Specific Targets in Sarcoma and Developmental Therapeutics

Medscape: Continuing Medical Education Online

Receptor Activator of NF-κB Ligand

The regulation of bone homeostasis has been examined in considerable detail over the past decade, and has been exploited therapeutically for a range of diseases from osteoporosis to giant cell tumor (GCT) of bone and metastatic cancer. Two main cell types are involved in bone homeostasis: the osteoblast and the osteoclast. ⁵ The receptor activator of NF-κB (RANK) pathway is required for osteoclast differentiation, along with its ligand (RANKL) and a decoy receptor, osteoprotegerin. ^6–9 Mice carrying loss-of-function alleles of rankl or RANK developed osteopetrosis, whereas those with mutant osteoprotegerin developed osteoporosis. ^8,10 Moreover, osteoclast-like cells can now be generated in vitro in the absence of stromal cells by adding recombinant RANKL to primary cultures of monocytic cells. ¹¹

GCT of bone is a benign, locally destructive neoplasm, characterized by the presence of”giant” osteoclast-like cells and a stromal mononuclear population. ¹² The stromal population in GCT expresses high levels of RANKL, which probably recruits mononuclear precursor cells to the tumor site where they mediate the bony destruction associated with the tumor. ^13,14 Unlike many targets discussed in this article, the overexpression of RANKL in GCT is not caused by known genetic or epigenetic events, and the molecular pathogenesis of GCT remains unclear.

GCT occurs in skeletally mature young people, and may cause pain, susceptibility to fracture, neurologic injury, and deformity. ¹² GCT may be associated with benign pulmonary metastases, usually after surgical instrumentation. Finally, between 1% and 3% of GCTs may undergo true malignant degeneration, which is most frequently seen after prior radiotherapy. ¹⁵ Treatment of GCT has typically involved surgical curettage and sometimes excision. ¹⁶ Local recurrence rates vary from 10% to 50%, depending on the surgery performed and the disease site. The surgical morbidity associated with achieving clear margins in tumors affecting the axial skeleton in particular can cause patients to have serious sequelae. Radiotherapy has been shown to be effective in up to 80% of cases but is associated with late effects, including infertility and the risk for second malignancies. Systemic therapies have had limited success.

The development of denosumab, a monoclonal antibody to RANKL, ¹⁷ led to a prospective, phase II clinical trial of this agent in patients with unresectable or recurrent GCT. ¹⁸ This study enrolled 37 patients, with a high proportion having extensive prior treatment, tumors located in the axial skeleton, and, in some cases, metastatic disease. Based on the criteria for response in this study (progression-free survival at 25 weeks or histologic evidence of eradication of giant cells within the tumor on biopsy), 86% of patients experienced response to denosumab. All patients who underwent histologic assessment showed eradication of giant cells. Although it was not formally assessed, the investigators reported that 84% of patients experienced clinical benefit, including reduced pain and improved range of movement and functional status. Almost a third of patients showed bone repair on study. The drug, which was administered as a subcutaneous injection, was well tolerated and no treatment-related serious adverse events were reported during the study.

The precise role of denosumab in GCT remains the subject of clinical research. A current clinical trial is evaluating the adjuvant use of denosumab in patients with locally advanced but resectable disease, because whether treatment may reduce the risk for recurrence or facilitate less-morbid surgery is unknown. Another question relates to the use of denosumab in other giant cell–rich tumors, such as giant cell reparative granuloma, which appears to overexpress RANKL. ¹⁶

Macrophage Colony–Stimulating Factor 1

Pigmented villonodular synovitis (PVNS) and tenosynovial GCT (TGCT) are closely related, benign, extraosseous neoplasms and, like GCT, are characterized by stromal cell proliferation accompanied by a giant cell infiltrate. ¹⁵ The proportion of giant cells is typically lower than that seen in GCT. The tumors are often seen around joints and involve tendon sheaths, particularly in the lower limb. Although these tumors typically do not metastasize, and rarely, if ever, have been reported to cause death, they are a source of recurrent morbidity and loss of function. Current standard of care, as for GCT of bone, is surgical extirpation. Although this procedure is usually effective, the infiltrating patterns of growth seen with PVNS/TGCT make complete excision difficult while preserving joint and tendinous structures, and local recurrences are common. Radiotherapy is occasionally used to control disease but may also be associated with long-term sequelae.

Recent progress has been made in understanding the pathogenesis of PVNS/TGCT. In 2006, a group from Stanford reported a recurrent translocation involving macrophage colony–stimulating factor 1 (CSF1, also known as M-CSF) in PVNS/TGCT. ¹⁹ Chromosome 1p13 is fused to 2q35, linking the CSF1 gene to the COL6A3 promoter and leading to overexpression of CSF1. The authors postulate that overexpression of CSF1 leads to recruitment of monocytes to the tumor site, because CSF1 is required for monocyte lineage differentiation. The translocation is found in a minority of tumor cells, and most cells within a tumor express the CSF1 receptor (CSF1R). Imatinib mesylate has submicromolar activity against the ABL, KIT, PDGFR, and CSF1R kinases and clinically important activity in several sarcomas, including GIST ¹ and DFSP. ² A case report suggests that imatinib may have activity in PVNS/TGCT, presumably through targeting CSF1R. ²⁰ As a result of this observation, a clinical trial using a related compound (nilotinib) is under development for patients with otherwise refractory PVNS/TGCT.

p53 Pathway

The p53 gene, located at 17p13, is mutated in almost half of all cancers, and the pathway in which it plays a central role is dysregulated in almost all cancers. ²¹ One of the gene's transcriptional targets, MDM2, physically binds and inhibits p53. ^21,22 MDM2 is amplified in up to 10% of all cancers, and is especially frequently affected in some sarcomas (see later discussion).

A related gene, MDM4, may act in a similar fashion to MDM2. MDM2 is inhibited by p14ARF, located at 9p21. p14ARF sequesters MDM2 away from p53, thereby liberating p53 to act as a tumor suppressor. Good evidence suggests that this pathway is linear and nonredundant, because mutations in either p14ARF, MDM2, or p53 seem mutually exclusive, consistent with mutual interdependence of each component, ²² whereas preclinical studies indicate that p53 is required for the biologic effects of MDM2. ²³

Different tumors target specific individual components of the p53 pathway. ²¹ MDM2 is clearly frequently amplified in soft tissue and bone sarcomas (20%–30%), and almost universally amplified in certain sarcoma types, such as well- and de-differentiated liposarcoma (WDLPS/DDLPS). ¹⁵ Amplification of MDM2 has also been reported in 20% to 60% of malignant peripheral nerve sheath tumors and a high proportion of osteosarcomas. ^21,24 Mechanistically, the near-universal amplification of MDM2 in WDLPS/DDLPS is from the incorporation of regions of chromosome 12 into a cancer-associated neochromosome. ¹² Ring or accessory neochromosomes are a near-pathognomonic feature of WDLPS/DDLPS and other sarcomas, including DFSP and parosteal osteosarcoma. Amplification of MDM2 and mutations in p53 are rarely reported in the same sarcoma. ²⁴

Over the past 5 years a series of small molecule antagonists of MDM2 have been developed, including nutlin 3a. ²⁵ Nutlin sterically interferes with binding between MDM2 and p53. Treatment of cancer cell lines carrying amplified MDM2 in vitro and in vivo results in activation of p53, leading to growth arrest and apoptosis. Importantly, nutlin 3a seems to be ineffective in cell lines carrying mutations in p53 but may be overcome by cotreatment with DNA-damaging agents. ²⁶ Significant interest has been shown regarding the activity of MDM2 antagonists in diseases in which MDM2 amplification is extremely common, such as WDLPS/DDLPS and malignant peripheral nerve sheath tumors. The clinical role of MDM2 antagonists is an area of active investigation, with early-phase clinical trials underway.

Retinoblastoma Pathway

Like p53, the Rb pathway is almost universally dysregulated in cancer, although (unlike p53) it has important developmental as well as tumor suppressor roles. ²⁷ The Rb pathway controls the cell cycle at the G1/S checkpoint.

A range of CDK4 antagonists are in clinical development, most in early-phase study, and are fairly nonselective, with effects on several CDKs. ²⁸ Genetic ablation of CDK2, -4, and -6 in mice seems dispensable in most normal cell cycles, but depends on CDK1. ²⁹ The first generation of nonselective CDK inhibitors was associated with significant toxicities, ^28,30 which may be ameliorated by the development of more selective antagonists of CDK4 (e.g., P1446A-05, Piramal Healthcare, Mumbai, India; NCT007728766). The frequent coamplification of MDM2 and CDK4 in WDLPS/DDLPS is of particular interest because it presents an opportunity for combined use of targeted therapeutics. ³¹

Vascular Endothelial Growth Factor Receptor Pathway

Vascular endothelial growth factor (VEGF) is a protein ligand produced primarily in response to tissue hypoxia in normal cells. VEGF binds to members of the VEGF receptor (VEGFR) family on endothelial cells and stimulates receptor tyrosine kinase activity. Endothelial cell proliferation results, and thus VEGF enhances local delivery of oxygen and other nutrients through neoangiogenesis. Various cancers have been shown to co-opt this pathway, and production of VEGF can accelerate tumor growth. Several agents, including the anti-VEGF antibody bevacizumab and the VEGFR2 small molecule inhibitors sunitinib, sorafenib, and pazopanib, target the VEGF/VEGFR axis and impair tumor progression, and have been approved by regulatory authorities for the treatment of cancers.

Many sarcoma cell lines or tumors have been shown to produce VEGF or show evidence of activation of VEGFR, although for most cases the precise mechanisms remain unclear. ³² Some sarcoma subtypes are highly vascular on gross or microscopic examination, and several agents have been studied for their efficacy against sarcomas. When examined against sarcomas as a whole, VEGFR inhibition seems to have minimal impact on sarcoma growth. A phase II study of 17 patients treated with combination doxorubicin and bevacizumab, for example, observed only 2 responses (12%), and 13 patients had stable disease for more than 12 weeks, ³³ not significantly different from what might be expected from doxorubicin alone.

However, some select subtypes may be uniquely sensitive to VEGFR inhibition, and if confirmed in larger studies this may lead to new histotype-specific treatment advances and a better molecular genetic understanding of tumor pathogenesis. In a phase II study of sunitinib, for example, 10 of 49 patients (20%) developed stable disease for more than 16 weeks, including 2 of 3 patients with solitary fibrous tumor/hemangiopericytoma (SFT/HPC) who had disease control for 24 and greater than 59 weeks, respectively. ³⁴ Although sunitinib inhibits several kinases other than VEGFR, the potential for a role of VEGFR in SFT/HPC is strengthened by the results of a retrospective analysis of patients with this disease who were treated with temozolomide and bevacizumab. ³⁵ In this case series, 11 of 14 (79%) patients experienced a partial response according to Choi criteria, and 2 other patients (14%) had stable disease. The role of each individual drug compared with the combination is unclear, but these observations, in conjunction with the suggestive results from the prospective phase II study of sunitinib, suggest that blockage of the VEGF/VEGFR axis may be an effective strategy for treating SFT/HPC.

Angiosarcoma, a mesenchymal cell–derived neoplasm with endothelial differentiation, is also an attractive target for VEGFR-directed therapy. Many angiosarcomas express VEGF or its receptors. ³⁶ Complete pathologic responses were observed in 2 patients treated with radiation and bevacizumab. ³⁷ Among 27 evaluable patients in an ongoing phase II study of single-agent bevacizumab, preliminary best results show 4 patients (15%) experiencing partial response and 13 (48%) with stable disease. ³⁸ In a phase II study of the multitarget kinase inhibitor sorafenib, 5 of 37 patients (14%) with angiosarcoma experienced partial responses and 21 (57%) developed stable disease, whereas patients with other sarcoma subtypes showed minimal activity. ³⁹

One study found that VEGFR2 and other vascular-specific tyrosine kinases were upregulated in angiosarcomas of the breast (mammary-type or radiation-associated), and mutations in the VEGFR2 gene were identified in 4 of these specimens, accounting for 10% of samples analyzed. ³⁹ Laboratory assessment showed that these VEGFR2 mutations result in constitutive activation of the receptor, which can be blocked by the small-molecule inhibitors sunitinib and sorafenib. Whether the presence of these mutations is required for or predictive of response to treatment in vivo remains to be determined in clinical studies.

In a series of 4 evaluable patients with advanced alveolar soft part sarcoma (ASPS), treatment with sunitinib resulted in radiographic responses in 2 patients and stable disease in 1. ⁴⁰ A biochemical analysis of frozen samples from these patients showed activation of VGFR2 in only 1 tumor, whereas other kinases sensitive to sunitinib were active, suggesting that the VEGFR pathway may not be a therapeutic target in this disease. However, early clinical studies of the more-selective VEGFR inhibitor cediranib have shown this agent to have significant activity against ASPS. ⁴¹ Of 7 patients treated in 2 studies, 4 experienced partial responses, 2 experienced minor responses, and 1 had stable disease. These observations suggest that further study of the role of VEFGR inhibition in ASPS is warranted.

Other small molecule inhibitors of VEGFR, including brivanib and pazopanib, are currently in phase II and III sarcoma-specific studies, respectively. Results from these trials may identify other sarcoma subtypes that may be particularly sensitive to the VEGFR blockade.

Mesenchymal-Epithelial Transition Factor Pathway

The mesenchymal-epithelial transition factor (MET) is a receptor tyrosine kinase that transduces extracellular signals mediated by hepatocyte growth factor (HGF) to the cytoplasm and nucleus through canonical signaling pathways. HGF is commonly produced by stromal cells and activates MET on epithelial cells, leading to cellular proliferation, survival, and migration. Although activating mutations in MET have been described in gastric, lung, and renal cancers, these alterations have not been observed in sarcomas. However, the coexpression or juxtaposition of HGF and MET in mesenchymally derived sarcoma cells could lead to autocrine or juxtracrine stimulation and promote tumor growth and metastasis. Investigational agents that could potentially disrupt the HGF/MET axis, including anti-HGF antibodies, anti-MET antibodies, and small molecule MET kinase inhibitors, are currently in clinical development. ⁴²

MET expression and activation has been shown in several sarcoma subtypes, including clear cell sarcoma (CCS), ASPS, alveolar rhabdomyosarcoma (aRMS), and osteosarcoma. ^43–54 With CCS and ASPS, MET expression seems to be driven at least partly by the activity of the MITF or TFE3 transcription factor oncogenes. In CCS, MITF is overexpressed as a direct transcriptional target of the pathognomonic EWRS1–ATF1 fusion gene product, which binds and activates the MITF promoter. MITF then induces expression of various genes, including MET. CCS cell lines and tumors also express high levels of HGF, and inhibition of HGF/MET binding or of MET activity itself significantly impairs CCS proliferation in vitro and in vivo. Similarly, TFE3, the product of the gene characteristically translocated in ASPS, has been shown to transcriptionally upregulate MET expression. These findings have provided the rationale for a phase II multi-institutional study of the investigational MET inhibitor ARQ197 in patients with advanced CCS and ASPS. With accrual ongoing, a partial response was reported in 1 patient with advanced CCS, and prolonged disease stability noted in several patients with ASPS, although this may be difficult to distinguish from the natural history of this often indolent disease.

MET has also been shown to be a transcriptional target of the t(2;13) PAX3–FOXO1A fusion gene product characteristic of aRMS. ^46,48 In a genetically engineered mouse model, PAX3–FOXO1A induced expression and ligand-independent activation of MET, ⁵⁵ which is essential to RMS cell line proliferation and tumor growth. ⁵³ MET is commonly expressed in human osteosarcomas, ⁴³ and its forced overexpression can transform osteoblasts into tumorigenic osteosarcoma-like cells that are dependent on MET kinase activity. ⁵⁰ Whether inhibition of MET activity plays a role in the clinical management of these diseases remains to be studied in prospective clinical studies.

Insulin-Like Growth Factor 1 Receptor Pathway

Insulin-like growth factor 1 receptor (IGF1R) activity provides key antiapoptotic/prosurvival signals to cancer cells in many model systems. Elevated IGF1 or IGF2 levels have been correlated with cancer risk and risk for cancer progression, and numerous preclinical models have validated the importance of this pathway for in vitro cell growth. In sarcomas, this is particularly apparent in aRMS, in which IGF1R is a transcriptional target of the oncogenic PAX3–FOXOA1 gene product, and in Ewing's sarcomas, which are highly sensitive to IGF1R inhibition in vitro, especially in combination with conventional chemotherapy. Other sarcoma subtypes overproduce IGF2, which can result in symptomatic hypoglycemia (Doege-Potter Syndrome), notably in patients with solitary fibrous tumors and GISTs. ^56,57

Intriguing activity of IGF1R antagonistic antibodies against Ewing's sarcoma has been observed in phase I clinical studies. In a phase I study of AMG479, one patient experienced a complete response for more than 28 months, and another experienced a partial response. ⁵⁸ Phase I and II studies of IGF1R antagonists AMG479, R1507, IMC-A12, SCH 717454, and CP-751,817, either alone or in combination with other agents, are currently underway for patients with sarcomas. IGF1R overexpression and activation have been observed in “wild-type” GISTs (tumors without identifiable mutations in KIT or PDGFR-alpha), leading to the design of clinical studies proposed to investigate the role of IGF1R inhibition in this genetically selected subset of tumors. ⁵⁹

Mammalian Target of Rapamycin Pathway

Signals from the cell surface transduced by diverse receptor tyrosine kinases converge on common cytoplasmic signaling pathways, including the serine/threonine kinase mammalian target of rapamycin (mTOR). Activation of mTOR plays critical roles in the control of cell metabolism and protein translation, integrating signals from PI3 kinase, hypoxia, amino acid and nutrient concentration, cell stress, and other modifiers of the cellular milieu. ⁶⁰ Additionally, mTOR is activated in malignancies through various genetic mechanisms, including amplification or mutation of upstream stimulatory proteins, such as growth factors, receptors, PI3 kinase, LKB1, and RHEB, or loss or inhibitory mutations of tumor suppressors, such as PTEN, TSC1, and TSC2. ⁶¹ Because of its key role as an integrator of proliferative and survival signals, inhibitors of mTOR have been studied for efficacy in the treatment of malignancies, and 3 are currently available commercially: sirolimus (approved for post-renal transplant immunosuppression), temsirolimus, and everolimus (CCI-779 and RAD001, respectively; both approved for treatment of advanced renal cancer), with the latter 2 being derivatives of sirolimus, or “rapalogues.” These drugs and the investigational agent ridaforolimus (AP23573) inhibit mTOR through interaction with FKBP12 in the mTORC1 complex, and all have been studied for their efficacy in the treatment of sarcomas.

Overall, response rates for nonselected sarcomas have been relatively low. A phase II study of temsirolimus in 41 patients with sarcoma observed only 1 partial response in a patient with fibrosarcoma. The median time to progression in this study was 2 months, ⁶² and 23 patients had grade 3 toxicities. Results of a phase I study of ridaforolimus involving 7 patients with sarcoma showed a partial response of more than 4 years in a patient with malignant mixed müllerian tumor (carcinosarcoma) and a partial response in a patient with Ewing's sarcoma. ⁶³ Six patients had a progression-free survival of greater than 6 months, prompting a sarcoma-specific phase II study involving 212 patients. ⁶⁴ This study showed 4 partial responses among 54 patients with bone sarcomas, and 1 partial response among 57 patients with soft tissue sarcoma, but overall 29% of participants experienced either a partial response or stable disease of more than 4 months.

To more fully evaluate whether ridaforolimus alters the rate of disease progression, a phase III placebo-controlled study is currently underway to study its efficacy in maintaining disease control after conventional chemotherapy.

The apparent lack of efficacy of mTOR inhibitors in the treatment of sarcoma seems at odds with in vitro models and strong rationale in subtypes driven, for example, by activating mutations in tyrosine kinase receptors such as KIT in GIST. Correlative studies have validated adequate plasma drug levels and target modulation in surrogate tissues. ^62,63 Biomarkers, such as the presence of phosphorylation of S6 in archival tumor specimens, may predict responsiveness to mTOR inhibition. ⁶⁵ Although more upstream oncogenic signals also may activate multiple parallel pathways that supersede inhibition of just one pathway controlled by mTOR. Dramatic responses to mTOR inhibition have been observed in select subtypes of sarcoma with genetic or biochemical alterations more directly tied to mTOR, such as with loss of the mTOR inhibitory proteins TSC1 and TSC2 in malignant PEComas. ⁶⁶ These findings suggest that mTOR inhibitors may have substantial activity in the appropriate histologic or genetic context.

Lastly, compensatory upregulation of AKT signaling has been reported as a possible mechanism of resistance after treatment with rapalogues. ^67,68 In certain cell models, this may be blocked by inhibition of IGF1R signaling, ⁶¹ and clinical studies of anti-IGF1R antibodies in combination with mTOR inhibitors against sarcomas are currently underway. Additionally, the efficacy of combined PI3 kinase/mTOR inhibitors and mTORC1/mTORC2 inhibitors will be interesting to evaluate in sarcomas, because drugs of this kind are currently in early clinical development and may abrogate feedback upregulation of AKT.

Conclusions

The paradigm of oncogene addiction is widely believed to explain why the presence of mutations in key driver genes predicts for clinical response in patients treated with cognate molecular therapeutic agents. This paradigm has held true for several sarcomas and other connective tissue tumors, including GIST, DFSP, and perhaps also PVNS. However, additional targets were recently identified that reflect key roles in the biology of the tumor, even when the targets themselves are not directly mutated. This class of targets, which may be termed lineage-addicted, also seem to predict for response to treatment in diseases, such as GCT of bone and a subset of angiosarcomas. Finally, another set of targets seems to exist with roles that are not well understood, either because they operate in nonredundant pathways in which the activating mutations remain to be identified, or because their obligatory role in the tissue of origin is not well understood. Why the tumor behavior depends on these pathways is important to understand, because how to predict which individuals will experience a dramatic response to a drug is not yet known. In an age of personalized medicine, biomarkers predictive of response (or nonresponse) will be critical to optimize outcomes and direct the appropriate use of high-cost therapeutic agents.