In the last 30 years, the use of adjuvant chemotherapy has led to dramatic improvement in the survival of children with previously lethal sarcomas. While 30 years ago, 80% of children with a primary bone sarcoma died, now at least that same number will survive (1, 2). One of the intriguing aspects of childhood sarcomas is that, despite similar histologies, stages, and prognostic factors, some patients respond well to treatment, whereas others seem to be resistant to chemotherapy. To date, patients with good prognoses cannot be distinguished from those with poor prognoses except by crude clinical characteristics, such as the presence of metastatic disease at diagnosis or the histologic response to preoperative chemotherapy (3). Recent molecular findings in sarcomas may shed light on their biologic behavior and their response to chemotherapy.

One method of looking for genetic alterations in tumors is to examine the chromosomes by karyotype analysis. The identification of recurrent chromosomal abnormalities provides clues regarding sites of potential gene mutations. Normally, there are 23 pairs of chromosomes in the nucleus of the human cell. Osteosarcomas in general have multiple, bizarre karyotypic abnormalities: some chromosomes are missing, some are duplicated, and some are grossly altered. To date, all studies of high-grade osteosarcomas have shown complex karyotypes and nonclonal chromosome aberrations superimposed on complex clonal events (4, 5). Low-grade juxtacortical osteosarcoma, on the other hand, is characterized by the presence of a ring chromosome accompanied by few other abnormalities or none at all (6). Although it is usually possible to distinguish high-grade from low-grade osteosarcoma by standard histology, the karyotype information may be diagnostically useful in the case of other tumors. In addition to possibly providing prognostic information, the specific chromosomal aberrations provide clues that assist molecular biologists who are looking for gene mutations (6).

In contrast to osteosarcoma, Ewing sarcoma (EWS)/peripheral neuroectodermal tumors (PNETs) and alveolar RMSs have single chromosomal translocations characteristic of their respective histologies. In these tumors, part of one chromosome is transposed to part of another chromosome through a breakpoint. A novel gene and gene protein product are created that presumably give the cell a growth advantage. The most common translocations for these tumors are listed in Table 13-1 (7, 8).

The demonstration of translocations has been useful in the differential diagnosis of round cell tumors. Under the light microscope, there is little to distinguish one of these tumor types from another, and although immunohistochemistry helps to a certain extent, it is at times difficult to be sure of the diagnosis. Demonstration of these characteristic karyotypic findings makes pathologists more secure in their diagnosis and has helped with the classification of these tumors. To perform a karyotype analysis, short-term cultures and metaphase spreads are necessary, but these are labor- intensive and require fresh tissue (7). Fluorescent in situ hybridization and reverse transcriptase-polymerase chain reaction (RT-PCR) allow rapid analysis for the presence of translocations; these techniques can be performed on frozen tissue and sometimes even on paraffin-embedded tissue (8, 9 and 10). Therefore, it is important to give the pathologist appropriate fresh tissue to be snap frozen to preserve messenger ribonucleic acid (mRNA) and allow these studies to be performed (11).

These translocations have significance beyond merely establishing the diagnosis. These rearrangements lead to novel proteins that give the tumor cell a growth advantage. In EWS/PNET, for instance, a fragment of the EWS gene contains DNA-binding domains of the FLY1 gene. The protein acts by disrupting pathways that regulate DNA transcription (12). For several years, it was difficult to make the distinction between EWS and PNET, and clinicians were not sure whether to treat them differently. The observation that both EWS, a poorly differentiated mesenchymal tumor of uncertain cell lineage, and PNET, a tumor believed to be of neural crest origin, shared the same chromosomal translocation led pathologists to believe that both were related neuroectodermal tumors (13). As noted in Table 13-1, further studies revealed other translocations in several of these tumors, each such translocation specifying a different novel protein. There is debate regarding whether one or the other of these is associated with a better prognosis, but the treatment strategies used today are the same for both tumors. While some authors suggest that tumors with the type 1 transcript (EWS-FLY1) are associated with a better prognosis than those with other transcripts, others have disputed this (14, 15).

More recently, these markers have been used in staging and follow-up of high-risk patients (16). Using RT-PCR technology, one can detect small numbers of tumor cells in a bone marrow or a peripheral blood cell population (17). This makes the interpretation of bone marrow aspirates more precise and may provide a method for the earlier detection of relapses after treatment. It is hoped that the gene products of these translocations can also be used in treatment strategies. Because the novel genes formed from the translocation make a novel protein that normal cells do not make, antibodies or targeted T cells can be generated to specifically kill tumor cells. This is being tried in early-phase trials of relapsed patients with RMS and EWS/PNET, and if it works, it may be a way of treating patients who fail standard drug therapy.

Genetic alterations in the DNA of sarcomas have been well demonstrated. Mutations in genes, called oncogenes, give some evidence about the pathogenesis of these tumors and may have some prognostic and therapeutic import (4, 18, 19). Oncogenes are normal cellular genes (protooncogenes) that are necessary for the normal development and functioning of the organism (20). When they are mutated, they may produce a protein that is capable of inducing the neoplastic state. Oncogenes act through a variety of mechanisms to deregulate cell growth. This is obviously a very complex process and may involve more than one genetic event.

There are two categories of oncogenes: dominant oncogenes and tumor-suppressor genes (20). The cumulative effect alters proteins that function as growth factors and their receptors, kinase inhibitors, signal transducers, and transcription factors (12). The dominant oncogenes encode proteins that are involved in signal transduction, that is, in transmitting an external stimulus from outside the cell to the machinery that controls replication in the cell nucleus. Mutant cellular signal transduction genes keep the cell permanently “turned on.” The protein products of oncogenes also function as aberrant growth factors, growth factor receptors, or nuclear transcription factors. These types of genes seem to have less of a role in osteosarcomas. One exception is amplification of the HER-2/NEU/ERBB-2 protooncogene in patients with breast cancer, which confers a poorer prognosis. Patients with this amplification are treated with a monoclonal antibody to this protooncogene [MAb45D5, trastuzumab (Herceptin)]. Overexpression of HER2-NEU in osteosarcoma has been reported and is associated with advanced disease and poorer prognosis (21, 22). Although this has been disputed by some studies (23, 24), it provides the potential for treatment strategies in patients with osteosarcoma who have amplification of HER2-NEU.

A second class of genes are the tumor-suppressor genes, which encode proteins whose normal role is to restrict cell
proliferation (25, 26). They act as brakes rather than as accelerators of growth. Their normal role is to regulate the cell cycle and keep it in check. The retinoblastoma gene (RB) was the first gene recognized in this class (27). Osteosarcomas are very frequent in patients with hereditary retinoblastoma (1000× increased chance), both in the orbit and in the extremities, and are unrelated to irradiation. It was subsequently learned that osteosarcoma in these patients, as well as spontaneously occurring osteosarcomas, carries mutations or deletions of the RB gene. It was one of the first clues to the finding that osteosarcomas have a genetic cause. It is estimated that approximately 60% to 75% of sporadic osteosarcomas either have an abnormality of the RB gene or do not express a functional RB product (19). The RB gene is located on the long arm of chromosome 13 (13q14) and is 200 kb in length. Its product is a 105- to 110-kDa nuclear phosphoprotein (pRB) that appears to have a cell cycle regulatory role. The retinoblastoma protein acts as a signal protein, or a gatekeeper, to regulate the cell cycle through the transcription of genes that mediate the cell cycle. Deactivation of the RB gene or absence of pRB allows cells to enter the cell cycle in an unregulated fashion, a condition that imparts a growth advantage to the affected cell. It should be noted that one copy of the gene is sufficient for a normal phenotype. A child born with a normal allele and a mutant or an absent allele will not manifest retinoblastoma until some event occurs in retinoblasts to alter the normal allele. If both copies become deranged, the normal check on the cell cycle disappears, and the conditions for the neoplastic state are met. There are several other mechanisms by which the function of the RB protein can be altered; for instance, viral proteins may bind to the RB protein and inactivate it (5).

The second tumor-suppressor gene to be identified was the p53 gene (28, 29 and 30). Located on the short arm of chromosome 17 (17p), its product is a nuclear phosphoprotein that has a cell cycle-regulatory role similar to that of the RB protein. As in the case of RB, inactivation of p53 gives the cell a growth advantage, probably because of loss of cell cycle regulation. The p53 phosphoprotein may be inactivated by a variety of mutations, including a single base change (point mutation) that increases the half-life of the protein, allelic loss, rearrangements, and deletions of the p53 gene. Each of these mechanisms can result in tumor formation by loss of growth control. The p53 protein functions as an extremely important cell cycle checkpoint that blocks cells with DNA damage until they can be repaired or directs damaged cells into apoptosis (programmed cell death) if they cannot be repaired. Cells lacking this checkpoint can accumulate successive genetic abnormalities and possibly become malignant. It is estimated that approximately 25% of osteosarcomas have detectable mutations of the p53 gene (31).

The p53 protein is a transcription factor, meaning that it binds to regions of other genes (DNA) and controls the expression of genes responsible for cell cycle control (cell growth), apoptosis (programmed cell death), and other metabolic functions, such as control and repair of DNA damage. In concert with RB and a variety of other proteins, p53 acts to regulate the cell cycle through a complex cascade of enzymes, in which RB probably plays the central role. Apoptosis has recently become recognized as an important mechanism by which chemotherapy and radiotherapy kill cancer cells. p53 is involved in this process and appears to arrest cell division after sublethal damage (e.g., by radiation), to give the cell time to repair DNA defects before the next division (32, 33 and 34). If repair does not take place, the cell undergoes apoptosis and dies. If p53 is not functional, the cell may survive and accumulate genetic defects, leading to malignant transformation. Osteosarcomas have been shown to have a variety of mutations of the p53 gene (35, 36 and 37). Preliminary evidence suggests that overexpression of mutant p53 protein (detected by immunohistochemistry) or loss of heterozygosity of the p53 gene is related to human osteosarcoma (38, 39).

In sarcomas, genetic defects other than p53 and RB have also been detected. One example is a gene called mdm-2, which is a zinc finger protein that is amplified in some sarcomas (28, 40, 41). It inactivates p53 protein by binding to it, preventing its transcription factor activity. Cordon-Cardo et al. (42) studied 211 adult STSs by immunohistochemistry, using monoclonal antibodies to mdm-2 and p53, and demonstrated a correlation between overexpression of mdm-2/p53 and poor survival rates. Patients without mutations in either gene (mdm-2/p53-) had the best survival rates, those with one mutation (either mdm-2+/p53- or mdm-2-/p53+) had intermediate rates of survival, and those with mutations in both genes (mdm-2+/p53+) had the lowest survival rates. Another mechanism in which p53 protein can be inactivated is by viral proteins that bind and inactivate both RB and p53 protein (43).

Not only are genetic mutations found in the tumors of patients with sarcomas, but mutations may also be present in all somatic cells (germ-line mutations) in patients with heritable cancer (44, 45 and 46). Although such defects do not appear to be common in the general population, germ-line p53 mutations are present in patients who are part of a familial cancer syndrome. These families have a variety of cancers, often at an early age, and osteosarcomas and STSs are a fairly common occurrence in these kindred. Identification of patients with p53 germ-line mutations can be useful in determining which patients in an affected family are at risk for developing cancers, but much more work is needed in the area of genetic counseling to determine how best to use this information. One study showed that germ-line mutations were present in approximately 3% to 4% of children with osteosarcoma, and that the detection of these mutations was more accurate than family history in predicting the family’s susceptibility to cancer (47).

How is this information useful for treatment? One possibility is that the p53 mutations may be potential biologic markers of prognosis and response to treatment (chemotherapy). There is some preliminary evidence that p53 mutations in the tumor may portend a worse prognosis in osteosarcoma. More recently, the association of p53 with apoptosis has suggested possible strategies for chemotherapy, on the basis of the status of the p53 pathway (33, 34). Gene therapy (replacing the missing or mutated gene by transfection with viral carriers) is often discussed, but there are major technical hurdles to overcome before this technology can be used for treating cancers
in humans. However, it might be possible to make tumor cells more antigenic, or to make them more sensitive to antineoplastic drugs, by gene transfer. Another strategy would be to alter normal cells to make them less sensitive to damage by chemotherapeutic agents. Currently, these techniques pose technical challenges, but they offer realistic promise for the near future.

Another exciting area of research in the molecular biology of sarcomas is multidrug resistance (MDR). MDR probably explains why some patients respond to chemotherapy and others do not. Drug resistance may be intrinsic (present at diagnosis) or acquired (appearing after treatment of a tumor) (48, 49). At least four basic mechanisms of drug resistance are now recognized under the category of the MDR phenotype. They are (a) changes in glutathione metabolism, (b) alterations in topoi-somerase II, (c) non-P-glycoprotein (P-gp)-mediated mechanisms, and (d) P-gp-mediated mechanisms (6, 7, 48, 49 and 50). Recent evidence has suggested that P-gp may be of particular relevance to osteosarcoma. P-gp is a glycoprotein encoded by the MDR-1 gene on the long arm of chromosome 7 in humans (48, 49). MDR-1 is one member of the aneurysmal bone cyst (ABC) superfamily of genes that encode membrane transport proteins; these proteins function as unidirectional membrane pumps using adenosine triphosphate hydrolysis to work against a concentration gradient. P-gp is a 170-kDa protein that is located in the cell membrane and functions as an energy (adenosine triphosphate)-requiring pump that excludes certain classes (amphipathic compounds) of drugs from the cell. This physiologic mechanism is believed to be important in certain organ systems, such as the blood-brain barrier, placenta, liver, kidney, and colon, for ridding the cell of unwanted toxins, but it is also responsible for actively excluding chemotherapeutic agents, such as Vinca alkaloids, anthracyclines, colchicine, etoposides, and taxol (many of which are active in osteosarcoma protocols) from the cancer cell. Another feature of the P-gp mechanism that may have some relevance to therapeutic strategies is that some classes of drugs can reverse the MDR phenotype by blocking the action of the pump. These drugs include verapamil, cyclosporin A, tamoxifen, and others.

Several studies have demonstrated that some sarcomas (25% to 69%) display the MDR phenotype at diagnosis, and that relapsed sarcomas show higher incidence and intensity of MDR expression (48, 49, 51, 52). Because of the small numbers of patients in these studies, and the variety of the methods by which MDR expression was tested, comparisons of the studies and an accurate determination of the incidence of MDR expression are difficult to accomplish. In addition, the age of the patient and the type of sarcoma appear to be related to the incidence of detectable P-gp at diagnosis. One study showed that osteosarcomas have a higher incidence of MDR than other types of adult sarcomas (51). Serra et al. (53) demonstrated that overexpression of P-gp protein was evident in 23% of primary and 50% of metastatic osteosarcomas.

Baldini et al. (54) reported on 92 patients with non-metastatic osteosarcoma of an extremity who had been treated with chemotherapy and surgery. The study demonstrated that an immunohistochemically determined expression of P-gp predicted a decreased probability of the patient having an event-free survival, and was more accurate in prediction than histologic response to preoperative chemotherapy. Another study failed to find a relation between MDR-1 mRNA expression and outcome in patients treated for osteosarcoma (55).

Findings such as these are important in planning future protocols in human osteosarcoma. The drug-resistant tumor is becoming better identified as one that has a poor histologic response to preoperative chemotherapy and that expresses P-gp. Undoubtedly, it is more complex than this, and other mechanisms will pertain. Several caveats exist. One is the complexity of defining the resistant tumor. Preoperative chemotherapy requires 10 to 12 weeks to provide an estimate of histologic necrosis, unless ways can be found to accurately predict percentage of necrosis by positron emission tomographic (PET) scans, thallium scans, and/or gadolinium-enhanced magnetic resonance imaging (MRI). Detection of P-gp at diagnosis is difficult, and no one method has proven superior. It is probably not sufficient to demonstrate the presence of P-gp; also important is whether the pump is functioning to exclude cytotoxic agents from the tumor cell. Ideally, one would like to reverse the action of the P-gp mechanism but, just as there are no new agents to rescue patients who show poor histologic response, the agents currently available to reverse MDR are of limited benefit. They are potentially problematic in that they make normal cells less tolerant of chemotherapy, and thereby increase toxicity; and in other tumors they have not proven to be effective. The future probably lies in developing more effective reversing agents and in defining other drug-resistant mechanisms.

A thorough evaluation is necessary for any child presenting with a bone or soft-tissue mass. Although infection and trauma are much more common than a neoplastic process, the consequences of the mismanagement of a patient with a musculoskeletal tumor can be grave (Fig. 13-1).

This chapter is not designed to be a definitive musculoskeletal pathology text, and only those tumors that are commonly seen in the pediatric orthopaedic practice are discussed. The authors have tried to confine the discussion to pertinent information regarding the tumors, their evaluation, and particularly their treatment.