic metastatic disease. Chemotherapy — Conventional low-grade and clear cell chondrosarcomas are chemotherapy-resistant. Chemotherapy is possibly effective in the mesenchymal subtype, particularly those that contain a high percentage of round cells, and is of uncertain value in dedifferentiated chondrosarcoma; both subtypes are rare and bear a poor prognosis. (See 'Systemic treatment' above.) Adjuvant chemotherapy — There is no role for adjuvant chemotherapy after complete resection of a conventional low-grade chondrosarcoma, and we recommend not pursuing this approach (Grade 1C). The role of chemotherapy after complete resection of a dedifferentiated chondrosarcoma is unknown. Retrospective studies suggest that patients who are managed with surgery and chemotherapy have a better outcome than those managed with surgery alone. We recommend that eligible patients be enrolled on clinical trials testing adjuvant chemotherapy, such as the Euroboss I trial [153]. If the protocol is not available or if patients are unable or unwilling to participate, we manage these patients on a case by case basis. If the high-grade component is an osteosarcoma, we treat the patient as per osteosarcoma protocols. (See 'Adjuvant chemotherapy' above.) For patients with a completely resected mesenchymal chondrosarcoma, retrospective studies suggest that those with a substantial round cell component are sensitive to doxorubicin-based combination chemotherapy. The poor prognosis seen with surgery alone provides a strong argument to discuss the risks and potential benefits of adjuvant doxorubicin-based chemotherapy with these patients as long as they are medically fit and able to tolerate the therapy. Initial induction chemotherapy is a reasonable option if the disease is locally advanced, and a response to therapy might increase the likelihood of complete resection or function-preserving surgery. (See 'Adjuvant chemotherapy' above.) Advanced disease — The benefit of conventional chemotherapy is limited in advanced disease, with the exception of mesenchymal osteosarcomas. We suggest that patients be enrolled in clinical trials testing new strategies. If trial enrollment is not available or not feasible, we use a doxorubicin plus cisplatin combination as is used for other bone sarcomas. 138