2. Select the appropriate diagnostic work-up for patients with newly diagnosed osteosarcoma.

Transcription

1 The Oncologist Pediatric Oncology Biology and Therapeutic Advances for Pediatric Osteosarcoma NEYSSA MARINA, a MARK GEBHARDT, b LISA TEOT, c RICHARD GORLICK d a Department of Pediatrics, Division of Hematology-Oncology, Stanford University Medical Center, Stanford, California, USA; b Department of Orthopedic Surgery, Children s Hospital, Boston, Massachusetts, USA; c Department of Pathology, Children s Hospital of Pittsburgh, Pittsburgh, Pennsylvania, USA; d Department of Pediatrics, Memorial Sloan-Kettering Cancer Center, New York, New York, USA Key Words. Pediatric osteosarcoma Therapy Biology Multidisciplinary treatment Therapeutic advances LEARNING OBJECTIVES After completing this course, the reader will be able to: 1. Discuss etiologic and biologic factors in osteosarcoma. 2. Select the appropriate diagnostic work-up for patients with newly diagnosed osteosarcoma. 3. Select appropriate multidisciplinary treatment for patients with newly diagnosed osteosarcoma. 4. Discuss various surgical approaches to the management of newly diagnosed osteosarcoma. CME Access and take the CME test online and receive 1 hour of AMA PRA category 1 credit at CME.TheOncologist.com ABSTRACT Osteosarcoma is the most common malignant bone tumor in children and adolescents. Survival for these patients was poor with the use of surgery and/or radiotherapy. The introduction of multi-agent chemotherapy dramatically improved the outcome for these patients and the majority of modern series report 3-year diseasefree survival of 60%-70%. This paper describes current strategies for treating patients with osteosarcoma as well as review of the clinical features, radiologic and diagnostic work-up, and pathology. The authors review the state of the art management for patients with osteosarcoma in North America and Europe including the use of limb-salvage procedures and reconstruction as well as discuss the etiologic and biologic factors associated with tumor development. Therapy-related sequelae and future directions in the biology and therapy for these patients are also discussed. The Oncologist 2004;9: INTRODUCTION Osteosarcoma is the most common primary malignant bone tumor in children and adolescents [1]. It is a highly aggressive neoplasm typically composed of spindle cells producing osteoid. The outcome for patients with osteosarcoma was poor before the use of effective chemotherapy, with 2-year overall survival rates of 15%-20% following surgical resection and/or radiotherapy [2-4]. Although osteosarcoma is primarily a disease of adolescents and young adults, a different type of osteosarcoma linked to Paget s disease occurs in older adults. This manuscript, however, focuses exclusively on osteosarcoma in the younger age group. Correspondence: Neyssa Marina, M.D., Stanford University Medical Center, 300 Pasteur Drive, Room G313, Stanford, California , USA. Telephone: ; Fax: ; Neyssa.marina@stanford.edu Received October 15, 2003; accepted for publication March 30, AlphaMed Press /2004/$12.00/0 The Oncologist 2004;9:

2 Marina, Gebhardt, Teot et al. 423 CLINICAL FEATURES Most patients with osteosarcoma present with pain and swelling in the involved region and usually seek medical attention following trauma or vigorous physical exercise, both of which are common in this population [1, 5, 6]. Patients generally have symptoms for several months (average, 3-4 months, but frequently exceeding 6 months) before a diagnosis is made. Although osteosarcoma can occur in any bone, it is most common in the metaphysis of long bones. The most common primary sites are the distal femur, proximal tibia, and proximal humerus [1, 4, 7, 8], with approximately 50% of cases originating around the knee area [1, 4, 7]. However, osteosarcoma can also occur in the axial skeleton (<10% of cases in the pediatric age group), most commonly the pelvis [9, 10]. Approximately 15%-20% of patients present with radiographically detectable metastases [1, 6, 11]. However, since about 80% of patients with localized osteosarcoma develop metastatic disease following surgical resection [2-4, 7], virtually all patients are presumed to have subclinical, microscopic metastases [12]. The most frequent site for metastatic presentation is the lung [13], but respiratory symptoms only develop with extensive involvement. However, metastases can also occur in other bones and soft tissues [11, 14]. Arguably, presentations with multiple bone metastases may actually represent multifocal primary tumors [13]. When osteosarcoma is widely metastatic, more frequently at recurrence than at the time of initial diagnosis, it can spread to the central nervous system [15] or other sites [14]. Death from osteosarcoma is usually the result of progressive pulmonary metastasis with respiratory failure due to widespread disease [1, 5, 6]. The evaluation of a patient with suspected osteosarcoma begins with a full history, physical examination, and plain radiographs. The history is usually remarkable for the presence of pain and swelling at the primary tumor site. The presence of pain at other sites may suggest the presence of metastatic involvement. Physical examination usually reveals a soft tissue mass at the primary tumor site, and laboratory work-up is seldom remarkable except for elevations of alkaline phosphatase and lactate dehydrogenase (LDH), which have been reported to have prognostic significance [12, 16]. RADIOGRAPHIC WORK-UP AND STAGING The radiographic appearance of the tumor usually suggests the diagnosis of osteosarcoma, as it can present as a lytic, sclerotic, or mixed lytic-sclerotic lesion. As with other malignant bone tumors, plain films reveal permeative destruction, with poorly defined zones of transition along with surrounding normal bone, and lack of endosteal response. Ossification in the soft tissue in a radial or sunburst pattern is classic for osteosarcoma but is neither a sensitive nor specific feature. Periosteal new bone formation with lifting of the cortex leads to the appearance of a Codman triangle [1, 5, 6]. Radiological work-up usually includes computed tomography (CT) or magnetic resonance imaging (MRI) of the primary to assess the extent of bone and soft tissue involvement. Defining the local tumor extent with CT or MRI has been shown to be an accurate predictor of tumor extent determined at the time of surgical resection [17, 18] and is, therefore, an essential part of the work-up. MRI is also accurate at assessing the intraosseous extent of tumor and its relationship to muscle groups, subcutaneous fat, joints, and major neurovascular structures [19]. This modality has become standard in the work-up of osteosarcoma patients and, in most institutions, has replaced CT to define local tumor extent. The differential diagnosis for most bone lesions includes infection as well as other tumors, such as aneurysmal bone cyst, Ewing s sarcoma, and chondrosarcoma. The location of the tumor within the bone (i.e., metaphysical) and the skeletal location help to distinguish osteosarcoma from Ewing s sarcoma, the second most frequent type of bone tumor in this age group. Metastatic bone disease from other primary tumors, such as lymphoma, neuroblastoma, and rhabdomyosarcoma, although not frequent in this age group, is also part of the differential diagnosis. In addition to imaging the primary tumor, various other radiological studies help determine the extent of disease at presentation. These include a radionuclide bone scan with methylene diphosphonate labeled with technetium-99m, which helps define the extent of the primary tumor [20]. Radionuclide bone scanning is also useful for the detection of skip lesions within the same bone and distant bone metastases [21]. Postero-anterior and lateral radiographs of the chest allow detection of lung metastases in the majority of cases. However, chest CT is more sensitive in detecting pulmonary metastases and has become the imaging procedure of choice [22], especially since up to 30% of patients with metastases are cured with chemotherapy and multiple surgical resections [23-25]. However, there are false-positive results, particularly with smaller lesions, and biopsy confirmation of lung disease is usually required. The diagnosis of osteosarcoma can be predicted by radiographic appearance and location in about two-thirds of cases. However, a diagnosis should never be made from radiographs [1, 5, 6], and a biopsy for pathologic confirmation is mandatory. Details regarding biopsies are discussed in a separate surgical section. Pathologically, osteosarcoma is a pleomorphic tumor producing extracellular osteoid. Details regarding histology are also discussed in a subsequent section. Rarely, the diagnosis of osteosarcoma is difficult because the sample is not representative of the entire tumor. Repeat biopsy may be necessary in those cases.

3 424 Advances in Pediatric Osteosarcoma The most widely used staging system is the one developed by Enneking and associates based on a retrospective review of cases of primary malignant bone tumors treated by primary resection [26-28]. This system categorizes localized malignant bone tumors by grade (low grade: stage I; high grade: stage II) and by the local anatomic extent (A: intracompartmental; B: extracompartmental). The compartmental status is determined by whether or not the tumor extends through the cortex. Patients with distant metastases are stage III. There are very few high-grade intramedullary lesions (i.e., stage IIA), because most high-grade osteosarcomas break through the cortex early in their natural history. In the younger age groups, the vast majority of osteosarcomas are high-grade lesions; hence, virtually all patients are stage IIB or III distinguished by the presence or absence of detectable metastatic disease. EPIDEMIOLOGY In general, bone tumors in children are rare, with an estimated 8.7 per million in children younger than 20 years of age [29], representing new bone cancer patients a year. The two most common tumors are osteosarcoma (400 cases per year) and Ewing s sarcoma (200 cases per year). Osteosarcoma has a bimodal age distribution, with a first peak during the second decade of life (during the adolescent growth spurt; modal age: 16 years in girls and 18 years in boys) and a second peak in older adults. Boys are reported to be affected more frequently in most series [7, 8, 29], and the incidence of osteosarcoma in African-American children is slightly higher than it is in Caucasians. It is extremely rare before 5 years of age. PATHOGENESIS We have a limited understanding of the etiology of osteosarcoma. The peak incidence coincides with a period of rapid bone growth, suggesting a correlation between rapid bone growth and the evolution of osteosarcoma. Other evidence supporting this relationship includes the higher incidence of osteosarcoma in large dog breeds as compared with small breeds [30], and the earlier peak age in girls as compared with boys, corresponding to the earlier age of their growth spurt [31]. However, although initial evidence suggested a higher incidence in tall people than in short people, this has not been confirmed. Additionally, osteosarcoma arises in many patients before and after the adolescent growth spurt. Radiation exposure is a well-documented risk factor for osteosarcoma [32-36], but since the interval between radiation exposure and tumor appearance is long, this is not relevant to most de novo osteosarcoma patients. Genetic alterations in sarcomas generally fall into two categories: A) reciprocal translocations with balanced karyotypes and alterations of tumor suppressor gene pathways, and B) complex unbalanced karyotypes with alterations of the p53 and retinoblastoma (Rb) pathways [37]. Despite the complexity of the karyotype and the absence of characteristic reciprocal translocations, numerous, nonrandom chromosomal abnormalities are observed in osteosarcoma. Numerous recent studies describe cytogenetic abnormalities in osteosarcoma, and the majority of samples are clonally abnormal [38, 39], with heterogeneity within the same patient. The majority of osteosarcoma samples (58%) have marker chromosomes (structurally abnormal chromosomes in which no part is identified). Ring chromosomes (7%) accompanied by multiple numerical (65%) and structural (72%) abnormalities are also frequently observed [38-40]. Additionally, there is evidence of genomic amplification (homogeneously staining regions or double minutes) in over one-third of cases. The observed cytogenetic abnormalities are both numerical and structural. Common numerical abnormalities include: gains of chromosome 1, losses of chromosomes 9, 10, 13, and/or 17, and partial or complete losses of the long arm of chromosome 6 [38, 39, 41-44]. Frequent structural abnormalities include rearrangements of chromosomes 11, 19, and 20 [38, 39, 41-45]. The p53 and Rb tumor suppressor pathways are clearly involved in the pathogenesis of osteosarcoma. Most tumor samples have some type of combined inactivation of the Rb and p53 tumor suppressor pathways [46], and about 3% of patients with sporadic osteosarcoma harbor germline mutations in p53 [47]. In studies of osteosarcoma, a number of loci were demonstrated to have loss of heterozygosity (3q, 13q, 17p, 18q), including the locations of the Rb and p53 tumor suppressor genes [48-50]. Retinoblastoma survivors have an increased incidence of second malignancies, the majority of which are osteosarcomas [51-54]. In the hereditary form of retinoblastoma, germline mutations of the Rb gene are common, likely forming the basis for the greater frequency of secondary cancers, since the rate of second malignancies in survivors of unilateral sporadic retinoblastoma is much less. Germline mutations in the p53 gene can also lead to a high risk of malignant tumor formation, including osteosarcoma (Li- Fraumeni syndrome) [55]. The p53 gene product in normal cells increases in response to DNA damage and directs the cell to either stop progression through the cell cycle or undergo apoptosis [56, 57]. The Rb gene product likewise regulates cell cycle progression [58-60]. Although germline mutations of either the p53 or the Rb gene are rare, these genes are altered in the majority of osteosarcoma tumor samples [61, 62]. Because these tumors almost universally have genetic alterations that inactivate the Rb and p53 tumor suppressor pathways, gene inactivation by itself may not be

4 Marina, Gebhardt, Teot et al. 425 a strong prognostic factor. Indeed, new data show low prognostic significance of p53 mutations in sporadic osteosarcoma. In a recent study [63], 22% of osteosarcoma samples showed p53 mutations, but there was no relationship to distant recurrence. There are other altered oncogenes in osteosarcoma tumor cells. These include amplifications of the product of the murine double minute 2 (MDM2) gene, amplification of cyclin dependent kinase 4 (CDK4), and overexpression of human epidermal growth factor receptor 2 [1, 5, 6, 64-66]. Amplification of the 12q13 region (containing MDM2 and CDK4) or INK4A deletion can affect both the p53 and Rb pathways, and indeed, these alterations seldom coexist with Rb or p53 alterations [67]. Although it is clear that alterations in tumor suppressor genes and oncogenes are necessary to produce osteosarcomas, it is not clear which of these events occurs first. The complex unbalanced karyotypes that characterize osteosarcoma may reflect its pathogenesis. Karyotypic complexity may reflect chromosomal fusion-bridge-breakage cycles resulting from advanced telomere erosion. A potential etiology of this chromosomal instability is telomere dysfunction (implicated in epithelial cancers) [68]. Telomeres are nucleoprotein structures that cap chromosome ends and serve at least three protective functions: preventing recognition of chromosomes as damaged DNA, preventing chromosomal end-to-end fusions and recombinations, and accommodating the loss of DNA that occurs with each round of replication. Normal human somatic cells have a finite proliferative capacity, and telomere length is one of the checkpoints that determines when a cell stops dividing [69, 70]. As cells divide, the telomere length gradually decreases to a critical size, at which point senescence is triggered by a p53-dependent process. Human cells may bypass this checkpoint by inactivating the p53 pathways (divide until telomeres become very short, chromosomal instability ensues, and apoptosis is triggered). Rare cells bypass this second checkpoint by activating mechanisms that lengthen telomeres. About 85% of cancers activate an enzyme called telomerase, which lengthens telomeres, and the other 15% of cancers use a recombination-based method called alternative telomere lengthening [71, 72]. At least 50% of osteosarcomas are dependent upon the alternative telomere-lengthening mechanism to maintain telomeres [73-75]. Alternative telomere-lengthening cell lines have greater genetic instability and more translocations than telomerase-positive cell lines [73]. In mouse models, they were unable to generate macroscopic lung metastases, despite robust subcutaneous tumor growth [76]. Upon telomerase reconstitution, they formed massive pulmonary nodules after tail-vein injection. It is hypothesized that alternative telomere lengthening-dependent human osteosarcomas have different clinical behaviors from telomerase-dependent osteosarcomas. Other authors have suggested that osteosarcomas have a viral etiology, based on the fact that bone sarcomas can be induced in selected animals by viruses [77, 78]. For example, hamsters injected with cell-free extracts of human osteosarcoma develop osteosarcomas [79], and some human osteosarcomas contain simian virus 40 (SV40)-like sequences. SV40, a DNA polyomavirus, consists of three major nonstructural proteins: the small T antigen (enhances the ability of large T antigen to transform cells), the smaller T antigen (function unknown), and the large T antigen (assists in viral replication, interacts with p53 and Rb, and promotes cell proliferation by inhibiting p53 function) [80]. A 1996 study showed that 11 of 18 osteosarcoma samples had evidence of incorporated SV40 DNA [80]; a 1998 study showed no correlation between the presence of SV40 and p53 or Rb mutation status [81]; and a 1997 study showed that 50% of osteosarcoma samples had incorporated SV40 DNA from each of the four regions of the viral genome [82]. However, there is no convincing data that viruses are a major etiologic factor in osteosarcoma. HISTOMORPHOLOGY Osteosarcoma has a broad spectrum of histologic appearances, which have in common the proliferation of malignant mesenchymal cells and the production of osteoid and/or bone by tumor cells. The amount of osteoid and/or bone production varies greatly both among tumors and within an individual tumor and, thus, identification of diagnostic osteoid may require extensive sampling. Chondroid and fibrous matrix may also be present, reflecting the mesenchymal origin of the malignant cells and their consequent ability to differentiate into various cell types. Conventional osteosarcoma is a primary intramedullary high-grade sarcoma and comprises the vast majority of osteosarcoma in children and adolescents. The current World Health Organization (WHO) classification of osteosarcoma recognizes three major subtypes of conventional osteosarcoma: osteoblastic, chondroblastic, and fibroblastic, reflecting the predominant type of matrix in the tumor [44]. Osteoblastic osteosarcoma, which has osteoid or bone as the predominant type of matrix (Fig. 1A and 1B), is composed of malignant plasmacytoid to epithelioid osteoblasts, with variable numbers of smaller round to ovoid cells, spindle cells, and anaplastic mono- or multinucleated giant cells. The appearance of the matrix may vary from dense sheets of osteoid and/or woven bone to interlacing trabeculae to delicate, arborizing wisps of osteoid. Chondroblastic osteosarcoma has a predominance of chondroid matrix, which usually resembles hyaline cartilage, with obviously malignant cells within the lacunae, whereas fibroblastic osteosarcoma is composed of malignant spindle cells with only scant osteoid. The presence of osteoid, even

5 426 Advances in Pediatric Osteosarcoma floating in the blood within the cyst-like spaces and, in small core biopsies, may be the only clue to the correct diagnosis. Small cell osteosarcoma [84, 86] is composed of malignant, small, round or, less often, spindled cells, which may mimic those of Ewing s sarcoma/peripheral neuroectodermal tumor (PNET). In biopsies in which osteoid is extremely scant or absent, immunohistochemical stains and/or genetic analyses may be necessary to distinguish it from these tumors. Although small cell osteosarcoma may be immunoreactive for MIC-2, which is typically seen in Ewing s sarcoma/ PNET, it lacks the t(11;22)(q24;q12) translocation observed in that entity [87, 88]. Low-grade intramedullary osteosarcoma and surface osteosarcomas, which include high-grade surface, periosteal, and parosteal types, are exceedingly rare in children and adolescents and are not included in this review. Figure 1. A) 20 magnification of osteosarcoma producing osteoid. B) 40 magnification of osteosarcoma. if minimal, distinguishes the chondroblastic and fibroblastic subtypes of osteosarcoma from chondrosarcoma and fibrosarcoma/malignant fibrous histiocytoma, respectively, with which they may be confused histologically. The value of these subtypes is unclear, as their recognition is very dependent on sampling, with mixed histologies frequently present in a given tumor. Moreover, there is no convincing evidence of a difference in clinical behavior or outcome based on histologic subtype. In addition to conventional osteosarcoma and its subtypes, the WHO classification recognizes two additional histologic variants, telangiectatic and small cell osteosarcoma [83, 84]. Telangiectatic osteosarcoma [83, 85] typically appears as a purely lytic lesion on plain films and is characterized histologically by cyst-like spaces often containing blood and separated by fibrous septae lined by bland osteoclast-like giant cells, thus simulating aneurysmal bone cyst. Within the fibrous septae and at the periphery of the lesion are variable numbers of highly pleomorphic malignant cells and variable, but typically scant, amounts of delicate osteoid. Small numbers of single malignant cells may also be found TREATMENT Successful treatment of osteosarcoma requires the use of systemic chemotherapy. Early results following treatment with either surgery and/or radiation therapy provided 2-year overall survival rates of 15%-20% [2-4, 7]. Almost all patients have microscopic metastases at the time of diagnosis, as evidenced by the fact that 80%-90% develop metastatic recurrence if treated with surgical resection and/or radiotherapy [3, 4, 7, 89]. Two different studies definitively proved the need for adjuvant chemotherapy to improve outcome for patients with localized extremity osteosarcoma [12, 90]. The most active agents include cisplatin (Platinol ; Bristol-Myers Squibb; Princeton, NJ) [91-93], doxorubicin (Adriamycin ; Pharmacia & Upjohn; Kalamazoo, MI) [94, 95], and high-dose methotrexate [96-98], and the management of these patients involves the use of these three agents along with surgical resection with adequate margins [12, 99, 100]. The best method of local control involves surgery with adequate margins, since this tumor is relatively radioresistant. However, a recent study suggests that patients with microscopically positive margins following resection or those unable to undergo surgical resection may benefit from the use of high-dose radiotherapy, as evidenced by a superior outcome in that series for patients given radiotherapy compared with patients who did not receive radiotherapy (p = ) [101]. Early, nonrandomized trials suggested that systemic chemotherapy produced better outcomes in osteosarcoma patients compared with historical controls [ ]. However, not all investigators were convinced that the better outcome resulted from the use of chemotherapy. At that time, most trials were limited to patients without clinically detectable metastases, and the superior outcome could have been the result of the selection of a cohort of patients with better outcomes. Additionally, it could also be explained by earlier diagnosis resulting from the routine use of CT to

6 Marina, Gebhardt, Teot et al. 427 assess for pulmonary metastasis or improvements in surgical techniques [ ]. In the early 1980s, investigators at the Mayo Clinic carried out the first randomized trial of adjuvant chemotherapy for osteosarcoma [90, 110]. In that study, following surgical resection, patients were randomly assigned to either observation or chemotherapy. There was no difference in outcome between the two groups, and the disease-free survival rate was 40%, suggesting that the natural history of the disease had changed and that this accounted for the difference in outcomes observed in the adjuvant chemotherapy trials [90, ]. That particular study raised substantial controversy, as it suggested that historical controls were not valid and that randomized trials were essential [108, 109]. However, other investigators vehemently resisted this idea and argued that historical controls were appropriate and that it was unethical to conduct a randomized trial that included observation following surgery [109, 111, 112]. Two subsequent randomized studies clarified this controversy [12, 90]. Link and coinvestigators developed a randomized study of observation and adjuvant chemotherapy. Patients treated with surgery alone had a 2-year relapse-free survival (RFS) probability of 17%, versus 66% for those receiving adjuvant chemotherapy. With longer follow-up, the 6-year RFS rate for the observation group was 11%, while for those receiving adjuvant therapy it remained at 66% [113]. An overall survival advantage with adjuvant chemotherapy also became apparent in accordance with the RFS rate [114]. Eilber et al. reported similar results, definitively proving that adjuvant chemotherapy produced higher disease-free survival rates for patients with nonmetastatic osteosarcoma [90]. Rosen et al. introduced the concept of chemotherapy administration prior to definitive surgery [115, 116]. This approach offered the opportunity to develop a custom endoprosthesis for limb-salvage procedures and offered the theoretical advantage of early treatment of micrometastases while facilitating the surgical procedure. It also provided the opportunity to examine the histologic response of the tumor to preoperative therapy and assess its effectiveness. A strong correlation between the degree of necrosis (Huvos grade) and subsequent disease-free survival (DFS) was observed [115], which has been confirmed in a number of subsequent clinical trials [ ]. A theoretical concern with this approach is that the delay in removal of the bulk tumor could lead to the emergence of chemotherapy resistance. However, a prospective Pediatric Oncology Group trial demonstrated no difference between treatment using immediate definitive surgery and treatment with neoadjuvant chemotherapy followed by definitive surgery [120]. Given the advantages in facilitating limb-salvage procedures and assessing chemotherapy response, the use of preoperative chemotherapy has become the standard approach to treatment. The identification of the prognostic value of the degree of necrosis following chemotherapy led to the suggestion that chemotherapy be modified for patients with less necrosis (currently referred to as either standard or poor responders, and variably defined as <90% through <98% tumor necrosis or the persistence of more than rare viable tumor cells or clumps) in an attempt to increase the probability of DFS. Investigators at Memorial Sloan-Kettering Cancer Center, using the T-10 protocol, reported an improved outcome for patients with poor histologic responses following a change in postoperative therapy [115]. Longer follow-up of that patient population, however, showed no benefit to therapy intensification [121, 122]. Numerous other investigators have undertaken studies using a similar strategy, delivering a variety of intensified regimens to patients with standard responses in an attempt to improve their outcomes. However, the majority of these studies have not been able to reproduce the initial results reported by Rosen et al. [117, 119, 121]. Intensification of therapy during preoperative treatment to increase the number of patients with good responses (favorable responders) likewise did not change the long-term outcomes of these patients [123] and, when preoperative therapy is lengthened, histologic response loses its prognostic value [123]. The specific roles of various chemotherapeutic agents in the treatment of osteosarcoma have been the subject of many studies. For example, the role of high-dose methotrexate remains controversial [124], with a few randomized studies reporting it not to be an important component of therapy [125, 126], while others reported that it was [127]. Unfortunately, the European study [126] was compromised by the study design, and the overall outcome was markedly inferior to that of other contemporary studies. However, in spite of these pitfalls, the standard chemotherapy for the European Osteosarcoma Intergroup (EOI) has continued to be the twodrug combination of cisplatin and doxorubicin [128, 129], since there was no survival advantage to the use of more complex regimens observed in their studies. Additionally, although the use of bleomycin, cyclophosphamide, and actinomycin D was common in osteosarcoma, subsequent studies have demonstrated the combination to be ineffective [130], and these drugs are no longer included in the treatment of osteosarcoma. Intra-arterial administration of chemotherapy offers the theoretical advantage of maximizing drug delivery to the tumor vasculature [118, 131, 132], and pharmacokinetic studies demonstrate high local drug concentrations with dramatic clinical responses [133]. Although theoretically appealing, and effective in inducing responses [132], the use of this approach in the context of multiagent chemotherapy does

7 428 Advances in Pediatric Osteosarcoma not appear to offer a significant advantage over systemic chemotherapy [134]. Ifosfamide (Ifex ; Bristol-Myers Squibb) has, relatively recently, been shown to have activity in osteosarcoma [ ] and, when incorporated either alone or in combination with etoposide into the treatment of patients with metastatic disease, the results appear promising [ ]. The last national North American randomized study (INT-0133) was designed to address whether the addition of ifosfamide and/or muramyl tripeptide-phosphatidyl ethanolamine (MTP-PE) to the three other agents used in the standard treatment of osteosarcoma (doxorubicin, cisplatin, high-dose methotrexate) could improve DFS [99]. MTP, a component of the bacillus Calmette-Guerin cell wall, is conjugated to PE and encapsulated in liposomes to improve delivery to the reticuloendothelial system [141, 142]. The rationale supporting the use of this immune adjuvant was the encouraging results obtained in a prospective randomized trial of this compound in canines [143], as well as its apparent efficacy in relapsed patients [141, 144, 145]. Preliminary results of the INT-0133 trial did not demonstrate a survival advantage for patients treated with either ifosfamide or MTP-PE alone [99]. However, there appeared to be an interaction between ifosfamide and MTP- PE, and further investigations, which attempt to exploit this interaction, are ongoing. Parallel to the North American developments in osteosarcoma, the EOI conducted a series of studies based on six cycles of the two-drug regimen of cisplatin and doxorubicin [126, 128]. The German-Austrian-Swiss Cooperative Osteosarcoma Study Group (COSS) also performed a series of studies incorporating multiagent chemotherapy and surgical resection. The best results for this group resulted from the use of methotrexate, cisplatin, doxorubicin and ifosfamide, with a 10-year survival rate of 71% [146]. The Scandinavian Sarcoma Group (SSG) has also performed various nonrandomized neoadjuvant chemotherapy trials for high-grade osteosarcoma. Their second osteosarcoma trial, using a three-drug combination of high-dose methotrexate, doxorubicin, and cisplatin up front and replacement with ifosfamide and etoposide for poor responders, resulted in a 5-year overall survival rate of 74%. Although the ifosfamide/etoposide combination failed to improve outcome, this drug pair replaced the standard agents postoperatively, making it difficult to determine whether the addition of this combination improved outcome. The event-free survival time for patients treated by COSS investigators was superior when ifosfamide was incorporated into the standard three-drug regimen, and a previous nonrandomized Italian trial reported that the addition of ifosfamide and etoposide to standard chemotherapy for patients with poor histologic responses resulted in an outcome for those patients that was similar to that reported for patients with good histologic responses [127]. In addition, although the INT-0133 trial concluded that the addition of ifosfamide did not improve outcome, this drug was administered at a lower dose than the one administered to patients with metastatic osteosarcoma [138, 140], and studies in those patients suggested the presence of a dosedependent effect. Taken together, these findings suggest that the combination of ifosfamide and etoposide has significant activity and might improve the outcome for patients with poor histologic responses. Although a few studies have evaluated the role of altering postoperative therapy in poor histologic responders, the role of high-dose ifosfamide and etoposide in this setting has not been investigated in a large controlled trial. The North American Children s Oncology Group (COG) has recently completed a series of three pilot studies using a backbone of cisplatin, doxorubicin, and high-dose methotrexate. The purpose of these pilots was to develop a chemotherapy regimen that could subsequently be tested in a randomized study. The pilots evaluated three different strategies. Pilot 1 was based on the premise that doxorubicin is an essential component of osteosarcoma therapy [147], and its use has been limited by the potential for cardiotoxicity. This complication appears to be at least partially ameliorated with dexrazoxane (Zinecard ; Pfizer Pharmaceuticals; New York, NY) [148]. Hence, pilot 1 evaluated the feasibility of increasing doxorubicin dose intensity by administering dexrazoxane. Pilot 2 evaluated the feasibility of combining standard-dose ifosfamide with dose-intensive doxorubicin with dexrazoxane and pilot 3 evaluated the feasibility of increasing the dose intensity of ifosfamide and etoposide. It appears that we have reached the limit in the survival of osteosarcoma patients achievable with currently available chemotherapy. Since further improvements in outcome will depend on refinements of therapy, the impact of which will be assessable only in large patient groups, four major research groups in osteosarcoma, the COG, the COSS, the EOI, and the SSG, have agreed on trying to conduct an intergroup randomized study. The power of such a collaboration lies in the ability to conduct large trials with rapid accrual, allowing investigation of new agents quickly and effectively. Acknowledging the difficulties that face the establishment of such a collaboration and recognizing that there are no available new agents, the group has agreed on a relatively simple randomized study to determine whether ifosfamide and etoposide improve the outcome for patients with poor histologic responses. Patients with good histologic responses have a 3-year event-free survival rate of 75%, and the use of ifosfamide and

8 Marina, Gebhardt, Teot et al. 429 etoposide results in an increased risk of late sequelae. In these patients, the group proposes to determine, in a randomized comparison, whether interferon-α improves event-free survival. The rationale for using interferon-α is to maintain remission in a significant proportion of patients who have previously had good responses to chemotherapy. The in vitro effects of interferon-α on osteosarcoma cells were demonstrated more than 20 years ago, and observations since have consistently supported its growth inhibitory effect on osteosarcoma both in cell lines and in animal models [ ]. Although interferon-α has not been widely tested in clinical trials in osteosarcoma, its role as maintenance in other tumors has been extensively studied [152, 153]. Most information on patients with osteosarcoma comes from a Scandinavian series in which 64 patients received interferon-α as a single adjuvant to surgery, and 69% remained in complete remission [154]. A pegylated preparation of interferon-α, with an extended halflife, offers the advantages of less frequent administration and improved dose delivery [155]. The tolerability of this preparation has now been demonstrated, and there is additional extensive data on the tolerability of interferon-α in children treated for chronic hepatitis [156, 157]. Although adjuvant chemotherapy is effective in the setting of localized osteosarcoma, the outcome for patients with clinically detectable metastases at diagnosis continues to be suboptimal [23-25, 158]. The standard management of these patients follows the same principles as the management of those patients who present with localized disease and, with this approach, a small subset of patients achieves prolonged disease-free survival [23-25, 158]. The treatment of patients who develop recurrent osteosarcoma depends on the initial therapy, time to recurrence, and the site and number of recurrent tumors. With aggressive treatment, as many as 40% of patients who develop lung metastases survive more than 5 years after relapse [ ]. Patients who relapse following the use of modern treatment approaches, including chemotherapy and surgery, have a significantly lower probability of survival. THERAPY-RELATED SEQUELAE Unfortunately, the use of multiagent treatment for osteosarcoma is associated with acute and long-term toxicities. These include the potential for hearing loss [91, 163] and hypomagnesemia [164] associated with the administration of cisplatin. Therefore, it is essential to obtain baseline audiograms prior to initiation of treatment to monitor for hearing loss. It is also important to monitor electrolytes secondary to the potential for abnormalities even years after treatment completion [165]. Other treatment-related complications include anthracycline-induced cardiomyopathy [ ], which is typically observed with high cumulative doses [170]. To monitor for this complication, patients usually have baseline cardiac evaluations with an echocardiogram or radionuclide scan. Cardiac function is usually followed closely during treatment. Since doxorubicin appears to be an important component of therapy [147], methods to minimize the potential for this complication are under evaluation. These include the use of dexrazoxane [148], continuous-infusion doxorubicin, and pegylated liposomal doxorubicin [171]. Both pegylated liposomal doxorubicin and dexrazoxane appear effective at minimizing acute cardiac toxicity, but there is limited information regarding their long-term efficacy. In addition, postpubertal males should be given the opportunity to carry out sperm banking [1, 5, 6], since chemotherapy for osteosarcoma has the potential to produce sterility [172]. Although newer techniques for maintaining fertility in women are under development, their indications are not well established. SURGERY Biopsy The initial surgical procedure for any patient with a bone tumor is usually a biopsy. There are two types: incisional and excisional; but, in general, when cancer is suspected, excisional biopsies are rarely performed. Incisional biopsies can be needle (closed) or open, and needle biopsies can be either fine or core. The type of biopsy performed must be carefully determined by evaluating the size and location of the tumor, the differential diagnosis, and the age of the patient. The placement of the biopsy site relative to the location of the tumor and the anatomic structures is of critical importance. Primary excision of expendable bones (rib, clavicle, sternum, ilium, scapular body, distal ulna) at diagnosis should only be considered by an experienced musculoskeletal oncologist, but in virtually all cases, patients receive neoadjuvant chemotherapy prior to definite resection [173, 174]. Most patients with bone tumors undergo diagnostic biopsies using an incisional approach, and the biopsy site is determined by assessment of the extent of local disease and the relationship of the tumor to critical structures (i.e., neurovascular bundle). It is strongly recommended that the biopsy be performed by the surgeon who will be performing the definitive resection so that the biopsy tract can be excised en bloc with the planned surgical resection [175]. Although needle (closed) biopsies can potentially expedite the diagnostic process, especially when performed as an outpatient in the doctor s office (using a local anesthetic), they are not recommended in children. Most malignant bone tumors have a soft tissue component, which is also the most representative tissue and, thus, deep deployment of the

9 430 Advances in Pediatric Osteosarcoma needle within the tumor is unnecessary. When a well-trained cytopathologist is available, fine-needle biopsy is definitely an option. This procedure is performed using a 0.7-mm diameter needle and has up to a 90% diagnostic accuracy [176], exceeding 80% in bone sarcomas [177]. The drawback to this technique is that it may not be possible to obtain sufficient material to perform cytogenetics, flow cytometry, gene profiling, and other helpful diagnostic tests. While needle biopsies can facilitate diagnosis, they can also lead to diagnostic delays (if the specimen is nondiagnostic) in up to 33% of cases, even at experienced centers [178]. In contrast, core biopsies are minimally invasive, can be performed under local anesthesia, maintain the architecture of the tissue, and provide an adequate specimen even for ancillary studies. This procedure has a diagnostic accuracy of over 95% [179]. Generally, open incisional biopsies are performed in the operating room employing longitudinal incisions, since the use of transverse incisions can contaminate flap planes and compromise subsequent resection. The area where the tumor is most superficial is preferable unless other factors, such as an overlying vessel or nerve, preclude it. Preoperative diagnostic imaging may help guide the biopsy. Deep sampling of the tumor is again not necessary and, once the tumor is reached, the biopsy should involve the periphery. Obtaining a frozen section is important in deciding if diagnostic tissue has been obtained. There should also be careful communication between the surgeon and pathologist preoperatively to clarify the amount of tissue necessary for special studies as well as the optimal way of processing the sample. Surgical Treatment of Malignant Bone Tumors The surgical management of malignant bone tumors has evolved into a complex field. Surgical approaches are a function of tumor type, location, and extent of disease. The two most common malignant bone tumors in children are osteosarcoma and the Ewing s sarcoma family of tumors, the management of which requires a multidisciplinary team approach. Surgical extirpation of the primary tumor generally occurs after treatment with neoadjuvant chemotherapy, which is followed by postoperative chemotherapy. The goal of any malignant tumor operation is to perform a complete, en bloc removal of the lesion with adequate margins. The use of neoadjuvant chemotherapy along with advances in imaging techniques has enabled the oncologic surgeon to obtain local control rates equivalent to amputation using limb-salvage surgery. Therefore, limb-salvage has become the standard of care, except in situations where it may compromise oncologic outcome. Because of the complexity of the musculoskeletal system, different reconstructive options are available depending on the site of involvement. Types of Reconstruction The main reconstructive options include autogenous bone grafts, structural bone allografts (intercalary or osteoarticular), and metallic endoprosthetics. All three have inherent advantages and disadvantages, and the choice of procedure depends on the location of the tumor and the age of the patient. In general, however, large structural allografts and endoprosthetics should be reserved for children older than 8 years of age. Nonvascularized autografts from the pelvis or other sites may be used in a limited fashion for relatively small defects and work well [180]. The advantage is a high incorporation rate, but with potential donor site morbidity and limited supply. Vascularized autografts such as those taken from the fibula are attractive because, when successful, the graft incorporates and may even remodel secondary to the forces exerted across it. Again, donor site complications can occur. Structural allografts have no donor site morbidity. The major drawback is the difficulty of the graft incorporating with the host bone (nonunion) and fracture. Their advantage is that they are a biologic solution and, if they heal and do not fracture, may last the lifetime of the patient. They also spare the adjacent epiphysis and growth plate and preserve bone stock for future reconstructions. Osteoarticular allografts may be used in the reconstruction of the proximal humerus, distal femur, and proximal tibia as well as potentially any joint [ ]. Diaphyseal tumors can be reconstructed with intercalary allografts [ ] and, in these instances, the physis may be spared allowing preservation of the epiphysis and the joint surface. Infection can occur in 10%-15% of allograft reconstructions [180, 185, 186, ], and nonunion at the osteosynthesis can occur in another 10%-25% [186, 187]. Infection often requires graft removal, whereas nonunions are managed by revision fixation and autogenous bone grafting. Both these complications are more likely in patients receiving chemotherapy. Augmentation of the allograft with a vascularized fibula, while adding its own set of potential complications, may facilitate osseous integration of the structural graft and prevent some of the inherent problems associated with allografts [194, 195]. The use of osteoarticular allografts results in satisfactory function in 60%-70% of cases following chemotherapy in high-grade sarcomas [181, 182, 185, 192, 196, 197]. Function is generally better with intercalary reconstructions [180, 189]. Although patients are discouraged from participation in sports and impact activities, compliance with this policy is not uniform, and fractures occur in about 20% of cases [181]. Fractures may require the use of grafts or implant removal and replacement. Vascularized fibular augmentation can be utilized as well [194, 195].

10 Marina, Gebhardt, Teot et al. 431 Metallic endoprosthetics provide an immediate stable reconstruction. Their major drawback is that they eventually fail due to loosening or failure of components. Modular components are now available for both pediatric and adult patients, making delay for customization the exception. Infections are a significant risk with endoprostheses, with rates ranging from 0%-35% [ ]. The durability of endoprostheses is influenced by many factors, but the anticipated event-free 5-year survival rate for proximal femur reconstructions approaches 90%; it is about 50% for the distal femur and just over 50% for the proximal tibia [204]. Prosthetic reconstructions of the proximal humerus tend to be more durable since they are not subjected to weight-bearing stresses. Failure can result from loosening at the prosthesis-host interface or from infection. Sometimes failure can be catastrophic, requiring delayed amputation. The development of newer, expandable prostheses has allowed limb-sparing procedures in younger children. The 5-year revision-free survival rate for these complex prostheses can be as low as 15% [203]. As with any reconstruction in the skeletally immature patient, the construct must be dynamic, so that it facilitates skeletal growth. Expandable prosthetics are available with a variety of mechanisms for expansion [196, 198, 203, ]. While some of these devices may be easy to expand, their longevity is poor in younger children. In younger children, the expandable prosthesis will need to be revised as the child reaches skeletal maturity. Although there are newer cementless, porous ingrowth systems [199, 212, 213], these have not yet replaced cemented stems in most centers. A novel prestress compliant fixation device (designed to facilitate osseous integration at the bone-implant interface) currently in development obviates the need for long intramedullary stems [214]. This device is currently undergoing U.S. Food and Drug Administration testing and is available at only a few centers. Allograft-prosthetic composites are another alternative for limb salvage surgery [181, 184, 188, 196, ]. Their advantage is the hybridization of a more conventional arthroplasty with the potential incorporation of an allograft for future bone stock. The construct may also prevent delayed allograft fracture. Although the use of arthrodesis remains an option in limb-preservation surgery, it is used with diminishing frequency as experience with endoprostheses and allografts has improved. The advantage to fusion is that, once healed, the construct is very durable and may endure heavy labor [180, ]. This procedure is better tolerated in the shoulder but, because of the lack of joint motion, many patients are dissatisfied. Reconstruction as a Function of Location For tumors of the shoulder and proximal humerus, limb salvage is generally possible with preservation of the neurovascular structures. Arthrodesis rather than arthroplasty is also a viable alternative. Preservation of hand function is preferable to amputation with the goal of positioning the hand in front of the patient so that they are able to feed and groom. Overhead activity, however, is frequently not recovered, even with the best reconstructions. Prosthetic replacements for the hand and upper extremity are not as functional as those for the lower extremity, and attempts at preserving the hand are worthwhile, even if good shoulder function cannot be preserved. The proximal fibula, rib, clavicle, scapular body, and most areas of the ilium, ischium, and pubis (pelvis) can be removed without reconstruction with good functional results. Periacetabular pelvic lesions, however, can be particularly troublesome in terms of resection and reconstruction. External hemipelvectomy (hind quarter amputation) historically was the only surgical option; however, internal hemipelvectomy (removal of the bony pelvis leaving an intact limb with neurovascular structures in place) can frequently be performed [225]. Reconstruction of the defect can be a challenge. Without reconstruction, the space between the hip and residual pelvis/sacrum produces significant limb shortening (up to 4 inches) and poor function, albeit better than external hemipelvectomy. An attempt at creating a sling from synthetic material to prevent proximal migration may assist in minimizing limb length inequality. Saddle endoprostheses, allograft-hip arthroplasty composites, and complete endoprosthetic replacement have all been performed. Each has significant disadvantages and complications, however [ ]. The type of reconstruction used is at the discretion of the surgeon and the patient. If internal hemipelvectomy is attempted, the goal must be an adequate resection of the tumor, which can prove to be difficult [10, ]. Special Considerations for the Skeletally Immature The skeletally immature patient presents a particular challenge in that reconstruction must be dynamic in order to accommodate future growth. In girls, the growth spurt occurs in pre- and early adolescence (12-14 years), while in boys it happens 1-2 years later (14-16 years). Therefore, girls reach skeletal maturity by age years, while this occurs at age in boys. Most of the growth in the lower extremity occurs at the physes about the knee joint (distal femur 40%, proximal tibia 30%), while the upper femur and lower tibia have contributions of about 15% each. Since limb-sparing procedures usually result in resection of a major growth center, other alternatives need to be considered in the skeletally immature patient. These include the available expandable