Osteosarcoma

Authors

  • Richard Gorlick,

    Corresponding author
    1. Department of Pediatrics and Molecular Pharmacology, The Albert Einstein College of Medicine, Yeshiva University, Bronx, NY, USA
    • Associate Professor of Pediatrics and Molecular Pharmacology, Albert Einstein College of Medicine of Yeshiva University, Department of Pediatrics, Children's Hospital at Montefiore, 3415 Bainbridge Avenue, Rosenthal, 3rd Floor, Bronx, NY 10467, USA.
    Search for more papers by this author
  • Chand Khanna

    1. Tumor and Metastasis Biology Section, Pediatric Oncology Branch, Center for Clinical Research, The National Cancer Institute, Washington, DC, USA
    Search for more papers by this author

Abstract

It has been difficult to identify the molecular features central to the pathogenesis of osteosarcoma owing to a lack of understanding of the cell or origin, the absence of identifiable precursor lesions, and its marked genetic complexity at the time of presentation. Interestingly, several human genetic disorders and familial cancer syndromes, such as Li-Fraumeni syndrome, are linked to an increased risk of osteosarcoma. Association of these same genetic alterations and osteosarcoma risk have been confirmed in murine models. Osteosarcoma is associated with a variety of genetic abnormalities that are among the most commonly observed in human cancer; it remains unclear, however, what events initiate and are necessary to form osteosarcoma. The availability of new resources for studying osteosarcoma and newer research methodologies offer an opportunity and promise to answer these currently unanswered questions. Even in the absence of a more fundamental understanding of osteosarcoma, association studies and preclinical drug testing may yield clinically relevant information. © 2010 American Society for Bone and Mineral Research

Introduction

Osteosarcoma is a well-defined clinical entity with a characteristic radiographic appearance, histologic features that can be readily recognized by pathologists, a relatively consistent spectrum of clinical presentations, and established standard treatments. These features have been the subject of many prior book chapters and reviews and are very briefly summarized in Table 1.1–6 A number of recent reviews have focused on the genetic complexity of osteosarcoma, lamenting the inability to use the numerous abnormalities present in tumors either for diagnostic purposes or for prognostication.5, 6 They have highlighted the inability to identify a precursor lesion and perhaps even the cell of origin, which has prevented the use of many laboratory methods employed to study malignancies with a clear multistep progression such as colon cancer. Despite these limitations, progress has been made through the use of epidemiologic features, genomic and molecular characterizations of available tumor samples, consideration of human predisposition syndromes, and the study of osteosarcoma in animal models. A consensus of these studies suggests that p53 and Rb gene/pathway dysregulation is central to the formation of osteosarcoma in human patients and animal models.5–7 It is unclear, however, how p53 and Rb pathway alterations are linked to the initial development or progression of osteosarcoma.

Table 1. Clinical Features of Osteosarcoma
Histologic appearanceMalignant spindle cell tumor that produces osteoid
Radiographic appearanceLytic and blastic bone lesion classically described as a “sunburst”; periosteal elevation related to a soft tissue mass producing a “Codman's triangle”
Clinical presentationPain most common symptom
LocationAny bone in the body but most commonly metaphyseal portion of appendicular bones; approximately 50% arise around the knee, with the proximal humerus the next most common site
AgeBimodal age distribution, with first and larger peak incidence in the second decade of life
DisseminationMetastases to lungs and other bones; approximately 80% present with localized disease, and i In approximately 90% of patients with metastatic disease, it will be in the lungs only
TreatmentSurgery to resect all sites of bulk disease for cure; all high-grade osteosarcomas, regardless of stage, are treated with chemotherapy, which most typically includes cisplatin, doxorubicin, and high-dose methotrexate; the vast majority of osteosarcomas in younger individuals are high grade

Etiology of Osteosarcoma

An extensive list of genetic abnormalities and environmental exposures, with examples, including fos overexpression and radiation exposure, has been associated with the development of osteosarcoma in laboratory models as well as humans.1–6 Interestingly, a single recurrent genetic event does not seem to define this cancer. A unifying approach to bring these multiple genetic risk factors together with epidemiologic and association data includes grouping gene or gene families into two categories: those which drive proliferation (growth) and those associated with an inability to repair DNA damage with associated loss of cell cycle regulatory control. This is summarized in a simplified manner in Table 2. Since both growth and loss of regulatory control are intrinsic properties of all cancers, it is not surprising that most osteosarcoma features can be categorized in this manner. Indeed, when one reviews the list of associated genetic abnormalities, it is clear that the vast majority of deranged oncogenes and tumor-suppressor genes associated with osteosarcoma are also common in the most prevalent cancers. A corollary assessment of this same result is that a distinctive and osteosarcoma-specific genetic dysregulation has not been found and, as such, suggests that the underlying and primary cause of osteosarcoma is a defect in one of these basic processes, that is, cell proliferation and cell repair. As an example, abnormal β-catenin signaling resulting from the loss of the adenomatous polyposis coli (APC) gene is believed to drive the formation of colon adenomas, with the end result being over 200,000 cases per year of colon cancer in the United States.8, 9 Osteosarcoma, in contrast, occurs in less than 1000 patients per year.10, 11 Therefore, these oncogenic events do not uniquely define osteosarcoma, nor is the frequency at which these derangements occur likely to define the incidence of this disease.

Table 2. All Epidemiologic and Genetic Alterations Associated with Osteosarcoma Are Related to Growth/Proliferation or DNA Damage/Cell Cycle Checkpoint Control
 Growth/proliferation-relatedDNA damage/cell cycle checkpoint control
Epidemiologic associationsAssociation with greater heightRadiation exposure
 Peak incidence in second decade of lifeHereditary retinoblastoma (Rb genetic association)
 Earlier peak incidence in females who have earlier pubertal growth spurtsLi-Fraumeni syndrome (p53 genetic association)
 Paget's diseaseRothmund-Thomson syndrome (RECQL4 genetic association)
 Occurrence selectively in large-breed caninesWerner syndrome (WRN genetic association)
 Parathyroid hormone exposure 
Genetic associationsMET/HGFOther means of abrogating p53 function (MDM2 amplification, INK4 deletion, COPS3 amplification)
 EGFR/EGF
 HER-2
 ErbB-4
 MAPK
 IGF-1R/IGFOther means of abrogating Rb function (INK4 deletion, CDK4 amplification)
 BMP
 WNT
 ß-cateninMYC
 VEGFR/VEGFFOS
 PDGFR/PDGFSV40
 TGFR/TGF 

What sets osteosarcoma apart from other malignancies in large part are the features that are immediately apparent: the organ of origin and the age of peak onset.10, 11 Although osteosarcoma comprises less than 1% of cancers diagnosed in the United States, it is the most common primary malignancy of bone.10, 11 Bone is an unusual site of cancer formation, and many of the unique properties of this disease may be related to either the cell of origin or unique features related to that environment. Osteosarcoma has its highest incidence during the second decade of life.10, 11 This is a period of time in which the incidence of cancer is low, and although surpassed in incidence by leukemia and lymphoma, it is one of the more common cancers during adolescence.1, 3, 4 The vast majority of osteosarcomas in children, adolescents, and young adults are high grade and begin in the intramedullary space of metaphyseal locations in long bones of the lower extremity, suggesting a relationship to expanding growth plates.1, 3, 4 After a lower incidence in individuals 25 to 59 years of age, the incidence of osteosarcoma rises once again in individuals over age 60 to approximate the incidence in adolescence.10, 11 When osteosarcoma in older individuals is compared with that in adolescents, a greater proportion is found to have tumors in axial sites, low-grade tumors, surface lesions, and rare presentations such as at extraosseous sites.10, 11 These differences in presentation may suggest that the underlying pathogenesis of these entities is not identical independent of epidemiologic features such as Paget's disease and prior radiation exposure, which are risk factors unique to older patients.10, 11

Osteosarcoma arises from a mesenchymal cell that has or can acquire the capacity to produce osteoid.1, 3, 4 Cells of this type exist in multiple compartments and include the adherent cell populations obtained in a bone marrow aspiration, hence the intramedullary space within bones, with cells of this type generally referred to as mesenchymal stem cells.12 They also include osteoblasts, which are concentrated along both the endosteal and periosteal surfaces of the bone and involved in fracture repair and remodeling. Longitudinal growth of long bones is accomplished by proliferation of chondrocytes at the growth plates with subsequent bone mineralization as part of endochondral ossification, although the site of initial tumor formation (intramedullary or surface lesions) is a matter of speculation based on the radiographic appearance of the primary tumor. Most osteosarcomas are defined as high histologic grade. These tumor originate most often in the intramedullary space; however, low-grade intramedullary osteosarcomas are seen rarely.13 Similarly, surface osteosarcomas, which presumably arise from cells in the periosteum, can occur in several variants, including parosteal lesions, which are low-grade osteosarcomas, high-grade surface osteosarcomas, and periosteal oteosarcomas that are intermediate in grade.14 The treatment of both intramedullary and surface lesions is based primarily on histologic grade. Little is known about the comparative biology of these distinctive lesions.14 Similarly, axial lesions that arise from bones, which form through intramembranous ossification, have not been compared in a comprehensive manner with tumors arising in appendicular sites. Although osteosarcoma's bone derivation is likely to be a critical feature, it is unclear if and how the cellular compartment from which the tumor arises influences its development and progression.

Even the cell of origin of osteosarcoma has been the subject of extensive recent discussion. A number of recent reviews have challenged the historical belief that osteosarcoma is derived from osteoblasts. Rather, newer data provide an argument that the presence of distinct histologic forms of osteosarcoma is a result of the tumor's retained potential for pluripotent differentiation.1, 3, 4 Nonetheless, osteosarcomas are categorized as various histologic subtypes based on their predominant pattern of differentiation. Histologic subtype does not appear to influence the clinical behavior of the disease.1, 3, 4 Knockout mice with conditional loss of Rb and p53 function in a preosteoblastic cell develop a tumor that, based on appearance, clinical features, and gene expression profile, is an osteosarcoma.15 This suggests that the cell of origin is this preosteoblast.15 However, the tumor that arises in this model system has been noted to express more primitive lineage markers than preosteoblasts.16 Although these markers may be acquired in the transformation process, it also has been suggested that a rare precursor cell that expresses markers of both mesenchymal stem cells and preosteoblasts may be the cell of origin.16 Targeted disruption of p53 and Rb in the mesenchymal cells of the murine limb bud also produces sarcomas.17 A separate line of evidence comes from the observation that serially passaged murine mesenchymal stem cells will undergo rare spontaneous transformation resulting in an osteosarcoma-like tumor, suggesting that these cells are the cell of origin.18, 19 With the existing available data, it is not possible to define whether osteosarcoma derives from mesenchymal stem cells or a more differentiated progenitor.

The rarity of osteosarcoma may be related to a small pool of cells being capable of producing the cancer. Alternatively, events occurring at the time of maximal longitudinal bone growth may be necessary for developing osteosarcoma, and this may be restricting its formation. The events may be related to aberrations in the differentiation program necessary for ossification and longitudinal bone growth or may be related to the need for an external signal initiating cellular proliferation. These events may trigger the proliferation of a mutated cell, or it may be proliferation that drives the initial mutation event. Although limited data exist in this regard, it is not clear that patients with hereditary retinoblastoma or Li-Fraumeni syndrome develop osteosarcomas markedly earlier than individuals who do not have a germ-line mutation resulting in an osteosarcoma predisposition despite their markedly higher incidence of osteosarcoma.1, 4, 11 This perhaps is suggestive of the former scenario, but few data exist to define what initiates and is necessary for osteosarcoma formation.

A number of mechanistic studies have been performed over the past 2 years in an attempt to decipher pathways associated with the pathogenesis of osteosarcoma. Two of the pathways that have been among the more extensively studied include Wnt and Notch. The aforementioned observation of spontaneous transformation of murine mesenchymal cells into osteosarcoma has been used to study pathogenesis.18 A parallel and functional phenotypic analysis of the parental mesenchymal stem cells, transformed mesenchymal stem cells, and resulting osteosarcoma suggested aneuploidization, translocations, and homozygous loss of the cdkn2 region as the major mediators of malignant transformation.18 This same research group has performed expression profiling of these cell types and has suggested that downregulation of Wnt signaling has an important role in osteosarcoma pathogenesis.20 When the Wnt pathway was activated using a GSK3β inhibitor in osteosarcoma cell lines, proliferation was inhibited, and osteogenic differentiation was observed.21 This is further supported by the observation that patients with osteosarcoma have higher serum levels of Dkk-1, a secreted inhibitor of the canonical Wnt pathway. Dkk-1 and RANKL also were noted to be coexpressed by rapidly proliferating osteosarcoma cells.22

In the majority of malignancies, activation of the Wnt pathway and β-catenin promotes tumorigenesis, in contrast to the aforementioned line of evidence in osteosarcoma.23 Indeed, even in osteosarcoma, a number of experiments have been performed downregulating the Wnt pathway using genetic or drug-treatment approaches, and in contrast to the prior observations, enhanced invasion and motility were observed, with the related clinical observations being an increased frequency of pulmonary metastases or decreased survival.24–28 It has been noted that Wnt inhibitory factor 1 is epigenetically silence in human osteosarcomas, with targeted disruption in mice accelerating osteosarcoma development.29 With the existing available data, it appears likely that the Wnt signaling pathway has an important role in osteosarcoma, but it is unclear whether up- or downregulation promotes its formation and malignant behavior.

The Notch receptors have an important role in cell fate decision making and are closely involved in mesenchymal stem cell differentiation. Since the pathway is involved in numerous skeletal diseases, it is logical to presume its involvement in osteosarcoma.30 Indeed, data from murine models and human cell lines suggest Notch involvement in osteosarcoma pathogenesis as well as invasion and metastases. Studies thus far on the Notch pathway in osteosarcoma are rather limited.31, 32

Metastases

A defining feature of osteosarcoma is the high rate of metastasis that results from the primary bone tumor disseminating to distant secondary sites by a hematogenous route. Despite the use of multimodality and multiagent systemic cancer therapy before and after management of the primary tumor, the vast majority of deaths seen in osteosarcoma patients occur as a result of metastases. The most common site of metastasis is the lungs.1, 4 In the era before the use of chemotherapy, most osteosarcoma patients still had successful management of the primary tumor (most often through limb amputation). Nonetheless, over 85% of patients continued to develop metastases. This suggests that in osteosarcoma patients, microscopic spread of cancer cells has occurred at the time of original presentation. The emergence of visible metastases can occur within months or after a prolonged period of “dormancy.” Dormancy in cancer and in osteosarcoma is poorly understood.33 First, it is unclear where in the body these dormant metastatic cells exist. Recent studies support a hypothesis that the bone marrow may be a site where dormant metastatic cells reside. Support for this hypothesis in osteosarcoma includes the identification of osteosarcoma cells in the bone marrow of patients.34 Following the rationale of this hypothetical model, cells would disseminate from the primary tumor early during tumor formation to the bone marrow. Metastatic cells then would persist during the period of dormancy in the bone marrow and then emerge subsequently and colonize distant secondary sites at the break in dormancy. The presumed mesenchymal stem cell origin for sarcomas (discussed earlier) and their ability to traffick to the bone marrow provide circumstantial support for this hypothesis. The determinants that result in a break in dormancy are similarly poorly understood. Experimental data suggest a link between bursts in angiogenesis and breaks in dormancy in sarcoma and other cancer models.35 The causes for such bursts or the clinical scenarios that may be linked to these events are not understood.

Once established, metastatic lesions become increasingly difficult to manage. Unlike most other cancers, the resection of pulmonary metastasis, when possible, is the common second-line treatment. Metastectomy is associated with 5-year survival rates of up to 40% of patients. Unfortunately, subsequent recurrence in most metastatic patients will be seen and eventually will require systemic treatment. The eventual resistance of these pulmonary metastases to currently available systemic therapy is common. The failure of treatment may be the result of acquired resistance to chemotherapy and/or to the ability of metastatic cells to develop “protection” within their microenvironment in the lung.

The process of metastasis includes tumor cell migration, invasion, entry into the circulation, and eventual arrest and extravasation at distant secondary sites, which has been reviewed extensively.36, 37 Many of the genes that have been associated with osteosarcoma formation (i.e., oncogenesis) are likely to contribute to progression and metastases. Unfortunately, as discussed earlier, the underlying genetic complexity has complicated efforts to identify driving and causal genetic drivers for metastasis in osteosarcoma despite a number of biologic motifs (i.e., growth factor signaling paths, angiogenic phenotype, and mesenchymal stem cell origin) that are consistently associated with osteosarcoma progression and may be described as metastasis “virulence” factors, as recently coined by others.36, 37 The following progression (i.e., virulence) factors in osteosarcoma have been consistently held across several investigative platforms and studies38:

  • Angiogenesis. The development of an angiogenic phenotype is a recognized determinant of metastatic cells. In osteosarcoma, several associations with such an angiogenic phenotype have been defined. This includes an association with metastatic risk and primary tumor microvessel density, expression of angiogenesis-associated growth factors, and the use of inhibitors of angiogenesis in osteosarcoma model systems.39–41

  • Ezrin. The cytoskeleton linker protein ezrin, a member of the ezrin, radixin, and moesin (ERM) family, has been connected to the metastatic phenotype in murine, canine, and human osteosarcoma.42 It is reasonable that a physical connection between the actin cytoskeleton and the cell membrane is of value to a metastatic cell as it engages its microenvironment in cancer. Studies of ezrin in osteosarcoma have demonstrated a functional efficiency provided by the linkage between the cell membrane and actin cytoskeleton that is related to the signal-transduction activity of membrane proteins that are associated with metastasis.43–45 The specific mechanisms associated with ezrin's role in metastasis is not known; however, a uniting hypothesis suggests that ezrin is part of a complex solution used by metastatic cells to deal with the stresses of the process of metastasis.

  • Integrins. The integrins are a large family of membrane-associated receptors that interact primarily with matrix associated proteins.46 Integrin signaling has been suggested to be a primary mechanism whereby cancer cells interact with the cellular microenvironment. In osteosarcoma, the expression of specific integrin family members has been linked to metastasis.45

  • Chemokines. Similar to the integrin family of proteins, expression of chemokines and the chemokine receptors have been linked to osteosarcoma progression and metastasis in a number of preclinical and correlative studies.47, 48 Interactions between chemokines and integrin family members further contribute to the ability of metastatic cells to interact with their microenvironment and promote metastatic cell survival.49

  • The insulin-like growth factor 1 (IGF-1) pathway. The IGF-1 pathway has been linked to the development and progression of many sarcomas, including osteosarcoma.50 The growth and development of adult mesenchymal tissues are largely the result of growth hormone–induced release of IGF-1 and its interaction with the IGF-1 receptors present on osteoblasts and other mesenchymal cells. Proliferation and survival of normal and malignant osteoblasts have been linked to activation of the IGF-1 pathway.50 Furthermore, the roles of the IGF-1 pathway in osteosarcoma include direct associations with the metastatic phenotype.51, 52 Recent opportunities to target the IGF-1 receptor have been possible through humanized and fully human antibodies that target the IGF-1 receptor and small-molecule inhibitors directed against IGF-1 receptor kinase. A number of therapeutic antibodies targeting the IGF-1 receptor are in various stages of preclinical and clinical development in cancers, including osteosarcoma, and have been associated with single-agent activity in preclinical models.53 Additional agents target downstream components of the IGF-1 receptor pathway, including PI3 kinase and Akt kinase, which are connected to progression and metastasis in sarcoma.

  • c-MET. c-MET is the receptor for hepatocyte growth factor. The identification and description of c-MET, as an oncogene, came from studies in a chemically transformed model of osteosarcoma. Furthermore, preclinical studies in vitro and in vivo support the role of c-MET signaling in cancer progression and specifically metastasis.54–56c-MET has been shown to be expressed in sarcoma primary tumors and metastatic lung nodules.57 It is likely that several metastatic processes are linked to c-MET signaling, including cell motility, invasion, proliferation, and survival.58 Since c-MET is a growth factor receptor with an intracellular tyrosine kinase activity, the development of small-molecule inhibitors of c-MET has been possible. The inhibition of c-MET has been effective in suppressing metastatic phenotype in osteosarcoma cells and preclinical models.59

  • Mammalian target of rapamycin (mTOR). mTOR is a critical node in a signaling pathway that connects many growth factor receptors through intermediaries, including AKT and MAPK, to the translational machinery of the cell.60 As a result, mTOR is able to convert signals that sense the nutritional and stress status of a cell (i.e., in the cell's microenvironment) into specific proteins that can manage the stress.61 Many of the known translational targets of mTOR have been connected to cancer, including c-myc, VEGFR, HIF, and TGFβ. The importance of mTOR in osteosarcoma is also supported by mTOR's importance in mesenchymal stem cells.62 Preclinical studies with agents that block mTOR have been shown to reduce metastases in a murine model of osteosarcoma.63 Early human clinical data with mTOR inhibitors support the therapeutic value of this target in many sarcoma histologies.64

New Resources and Methods

Much has been written thus far about the complexities in understanding osteosarcoma, but much of the promise lies in the new resources that exist for studying the disease. In the 1980s and 1990s, the material that was available to study osteosarcoma was predominantly cell lines.7 Not minimizing the importance of human osteosarcoma cell lines such as the SaOS-2, HOS, and U-2 OS cell lines (American Tissue Type Culture Collection, ATTCC), both issues of selection for growth in vitro and the long duration with which they have been passaged, among other issues, limit their relevance for studying the human disease.7 In the late 1990s, the Children's Oncology Group launched a successful national osteosarcoma tissue-banking effort. At present, over 1000 patients have been enrolled in the study, with blood and serum available on over 900 individuals, frozen tumor tissue from over 500, and paraffin-embedded tumor from over 600. A broad range of assays is planned for these tissues, but requests for banked material can be made by all investigators using a Web-based application (https://ccrod.cancer.gov/OsteosarcomaSampleRequest/). Applications are subjected to a peer-review process, but being a part of the Children's Oncology Group is not a review criterion. This bank of tissue, although a tremendously valuable resource, is limited by the fact that it has been acquired solely through a pediatric cooperative group and therefore includes only patients up to age 40. Interactions with other cooperative groups such as the Sarcoma Alliance for Research Though Collaboration (SARC) ultimately may permit banking of osteosarcoma tissue from older individuals, but at present, that tissue resource is not readily available. In addition to tumor tissue, human osteosarcoma has been serially passaged in heterotopic sites in immunocompromised mice as another resource for studying the disease.65

New technologies continue to emerge that permit more comprehensive assessments of tumor tissue. Massively parallel sequencing is permitting whole-genome sequencing of tumors. These technologies hold the promise that they may allow an acquisition of a deeper understanding of the pathogenesis of osteosarcoma. The prior development of techniques such as oligonucleotide microarrays were believed previously to hold significant promise.7 Unlike sarcomas derived from recurrent chromosomal translocations, the signatures produced by expression profiling were exceedingly heterogeneous. Simple classifications that were possible for the translocation-associated sarcomas were not possible for osteosarcoma, and these analyses thus far have produced few tangible results.66, 67 Perhaps suggesting whole-genome sequencing may be of value in osteosarcoma is the ability of this approach to identify consistently mutated pathways in genetically complex malignancies such as lung adenocarcinoma.68 Whether a consistent pattern will emerge from whole-genome sequencing analyses of osteosarcoma awaits completion of these studies.

Preclinical Screening

Even in the absence of understanding osteosarcoma's pathogenesis, laboratory studies can be performed that may yield information that is useful clinically. Identifying biologically based prognostic factors may not require a fundamental understanding of the tumor's initiation because invariable events are unlikely to discriminate clinically distinct patient subsets.7 Beyond prognostic factors, one can perform empirical preclinical screening through model systems to identify therapeutically relevant drugs. This effort is facilitated by the fact that few drugs are likely to be developed specifically for osteosarcoma. Therefore, one does not need to identify what is the optimal therapeutic target in osteosarcoma but rather can focus on screening the drugs that are likely to be available clinically to determine which of them, if any, may be effective. Using model systems is much more efficient, cost-effective, and rapid than performing human clinical trials, particularly given the rarity of osteosarcoma.

The Pediatric Preclinical Testing Program is one such effort under way to facilitate the introduction of new, active agents into clinical trials for all childhood cancers. With a consortium of laboratories in the United States and abroad, the program is able to quickly screen a large number of agents using in vitro and in vivo models.65, 69–71 The in vitro models are cell lines, with the in vivo models typically human patient-derived tumors grown as heterotopic xenografts in severe combined immunodeficiency mice. Although osteosarcoma can be grown in orthotopic as well as heterotopic sites, the heterotopic site permits frequent tumor size measurements using a caliper, which is feasible and cost-effective. Preclinical testing potentially may predict the activity of new agents in patients with childhood cancers, allowing identification of active agents more rapidly. Some believe that preclinical testing needs to be validated as accurately representing responses in human clinical trials prior to use as a means of prioritizing clinical trials. Others believe, in the absence of other data, that preclinical testing should be used as a basis of prioritization because it is more likely to be predictive than intuitive or random selection. This program has generated a large amount of data, including characterizations of the model systems, that has rapidly been published or made available through Web-based applications (http://home.ccr.cancer.gov/oncology/oncogenomics/).65, 69–71

Canine Models

An important opportunity to extend our understanding of cancer biology and therapy through preclinical studies is provided by the natural development of osteosarcoma in pet dogs.72 Naturally occurring tumors in dogs and other animals have clinical and biologic similarities to human cancers that are difficult to replicate in other model systems. A recently launched cooperative effort, the National Cancer Institute's (NCI's) Comparative Oncology Trials Consortium (COTC; http://ccr.cancer.gov/resources/cop/COTC.asp), provides an infrastructure and the resources needed to integrate these naturally occurring cancer models into the development of new human cancer treatments.73 The study of cancer biology and therapy in animals with naturally occurring cancers, referred to as comparative oncology, is not a novel concept. Indeed, over the last 30 to 40 years, investigators have used this approach to make important contributions to the understanding and practice of human oncology in fields such as basic tumor biology and immunology, radiation biology, hyperthermia, and systemic therapies for a variety of cancers, including osteosarcoma, lymphoma, melanoma, and others. The parallels between canine and human osteosarcoma are perhaps the strongest across the comparative oncology opportunities. Both diseases are characterized by primary tumor growth in the appendicular skeleton and a high risk for metastasis to the lungs. The canine disease is indistinguishable from the human disease at the histologic and gene expression levels. Indeed, both conventional and investigational treatments for both the primary tumor and the metastatic disease are associated with similar response features in both species. The primary differences between the models is the age of development and the prevalence of disease. In dogs, osteosarcoma is a disease of older, large breed dogs (i.e., 6 to 12 years of age), whereas osteosarcoma occurs most commonly in the second decade of life in humans. The incidence of osteosarcoma in dogs is not known; however, some estimates suggest well over 10,000 cases annually in the United States. This high prevalence and the relatively rapid rate of disease progression (median disease-free interval following surgery alone is 4 months; with surgery and chemotherapy, 13 months) provides the opportunity to evaluate novel treatment options in dogs in a relatively compressed time period. Current studies in collaboration between the Children's Oncology Group and the COTC will attempt to rank the most active agents evaluated in canine osteosarcoma as part of future consideration in pediatric osteosarcoma clinical trials.

Osteosarcoma and Its Relationship to Normal Bone Biology

Throughout this review it has been highlighted that osteosarcoma is not well understood within the context of normal bone biology. The National Cancer Institute, among other groups, has sponsored several conferences bringing together sarcoma researchers with individuals who research mesenchymal stem cells as well as normal bone physiology attempting to foster collaborations (http://rarediseases.info.nih.gov/ASP/html/conferences/conferences/sarcoma20040927.html#report). Despite these conferences, few collaborations of this type are evident in the literature. The roles of the Wnt and Notch pathways in osteosarcoma, which are important in mesenchymal stem cell differentiation and bone formation, have been clarified only to a limited extent thus far, as has been described previously. Several researchers explore the tumor-environment interactions for the more common bone-metastasizing cancers such as breast and prostate adenocarcinoma.74–78 It is unclear to what extent these same mechanisms are operative in osteosarcoma, which arises from as well as disseminates to bone. This remains a critical direction for future research efforts. Largely based on the studies of epithelial cancers that are metastatic to bone, current clinical and translational studies on the bone microenvironment and osteosarcoma are limited to the study of the bone osteoclast. Unfortunately, the current understanding of the interaction between the bone osteoclast and osteosarcoma is incomplete. Such an understanding is necessary for optimal development of treatment strategies that target the bone osteoclast and bone osteoclast activation, such as the bisphosphonates and RANK ligand antibodies in the treatment of osteosarcoma.79–83

Conclusions

Although many questions remain unanswered with regard to understanding the fundamental biology of osteosarcoma, the availability of a large tissue bank, along with xenograft and canine model systems, offers considerable promise for the future. Clinically useful information may be derived from empirical screening approaches as well as the application of new laboratory methodologies focused on defining rare but recurrent genetic drivers of osteosarcoma development and progression. A critical need for future research efforts will be to understand osteosarcoma in the context of normal bone physiology and the environment in which it arises and progresses. Use of osteosarcoma-related tissue resources that are available to the entire research community and approaches to foster collaborations between the sarcoma and bone research disciplines is encouraged.

Ancillary