Targeting tumor-stromal interactions in bone metastasis

Abstract

Bone metastasis is a frequent occurrence in late stage solid tumors, including breast cancers, prostate or lung. However, the causes for this proclivity have only recently been elucidated. Significant progress has been made in the past decade toward understanding the molecular underpinnings of bone metastasis, and much of this research reveals a crucial role of the host stroma in each step of the metastatic cascade. Tumor-stromal interactions are crucial in engineering a pre-metastatic niche, accommodating metastatic seeding, and establishing the vicious cycle of bone metastasis. Current treatments in bone metastasis focus on latter steps of the metastatic cascade, with most treatments targeting the process of bone remodeling; however, emerging research identifies many other candidates as promising targets. Host stromal cells including platelets and endothelial cells are important in the early steps of metastatic homing, attachment and extravasation while a variety of immune cells, parenchymal cells and mesenchymal cells of the bone marrow are important in the establishment of overt, immune-suppressed metastatic lesions. Many participants during these steps have been identified and functionally validated. Significant contributors include integrins, (α_vβ₃, α₂β₁, α₄β₁), TGFβ family members, bone resident proteins (BSP, OPG, SPARC, OPN), RANKL, and PTHrP. In this review, we will discuss the contribution of host stromal cells to pre-metastatic niche conditioning, seeding, dormancy, bone-remodeling, immune regulation, and chemotherapeutic shielding in bone metastasis. Research exploring these interactions between bone metastases and stromal cells has yielded many therapeutic targets, and we will discuss both the current and future therapeutic avenues in treating bone metastasis.

1. Introduction

Metastasis to the bone is one of the most common and devastating complications in patients with advanced cancers of the breast, prostate or lung. Also manifest in other cancers (thyroid, renal cell, colon, esophageal or rectum), bone metastasis is a pathological process notable for the ability of tumor cells to exploit endogenous stromal environments and coerce other host cell types into cooperation. Despite gross morphological differences between the bone and the soft tissues from which bone metastases originate, the underlying molecular interactions between disseminated tumor cells (DTCs) and bone tissues make bone a particularly attractive niche for the growth of metastatic lesions. Lending credence to this idea, many of the genes associated with breast cancer metastasis to bone are surface interaction proteins or secreted growth factors, demonstrating that mechanisms extrinsic to the tumor cells are paramount to metastatic progression (Kang, et al., 2003). From instructing the pre-metastatic niche to establishing a vicious cycle of bone remodeling and tumor growth, tumor-stromal interactions are crucial to metastatic expansion in bone. Although this tumor-stromal relationship endows resistance to many conventional therapeutic approaches, exploiting the crosstalk between tumor cells and the bone stromal compartment may provide an effective means to thwart cancer metastasis to bone.

Myriad evidence has emerged within the last decade which indicates the need to target early tumor-stromal interactions (pre-metastatic niche conditioning, seeding and dormancy) to best treat bone metastasis. Current treatments target overt, established metastases and the symptoms associated with increased bone remodeling (Ell & Kang, 2012; Roodman, 2004; Weilbaecher, Guise, & McCauley, 2011). Meanwhile, recent research implicates multiple novel therapeutic opportunities within the priming of the pre-metastatic niche, metastatic seeding, micrometastatic dormancy and immune surveillance (Catena, et al., 2010; Sipkins, et al., 2005) (Table 1). Knowledge regarding the later steps of metastasis, such as bone-remodeling and the establishment of an immune-suppressive environment, has also advanced in recent years and future treatments may strive to transform overt metastasis into a chronic, treatable condition. Furthermore, select patients are receptive to current immune-modulatory therapies, yet the factors that govern a positive response are unknown and require further research to elucidate.

Table 1.

Current and potential therapeutics targeting tumor-stromal interactions in bone metastasis

Therapeutic Target	Mechanism	Agent	Reference
FDA-approved therapies
Prenylation	Osteolysis inhibitor	Zoledronic Acid	Rosen, et al., 2001
RANKL	Osteoclastogenesis inhibitor	Denosumab	Fizazi, et al., 2009
CTLA4	Cytotoxic T cell activator	Ipilimumab	Hodi, et al., 2010
Cytotoxic cells	NK cell activator	Interleukin 2	Rosenberg, et al., 2004
Potential therapies
EGFR	Osteoclastogenesis inhibitor	Gefitinib	Normanno, et al., 2005
PD1	Immunosuppression antagonist	CT-011	Dulos, et al., 2012
CD137	Immunosuppression antagonist	BMS-663513	Simeone & Ascierto, 2005
Gal3	CTC Attachment ligand mimic	Lactulose-L-Leucine	Heimburg, et al., 2006
c-MET	Pre-metastatic conditioning Inhibitor	Tivantinib	Previdi, et al., 2011
CXCR4	Homing and dormancy inhibitor	AMD3100	Shiozawa, et al., 2011
CCL2	Homing and growth antagonist	Carlumab	Loberg, et al., 2008
ET1	Osteogenesis antagonist	Atrasentan	Yin, et al., 2003
Metastatic cells	Suppressive pathway activator	Interferon 7	Bidwell, et al., 2012
PSA+ cells	PSA based vaccine	PROSTVAC-VF	Kantoff, et al., 2010
VWF	Platelet shielding inhibitor	ARC1779	Karpatkin, et al., 1988
β3 integrin	Seeding and growth antagonist	RGD-mimetic/ mAb	Matsuura, et al., 1996
Cathepsin B	Osteolysis inhibitor	CA-074	Withana, et al., 2012
c-FMS	Osteoclastogenesis inhibitor	Ki-20227	Ohno, et al., 2006
SRC	Proliferation inhibior	Dasatinib	(Zhang, et al., 2009)
Cathepsin K	Osteolysis inhibitor	Odanacatib	Le Gall, et al., 2007
TGFβ	Osteoclastogenesis inhibitor	Ki26894, SD-208, LY2109761	(Ehata, et al., 2007;Korpal, et al., 2009; Mohammad, et al., 2011)

2. Educating the bone: forming the pre-metastatic niche

The discovery that primary tumor cells are able to instruct the adaptation of foreign sites for future colonization represents a paradigmatic shift in cancer research. Rather than a stochastic process through which a certain proportion of CTCs are able to colonize sites of distant metastasis, the ability of the primary tumor to influence future routes of metastasis both supports Stephen Paget’s well established seed and soil hypothesis (Paget, 1989), and presents new opportunities for therapeutic intervention. Considerable evidence points to the formation of perivascular pre-metastatic niches in the lung by bone-derived cells such as Toll-like receptor 4⁺ (TLR4⁺) myeloid cells or vascular endothelial growth factor receptor-1⁺ (VEGFR⁺) hematopoetic progenitor cells (Hiratsuka, et al., 2008; Kaplan, et al., 2005). Similar mechanisms are observed in liver metastasis as cytokine secretions from metastatic cells rapidly upregulate intercellular adhesion molecules in the liver and allow for enhanced adherence (Khatib, et al., 1999). Evidence also exists for the presence of pre-metastatic conditioning in bone marrow (Kelly, et al., 2005; Peinado, et al., 2012), however the results of many studies suggest that the bone already possesses many features ideal for fostering metastatic colonization.

Peinado and colleagues recently demonstrated the ability of tumor-derived exosomes, or small lipid coated vesicles containing tumor-derived proteins, to educate bone marrow derived cells, thus facilitating metastasis to the bone or lung in a melanoma model (Peinado, et al., 2012). This conditioning was attributed to increased Met signaling in bone marrow stromal cells. Likewise, inhibition of the hepatocyte growth factor (HGF) receptor (c-MET) interaction has shown promise in weakening the migratory phenotype of breast cancer metastatic cells, and in vivo administration of Tivantinib is able to significantly delay bone metastatic progression (Previdi, Abbadessa, Dalo, France, & Broggini, 2011). Changes in bone marrow structural components has also been observed (Figure 1), as Heparanase (HPSE) secreted by primary tumors increases bone degradation in the absence of metastatic lesions (Kelly, et al., 2005).

Figure 1. Primary tumor-derived factors predispose the bone stroma for colonization.

Tumor secretion of enzymes, growth factors and cytokines alters the distant bone environment to accommodate metastatic seeding. Tumor-derived exosomes enhance MET signaling in bone marrow-derived cells while unknown factors inhibit osteoblast differentiation by upregulating DKK1 plasma levels. Proteases such as HPSE alter the bone stroma by digesting extracellular matrix proteins.

Understanding the role of mesenchymal stem cells (MSCs) in tumor-stromal interaction has become an important field of study (Koh & Kang, 2012). Patients with advanced lung or breast cancer, but without bone metastasis, exhibit changes in MSC plasticity which predisposes the bone toward enhanced osteolysis (Fernandez Vallone, et al., 2013). This predisposition was accompanied by altered serum levels of Dickkopf 1 (DKK1), an inhibitor of osteoblast differentiation, reflecting perturbations in bone marrow homeostasis prior to metastatic seeding (Fernandez Vallone, et al., 2013).

Contrary to the hypothesis of pre-metastatic conditioning, the existence of sites permissive for tumor engraftment in healthy mice has also been established. Work by Sipkins et al utilized in vivo imaging to show that both leukemic cells and hematopoetic stem cells (HSCs) home to discrete bone marrow sites expressing stromal-derived factor-1 (SDF1/ CXCL12) and E-selectin (SELE) (Sipkins, et al., 2005). Cell homing and attachment was reduced by 80% upon ablation of SDF1 and 20% upon loss of SELE. Implicated elsewhere in metastasis, cell adhesion molecules VCAM1, ICAM1 and PECAM-1 were not directly associated with metastatic seeding. This work demonstrates that bone marrow may not be conditioned as extensively as in pulmonary metastasis, and rather that the same mechanisms that govern HSC homing in healthy individuals are co-opted by tumor cells.

3. Metastatic seeding: survival in circulation, homing, and attaching to bone parenchyma

Metastatic cells are especially vulnerable during transit from the primary tumor to distant metastatic sites. Selective pressures placed on metastatic cells during seeding and extravasation results in an high attrition rate—only an estimated 0.2% of experimentally introduced circulating tumor cells (CTC) successfully accomplish distant colonization (Chambers, Groom, & MacDonald, 2002). During the traverse from primary tumor to bone marrow, circulating tumor cells (CTCs) must both evade immune surveillance and breach the normal vascular endothelium. This process is accomplished by co-opting circulating platelets and leukocytes, avoiding recognition by immune cells, and deploying immune-like strategies to attach and extravasate into foreign sites (Figure 2).

Figure 2. Survival, seeding and arrest are accomplished by co-opting immune cells and mimicking immune-based strategies of intravasation.

Circulating tumor cell survival is enhanced by platelet secretion of TGFβ and formation of platelet aggregates. In a multistep cascade, metastatic cells home to the bone upon exposure to bone endogenous proteins such as SDF1, SPARC and CCL2. Initial attachment to the endothelium is then mediated by lectin-family proteins, leading to integrin clustering. Engagement of integrins to bone resident proteins enhances both migration and survival, allowing for trans-endothelial extravasation and colonization.

Survival

Survival in circulation is attributed to a combination of cell intrinsic programs, such as decoupling of the anoikis pathway (Demers, et al., 2009), and interactions with circulating stromal cells, including platelets and natural killer (NK) cells. The role of platelet adherence in metastasis was recognized early (Gasic, Gasic, & Stewart, 1968). Multiple genetic deficiency models have identified crucial molecular mediators of CTC interaction with platelets; these interactions are largely dependent on integrin, membrane glycoprotein complex Gp11b-Gp111a, and platelet selectin (SELP) engagement to ligands on metastatic cells. Bone metastasis in β₃ integrin-null mice exhibited a 95% decrease in bone tumor burden (Bakewell, et al., 2003), and antibody blocking of the β₃ ligand Fibronectin (FN) demonstrated similar reductions. Antibodies to mouse von Willebrand factor (VWF), the ligand for the Gp11b-Gp111a complex, demonstrate a 50–75% reduction in metastatic tumor burden that can be reconstituted through infusion with human platelets (Karpatkin, Pearlstein, Ambrogio, & Coller, 1988). Selp-deficient mice also show attenuated metastatic progression, a result which the authors attribute to reduced platelet shielding (Y. J. Kim, Borsig, Varki, & Varki, 1998).

Surface shielding of CTCs from NK cell-mediated lysis is thought to be the major functional role for platelets during metastasis, however recent studies have begun to challenge this notion. While metastatic cells sensitive to NK cell-mediated lysis are more efficient in seeding when bound to platelets (Nieswandt, Hafner, Echtenacher, & Mannel, 1999; Palumbo, et al., 2007; Palumbo, et al., 2005), models of breast cancer and melanoma show platelets enhance metastasis independently of NK cells (Coupland, Chong, & Parish, 2012). P2Y₂ receptor is an endothelial receptor activated by platelet-derived ATP which in turn facilitates trans-endothelial migration of CTCs (Schumacher, Strilic, Sivaraj, Wettschureck, & Offermanns, 2013). Perhaps most significantly, platelets in contact with CTCs during transit are sources of both TGFβ1 and activators of the NF-κB pathway. These factors are responsible for a transition to a mesenchymal morphology and enhanced metastasis (Labelle, Begum, & Hynes, 2011).

Homing

Multiple cytokines are responsible for increased chemotaxis and homing to the bone marrow (Figure 1). The most strongly associated protein receptor, CXC chemokine receptor 4 (CXCR4), provokes actin polymerization and pseudopodia formation resulting in migration upon exposure to bone endothelial cell secretions (A. Muller, et al., 2001). CXCR4 is normally expressed on HSCs, and this expression is important for guiding marrow engraftment of HSCs to organs with high levels of the CXCR4 ligand, SDF1 (Brenner, et al., 2004). In a functional screen for bone metastatic variants, CXCR4 expression was strongly selected for in bone metastatic cells (Kang, et al., 2003), and synthetic peptide inhibition of CXCR4 binding is able to reduce breast cancer metastatic burden (Liang, et al., 2004). Another ligand in the C-C class of chemokines, Chemokine C-C motif ligand 2 (CCL2), is a chemoattractant for myeloma cell homing and plays an instructive role in trans-endothelial migration to the bone marrow via the CCR2 receptor (Craig & Loberg, 2006). Supporting the role for CCL2 in bone metastasis (X. Lu & Kang, 2009a), CCR2 expression in prostate cancer correlates with progression (Yi Lu, et al., 2007), and systemic administration of CCR2-neutralizing antibodies lessens metastatic burden after intracardiac injection of prostate metastatic cells (Loberg, et al., 2007).

In addition to the influence of chemokines, secreted proteins and lipids normally expressed in bone have demonstrated a strong ability to induce bone metastatic cell chemotaxis. For example, the glycoprotein osteonectin (SPARC) acts as a chemoattractant for both breast and prostate cancer cells (Jacob, Webber, Benayahu, & Kleinman, 1999). Preliminary evidence has established a possible role for adipocytes in metastatic homing to the bone. Prostate cells preferentially home to lipid rich regions of the bone stroma; this effect was attributed to Omega-6 exposure and could be inhibited by increased Omega-3 concentrations (Brown, Hart, Gazi, Bagley, & Clarke, 2006). Furthermore, adipose tissue enhances both primary tumor growth and metastasis in an estrogen dependent manner (Elliott, Tam, Dexter, & Chen, 1992). Much work remains in elucidating how the bone marrow stroma affects tumor homing. For example, bone marrow-derived cells home to soft tissues following chemotherapy or radiation (Epperly, Guo, Shields, Zhang, & Greenberger, 2004; Ortiz, et al., 2003), how this affects bone metastasis is yet unexplored.

Attachment and escape to bone parenchyma

Numerous molecular mechanisms have been elucidated which endow bone metastatic CTCs with the ability to attach to and invade bone tissue (Figure 2). Broad classes of stromal interactions, such as integrin- and lectin-mediated attachment or protease-dependent invasion have been characterized. Whereas multiple trials with Matrix Metalloprotease (MMP) inhibitors have met with failure (Coussens, 2002), many known inhibitors of surface adhesion molecules have not yet been tested in clinical trials.

With the ability to induce mitogenic intracellular signaling via outside-in activation, integrins are broadly implicated in bone metastasis. Three major integrins have demonstrated instructive roles in bone metastatic seeding: integrins α_vβ₃, α₂β₁, and α₄β₁ (Schneider, Amend, & Weilbaecher, 2011). These integrins are expressed by bone metastatic cells and bind extracellular matrix proteins normally expressed by bone-associated cells. Following binding, pleiotropic intracellular signaling programs induce migration, invasion and proliferation.

The α_vβ₃ integrin functions in breast cancer bone metastasis through binding either osteopontin (OPN), bone sialoprotein (BSP), or CD44 (Hayashi, et al., 2007; Senger, Perruzzi, & Papadopoulos, 1989; Takayama, et al., 2005; Zohar, et al., 2000), and targeted α_v integrin-inhibitory peptide mimics are effective in attenuating both early metastatic seeding and proliferation (Shannon, et al., 2004). OPN is a major extracellular component of multiple bone tissue cells, and engagement by α_vβ₃ results in MEK-induced upregulation of matrix metalloproteinase 9 (MMP9) (Chen, et al., 2009). Additionally, α_vβ₃ binding to serum OPN induces the migration of B16 melanoma cells, and forced expression is able to redirect a spontaneous lung metastasis model to the bone marrow (Hayashi, et al., 2007; Sloan, et al., 2006). BSP is a distinct α_vβ₃ ligand normally expressed in bone tissue which is upregulated in bone metastatic lesions compared to metastases in other organs (Waltregny, et al., 2000). BSP, MMP2 and α_vβ₃ may form a tripartite complex responsible for increased invasiveness of thyroid carcinoma cells (Karadag, Ogbureke, Fedarko, & Fisher, 2004) which can potentially be abrogated by overexpression of tissue inhibitor of MMP2 (TIMP2) (Yoneda, et al., 1997).

The α₂β₁ and α₄β₁ integrins bind Collagen 1 (COL1) and VCAM1 in bone metastasis. COL1 is the major structural component of the bone matrix; when bound by α₂β₁, subsequent RhoC activation primes cell morphology for invasion (Hall, 2006; Hall, et al., 2008). VCAM1 constitutively expressed by bone endothelial cells binds α₄β₁ (Miyake, et al., 1991), and forced expression of α₄β₁ is sufficient to endow Chinese hamster ovary (CHO) cells with bone metastatic ability. Preclinical evidence suggests that antibody blocking of this interaction is able to suppress the establishment of osteolytic myeloma (Matsuura, et al., 1996).

Whereas integrins mediate mitogeneic and migratory signaling, the initial attachment of metastatic cells to bone endothelial cells is largely attributed to the lectin class of protein adhesion molecules (Barthel, et al., 2013; Lehr & Pienta, 1998). Glycosylated ligands present on CTCs such as PSGL1 and CD44 engage endothelial selectin (SELE), and mediate initial cell attachment to and subsequent rolling along bone endothelial cells in both HSCs and CTCs (Barthel, et al., 2013; Dimitroff, Lee, Rafii, Fuhlbrigge, & Sackstein, 2001; Katayama, et al., 2003). This initial attachment facilitates integrin clustering and subsequent trans-endothelial migration. Two other selectins, SELP and leukocyte selectin (SELL), are also shown to engage metastatic ligands and mediate CTC-endothelial cell attachment through the action of platelets or circulating leukocytes (Borsig, Wong, Hynes, Varki, & Varki, 2002). Galectin 3 (GAL3), an additional carbohydrate binding protein, is able to engage in heterotypic interactions between CTCs and bone endothelium (Heimburg, et al., 2006; Lehr & Pienta, 1998; Pienta, et al., 1995). Therapeutics targeting lectin family proteins are able to diminish metastasis in both animal models and human patients. The GAL3 ligand mimic lactulose-L-leucine has shown therapeutic promise in models of breast, prostate and lung metastasis to bone (Glinskii, et al., 2012; Heimburg, et al., 2006), while the mechanism of cimetidine-induced survival in colorectal patients was attributed to decreased SELE expression (K. Kobayashi, Matsumoto, Morishima, Kawabe, & Okamoto, 2000).

4. Metastatic niches and metastatic dormancy

The existence of dormant cells in bone metastasis has long been hypothesized; however, research on the molecular regulation of metastatic dormancy only started to gain momentum in recent years (Aguirre-Ghiso, 2007). Existing knowledge regarding dormancy indicates that metastatic quiescence is largely dependent on the stromal environment and specifically on the niche where the DTCs reside. How stromal cells affect cell dormancy and the extent to which either the HSC niche or perivascular niche regulates cell proliferation is a promising new avenue for the development of novel therapeutics.

Early work on cancer depicted metastasis as the final step in primary tumor progression, however, clinical evidence has established that bone metastasis can be an early event in the progression of cancer. Using cytokeratin staining of bone marrow aspirates, studies have shown that DTCs are present in patients with locally restricted prostate or breast cancer (Banys, et al., 2011; Braun, et al., 2005; Hüsemann, et al., 2008; Sänger, et al., 2011; Solakoglu, et al., 2002). The link between DTCs and metastasis is not straightforward as detection rates of bone marrow DTCs in patients with localized disease (55%) do not reflect the overall relapse rates (12–30%) (Melchior, et al., 1997), indicating that exit from dormancy cannot be achieved by all DTCs within the bone marrow. Reflecting the complexity of connection between DTC and metastasis, substantial proportion of CTCs lose epithelial traits such as cytokeratins, via the epithelial to mesenchymal transition (Kang & Pantel, 2013; Yu, et al., 2013), and the presence of epithelial DTCs in bone marrow may not accurately depict the most invasive population (Gradilone, et al., 2011).

Programs mediating dormancy

Current evidence suggests that metastatic dormancy is the result of both a permissive endogenous bone marrow environment as well as a specific reaction to the presence of invading metastatic cells (Figure 3). For example, bone endothelial cells normally secrete basic fibroblast growth factor (FGF2) which induces breast cancer cell dormancy. FGF2-stimulated cells are able to maintain a dormant state upon α₅β₁ attachment to fibronectin and will otherwise undergo apoptosis (Korah, Boots, & Wieder, 2004). Bone endothelial cells also express Duffy antigen receptor for chemokines (DARC) which causes growth arrest upon binding to KAI1 expressed by metastatic cells (Bandyopadhyay, et al., 2006). Alternately, the identification of several metastasis suppressor genes has demonstrated that a stromal reaction to tumor cells induces dormancy. Kisspeptin-1 (KiSS1) is a G-protein-coupled receptor ligand which is secreted by tumor cells and processed by stromal cells which functions to suppress metastasis (Lee, et al., 1996; Nash, et al., 2007). Furthermore, the serine protease inhibitor Maspin is responsible for attenuating uPA activity on the surrounding stroma (Cher, et al., 2003). Taken together, these data indicate that perturbations in the normal bone environment may induce dormancy.

Figure 3. Metastatic dormancy is dependent on bone marrow stromal elements.

Two distinct niches are host to invading tumor cells in the bone marrow: the hematopoetic stem niche or the perivascular niche. In either niche, the bone microenvironment invokes tumor dormancy through growth factor secretion, activation of metastasis suppressor proteins, or binding to anti-proliferative signals. Metastatic cells respond and proliferate by secreting proteases or growth factors to accommodate growth and niche remodeling.

Regardless of the programs which induce dormancy, survival and exit from dormancy is achieved through a complex series of steps such as adaptation of dormant cells, changes in systemic endocrine levels by stress, or the effects of distant established tumors (Ben-Eliyahu, Shakhar, Page, Stefanski, & Shakhar, 2000; McAllister, et al., 2008). For example, distant metastatic cells are able to mobilize bone marrow endothelial cells through OPN secretion. Mobilized cells migrate into dormant lesions and lead to proliferation (McAllister, et al., 2008). Alternately, formation of a complex between urokinase type plasminogen activator (uPAR) and the dormancy-sustaining α₅β₁-Fn interaction may be responsible for an exit from dormancy (Aguirre Ghiso, Kovalski, & Ossowski, 1999). The dormancy state of DTCs is further influenced by the balance of ERK1/2 and p38α/β signaling activity (Sosa, Avivar-Valderas, Bragado, Wen, & Aguirre-Ghiso, 2011). In bone metastasis, early osteoclast activity is potentially important, and engagement of α₄β₁ to VCAM1 expressed on osteoclasts provokes growth of indolent breast cancer metastases (Xin Lu, et al., 2011). The α_vβ₃ integrin is also important in tumor proliferation and angiogenesis (Nemeth, et al., 2003). As a potential ligand of α_vβ₃ integrin, periostin is associated with breast cancer metastasis (Sasaki, et al., 2003), and enhances endothelial cell vascularization as well as tumor intrinsic survival by binding α_vβ₃ (Bao, et al., 2004).

Members of the TGFβ family are known to have opposing roles in tumor growth. Bone morphogenic protein 7 (BMP7) secreted from endothelial cells effects a reversible cell cycle block in prostate cancer stem cells through p38 induction. Constant exposure of BMP7 leads to senescence and death (A. Kobayashi, et al., 2011). Alternatively, upon induction of angiogenesis, TGFβ1 from sprouting vascular tips is linked to an exit from dormancy and proliferation (Ghajar, et al., 2013). Administration of a TGFβ1 receptor inhibitor or genetic ablation of the TGFβ signaling pathway reduces proliferation of bone metastasis by inhibiting the production of pro-bone metastatic genes, including Interleukin-11 (IL11), Jagged1 and Parathyroid hormone-related protein (PTHrP) (Ehata, et al., 2007; Korpal, et al., 2009; Mohammad, et al., 2011; Sethi, Dai, Winter, & Kang, 2011).

Tumor cell niches

Two distinct stromal environments have been depicted as potential niches to invading metastatic cells—the hematopoetic stem cell (osteoblastic) niche and the perivascular niche. Both niches are theorized to impact metastatic survival and proliferation, albeit through distinct mechanisms. The HSC niche shares many of the same programs characterized in bone metastases, and direct competition has been observed between HSCs and bone metastatic cells (Shiozawa, et al., 2011). Recently, Shiozawa and colleagues demonstrated that metastatic cells and HSCs utilize SDF1/CXCR4 signaling for ingress and egress to the endosteal compartment. Remarkably, antagonism with either the CXCR4 inhibitor AMD3100 or recombinant GM-CSF is able to mobilize both HSCs and DTCs from bone marrow (Shiozawa, et al., 2011). Interestingly, the HSC niche, which is composed of MSCs and osteoblasts, is stimulated by many of the same signals which cause bone metastatic cells to proliferate (Adams, et al., 2007; Calvi, et al., 2003; Méndez-Ferrer, et al., 2010). The similarities between HSCs and metastatic cell behavior in the niche have profound implications on strategies for therapeutic intervention—namely, the same agents which affect HSCs may also have therapeutic values in controlling bone metastasis (Shiozawa, et al., 2011). While this increases the arsenal of drugs to treat bone metastasis, targeting the niche may also exacerbate HSC death and hematopenia. Compounding this issue, active TGFβ signaling by myeloma cells reduces the plasticity of hematopoetic stem progenitor cells and MSCs, resulting in diminished ability for hematopoesis (Bruns, et al., 2012).

Interestingly, breast or prostate cancer patients with locally restricted disease who coincidentally received radiation to lumbar or thoracic vertebrae experienced a significant reduction in subsequent metastasis to these regions (Bagshaw, Kaplan, Valdagni, & Cox, 1992; Hercbergs, Werner, & Brenner, 1985). The cause for the phenomenon remains unidentified, yet may be linked to changes in HSC niche composition.

Angiopoietin-2 (ANG2) is a potential regulator of the HSC niche and metastatic dormancy. Expression of ANG2 is upregulated in metastatic prostate and breast tissue (Imanishi, et al., 2007; Lind, et al., 2005), but expression has not been tied to metastatic dormancy and the HSC niche. ANG2 and ANG1 directly compete for binding to the TIE2 receptor (Maisonpierre, et al., 1997). Upon binding, TIE2 signaling regulates HSC quiescence (Arai, et al., 2004), sensitizes endothelial cells to inflammation (Fiedler, et al., 2006), and disrupts angiogenesis (Maisonpierre, et al., 1997). Similar pathways are instrumental in metastasis, and recently developed TIE2 inhibitors (rebastinib) may be effective in attenuating bone metastasis (Lamontanara, Gencer, Kuzyk, & Hantschel, 2013).

The perivascular niche has been characterized as an alternate site for bone metastatic colonization. This niche bears similarity to the sites of metastatic seeding in pulmonary, brain, or hepatic metastasis: endothelial progenitor cells, pericytes and other cell types are recruited to form a supportive stroma around the tumor cells. High resolution imaging of DTCs directly associated with bone marrow microvasculature after intracardiac injection provides direct evidence for tumor cells residing in the perivascular compartment. Furthermore, engineered culture conditions mimicking bone microvasculature were able to support metastatic cell attachment and induce dormancy (Ghajar, et al., 2013). Only cells in direct contact with bone endothelial cells were negative for Ki67 staining, and the metastasis suppressor Thrombospondin-1 (TSP1) was shown to be causal in this dormancy by inhibiting angiogenic programs (Ghajar, et al., 2013). Many protein dynamics which have been implicated in DTC progression independent of niche specification are likely important in the perivascular niche, for example periostin and VEGF are strong candidates for master regulators of this niche.

5. The osteolytic/osteogenic axis in bone metastasis

The bone is best known for two of the most vital roles in normal physiology: structural support and hematopoesis. The ability of bone metastases to subvert both processes is the primary cause of morbidity and mortality in cancer patients with bone metastasis. The bone is in a constant state of dynamic remodeling by a balance between osteolytic and osteogenic programs. Metastatic cancer cells often successfully exploit the normal bone homeostatic process and tip the equilibrium toward either bone lysis or bone growth to facilitate the formation of bone metastases (Ell & Kang, 2012; Mundy, 2002; Weilbaecher, et al., 2011).

Osteolytic lesions

The paracrine regulatory program of osteoclastogenesis controlled by metastatic cells is the most researched and targeted tumor-stromal interaction in bone metastasis. Multiple cytokines and growth factors have shown pleiotropic effects in stimulating hyperactive osteoclast activity and promoting osteolytic metastasis (Ell & Kang, 2012; Weilbaecher, et al., 2011).

The most prominent cytokine inducer of osteoclastogenesis is receptor activator of NF-κB ligand (RANKL) (Mundy, 2002; Weilbaecher, et al., 2011). RANKL binding to its receptor (RANK) expressed on both osteoclast precursors and metastatic cells activates NF-κB signaling (Eghbali-Fatourechi, et al., 2003), leading to both the transcription of key osteoclastogenic factors (Park, et al., 2007) and enhanced breast cancer cell migration selectively to bone marrow (Jones, et al., 2006). Similar to other TNF family members, RANKL is bound to the extracellular matrix, and is activated and released upon cleavage by tumor-secreted MMP7 (Lynch, et al., 2005). RANKL activity is balanced in normal bone homeostasis by osteoprotegerin (OPG), a decoy receptor for RANKL secreted by osteoblasts. Experimental introduction of OPG results in a pronounced decrease in osteolytic tumor growth (Morony, et al., 2001). Multiple other cytokines or ligands have also shown importance in osteoclast activation; TNFα increases the preosteoclast pool in the bone marrow and synergistically complements RANKL-induced osteoclast activation (Lam, et al., 2000). Tumor-derived Jagged1 can directly stimulate osteoclast differentiation by activating Notch signaling (Sethi, et al., 2011). Both Granulocyte macrophage-colony stimulating factor (GM-CSF/CSF2) and Macrophage-colony stimulating factor (M-CSF/CSF1) are also shown to be necessary for the maturation of preosteoclasts into osteoclasts (Lacey, et al., 1998; Park, et al., 2007). M-CSF receptor (c-FMS) antagonists have shown a significant ability to suppress osteoclast differentiation and osteolytic bone destruction (Ohno, et al., 2006).

Following osteoclast maturation, the “vicious cycle” of lytic bone metastasis takes place whereby degradation of the bone matrix by osteoclasts releases various growth factors such as IGFs, TGFβ, EGFs and FGFs, spurring tumor cell growth and further promoting osteoclast activation (Hauschka, Mavrakos, Iafrati, Doleman, & Klagsbrun, 1986; Mundy, 2002).

Dynamic in vivo imaging coupled with genetic and pharmacological inhibition of the TGFβ pathway showed that TGFβ signaling is most important in the early establishment of breast cancer metastatic lesions (Korpal, et al., 2009), and a milestone study by Yin and colleagues demonstrated that blockade of TGFβ signaling in tumor cells inhibits PTHrP secretion and reduces osteolysis (Yin, et al., 1999). In addition to inducing RANKL production from osteoblasts to increase osteoclastogenesis, the pro-metastatic function of PTHrP has been attributed to other important functions. PTHrP increases the expression of CCL2 which functions to promote osteoclast activation and bone metastasis in both breast cancer prostate cancers (X. Lu & Kang, 2009b; Y. Lu, et al., 2008; Y. Lu, et al., 2007). PTHrP is also responsible for activating the transcription of the Wnt inhibitor DKK1. DKK1 is highly expressed in myeloma, a purely lytic cancer type, and prevents the maturation of osteoblasts (Tian, et al., 2003). Paradoxically, patients with high PTHrP levels in primary tumors experience increased survival, suggesting the potential differential role of PTHrP in different tumor microenvironments (Henderson, et al., 2006).

Other factors affecting osteoclastogenesis have also emerged. For example, the proteases MMP1 and ADAMTS1 were found to release EGF ligands from the metastatic cell matrix and inhibit osteoblast differentiation from MSCs, decoupling the ability to regulate osteoclast activation (X. Lu, et al., 2009). Many other inducers of RANKL have also been discovered, including Macrophage inflammatory protein 1α (MIP1α) (Oba, et al., 2005; Tsubaki, et al., 2007), and IL8 (Bendre, et al., 2003). Tumor associated fibroblasts are also sources of RANKL, and secretion of RANKL is able to induce osteoclast differentiation from tumor associated macrophages (Lau, et al., 2006).

Osteoblastic metastasis

Parallel to the RANKL/RANK/OPG system in osteolysis, the gene regulatory networks governed by the transcription factor RUNX2 are the most heavily implicated in osteoblastic bone metastasis. Normally expressed in mesenchymal progenitor cells during osteoblast differentiation, RUNX2 activates bone matrix protein transcription such as BSP and osteocalcin (Ducy, Zhang, Geoffroy, Ridall, & Karsenty, 1997). RUNX2 is also expressed in prostate cancer cells (Akech, et al., 2010), and is theorized to induce osteomimicry in these cells through expression of pro-metastatic, bone-specific matrix proteins (Koeneman, Yeung, & Chung, 1999) such as BSP, MMP9 and SDF1 (Baniwal, et al., 2010; Barnes, et al., 2003). Evidence for osteomimicry in advanced bone metastasis is observed in both osteolytic and osteoblastic lesions. Expression of bone specific markers likely has numerous effects in increasing invasion, angiogenesis, survival and any other numbers of functions necessary to endogenous bone cells. Proteins such as osteoblastic cadherin (CDH11) are also implicated in this role (Chu, et al., 2008).

Both Endothelin-1 (ET1), a secreted osteoblast mitogen, and Prostate specific antigen (PSA), a serine protease, have been very well established in the pathology of osteoblastic metastasis (Nelson, et al., 1995). ET1 is secreted from both breast and prostate cancers; Yin and colleagues demonstrated a causal role in increasing osteoblastic tumor burden through binding the ET_A receptor of osteoblasts, and orally available antagonists were able to reduce tumor burden (Yin, et al., 2003). A Phase III clinical trial demonstrated that ET1 antagonist Atrasentan did not decrease the risk of disease progression in metastatic prostate cancer despite reduced levels of PSA and bone alkaline phosphatase (BAP) (Carducci, et al., 2007). PSA is shown to activate growth factors such as PTHrP (Cramer, Chen, & Peehl, 1996; Iwamura, Hellman, Cockett, Lilja, & Gershagen, 1996), and clinical trials of PSA vaccines show strong promise in improving overall survival (Kantoff, et al., 2010).

Osteoclast and osteoblast activity are believed to be coupled in bone metastasis, with most bone lesions exhibiting varying degrees of both osteolytic and osteoblastic activities. It is perhaps not surprising that several factors are implicated in both osteolytic and osteoblastic metastasis, perhaps through activating different downstream genes in different cancer types. For example, TGFβ activates the expression of osteoclastogenic genes, such as Jagged1, PTHrP and IL-11 (Kang, et al., 2003; Sethi, et al., 2011; Yin, et al., 1999), as well as osteoblastic genes such as ET1.

6. Current therapies targeting bone remodeling

Therapeutics targeting the bone resorptive cycles of myeloma and breast cancer are a palliative rather than curative approach toward treatment of bone metastasis as these therapies slow osteoclast activity and thereby extend patient lifespan. Anti-resorptive therapies similarly extend time to relapse in bone when tested in patients with osteoblastic metastases, demonstrating that the cross communication between osteoclasts and osteoblasts is important to progression in both subtypes of bone metastasis (Figure 4). Of the osteoclast inhibitors, two have shown efficacy through Phase III clinical trials: bisphosphonates, and Denosumab (anti-RANKL antibody). Additional pharmacological agents have demonstrated promise in preclinical experiments but have yet to be tested in clinical trials.

Figure 4. The vicious cycle of osteolytic metastasis offers many targets for therapeutic intervention.

Tumor cell induction of RANKL/ GM-CSF/ M-CSF results in fusion of osteoclast precursor cells into multinucleated osteoclasts. Enhanced osteoclast-mediated lysis of the bone matrix releases cytokines such as TGFβ, which spur tumor cell growth and stimulate the expression of pro-metastatic factors such as Jagged1 and PTHrP from tumor cells to further induce osteoclast differentiation. Exposure to TNFα increases the pre-osteoclast pool, while osteoblasts are also important mediators of the tumor-osteoclast crosstalk.

The activity of bisphosphonates in diminishing bone resorption has long been known (van Holten-Verzantvoort, et al., 1987), and has become a standard of care in the treatment of osteolytic bone metastasis. Bisphosphonates, such as zoledronic acid, pamidronate or ibandronate, specifically bind to the bone matrix and are internalized by osteoclasts upon resorption. Once internalized, these drugs inhibit various metabolic processes such as prenylation, and lead to apoptosis (Mundy, 2002). Multiple bisphosphonates have been developed, and of these, zoledronic acid shows the highest efficacy in a Phase III clinical trial versus pamindronate for breast cancer and myeloma (Rosen, et al., 2001).

While treatment with bisphosphonates is considered the standard of care for bone metastasis, the mechanisms of bisphosphonate therapy are not fully understood. For example, the impact of ovarian activity in osteolytic bone metastasis for breast cancer patients was recognized, as treatment with zolendroate does not increase lifespan in pre-menopausal women (Gnant, et al., 2009) unless they are placed under ovarian suppression. Mechanisms for this response have not been identified (Steinman, Brufsky, & Oesterreich, 2012). Furthermore, limited studies have shown cytotoxic effects of bisphosphonates (Hiraga, Williams, Mundy, & Yoneda, 2001), however, the concentrations required for this effect may not replicate the local bone environment.

Compared to bisphosphonates, Denosumab and recombinant OPG demonstrate higher clinical efficacy by targeting RANKL. Both have shown similar promise in clinical trials, however only Denosumab has advanced through Phase III. Administration of Denosumab showed significant efficacy by inhibiting osteolysis in patients with prostate or breast cancer bone metastasis, based on the analysis of urinary N-telopeptide levels as a marker for bone turnover (Fizazi, Bosserman, Gao, Skacel, & Markus, 2013; Fizazi, et al., 2009). A single dosing system of recombinant OPG (AMGN-0007) in a Phase I trial demonstrated a similar sustained reduction in bone lysis (Body, et al., 2003).

A second cohort of lesser known anti-osteoclastogenic factors are also emerging for clinical trial. EGF signaling is able to induce RANKL and M-CSF secretion, and treatment with the EGFR kinase inhibitor Gefinitib suppresses osteoclast activity (Normanno, et al., 2005) and inhibits bone metastasis (X. Lu, et al., 2009). Similarly, small molecule antagonists of c-FMS can attenuate growth by 89% (Murray, et al., 2003; Ohno, et al., 2006). The vitamin K analogue from traditional medicine, plumbagin, inhibits NF-κB signaling through RANK and inhibits osteolytic bone metastasis (Sung, Oyajobi, & Aggarwal, 2011). Both Cathepsins B and K are functional in osteolytic bone metastasis, and specific inhibitors are available which show therapeutic promise in bone metastasis (Le Gall, et al., 2007; Withana, et al., 2012). Unexpectedly, the use of Inhibitor of apoptosis protein (IAP) antagonists has shown an activating effect on osteoclasts and preliminary trials showed accelerated metastatic growth in patients treated with IAP inhibitors (Yang, et al., 2012).

7. Immune regulation and immune suppression

The immune system plays a critical role in each step of tumor progression—from the original transformation to macrometastatic progression, immune suppression and/or immune evasion is required for the survival and expansion of tumors. In a successful metastatic progression, the immune system is often reconfigured to become a malignant accomplice able to assist in multiple functions. Several programs, such as NK cell- or effector T cell-mediated lysis, are antagonistic to progression. However the majority of immune cells, such as regulatory T cells (T_reg), dendritic cells (DC), MDSCs, and macrophages, can be converted to a tumor-promoting phenotype. Therapeutic strategies, such as DC vaccination or interferon therapy, which give the immune system stimulus, are currently under investigation and show clinical promise. The role of the immune system in the progression of primary tumors and dissemination has been extensively reviewed elsewhere (Shiao, Ganesan, Rugo, & Coussens, 2011; Smith & Kang, 2013). In this section, we will focus on immune targets which have shown direct impact in bone metastasis or in clinical studies involving cancers which show a predisposition toward bone metastasis.

Tumor-antagonistic functions

Both the innate and adaptive immune responses are attributed with tumor-antagonistic roles early in metastatic progression. These responses are often attenuated as the tumor develops the ability to suppress T cell activation or shield itself from NK cell activity.

Traditional antigen presenting machinery through the MHC class I and immune surveillance is sometimes active in delaying metastatic progression. A spontaneous mouse melanoma model which constitutively expresses the MHC antigen AAD has been shown to disseminate early in tumor progression but the development of lung metastasis was kept in check by immune surveillance by CD8⁺ T cells (Eyles, et al., 2010). A similar mechanism is seen in bone metastasis as lesions which rapidly proliferate in athymic mice are dormant when introduced into immune competent mice, and proliferation is restored upon CD8-specific depletion (M. Muller, et al., 1998). While platelets generally exhibit multiple pro-metastatic functions, bone marrow megakaryocytes, which are platelet precursor cells, are antagonistic to growth of metastatic lesions. This is attributed to the ability of megakaryocytes to inhibit osteoclast maturation from monocytes (Beeton, Bord, Ireland, & Compston, 2006).

Immune suppression

The establishment of immune suppression in both primary tumors and metastases is well recognized – while multiple cell-intrinsic programs exist which interfere with the antigen presenting pathway, crosstalk between tumor cells and the immune system is responsible for establishing a state of chronic inflammation and immune dysregulation (Smith & Kang, 2013). Much evidence indicates that tumor-antagonistic cells are present in the bone microenvironment, however, pervasive programs of immune suppression keep these cells at bay. Immune-modulatory therapies aimed at reactivating host immune response have demonstrated clinical success (Lode, et al., 1998; Quesada, Swanson, Trindade, & Gutterman, 1983).

Myeloid derived suppressor cells (MDSC), plasmacytoid dendritic cells (pDC), and T_reg cells are strongly implicated in bone metastasis. It is generally accepted that infiltrating macrophages, pDCs, T_reg cells, and MDSCs induce a sustained Th2 response, suppress CD8 or NK cell activity and exacerbate bone remodeling stresses (A. Sawant, et al., 2012; Anandi Sawant & Ponnazhagan, 2013; Sisirak, et al., 2013). Several major signaling hubs are identified in this network—notably, GM-CSF and IL3 are necessary for immune-suppressive activity (Salgia, et al., 2003; Young, Wright, & Young, 1991). On the other hand, Bidwell et al demonstrated that a loss of Interferon-7 (IRF7) pathways was responsible for mammary tumor progression to bone metastasis, and treatment with IRF7 was able to restore host CD8⁺ T cell or NK cell response (Bidwell, et al., 2012). The STAT6 pathway is important in CD8 T cell suppression, however this functions independently of T_reg cells (Ostrand-Rosenberg, Grusby, & Clements, 2000).

Successful metastatic lesions thrive in an environment replete with tumor-promoting myeloid cells and osteoclasts that secrete inflammatory cytokines. For example, the ECM molecule versican engages TLR2 and TLR6 to stimulate macrophage secretion of TNFα, which in turn stimulates tumor growth (S. Kim, et al., 2009). The creation of this environment can foster bone invasion, as bone-derived MSCs are able to infiltrate primary breast carcinoma and promote bone metastasis through IL17B secretion (Goldstein, Reagan, Anderson, Kaplan, & Rosenblatt, 2010). Similarly, Lysophosphatidic acid released from activated platelets in the bone microenvironment stimulates IL6 and IL8 secretion, enhancing tumor proliferation through increased osteolysis (Boucharaba, et al., 2004).

Paradoxically, tumor-associated antigen (TAA)-specific memory T cells selectively increase in the bone marrow of advanced breast cancer patients with bone marrow DTCs (Feuerer, et al., 2001). The bone marrow is also a significant reservoir of T_reg cells (Zou, et al., 2004), indicating that activity of these TAA-specific memory T cells is potentially suppressed by T_reg cells. Metastasis associated T_reg cells also specifically increase in prostate bone metastasis microenvironments, and inhibit bone resorption by exhibiting a broad immune-suppressive profile (Zhao, et al., 2012). The creation of an inflammatory microenvironment with an abundance of proliferative and angiogenic signals is conducive for both primary and metastatic cell growth. Indeed, even treatment with aspirin shows significant decreases in tumor osteolytic activity (Powles, Dowsett, Easty, Easty, & Neville, 1976).

While the establishment of immune suppression is a hallmark of most metastatic tumors, few immune-modulatory therapies have demonstrated clinical success. This type of therapy is theorized as a molecular reboot, the programs which maintain immune-suppression are temporarily disrupted allowing for either host response or response by allogenic cells. Both metastatic renal cell carcinoma and melanoma seem especially susceptible to immune-modulatory therapies and several therapies have resulted in complete relapse. In chemoresistant metastatic renal cell cancer or melanoma, treatment with interleukin-2 (IL2) provoked an anti-metastatic response in 20% of patients and is used in clinical practice currently (Rosenberg, et al., 1994). IL2 therapy was also proved effective exclusively through NK cell activity in a model of neuroblastoma bone metastasis (Lode, et al., 1998). Interferon α treatment is marginally effective in renal cell metastasis with 30% of patients showing a response (Quesada, et al., 1983). A separate trial with chemoresistant renal cell metastasis demonstrated that non-myeloablative immune suppression followed by stem cell transplantation from allogenic donors resulted in metastatic regression in multiple sites, including bone (Childs, et al., 2000). Similar responses have been anecdotally reported in advanced breast cancer patients (Eibl, et al., 1996). Taken together, these trials demonstrate promise for a specific cohort of patients, however, considerable work remains to identify patient populations by likelihood of clinical response.

Blockade of the specific molecules promoting immune suppression in advanced tumors has similarly received attention. CD-28 activates T cells when bound to CD-80 expressed on antigen presenting cells, cytotoxic T-lymphocyte associated antigen-4 (CTLA4) competes with CD-28 for CD80/86 binding and inhibits T cell activation (O’Day, Hamid, & Urba, 2007). Phase III clinical trials with Ipilimumab (Anti-CTLA4) in metastatic melanoma patients demonstrated a median lifespan of 10 months compared to 6 months for placebo (Hodi, et al., 2010). This lifespan extension is attributed to CD8⁺ T Cell activity at sites of metastasis (Demaria, et al., 2005). A similar mechanism is targeted with anti- Programmed Death-1 (PD1) or anti-CD137 antibodies, which shift T cell populations from a tumor-promoting Th2 population to a tumor-antagonistic Th1/ Th17 population (Dulos, et al., 2012; Simeone & Ascierto, 2012). Furthermore, dendritic cell vaccination in patients with metastasis demonstrates reactivation of CD4/CD8 T cell-mediated metastatic clearance (Gong, Chen, Kashiwaba, & Kufe, 1997). Further studies show that the machinery for antigen mediated clearance is intact in at least 7 of 17 patients (Kugler, et al., 2000).

8. Stromal-endowed therapeutic resistance

Bone stromal cells not only participate in the pathogenesis of bone lesions, but also make them refractory to traditional chemotherapies. Chemotherapy has no effect on dormant DTCs in the bone marrow (Braun, et al., 2000), and many groundbreaking studies have attributed this phenomenon to various stromal cell programs mediating cancer cell survival. Acharyya et al found that a chemoresistance cycle was instated when chemotherapy elicits TNFα secretion from the stroma – this then increases CXCL1/2 secretion from tumor cells which in turn attracts S100A8/9 secretion from myeloid cells to ensure survival (Acharyya, et al., 2012). Alternately, cellular contacts with components of the HSC niche or osteoclasts have also been shown to enhance tumor cell survival during adjuvant therapy (Abe, 2004). Inhibition of these tumor-stromal programs sensitizes DTCs to conventional therapeutic approaches. The use of AMD3100 to antagonize CXCR4 makes prostate or leukemic cells in the bone marrow more receptive to chemotherapy (Domanska, et al., 2012; Nervi, et al., 2009), while use of CXCR2 antagonists is similarly able to enervate the survival phenotype (Acharyya, et al., 2012).

9. Conclusions and future directions

The understanding of bone metastasis has progressed considerably in recent years, with particular emphasis on the contribution of the host to metastatic progression. Bone metastases often acquire a unique set of molecular traits which are able to temper the bone microenvironment to facilitate the growth of metastatic tumors in bone. These traits are only part of the pathogenesis of bone metastasis—contributions of associated stromal cells are active participants in every step of progression. While these interactions are potential therapeutic targets, the similarities between bone metastatic cells and the stem cells of the bone marrow may make the treatment of engrafted metastatic lesions difficult using conventional therapeutic approaches. New molecular targeting strategies should combine with immune modulation in order to achieve effective eradication of metastatic lesions with minimal side effects.

Current anti-resorptive treatments used in bone metastasis do not constitute a curative approach toward bone metastasis. Much progress needs to be made in the clinical setting to bring our improved basic understanding of bone metastasis into clinical practice, yet many hurdles exist in developing novel anti-metastasis drugs for clinical trial. Primarily, the endpoints used in treating primary tumors, such as shrinkage in tumor size, need to be adjusted for evaluating the therapeutic efficacy of targeting the metastatic process (Steeg, 2012). The clinical course of bone metastasis is highly variable, and early stage processes such as tumor seeding, dormancy, and local niche remodeling need to be assessed, mostly likely in an adjuvant setting. This also necessitates new surrogate markers for each stage of metastatic progression. While the use of urinary bone turnover markers provides an invaluable surrogate for osteolytic bone metastasis (Kamiya, et al., 2012), minimally invasive methods for the detection of CTCs or dormant metastatic cells must be developed.

Abbreviations

CTC

circulating tumor cell

DC

dendritic cell

DTC

disseminated tumor cell

HSC

hematopoetic stem cell

MSC

mesenchymal stem cell

NK

natural killer

T_reg

regulatory T cell