| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
B Ligand and Osteoprotegerin Regulation of Bone Remodeling in Health and DiseaseEndocrine Research Unit (A.E.K., S.K.), Mayo Clinic College of Medicine, Rochester, Minnesota 55905; and Metabolic Disorders Research (P.J.K.), Amgen Inc., Thousand Oaks, California 91320
Correspondence: Address all correspondence and requests for reprints to: Paul J. Kostenuik, Amgen Inc., 1 Amgen Center Drive, M/S 15-2-B, Thousand Oaks, California 91320. E-mail: paulk{at}amgen.com
 |
Abstract |
|---|
 Top Abstract I. IntroductionII. Discovery and Validation...III. Pharmacological Effects of...IV. Role and Regulation...V. Clinical Therapeutics...VI. SummaryReferences |
|---|
B ligand), a TNF family member that is essential for osteoclast formation, activity, and survival in normal and pathological states of bone remodeling. The catabolic effects of RANKL are prevented by osteoprotegerin (OPG), a TNF receptor family member that binds RANKL and thereby prevents activation of its single cognate receptor called RANK. Osteoclast activity is likely to depend, at least in part, on the relative balance of RANKL and OPG. Studies in numerous animal models of bone disease show that RANKL inhibition leads to marked suppression of bone resorption and increases in cortical and cancellous bone volume, density, and strength. RANKL inhibitors also prevent focal bone loss that occurs in animal models of rheumatoid arthritis and bone metastasis. Clinical trials are exploring the effects of denosumab, a fully human anti-RANKL antibody, on bone loss in patients with osteoporosis, bone metastasis, myeloma, and rheumatoid arthritis.
|
I. Introduction |
|---|
|
TopAbstractI. IntroductionII. Discovery and Validation...III. Pharmacological Effects of...IV. Role and Regulation...V. Clinical Therapeutics...VI. SummaryReferences |
|---|
In many pathological states, and perhaps in healthy adults, each bone remodeling cycle leads to a net loss of bone due to a modest deficiency in the amount of new bone formed relative to old bone removed (1, 2). This relative deficit in bone formation could be related at least in part to trabecular perforations that occur during the resorptive phase, leading to the loss of bone surface on which osteoblasts could deposit new bone (3). Acceleration of the bone remodeling rate, which is common in many metabolic bone diseases including postmenopausal osteoporosis, thereby leads to a greater rate of bone loss (1). Inhibition of bone remodeling has therefore become a major therapeutic strategy for preventing bone loss. This review will focus on the recent discovery of the OPG/RANKL/RANK signaling pathway (Fig. 1
), its critical involvement in the cellular regulation of bone remodeling, and the effects of bone remodeling suppression via RANKL inhibition.
|
|
II. Discovery and Validation of the OPG/RANKL/RANK Pathway |
|---|
|
TopAbstractI. IntroductionII. Discovery and Validation...III. Pharmacological Effects of...IV. Role and Regulation...V. Clinical Therapeutics...VI. SummaryReferences |
|---|
.
|
(adapted from Ref. 15). OPG is an atypical member of the TNF receptor family in that it is a secreted protein with no transmembrane domain and no direct signaling properties (4, 6). Secretion of OPG is mediated by a 21-residue signal peptide (4). OPG has three major structural motifs including four cysteine-rich TNF receptor domains, a heparin-binding domain, and two death domain homologous (DDH) regions. The DDH regions within native OPG have yet to be associated with any functional role (16), although recent evidence suggests that this domain contributes to the formation of homodimers (17). Unlike the intracellular DDH regions found in apoptosis-inducing proteins such as Fas, the DDH regions in OPG reside in an extracellular environment because OPG is secreted.
|
To enhance the pharmacological activity of native OPG, numerous constructs have been created wherein the signal peptide, heparin binding domain, and DDH domains were removed and the remaining peptide was fused to the Fc domain of human IgG1 [e.g., Fc-OPG (20, 21, 22) and OPG-Fc (23, 24, 25, 26); Fig. 2
]. The Fc fusion partner maintains the potent dimeric nature of OPG while significantly increasing its circulating half-life. The great majority of published studies describing the use of OPG in animals have relied on these types of recombinant OPG fusion proteins, whereas a few early studies demonstrated that native OPG was also capable of suppressing bone resorption in mice (7).
The major biological sources of OPG that modulate bone resorption are not well established. Osteoblasts or cells of that lineage have been shown to produce OPG (27, 28, 29, 30). The physical proximity of osteoblasts and osteoclasts during bone remodeling creates opportunities for osteoblast-derived regulation of bone resorption. However, the production of OPG by cultured osteoblasts was shown to increase with cell differentiation, suggesting that mature osteoblasts are not well suited to support osteoclastogenesis (27). OPG is also produced by endothelial cells (31, 32), vascular smooth muscle cells (32), and other cells (33). The high concentrations of OPG extracted from healthy and diseased human blood vessels suggest that vascular cells might be major contributors to the circulating pool of OPG. Circulating OPG would still have significant potential to inhibit bone resorption by suppressing the differentiation of circulating osteoclast precursors (34).
The ability of OPG to suppress bone resorption and increase bone mass is related to its ability to bind via its TNF receptor domains to TNF domains within its natural ligand (17), now known as RANKL. RANKL has a storied history in bone biology that precedes its positive identification by more than a decade. As early as 1980 it was appreciated by Chambers (35) that cells of the osteoblastic lineage might be involved in promoting osteoclast recruitment and activity. A more formal hypothesis (36) and experimental confirmation in various model systems (37, 38) further supported this notion, but the identity of this "osteoclast activating factor" remained elusive. The eventual identification of this protein involved the use of OPG as a probe to recover an OPG binding protein that was initially referred to as osteoprotegerin ligand (OPGL) (39). An identical protein [osteoclast differentiation factor (ODF)] was independently identified from a cDNA library prepared from ST-2 stromal cells (40). This cell line had been previously shown to support osteoclast formation in a manner that could be completely inhibited by OPG (5). Although the essential role of OPGL/ODF in osteoclast biology was an important finding, this protein proved to be identical to one previously identified and referred to as receptor activator of nuclear factor B ligand (RANKL) (41). RANKL, also referred to as TNF-related activation-induced cytokine (TRANCE) (42), was originally identified as a product of T cells. This TNF family member was able to promote dendritic cell survival and costimulation. The essential role of RANKL/TRANCE in bone resorption was confirmed soon thereafter (43). RANKL is the generally accepted acronym for this protein (14) (Table 1).
RANKL is a novel member of the TNF family of ligands. There are at least three forms of RANKL, two of which possess a transmembrane domain that positions the biologically active carboxy-terminus to the extracellular domain (i.e., a type II transmembrane protein) (Fig. 2). One of these forms, RANKL2, is a shorter alternative splicing variant of RANKL1 (44). Both of these variants can remain on cell surfaces or can be proteolytically cleaved into soluble forms that possess osteoclast-stimulating activities within their TNF-homology domains (44, 45). RANKL is produced by numerous cell types including cells of the osteoblast lineage (46, 47) and activated T cells (42, 48, 49). T Cells express both soluble and membrane-bound forms of RANKL (49, 50), and both forms are implicated in focal bone erosions associated with inflammatory arthritis (26, 49). Cells of the osteoblast lineage can express RANKL on their surface in a manner that facilitates osteoclastogenesis in vitro via cell-to-cell contact with osteoclast precursors (29). Differentiation of cultured osteoblasts was associated with reduced RANKL expression and decreased ability to support osteoclastogenesis (27, 51), suggesting that the mature bone-forming osteoblast might not be capable of directing osteoclast activity via RANKL. Membrane RANKL has been suggested to be somewhat more potent than soluble RANKL in stimulating osteoclastogenesis in vitro (45, 52). However, soluble RANKL is measurable in the circulation, and serum RANKL has been shown in some studies to increase with stimulated bone resorption (26, 50, 53). Soluble recombinant RANKL is also capable of causing severe skeletal catabolism in mice (54, 55) and in rats (56). These results indicate that soluble RANKL has the potential to be an important physiological and pathological mediator of bone resorption.
RANKL is involved in numerous aspects of osteoclast differentiation and function. RANKL was implicated in the fusion of osteoclast precursors into multinucleated cells (39), their differentiation into mature osteoclasts (39), their attachment to bone surfaces (12), their activation to resorb bone (9, 39), and their continued survival by avoiding apoptosis (11). In most situations, RANKL probably relies on macrophage colony-stimulated factor (M-CSF, also known as CSF-1) as a cofactor for osteoclast differentiation (57, 58). However, it is interesting to note that unlike RANKL knockout mice, the osteoclast population recovers over time in mice lacking functional M-CSF (59). Vascular endothelial growth factor has been implicated as a compensatory factor that might maintain osteoclasts in the absence of functional M-CSF (60). Recent preliminary evidence also suggests that RANKL can stimulate osteoclastogenesis and bone resorption in mice that lack functional M-CSF (47). These results suggest that RANKL plays a dominant role in the regulation of bone resorption, and no factor or combination of factors have been shown to be capable of restoring bone resorption when RANKL is absent.
The receptor that mediates all known activities for RANKL is called receptor activator of nuclear factor B (RANK) (Figs. 1 and 2). RANK is a homotrimeric TNF receptor family member that was initially discovered from a bone marrow-derived dendritic cell cDNA library. RANK was identified as a receptor that mediated the ability of RANKL to promote the survival of cultured dendritic cells (41). An important role for RANK in osteoclastogenesis was reported by Nakagawa et al. (61), who referred to this TNF receptor family member as osteoclast differentiation factor receptor. The essential role for RANK in bone resorption was demonstrated soon therafter by the high bone mass phenotype of RANK knockout mice, which were virtually devoid of osteoclasts (62, 63). RANK and RANKL knockout mice were virtual phenocopies of each other (62, 63, 64), which indicated that RANK and RANKL had few if any roles beyond their mutual interactions.
The binding and activation of RANK involve direct interactions between the extracellular receptor binding domain of trimeric RANKL and the extracellular cysteine-rich domains of trimeric RANK (17, 65). This interaction is thought to cause oligomerization of RANK and the subsequent activation of several signal transduction pathways. A simplified overview of major signaling pathways is described in Fig. 3. RANKL binding to RANK results in the recruitment of an adapter protein known as TRAF 6 (TNF receptor associated factor 6) to specific sites within the intracellular domain of RANK. TRAF 6 acts as a second messenger to activate various protein kinase pathways as well as transcription factors including nuclear factor-B (NF-B). Activated NF-B translocates to the nucleus and up-regulates the expression of c-fos, which then interacts with nuclear factor of activated T cells (NFAT)-c1 to induce the transcription of osteoclastogenic genes. RANK signaling is far more complex than what is described above, and more details of these and other important pathways were recently described (66, 67, 68, 69).
|
). OPG inhibition of RANKL leads to the rapid arrest of osteoclast formation, attachment to bone, activation, and survival.
B. Expression patterns and hormone/cytokine regulation of expression
With some notable exceptions, the expression patterns of OPG, RANK, and RANKL genes would not obviously predict their biological roles. Expression of each gene product has been shown in numerous cells and tissues in which no known role for that gene product has been ascribed. OPG mRNA, for example, is highly expressed in the lung, heart, kidney, and placenta (4), and OPG has yet to demonstrate any physiological or pathological role in those organs. OPG mRNA is also expressed in liver, stomach, intestine, skin, bone, and most other human tissue (6) with the exceptions of brain, skeletal muscle, thymocytes, or peripheral blood lymphocytes (6, 70). At the cell culture level, OPG is expressed by bone marrow stromal cells (27), fibroblasts (6), endothelial cells (31, 71), lymphoid cells (70), smooth muscle cells (70), and other cell types.
RANK mRNA is expressed in bone, bone marrow, spleen, skeletal muscle, brain, heart, liver, lung, mammary tissue, and skin (61, 72, 73) and at the cellular level by osteoclasts, osteoclast precursors, dendritic cells, chondrocytes, endothelial cells, fibroblasts and macrophages, and other cells (41, 61, 72, 74). RANKL is highly expressed in peripheral lymph nodes and bone, and also in spleen, thymus, Peyer’s patches, intestine, brain, heart, skin, skeletal muscle, kidney, liver, lung, and in mammary tissue of pregnant mice (39, 40, 73, 75). Cells that express RANKL include osteoblasts, bone marrow stromal cells, activated T cells, B cells, fibroblasts, endothelial cells, chondrocytes, and mammary epithelial cells (31, 40, 42, 76, 77). Regulatory elements involved in the transcription of the OPG gene (78, 79, 80, 81) and the RANKL gene (78, 79) have been recently described but are beyond the scope of this review.
Numerous growth factors, hormones, cytokines and drugs have been shown to influence the expression of OPG and RANKL, and an excellent summary of these regulatory responses was described recently (15) (Table 2). RANK tends to be minimally regulated, which is not unexpected because the modulation of OPG and/or RANKL levels is probably sufficient to control the signaling output of RANK. Numerous molecules that stimulate bone resorption have been shown to up-regulate RANKL expression and/or to down-regulate OPG expression. Reciprocal regulation of RANKL and OPG is often observed, a phenomenon that might serve to amplify proresorptive signals (82, 83, 84). Evidence in numerous disease models and settings suggests that the ratio of RANKL:OPG represents an important determinant of bone resorption (85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96), although exceptions to this concept have been noted (53, 97).
|
). Intermittent PTH treatment in humans was also shown to increase serum RANKL and to reduce serum OPG (93), suggesting that OPG and RANKL might mediate the resorptive response to PTH independent of the mode of PTH administration.
|
(52, 98), PTH (86, 88), PTHrP (101), and vitamin D (102, 103) (Table 2). Many molecules that suppress osteoclastogenesis and/or bone resorption have also been shown to regulate OPG and/or RANKL, including estrogen (46, 76, 84, 104, 105, 106, 107, 108), interferon-
(52), TGF-β (52), IL-4 (109), and IL-13 (52) (Table 2). These studies have provided important molecular insights into the mechanisms by which some of these molecules might regulate bone resorption. OPG prevents the ability of proresorptive agents (IL-1, TNF-, PTH, PTHrP, vitamin D) to increase bone resorption in mice (21), which is consistent with the interpretation that those agents increase bone resorption by increasing the RANKL:OPG ratio. This conclusion must be tempered, however, by the fact that OPG and RANKL also play dominant roles in the regulation of normal bone remodeling. In the absence of RANK or RANKL, osteoclasts cannot form, function, or survive (62, 63, 64, 110, 111). The dominant role of RANKL in osteoclast biology creates practical limitations in using OPG as a research tool to evaluate the extent to which different proresorptive factors work through the RANK-RANKL pathway. For example, cell culture studies have shown that proresorptive agents including IL-1 and TNF- can stimulate certain aspects of osteoclast activation or differentiation in a manner that cannot be entirely inhibited by OPG (112, 113, 114). Yet the direct injection of IL-1 and/or TNF- into RANK knockout mice induced the appearance of few if any osteoclasts, with no evidence of bone resorption (115) or hypercalcemia that was apparent in similarly challenged wild-type mice (63). A possible explanation for the apparent discordance between these in vitro vs. in vivo observations was provided when TNF- was shown to require very low "permissive" levels of RANKL to stimulate osteoclasts (116). This low level of signaling might persist in cell culture studies despite apparent saturation by OPG, or alternatively culture conditions might include stimuli for osteoclast formation and survival that do not exist in animals. In support of the latter notion, OPG treatment of mice was able to fully prevent the proresorptive effects of IL-1 and TNF- (21). Thus, whereas TNF-, IL-1, IL-8 (117), IL-6, IL-11 (118), and other molecules might have independent stimulatory effects on osteoclasts, the very existence of those osteoclasts remains dependent on RANK-RANKL signaling. Unexpected sources of regulation for OPG and RANKL have been suggested by studies of signaling pathways for osteoblasts and osteocytes. An important signal transduction node for osteoblasts is β-catenin, a cytoplasmic protein that mediates signaling by members of the wnt (wingless) family of ligands. The genetic ablation of β-catenin in mice led to an osteopenic phenotype that appeared to be driven primarily by increased bone resorption (119). OPG mRNA expression was significantly reduced in osteoblasts isolated from these β-catenin-ablated mice. In contrast, mice with a gain-of-function mutation in β-catenin had high bone mass, and their cultured osteoblasts had increased expression of OPG mRNA. Other studies showed that the conditional knockout of β-catenin in osteoblasts resulted in increased expression of RANKL and decreased expression of OPG. Conversely, increased β-catenin activity in osteoblasts was shown to reduce RANKL and increase OPG expression (120). These observations raised the interesting possibility that wnt/β-catenin signaling in osteoblasts regulates bone mass by coordinating the activity of both osteoblasts and osteoclasts, with osteoclast regulation occurring via changes in OPG and RANKL.
The osteocyte is by far the most abundant cell type in bone. These cells are interconnected by numerous cellular processes that permeate bone matrix and are thought to play an important role in sensing mechanical strain (121). The mechanosensory function of osteocytes might ultimately be mediated via their connection to bone surfaces, where they have the potential to regulate osteoblasts and osteoclasts. Although osteocytes do not appear to express RANKL mRNA (122), a recent study suggests that these cells might regulate RANKL expression by other bone cells. The conditional ablation of osteoctyes in mice was associated with rapid bone loss in association with increased bone resorption, increased skeletal expression of RANKL mRNA, and an increased RANKL:OPG ratio (123).
C. Role of OPG, RANK, and RANKL in normal bone turnover
The dominant physiological role of the RANK/RANKL/OPG pathway in the regulation of bone resorption was demonstrated in animal models wherein these molecules were genetically ablated or overexpressed or, in some cases, both. OPG overexpression was shown to result in increased cortical and cancellous bone mass in association with reduced osteoclast numbers (4). The essential physiological role of OPG in maintaining bone mass was shown by the severe osteopenic phenotype of OPG knockout mice (124, 125, 126). These mice exhibit increased bone turnover, cortical porosity, and reductions in cortical and cancellous bone volume and density. These deleterious changes are accompanied by spontaneous fragility fractures in the first weeks of life (124). Biomechanical testing revealed that cortical bone strength was reduced by 50% in OPG knockout mice compared with wild-type controls (125). The osteoporotic phenotype of OPG knockout mice tends to worsen with age, and the loss of even one OPG allele was associated with significant osteopenia (124). In addition to a deficit in bone quantity, OPG deficiency is associated with poor bone quality as well. Collagen within the cortical bone of OPG knockout mice has a woven rather than lamellar orientation, a phenomenon that is associated with pathologically elevated bone turnover (127). Mutations in the OPG gene in humans is associated with juvenile Paget’s disease, and bone biopsies from such patients reveal a severe disruption in the normal architecture of cancellous bone (128).
These genetically based models of OPG deficiency are influenced by skeletal growth and development, which confound attempts to deduce the consequences of OPG deficiency on the mature skeleton. In the absence of conditional OPG knockout models, investigators have relied on injections of RANKL into adult rats and mice to create a model of relative OPG deficiency. RANKL injections led to rapid skeletal changes that closely resembled those associated with the lifelong absence of OPG: increased bone turnover, increased cortical porosity, weakened cancellous microarchitecture, and reductions in bone volume, density, and strength (54, 55, 56). Interestingly, the typical age-related bone loss seen in normal mice was correlated with increased skeletal expression of RANKL mRNA and reduced expression of OPG (89). These studies and others reveal that the adult mouse skeleton continues to rely on OPG and a positive OPG:RANKL ratio to maintain bone mass and strength.
Based on the established role of OPG in maintaining bone mass and the mechanistic basis of that role, the skeletal phenotypes of RANK and RANKL knockout mice were not unexpected. Genetic ablation of either RANK or RANKL led to severe osteopetrosis in association with a near total lack of osteoclasts (62, 63, 64, 110). A few notable skeletal differences were apparent between RANK and RANKL knockout mice compared with OPG transgenic mice. OPG transgenic mice exhibited an osteopetrotic phenotype, with high bone mass but without significant changes in the external shape of their long bones or defective tooth eruption (4). In contrast, RANK and RANKL knockout mice had more classical osteopetrotic features including high bone mass, shortened club-shaped long bones, and the failure of teeth to erupt. These differences suggest that a small residual level of RANK signaling, which is present in OPG transgenics (but not in RANK or RANKL knockout animals), was sufficient to establish normal bone shape and tooth eruption. This notion is supported by the skeletal phenotype associated with transgenic overexpression of RANK-Fc, a soluble extracellular portion of RANK that functions as a RANKL inhibitor (72). Similar to OPG transgenic animals, RANK-Fc transgenic animals had high bone mass, normal bone shape, and normal tooth eruption. Unlike RANK and RANKL knockouts, RANK-Fc transgenics were able to maintain a greatly reduced but still present osteoclast population that may have permitted the establishment of normal bone shape and tooth eruption.
Two interesting hallmarks of recombinant OPG and RANKL are the rapidity of their effects on bone resorption and their ability to regulate osteoclast numbers. Soluble RANKL stimulates bone resorption within 1 h of injection in mice (9), and soluble OPG suppresses indices of bone resorption within 2 h of its injection (7). These kinetics are consistent with the demonstrated ability of RANKL and OPG to regulate the activity of mature active osteoclasts (9). OPG suppression of osteoclast numbers in mice and rats also occurs rapidly, with 50–60% reductions in osteoclasts noted within 12–24 h of treatment with an OPG-Fc fusion protein (11, 24). Maximal suppression of osteoclast numbers in these animal studies was greater than 90%. Reduced osteoclast numbers in mice was attributable to osteoclast apoptosis, which was apparent histologically within 6 h of OPG-Fc treatment (11). The prolonged duration of osteoclast suppression with OPG-Fc treatment was probably related to the long circulating half-life of the drug (24), as well as the time required to regenerate new osteoclasts from their precursor population once the drug is cleared (11).
The ability of OPG to cause consistent, rapid, and marked reductions in osteoclast numbers is relatively unique among antiresorptive agents. In contrast, bisphosphonates have a different mechanism of action that can be associated with normal or even increased osteoclast numbers despite clear antiresorptive effects (129, 130, 131, 132, 133, 134, 135, 136, 137, 138). This phenomenon is consistent with previous observations that bisphosphonates suppress bone resorption via injury to osteoclasts when they solubilize bisphosphonate-contaminated bone (139, 140). Bisphosphonates do not appear to suppress osteoclastogenesis (129, 140, 141) and in some cases can paradoxically stimulate the formation of osteoclasts (133). Although bisphosphonates are clearly capable of inducing osteoclast apoptosis, this apoptotic response does not appear to be an essential component of their ability to suppress bone resorption (142). Whether the ability of OPG to suppress osteoclast formation, function, and survival has any clinical advantages over agents that act primarily or exclusively through the suppression of osteoclast function is not known. But the ability of bisphosphonates to suppress bone resorption without consistent reductions in osteoclast numbers suggests that bisphosphonates might act through a mechanism that is distinct from that of OPG.
D. Non-osseous roles for OPG, RANK, and RANKL
1. The vascular system.
Although the skeletal phenotypes associated with genetic knockout of OPG, RANK, and RANKL were somewhat predictable, a few unexpected phenotypes were also observed in these animals. OPG knockout animals developed medial calcification of the aorta and renal arteries, in association with elevated bone turnover (124). This vascular calcification could be prevented via transgenic overexpression of soluble OPG, indicating that systemic RANKL inhibition could prevent the vascular consequences of endogenous OPG deficiency (127). Bone turnover and RANKL levels are both systemically elevated in OPG knockout mice (143), and it is unclear whether their vascular calcification is related to high turnover, or to a relative increase in RANKL activity within blood vessels, or to other causes. Evidence at this point seems to favor high bone turnover as an etiological aspect of vascular calcification, at least in animal models. Recombinant OPG was shown to prevent vascular calcification induced by warfarin and by vitamin D in rats (144), and in each of these model systems the prevention of vascular calcification by OPG was associated with strong suppression of bone turnover. Bisphosphonate treatment of rats also suppressed vascular calcification in this warfarin model, in association with bone turnover suppression (145). The ability of OPG to prevent apoptosis of cultured endothelial cells (71, 146) and arterial smooth muscle cells (143) suggests that bone turnover suppression might not fully explain the ability of OPG to prevent vascular calcification. However, although bisphosphonates suppress vascular calcification (145), these agents have also been shown to suppress the proliferation, adhesion, migration, and survival of endothelial cells in vitro (147, 148) while inducing endothelial cell apoptosis in vivo (149). Thus, although OPG and bisphosphonates can have opposite effects on the survival of endothelial cells, both agents suppress bone turnover and prevent vascular calcification when tested in a similar animal model (144, 145). These phenomena are consistent with the interpretation that bone turnover is an important aspect of vascular calcification.
More evidence for a turnover-related mechanism of vascular calcification comes from recent studies using knockout mice that lack the ApoE gene, which predisposes them to atherosclerosis and vascular calcification. When these ApoE knockout mice were cross-bred with OPG knockout mice, bone turnover increased in association with greater calcification and advanced plaque progression. The absence of OPG had no apparent influence on circulating levels of total or HDL cholesterol (143). These interventional studies in preclinical models support the possibility that OPG serves a protective role in vascular disease. Based on this premise, osteoprotegerin has also been alternatively described as "vasculoprotegerin" (150, 151).
Evidence from human studies also suggests a relationship between bone turnover and vascular disease. Observational studies in postmenopausal women showed that aortic calcifications were inversely related to bone density and positively associated with fractures (152). It was therefore surprising that observational studies frequently show positive correlations between serum OPG levels and vascular disease, although the correlation between serum OPG levels and fractures or BMD is inconsistent as discussed in Section IV.A (151, 153, 154, 155, 156, 157, 158, 159, 160). Recent review papers have summarized this literature more extensively (161, 162, 163). Possible interpretations of this positive relationship include OPG playing an active role in disease progression (157, 158), or OPG serving as a compensatory response to minimize disease progression (151, 154, 156, 164), or OPG representing a noncompensatory (neutral) response to disease (156, 165). Interventional studies with RANKL inhibitors will be needed to sort out these possibilities.
Although OPG therapy consistently suppresses vascular calcification in animal models (127, 144, 166), there are minimal data on the effects of OPG therapy on atherogenesis or other aspects of vascular disease. The clinical observational data described above highlight the need for further preclinical intervention studies on this subject. An ideal animal model would recapitulate the clinical picture of progressive atherosclerosis and vascular calcification in association with systemic increases in endogenous OPG levels. Those very conditions were recently described in an animal model involving LDL receptor knockout mice on a high-fat diet (167). Studies have been performed in similar mice to evaluate the consequences of therapeutic intervention with OPG in the presence of vascular disease and elevated endogenous OPG levels. LDL receptor knockout mice were put on a high-fat diet while being treated for 5 months with OPG. OPG treatment significantly reduced calcified lesion area in the aorta of these mice, without affecting atherosclerotic lesion size or number (166). OPG had no effect on concentrations of inflammatory cytokines in the aorta or on plasma cholesterol levels. These results support the hypothesis that OPG might reduce vascular calcification via suppression of bone turnover, with a neutral effect on atherosclerotic changes. Additional studies on RANKL inhibition in different atherogenesis models are warranted to corroborate these preliminary findings.
2. Immune system development and function.
Another unexpected phenotype found in RANK and RANKL knockout mice was the total lack of lymph nodes (62, 64). Several other knockout mouse models also exhibit lymph node agenesis, but those models also exhibited a lack of Peyer’s patches and follicular dendritic cells. Those findings led investigators to consider that these developmental processes were linked (168). However, RANK and RANKL knockout mice had normal Peyer’s patches and intact splenic architecture (49, 62), indicating that the development of lymph nodes and secondary lymphoid organs can be differentially regulated. The RANK and RANKL knockout phenotypes prove that some level of RANK signaling is essential for the development of lymph nodes in mice. But beyond this observation, it has been difficult to ascribe an important role for RANK signaling in the function of immune systems in genetically intact mice or rats. It is noteworthy that in contrast to the effects of total RANKL ablation, prenatal inhibition of RANKL in OPG transgenic mice and rats did not lead to any observed changes in lymph node development or architecture (4, 182). Normal lymph node development occurred in OPG transgenic rats despite evidence that OPG was overexpressed in embryos by 100-fold during the gestational days that preceded the typical appearance of early lymphoid nodules. The low level of RANKL activity that persists in the presence of 100-fold overexpression of OPG was apparently sufficient to permit lymph node development, suggesting that total genetic ablation of RANK-RANKL signaling might be necessary to cause lymph node agenesis. Although RANKL knockout mice had no lymph nodes, there was no apparent intrinsic defect in immune cells from those animals, because T and B cells from RANKL knockout mice were able to populate the lymph nodes of normal wild-type mice (64). Examination of mice genetically deficient in RANKL, RANK, or OPG also revealed no overt dendritic cell defects (62, 64, 170). These observations suggest that the immune phenotype of RANK and RANKL knockout mice might be largely restricted to lymph node agenesis and the consequences thereof.
These findings notwithstanding, there is still significant interest in the potential immunomodulatory effects of RANKL and RANKL inhibition. Within the developed immune system, RANKL is expressed by T cells (42), RANK is expressed by dendritic cells (49), and OPG is expressed by B lymphocytes and dendritic cells (4, 70). In vitro studies show that binding of RANKL from T cells to RANK on dendritic cells regulates dendritic cell function and survival (42). While RANKL has the ability to stimulate dendritic cells, there is little evidence that RANKL inhibition leads to dendritic cell suppression. This is probably related to the central role of CD40L and CD40, found on T cells and dendritic cells, respectively, in mediating cross-talk between T cells and dendritic cells. This important pathway is apparently able to compensate for the total absence of RANK-RANKL signaling in knockout mouse studies (171, 172).
Several studies have shown that RANKL inhibition had no significant effects on the in vivo responses to immune challenges in genetically intact animals (171, 172, 173, 174). Short-term (up to 2-wk) mouse studies have show that RANKL inhibition by OPG did not impair cell-mediated reactions such as contact hypersensitivity or granuloma formation, innate immunity, humoral responses to defined immune challenges (175), or influenza infection (173). Of greater interest, however, is the potential consequence of long-term RANKL inhibition on immune function. This question was addressed using OPG transgenic mice and rats, in which RANKL is inhibited continuously throughout life. Analysis of numerous cellular, innate, or adaptive immune responses in 4- to 6-month-old transgenic mice and rats showed no differences compared with normal wild-type controls (169, 174, 176). OPG transgenic rats in these studies had marked suppression of bone turnover and increased BMD, thereby demonstrating that RANKL inhibition can cause turnover suppression in the absence of measurable changes in the integrated response of the immune system (169). Similar findings have been noted in animal models of inflammatory bone loss, wherein RANKL inhibition effectively preserved bone while having no significant effects on the inflammatory component of disease (26, 110, 136, 177, 178, 179, 180, 181). These results suggest that RANKL likely plays a redundant role in the integrated function of the postnatal immune system, perhaps due to the overlapping and predominant role of the CD40-CD40L costimulatory pathway.
3. Mammary gland development during pregnancy.
The other significant extraskeletal manifestation of total RANK/RANKL ablation in knockout mice was a defect in mammary gland development during pregnancy and lactation (73). Neither RANK nor RANKL was required for mammary gland development in mice before pregnancy, whereas pregnant RANK knockout mice failed to lactate and their litters did not survive. The mammary gland defect in RANK knockouts was attributed to the inability of sprouted alveolar buds to differentiate and expand into mature lobuloalveolar structures. In contrast to the effects of total genetic ablation of RANK-RANKL signaling, the continuous inhibition of RANK-RANKL signaling does not appear to result in a failure of lactation. OPG transgenic rat and mouse pups are viable, show normal suckling behaviors, and survive to weaning without supplemental feeding (Ref. 4 , and P. J. Kostenuik, unpublished observations). OPG is overexpressed in these animals by 100-fold during gestation and throughout life, and this level of OPG was associated with marked suppression of bone turnover (169, 182). These observations suggest that, as with lymph node development, a small permissive level of RANKL activity is sufficient to allow for mammary gland development during pregnancy in mice.
|
III. Pharmacological Effects of RANKL Inhibition in Preclinical Models of Bone Disease |
|---|
|
TopAbstractI. IntroductionII. Discovery and Validation...III. Pharmacological Effects of...IV. Role and Regulation...V. Clinical Therapeutics...VI. SummaryReferences |
|---|
The suppression of bone remodeling in normal rats by OPG leads to increased bone volume and density despite the significant coupling-related suppression of bone formation parameters (54, 55, 56). Furthermore, stimulation of bone remodeling in normal mice or rats by RANKL leads to reduced bone volume despite the significant coupling-related stimulation of bone formation (54, 55, 56). These observations suggest that remodeling-related changes in osteoblast output did not adequately counter the changes in osteoclast activity resulting from OPG or RANKL administration. This hypothesis must be tempered by the possibility that OPG and RANKL also influenced osteoclasts participating in bone modeling, which is more prevalent in younger growing animals and which is not countered by subsequent bone formation (184). Modeling-dependent bone formation and growth appear to be spared during RANKL inhibition because OPG had no influence on longitudinal bone growth in normal young rats (185), mice (186), piglets (187), or cynomolgus monkeys (25).
A brief summary of data from healthy gonad-intact rats and cynomolgus monkeys provides an illustration of the consequences of RANKL inhibition on normal bone remodeling, density, and strength. In rats, a single injection of OPG-Fc resulted in a rapid (within 12 h) and profound (up to 95%) reduction in the percentage of bone surfaces occupied by osteoclasts (24). Osteoclast numbers showed a gradual recovery from d 10 to 30 after injection as the drug was gradually cleared from the circulation. Osteoblast surface and biochemical markers of osteoblast activity were more modestly suppressed relative to osteoclast inhibition. These changes were associated with a progressive increase in cancellous bone volume and a significant increase in bone mineral density (BMD) at d 30 compared with vehicle-treated controls (24).
Assessments of bone strength and bone quality in normal healthy rodents treated with RANKL inhibitors have been limited to short treatment periods of less than 1 month. The marked turnover suppression associated with RANKL inhibition raises questions about the consequences of long-term therapy to bone strength and bone quality. These questions are particularly relevant if the repair of skeletal microdamage is a significant feature of normal bone remodeling. One approach to address this question involved the use of transgenic rats that overexpressed soluble OPG from a prenatal stage and throughout life (188). The transgene construct in these rats was the same one used to identify the critical role of OPG in regulating bone mass in transgenic mice (4). In transgenic rats, continuous overexpression of OPG for the first 6 months of life led to a 98% reduction in osteoclast surface in the lumbar vertebrae. This marked and sustained suppression was associated with 2-fold increases in extrinsic strength parameters of the vertebrae relative to age-matched wild-type (normal) control rats. These strength increases appeared to be primarily related to increases in bone mass. When strength parameters are standardized to bone mass parameters, one can indirectly examine the strength of the bone tissue itself (intrinsic bone strength). Intrinsic strength parameters in OPG transgenic rats were similar to values obtained in normal controls (188). These data suggest that long-term RANKL inhibition improves bone strength in normal animals primarily by increasing bone mass, while preserving the quality of bone matrix. Similar conclusions were supported by analyses of bone mass and bone strength in 12-month-old OPG-Tg rats (189).
Although bone mass is clearly a major attribute of bone strength, there are other components of bone strength that are not effectively captured by standard bone densitometry modalities such as dual energy x-ray absorptiometry (DXA) or peripheral quantitative computed tomography. These "bone quality" parameters include matrix mineralization (190), cortical porosity (191), cortical geometry (192), and trabecular microarchitecture (193, 194). The effects of recombinant RANKL on these parameters were recently described. In normal healthy mice, 10 d of recombinant RANKL injections led to reduced matrix mineralization, an 85% reduction in trabecular bone volume, deleterious changes in cortical geometry, and significantly reduced bone strength (54). In normal healthy rats, 4 wk of RANKL infusion led to reduced matrix mineralization, decreased trabecular and cortical bone volume, reduced trabecular connectivity, and reduced bone strength (56).
Based on their mechanisms of action and their important roles in normal bone turnover, one might expect OPG and RANKL to have opposite effects on bone quality and strength parameters in normal animals. This seems to be the case because OPG treatment of normal rats was shown to increase matrix mineralization and improve bone strength (185). While RANKL caused deleterious effects on trabecular microarchitecture in normal rats (56) and reduced matrix mineralization in normal mice (56), OPG treatment was recently shown to have the opposite effects in normal mice. These data were presented in abstract form (195), and the lack of published data on the effects of OPG on trabecular microarchitecture in normal healthy animals warrants a brief description of those data. Briefly, 2- to 3-month-old Black Swiss/129 gonad-intact mice (n = 8 per group) were treated for 3 wk with either human OPG-Fc (5 mg/kg sc, twice a week) or with vehicle (saline). Distal right femurs and lumbar vertebrae were scanned ex vivo by microcomputed tomography (microCT) at 18-µm resolution for analyses of trabecular bone. OPG treatment was associated with significantly greater bone volume, volumetric BMD, bone matrix mineral density, trabecular thickness, number, and connectivity (Table 3). Representative microCT images of trabecular regions from the distal femur and lumbar vertebra are shown in Fig. 5. It is important to note that the magnitude of changes observed in this 3-wk study were greatly influenced by growth-related modeling that is typical of young mice. The magnitude and rate of such changes would not be achievable in adult humans, with or without bone disease. However, these data suggest that bone mass improvements with OPG are associated with evidence of improved trabecular architecture.
|
|
Denosumab (previously known as AMG 162) is a fully human monoclonal antibody that inhibits RANKL. Twelve months of denosumab treatment in gonad-intact cynomolgus monkeys was associated with significant increases in bone strength at the femoral shaft (a purely cortical site) and at the lumbar spine (a predominantly cancellous site) (196). At both sites, bone strength was highly correlated with BMC (R2 = 0.75–0.85), which suggests that the material properties of bone matrix (bone quality) were maintained during 12 months of denosumab therapy. Increments in bone strength in denosumab-treated animals were attributed primarily to increases in bone mass. It is important to confirm these findings in aged animals with high-turnover bone diseases, including animal models that are described in the next section.
B. RANKL and RANKL inhibition in ovariectomy models of bone loss
Ovariectomy (OVX) represents a model of surgical menopause, wherein estrogen levels drop more precipitously compared with the slower decline seen in natural menopause. OVX is associated with increased bone turnover and reductions in BMD, volume, and strength in numerous species, including mice, rats, and cynomolgus monkeys. The therapeutic rationale for RANKL inhibition in OVX models was first suggested in 1997 when OPG was shown to fully prevent the increased osteoclasts and reduced bone volume associated with OVX of young rats (4). Those findings sparked intensive research into the question of whether estrogen might regulate OPG and RANKL in a manner that could contribute to the ability of estrogen to preserve bone mass. Abundant data from cell culture studies demonstrate that estrogen can increase OPG production (46, 84, 104, 108, 197), and in some cases estrogen also causes suppression of RANKL expression (84). The withdrawal of estrogen from cultures of a stromal cell line that supports osteoclastogenesis resulted in a dramatic reduction in their expression of OPG (197). Whether estrogen withdrawal also leads to decreased OPG levels in humans is currently unknown.
OVX has been shown to increase RANKL levels in various animal models (50, 198, 199, 200, 201). OPG levels have been reported to increase (199) or decrease (198, 201) in association with OVX, and this difference in response could relate to variations in the animal model and/or the tissue source analyzed. In humans, estrogen deficiency is associated with increased RANKL expression by bone marrow osteoblastic, T and B cells, whereas estrogen treatment reversed this increase (76). These associations are consistent with the interpretation that OPG and RANKL play important roles in mediating the effects of estrogen and estrogen withdrawal on bone. Definitive proof of these phenomena will be challenging, however, because the interventional use of OPG or other RANKL inhibitors results in suppression of osteoclasts regardless of the stimuli that drove their activity. It is possible that estrogen suppresses osteoclasts in a manner that does not rely entirely on the regulation of RANKL and/or OPG expression. In support of this idea, estrogen has been shown to directly suppress RANKL-induced osteoclast formation in vitro in the absence of osteoclast-supporting stromal cells (105).
Preclinical data from OVX animal studies highlight the therapeutic promise of RANKL inhibition for postmenopausal osteoporosis. One OVX study involved the use of very old rats, which were ovariectomized at 3 months of age and allowed to lose bone for 15 months before OPG treatment (202). Substantial osteopenia was evident at several skeletal sites before the initiation of OPG treatment, which was delivered for 24 wk. OPG treatment was associated with marked reductions in osteoclast surfaces and increased cancellous bone volume in the vertebrae and tibiae. OPG treatment also significantly increased lumbar and leg BMD, while increasing biomechanical strength parameters at the lumbar vertebrae and femoral shaft. In a separate study, OPG was shown to preserve bone density in OVX rats while fully preventing the deleterious OVX-related changes in cancellous microarchitecture (200).
A recent study in young adult OVX mice was the first to compare OPG with a potent bisphosphonate (alendronate) (203). Both agents prevented OVX-related decreases in spine and femur BMD, while increasing vertebral cancellous bone volume. Both agents also caused significant suppression of osteoclast and osteoblast numbers, as well as biochemical markers of bone turnover. In this study OPG, but not alendronate, increased the mechanical strength of the femoral shaft. OPG also showed significantly greater changes in femur BMD, vertebral bone volume, and osteoclast numbers when compared with alendronate (203).
Other OVX mouse studies have corroborated the benefits of RANKL inhibition in estrogen ablation settings, using less direct means of targeting RANKL. An adenovirus gene therapy vector was used to deliver OPG at therapeutic levels that were sufficient to fully prevent bone loss in an acute OVX model (204). An adenovirus-associated gene therapy vector was subsequently shown to deliver OPG at levels sufficient to reverse established OVX-induced bone loss in mice (205). A peptidomimetic approach has been applied to disrupt RANK-RANKL interactions. This small molecule mimic of OPG was delivered continuously by osmotic minipump for 28 d immediately after mice were ovariectomized, and this approach led to partial prevention of OVX-related bone loss (206). An anti-RANKL vaccine has been tested in another acute OVX mouse model, and this approach also led to partial prevention of OVX-related bone loss (207). The safety and efficacy of these experimental approaches to RANKL inhibition would require rigorous preclinical safety testing in animals before their consideration for human use.
The only data describing the effects of RANKL inhibition in OVX primates were recently reported in preliminary form (208, 209). Adult cynomolgus monkeys were subjected to OVX or sham surgeries and allowed to recover for 1 month. OVX cynos were then treated with vehicle or denosumab once per month for 16 months. Denosumab treatment was associated with rapid and sustained suppression of biochemical parameters of bone resorption and formation (208). These changes were accompanied by significant increases in cortical and trabecular BMC and BMD at multiple skeletal sites (208). Denosumab treatment was also associated with significant improvements in biomechanical parameters of bone strength at the lumbar spine, femur neck, and femur diaphysis (209). Strong linear correlations were observed between bone mass parameters and bone strength parameters at all skeletal sites (209). These results suggest that denosumab improved bone strength primarily by increasing bone mass, and that bone formed during denosumab therapy has normal material properties.
C. RANKL and RANKL inhibition in orchiectomy models of bone loss
Preclinical orchiectomy (ORX) models have utility in examining the consequences of androgen ablation and androgen deprivation therapy on bone (210, 211, 212). An important distinction between standard ORX rat models and androgen deprivation therapy in humans is the existence of prostate cancer in the latter population and the lack thereof in ORX rats. Normal ORX rat models are appropriate to examine the isolated effects of androgen ablation on bone. The ORX rat can also be used as a model of late-onset male hypogonadism (andropause) (213). Within this context, one limitation of the ORX rat model is the immediacy of disease onset, which contrasts with the more gradual decline in androgen levels in aging men (214).
Androgens have been shown to directly suppress the formation (215) and the bone-resorbing activity of cultured osteoclasts (216). Androgens also suppress osteoclastogenesis in coculture model systems that are dependent on osteoblasts (217). Whether OPG and/or RANKL mediate the suppressive effects of androgens on osteoclasts in such coculture systems is not clear. Androgens have been shown to up-regulate (218) and down-regulate (219) OPG expression in cultured osteoblasts, whereas RANKL was unregulated by androgens in both studies. Skeletal mRNA levels of OPG and RANKL were recently shown to be increased by ORX in rats (220), but whether RANKL protein levels exceeded OPG levels after ORX is unknown.
Whether or not ORX causes bone loss via the OPG-RANKL axis, RANKL inhibition has clear therapeutic benefits in ORX rats. In a recent study, rats were sham-operated or orchiectomized and then treated with vehicle or with OPG for 6 wk. MicroCT assessment at wk 6 revealed significant deterioration of cancellous microarchitecture in vehicle-treated ORX rats, and OPG fully prevented these changes (221). DXA analysis at wk 6 revealed that vehicle-treated ORX rats had significantly decreased BMD, whereas histomorphometry showed significant ORX-related reductions in periosteal bone formation rate and cancellous bone volume. OPG treatment fully prevented each of these changes (222). Androgen replacement therapy has been shown to partially mitigate the ORX-related deficit in periosteal bone formation in rats (211). The ability of OPG also to prevent the ORX-related reduction in periosteal bone formation was an unexpected finding. OPG was recently shown to stimulate the proliferation and differentiation of cultured osteoblasts (203), which suggests a possible mechanism for this periosteal response. An alternative explanation is that OPG prevented ORX-related increases in endocortical and/or cancellous resorption, thereby preserving bone geometry in a manner that obviated the periosteal response that is typically associated with ORX. The ability of OPG to prevent bone loss in this study could be interpreted as evidence that ORX is associated with a functional increase in the RANKL:OPG ratio. This conclusion might be premature, however, because OPG suppressed osteoclast numbers in ORX rats to levels that were well below ORX or sham control values (222). This phenomenon once again emphasizes the essential role of RANKL for the existence of osteoclasts, independent of gonadal hormone status.
D. RANKL and RANKL inhibition in rheumatoid arthritis models
Abundant research in rheumatoid arthritis (RA) models has produced perhaps the best evidence for a causal role of excessive RANKL activity in a bone loss setting. The intense interest in OPG-RANKL biology in RA is related in part to the multiple forms of osteoclast-mediated bone loss that are characteristic of the disease, including focal bone erosions at the bone-pannus junction of inflamed joints, subchondral (periarticular) bone loss adjacent to inflamed joints, and generalized systemic bone loss (223, 224, 225, 226, 227). The alignment between clinical observations and the attributes of preclinical RA models is close in terms of theoretical mechanism of bone loss and their manifestations (228).
Animal models of RA consistently show up-regulation of RANKL in inflamed joints (26, 49, 181, 225, 229, 230), and in some cases in serum as well (26). There are fewer data on OPG levels in preclinical RA models, but OPG protein levels were recently shown to be decreased in the inflamed joints of two well-characterized arthritic rat models (adjuvant and collagen-induced arthritis). RANKL protein levels were concomitantly increased in inflamed joints, leading to marked increases in the RANKL:OPG ratio. This ratio was positively correlated with focal bone erosions and with osteoclast activity (231). A detailed time course assessment of inflamed joints from adjuvant and collagen-induced rat arthritis models showed that RANKL protein levels were significantly increased before or simultaneous with the earliest appearance of bone erosions and osteoclast responses (26). These data suggest that RANKL is well positioned to play a causal role in the progression of erosive disease in these commonly used models of RA. Additional support for this notion comes from numerous animal studies that consistently show that RANKL inhibition is associated with marked suppression of bone erosions. Total genetic ablation of RANKL in knockout mice rendered the animals resistant to arthritis-related joint destruction, and osteoclasts were essentially absent in their joints (110). RANKL inhibition via recombinant OPG treatment also results in marked suppression of bone erosions and osteoclasts in several rodent models, including adjuvant arthritis (26, 49, 177, 232, 233), collagen-induced arthritis (26, 181, 233), and TNF-mediated arthritis (136, 234). OPG was also shown to reduce periarticular bone loss in arthritic rats (235), and OPG also prevented systemic bone loss in arthritic mice (180, 234, 236) and arthritic rats (26).
Although bone preservation is an important and expected outcome for OPG therapy, cartilage preservation is a primary goal of RA therapy. Joint space narrowing is an important clinical endpoint for evaluating the efficacy of RA treatment, and maintenance of articular cartilage is a significant component of this endpoint. Cartilage damage in RA is thought to be mediated at least in part by degradative enzymes such as matrix metalloproteinases, which are released by synovial fibroblasts. Inadequate synthesis of type II collagen, aggrecan, and proteoglycans might also contribute to the progression of cartilage loss (237). OPG, RANK, and RANKL are all expressed at the protein level in human articular cartilage, but their functional roles in cartilage have not been clearly demonstrated (238). Nonetheless, interest in possible roles for this pathway in the maintenance of articular cartilage was fostered by early observations that OPG treatment provided substantial cartilage preservation in arthritic mice (49). Other arthritis models have confirmed the ability of OPG or RANKL ablation to preserve cartilage, as measured histologically by proteoglycan preservation (110, 177, 181). Subsequently, serum RANKL concentrations in RA patients were shown to correlate positively with urine levels of CTX-II, a type II collagen degradation product (239). This relationship is not necessarily a causal one, because serum RANKL levels are also correlated with overall disease progression in RA (53, 239). Evidence that RANKL plays a direct role in maintaining cartilage is currently minimal. Although human articular cartilage and primary articular chondrocytes express RANK protein, exogenous RANKL did not appear to activate signaling in cultured chondrocytes, suggesting that RANK might be functionally inactive in those cells (238). A more likely mechanism for chondroprotection is related to the well-established ability of OPG to preserve subchondral bone, which helps to preserve the geometry of the joint while preventing osteoclasts from undermining the epiphyses from within. This indirect mechanism of chondroprotection is consistent with observations that OPG has no significant impact on the inflammatory components of arthritis, such as joint swelling (49, 136, 177, 179, 180, 181, 235, 236, 240).
E. RANKL and RANKL inhibition in glucocorticoid-induced bone loss models
Glucocorticoids have antiinflammatory and immunosuppressive effects and are used in numerous disorders including rheumatoid arthritis, Crohn’s disease, and glomerulonephritis. Glucocorticoids also have remarkably complex effects on bone, virtually all of which are detrimental, and these influences have been summarized in recent reviews (241, 242). This section will focus on the potential roles of bone resorption and the OPG-RANKL axis in mediating the effects of glucocorticoids on bone. The most universally described skeletal consequence of glucocorticoid therapy is a reduction in BMD, which can lead to increased fracture risk (243, 244, 245). Mechanisms responsible for glucocorticoid-induced osteoporosis have been extensively studied, but consensus has yet to be achieved. It is widely believed that suppression of bone formation is a component of bone loss, whereas increased (or inappropriately "normal") bone resorption likely also contributes.
At the cellular level, glucocorticoids [mainly prednisolone, dexamethasone (DEX), and cortisol] have complex effects on osteoblasts and osteoclasts. Glucocorticoids have been shown to stimulate the proliferation of cultured osteoblasts and/or their precursors in some studies (246, 247), whereas other studies showed that osteoblast proliferation was inhibited by glucocorticoids (246, 248). These conflicting results may be related to differences in the stage of osteoblast differentiation, and the timing and duration of glucocorticoid exposure in different systems (246). Glucocorticoids can also inhibit osteoblast differentiation (249, 250, 251), osteoblast attachment to bone matrix (252), and osteoblast production of IGF-I (253). Bone marrow harvested from mice treated with prednisolone was shown to have markedly reduced osteoblastogenic capacity (251). In general, these studies point toward significant osteoblast suppression as a component of glucocorticoid-induced bone loss.
The reported effects of glucocorticoids on osteoclasts are only slightly less complex. In favor of a proosteoclast effect, DEX was able to directly promote osteoclastogenesis from bone marrow cells that were exposed in culture to the essential osteoclastogenic cofactors RANKL and M-CSF (254). Furthermore, conditioned medium from glucocorticoid-treated osteoblasts had increased ability to support osteoclastogenesis (254). However, cortisol has been shown to inhibit resorption in bone organ cultures (255), and bone marrow harvested from mice treated with prednisolone showed reduced osteoclastogenic potential (251). A bone xenograft model showed that glucocorticoids inhibited osteoclast recruitment and bone resorption, whereas delayed treatment with glucocorticoids resulted in stimulation of osteoclast activity and bone resorption (256). DEX was recently shown to prolong the lifespan of osteoclasts, while decreasing their individual capacity to degrade bone (257). Osteoclast responses to glucocorticoids appear to be greatly influenced by their stage of differentiation as well as the timing of glucocorticoid exposure.
To the extent that increased (or inappropriately "normal") bone resorption contributes to the skeletal effects of glucocorticoids, there is ample molecular evidence that OPG and RANKL might be involved in the process. Glucocorticoids have been shown to suppress OPG mRNA expression in primary human osteoblasts and in osteoblast cell lines (254, 258), while up-regulating RANKL expression (254). Glucocorticoids also suppressed OPG protein production by human fibroblasts (254). DEX and prednisolone were shown to inhibit OPG production and increase RANKL production by human osteoblasts, leading to increased RANKL:OPG ratio (259). Glucocorticoids also stimulate osteoblasts to produce M-CSF (260), which is an important cofactor for osteoclastogenesis (261).
The lack of concordance regarding the effects of glucocorticoids on bone formation and bone resorption in these model systems is multifactorial. One important component of glucocorticoid biology that is lacking in most cell culture models is the marked depletion of osteoblast numbers in animals treated with glucocorticoids (251). Osteoblasts are an important source of both OPG and RANKL, and the effects of glucocorticoid-mediated osteoblast depletion on the relative balance of OPG and RANKL have not been well established. Regardless of net effects of glucocortoids on the RANKL:OPG ratio in vivo, animal studies account for this variable and might therefore have greater relevance than cell culture models. A limitation with animal models of glucocorticoid-induced bone loss is that they typically fail to include underlying diseases for which glucocorticoids are typically prescribed. This can be an advantage as well because the lack of underlying disease provides an opportunity to examine the direct consequences of glucocorticoids on bone.
Glucocorticoids decrease BMD in rats (262, 263), mice (251), and minipigs (264). In mice, BMD loss with prednisolone was associated with decreased osteocalcin and decreased bone formation, with no histological or biochemical evidence for increased bone resorption (251). In minipigs, prednisolone decreased biochemical parameters of bone resorption and bone formation while decreasing bone growth and bone strength (264). Prednisolone infusion in rats resulted in suppressed bone growth and decreases in cortical area and cortical strength (263). In another rat study, prednisolone infusion led to reductions in biochemical markers of bone resorption and bone formation and a paradoxical increase in cancellous bone volume (265). The prednisolone-related increase in bone volume in rats, confirmed in a study described immediately below (262), is not consistent with histological data from patients with glucocorticoid-induced osteoporosis (194) and is a recognized limitation of the model (251). That being said, the only data on the effects of OPG on glucocorticoid-induced bone loss was generated in rats, using sc implanted prednisolone pellets. Those data were presented in abstract form (262), and the main findings are described in Fig. 6 due to the lack of published data on RANKL inhibition in other glucocorticoid-induced bone loss models. In this study, 45 d of prednisolone were associated with significant bone loss and increased osteoclast surface, and OPG treatment prevented these changes. Despite the limitations of this model, the skeletal response to prednisolone in these rats shared several similarities to changes observed in humans, including significantly decreased BMD (243), histological evidence of increased bone resorption (194), and biochemical evidence of suppressed bone formation (244, 267, 268). The ability of OPG to markedly reduce osteoclasts and to prevent the prednisolone-related loss of BMD suggests that RANKL inhibition represents a potential new strategy for the treatment and prevention of glucocorticoid-induced bone loss.
|
The therapeutic potential of RANKL inhibition in disuse osteopenia is of interest because bisphosphonate therapy does not appear to consistently preserve both cortical and cancellous bone mass in unloading models. Long-term disuse osteopenia in dogs was suggested to be less responsive to bisphosphonate therapy compared with the efficacy of bisphosphonates in osteoporosis models (282). There was no evidence of suppression of cortical bone resorption in unloaded dogs treated with bisphosphonates, and cortical bone strength in treated animals remained significantly lower than the strength of normally loaded control bones (282). Cortical bone seems to be particularly sensitive to unloading in rats (283), and bisphosphonates are frequently ineffective at improving cortical bone strength in rat unloading models (284, 285, 286), even when cancellous bone strength was significantly improved (286). Cortical bone resistance to bisphosphonates was also suggested in a human bed rest study, wherein bisphosphonate treatment suppressed osteoclast numbers in cancellous bone but not in cortical bone (276). The reasons for the apparent resistance to bisphosphonates in unloaded cortical bone are not clear. One possibility is that the lower surface area in cortical bone provides lower uptake of bisphosphonates.
RANKL inhibition for the prevention of disuse osteopenia is appealing also because there is evidence that RANKL could be involved in the skeletal response to loading and unloading. Mechanical loading was shown to suppress osteoclast formation by decreasing RANKL expression in bone marrow stromal cells (287). In a cell culture model of unloading, simulated microgravity via vector averaging resulted in increased RANKL expression by a bone marrow-derived stromal cell line (288). Regardless of whether skeletal unloading promotes bone resorption and bone loss via increases in the RANKL:OPG ratio, animal studies in numerous unloading models show that RANKL inhibition consistently prevents cortical and cancellous bone loss.
Preclinical models of disuse osteopenia have been recently and extensively reviewed (289). Common rodent disuse models include tail suspension, sciatic neurectomy or crush, and spaceflight. OPG reduced cortical bone resorption and increased bone density in mouse nerve crush models (290, 291). OPG also prevented decreases in BMD and bone strength in a rat sciatic neurectomy model. This study showed strong correlations between femur BMD and femoral neck strength, consistent with the preservation of bone quality and improvements in bone mass (292). OPG preserved bone mass and increased cortical bone strength in tail-suspended mice (186). Another mouse tail suspension study showed that OPG preserved BMD in unloaded bones to a greater extent than was observed with two bisphosphonates (293). OPG was also able to prevent bone loss and improve cortical bone strength under the extreme conditions of microgravity. In that study, a single injection of OPG before a 12-d spaceflight caused significant increases in BMD and improved the strength of the femoral diaphysis of mice (294, 295). An elegant new mouse model of disuse osteopenia was recently described, in which botulism toxin was injected into the hindlimbs to induce temporary muscle paralysis (296). In this model, OPG treatment was shown to prevent deficits in cancellous bone volume as well as the unloading-related decreases in cortical thickness, volume, and strength (297). Skeletal unloading models in rats also show consistent bone preservation with OPG. OPG treatment of neurectomized rats improved BMD and bone strength at the femoral neck (292). OPG significantly increased the density and strength of cortical bone (298) in a rat unloading model that was nearly identical in design to one where bisphosphonate treatment had no such effects (284). More head-to-head studies, with a robust range of doses, are required before conclusions can be made regarding the differential effects of bisphosphonates vs. RANKL inhibitors in the prevention of cancellous or cortical bone loss in disuse osteopenia.
G. RANKL and RANKL inhibition in bone metastasis and multiple myeloma models
Nearly 120 yr ago, British surgeon Stephen Paget observed that certain cancer cells had a predilection to metastasize to bone, and he proposed that the bone microenvironment provided a fertile "soil" for the localization and growth of particular types of cancer cells (299). Almost 100 yr later, it was reported that products released from resorbing bone provided a chemotactic stimulus for directed tumor cell migration (300). This provided clues as to possible mechanisms for tumor cell osteotropism, and shortly thereafter it was reported that bone resorption in bone organ cultures stimulated the proliferation of tumor cells that possessed bone metastasizing properties (301). Consistent with these culture studies, it was subsequently shown in rats that cancer cells immediately adjacent to bone surfaces had significantly greater proliferation rates compared with those that were distant from bone (302). Stimulation of bone resorption was shown to specifically increase the proliferation rate of metastatic cancer cells in bone, but not in other tissues (303). These observations, coupled with the ability of tumor cells to stimulate osteoclastic bone resorption (304), provided a powerful rationale for exploring bone turnover suppression as a therapeutic strategy for bone metastasis. This rationale is predicated on the interruption of a "vicious cycle" wherein bone resorption attracts tumor cells and stimulates their growth, whereas tumor cells promote osteoclast formation and activity, the result of which is more bone resorption that further attracts and stimulates cancer cells.
These seed-and-soil and vicious cycle theories have gained abundant experimental support, such that adjuvant therapies used today to control bone metastases target bone turnover rather than tumor cells. These lesions lead to skeletal-related clinical events (SREs) that include severe bone pain, pathological fracture, and spinal cord compression. Multiple myeloma also has deletrious effects on bone by virtue of increasing bone resorption while suppressing bone formation (305). Bisphosphonates are the only class of drugs currently approved for the reduction of SREs in patients with bone metastases or multiple myeloma (MM). These agents act by binding directly to mineralized bone matrix and directly inhibiting the activity of osteoclasts that attempt to degrade bisphosphonate-laden bone (140). In 1983 the bisphosphonate clodronate was shown to reduce hypercalcemia, bone pain, and osteolytic progression in breast cancer patients with bone metastases (306). In 1991, pamidronate was shown to reduce bone destruction in breast cancer patients with bone metastases and in myeloma patients (307) and prostate cancer patients with bone metastases (308). Ten years later, zoledronic acid (ZOL) was shown to reduce SREs in patients with bone metastases (309). These trials helped to establish the utility of bone turnover suppression as a therapeutic strategy for patients with bone metastases or myeloma.
Based on the dominant roles for OPG and RANKL in the regulation of bone resorption, numerous investigators have examined the potential role of this pathway in cancer cell regulation of osteoclasts. In the mid 1970s, the existence of "osteoclast stimulating factor" and "osteoclast activating factor" was theorized by Mundy et al. (310) and by Galasko (304), respectively. The existence of these factors was proposed to account for the local stimulation of osteoclasts that is associated with myeloma and other cancer cells in bone. The exact identity of these factors was not provided by these early studies, but the activities they described would later prove to have significant overlap with the biological attributes of RANKL. Some 30 yr later, the possible involvement of the RANK/RANKL/OPG pathway in bone metastasis has been described in numerous forms of cancer, including myeloma, and in dozens of preclinical bone metastasis models. In-depth review articles on these subjects have been recently published (311, 312, 313).
Cancer cells have been shown to regulate OPG and/or RANKL in a variety of ways, some of which are described in Fig. 7. With few exceptions, regulation of OPG and RANKL by cancer cells appears to increase the RANKL:OPG ratio in a manner that favors bone resorption. Moving clockwise from the cancer cell in Fig. 7, prostate cancer cells (CaP) (314, 315, 316) and MM cells (317, 318, 319, 320, 321, 322) have been shown to express RANKL directly. Although breast cancer cells (CaBr) do not typically express RANKL directly (101, 323, 324), they can up-regulate RANKL expression by osteoblasts (OB) (101, 323) and bone marrow stromal cells (BMSC) (323, 324, 325). CaP cells can also up-regulate RANKL expression in osteoblasts (326). MM cells have been shown to up-regulate RANKL expression by BMSCs (327, 328), by T cells (48), and by endothelial cells (329). CaBr cells (101) and MM cells (327, 328, 330) can also down-regulate OPG production by osteoblasts and/or BMSCs. MM cells express syndecan on their surface, which sequesters and degrades heparin-binding proteins including OPG (331).
|
H. RANKL and RANKL inhibition in animal models of breast cancer metastasis to bone
Human MDA-231 breast cancer cells have been widely used in immunodeficient mice as an experimental bone metastasis model. These cells are capable of stimulating osteoclastogenesis and osteoclast activity in vitro (334), and their direct injection into the left ventricle of nude mice leads to the establishment of skeletal metastases that induce intense focal osteolysis (335). MDA-231 cells and other human breast cancer cells have been shown to express RANK (101, 336), which raises the possibility that RANKL and OPG might exert direct effects on RANK-expressing breast cancer cells. RANKL has been reported to stimulate signaling and invasiveness in human breast cancer cell lines (337), and RANKL was recently shown to stimulate the directed migration of human breast cancer cells in a manner that was inhibited by OPG (337). Although these direct responses to RANKL and OPG are intriguing, most animal models of breast cancer metastasis support the notion that bone turnover suppression is the primary mechanism by which RANKL inhibition prevents bone destruction and reduces skeletal tumor burden.
The MDA-231 tumor model was used to examine the therapeutic potential of RANKL inhibition in the prevention of bone metastases and the control of skeletal tumor growth. OPG treatment at the time of tumor cell inoculation was able to completely prevent the radiographic appearance of lytic lesions (338). Histological assessment revealed that skeletal tumor burden was 75% lower in OPG-treated animals compared with vehicle controls. The mechanism for this reduction in tumor burden was probably related to OPG inhibition of bone destruction because tumor burden in numerous soft tissue organs was unaffected by treatment. OPG treatment virtually eradicated tumor-associated osteoclasts in this and other studies, which is a somewhat novel observation that relates to OPG’s mechanism of action. OPG functions systemically as a soluble osteoclast inhibitor, and unlike bisphosphonates OPG does not require binding to bone surfaces for efficacy. This feature allows OPG and other soluble RANKL inhibitors such as RANK-Fc to eliminate osteoclasts within tumor foci (314). Consistent with this differential mechanism of action, two recent studies showed that OPG caused greater suppression of tumor-associated osteoclasts compared with bisphosphonate therapy in mice (339, 340).
Although prevention of bone metastases is a laudable therapeutic outcome, it is important that therapies are able to control the progression of established metastatic lesions in bone. OPG was tested in an MDA-231 treatment model, wherein treatment was delayed until osteolytic bone lesions were already apparent. These results were previously described in abstract form (341), and some key data are presented here. Experimental details are described in the legend for Fig. 8. Briefly, MDA-231 human breast cancer cells were injected into the left ventricle of immunodeficient mice, which were left untreated for 3 wk to allow for the development of lytic lesions in bone. One group was killed at wk 3 as a baseline control, and two other groups were treated with either vehicle or OPG-Fc for 7 d. During those final 7 d, radiographic lesion area and histological tumor burden increased by 3-fold in vehicle-treated animals, but there was no radiographic or histological progression in mice treated for 7 d with OPG. OPG treatment was also associated with a nearly complete elimination of tumor-associated osteoclasts (Fig. 8). This powerful antiosteoclast effect was a likely explanation for the reduction in skeletal tumor burden, based on the seed-and-soil hypothesis and the lack of OPG effects on extraskeletal tumor burden in this MDA-231 model (338).
|
I. RANKL and RANKL inhibition in animal models of prostate cancer metastasis to bone
While bone metastasis lesions in prostate cancer patients typically have an osteoblastic (osteosclerotic) appearance, there is evidence that bone resorption is also frequently increased in these lesions (346, 347). In animal models of prostate cancer metastasis, skeletal lesions frequently exhibit increased osteoclast activation and osteolysis (316, 348). OPG treatment was shown to markedly reduce osteoclast numbers in skeletal lesions that resulted from the direct intratibial injection of human C4–2B prostate cancer cells into immunodeficient mice (316). This therapeutic effect was accompanied by significant reductions in skeletal tumor burden and the prevention of radiographic bone destruction. OPG treatment had no significant effects on the sc growth of C4–2B tumors, which is consistent with the interpretation that OPG reduced skeletal tumor burden via suppression of bone turnover (316). A different human tumor cell line (LNCaP) that produces an osteoblastic response was used to examine the therapeutic effects of RANKL inhibition in mice. In this study, OPG significantly decreased the development of bone tumors and also reduced the progression of established tumors (348). These benefits were accompanied by significant reductions in osteoclast numbers, whereas OPG did not inhibit LNCaP cells grown sc or in culture. RANK-Fc was also used as a RANKL inhibitor in a mouse model wherein human LuCaP prostate cancer cells were injected into bone. RANK-Fc treatment was associated with a significant reduction in serum prostate-specific antigen and decreased skeletal tumor burden, whereas RANK-Fc did not suppress the sc growth of LuCaP cells (314). Thus, in these two models, suppression of skeletal tumor burden was probably related to the antiresorptive effects associated with RANKL inhibition.
This notion is supported by recent studies that directly compared OPG and the bisphosphonate ZOL on the intratibial growth of human C4–2 prostate cancer cells. OPG and ZOL both suppressed tumor-associated osteolysis and decreased osteoclast surfaces in tumor-bearing bones (344). OPG and ZOL both significantly decreased skeletal tumor burden when delivered in a preventative manner, whereas only OPG caused significant suppression of skeletal tumor burden when treatment was delayed until lytic tumors were established (344). Although OPG had no significant effects on the proliferation of cultured C4–2 cells, high doses of ZOL (10–100 µM) were associated with significant C4–2 growth suppression. These data suggest that the ability of both OPG and ZOL to reduce skeletal tumor burden in this prostate cancer model was related to osteoclast suppression rather than by direct antitumor effects. Other human prostate cancer cell lines have been shown to express RANK and to respond directly to RANKL stimulation via activation of signal transduction and increased migration in culture (349). OPG inhibited RANKL-mediated migration of these RANK-expressing prostate cancer cells, so it is therefore possible that some human prostate tumors might respond directly to RANKL and therefore to RANKL inhibition.
J. RANKL and RANKL inhibition in animal models of multiple myeloma
MM cells appear to have the broadest array of mechanisms to up-regulate bone resorption via the OPG-RANKL pathway. Unlike bone metastasis, MM cells originate in bone marrow where they clonally expand and spread to cause widespread osteolysis (350). Bone formation is typically suppressed in MM (351), which creates an additional therapeutic challenge that is not typically associated with bone metastases. Furthermore, widespread skeletal expansion of myeloma cells can lead to significant systemic bone loss (350). Systemic bone loss was a feature of a murine MM model, wherein RANKL-expressing 5T2MM myeloma cells were shown to cause extensive bone lesions in mice with increased osteoclasts, decreased BMD, and cortical thinning of long bones. OPG treatment prevented each of these changes (352). Another murine myeloma cell line (5T33MM), which also directly expresses RANKL, caused extensive skeletal tumor burden in mice such that 85% of bone marrow cells were of myeloma origin (353). OPG treatment of tumor-bearing mice led to significant reductions in tumor burden, including a reduction in serum paraprotein levels, and OPG-treated animals showed significantly longer survival than untreated mice. OPG had no significant effects on the survival or proliferation of cultured 5T33MM cells, and this survival benefit was therefore attributed to the eradication of tumor-associated osteoclasts (353). To the extent that myeloma cells might rely on the presence of osteoclasts for their continued growth (354), the ability of RANKL inhibitors to eradicate tumor-associated osteoclasts could provide an indirect antitumor effect that is independent of their ability to prevent bone destruction.
RANK-Fc, a RANKL inhibitor with a mechanism of action that is similar to that of OPG, also has therapeutic effects in myeloma models. Primary human myeloma cells were injected into immunodeficient mice bearing human bone xenografts, and these myeloma cells infiltrated the xenografts and induced osteoclast-mediated lytic lesions (328). RANK-Fc treatment, initiated after the establishment of skeletal lesions, prevented radiographic progression of disease and significantly reduced osteoclast numbers and tumor cell numbers in the xenografts. The antitumor effect of RANK-Fc was probably related to osteoclast suppression because RANK-Fc did not show direct cytotoxic effects on cultured myeloma cells (328). A human plasma cell leukemia cell line (ARH-77) also created extensive osteolytic lesions when injected into nude mice, leading to hind limb paralysis in 80% of animals. RANK-Fc prevented osteolysis in these mice and prevented hind limb paralysis, but without evidence of changes in skeletal tumor burden (328). ARH-77 cells are less dependent on the bone marrow microenvironment compared with MM cells, which may explain the lack of reduced tumor burden with RANK-Fc. The ability of RANK-Fc to prevent osteolysis in this model indicates that these plasma cell leukemia cells still rely on RANKL to cause bone destruction.
A comparison of RANK-Fc and bisphosphonates was recently described in a murine model, wherein primary human MM cells were injected into human bone xenografts. When these xenografts were implanted into immunocompromised mice, myeloma cells stimulated bone resorption and decreased BMD of the xenograft. Treatment of mice with RANK-Fc or with bisphosphonates resulted in decreased bone resorption and reduced skeletal tumor burden (345). RANK-Fc had no direct effects on cultured myeloma cells, which suggests that bone turnover suppression was the primary mechanism by which tumor burden was inhibited.
In contrast to these in vivo results, cell culture studies have suggested that OPG might serve as a paracrine survival factor for human MM cells by inhibiting the proapoptotic effects of TNF-related apoptosis-inducing ligand (TRAIL) (330). This is an intriguing observation, particularly in light of the ability of MM cells to down-regulate OPG production by osteoblasts or stromal cells (327, 328, 330) and to also sequester and degrade OPG via syndecan expression (331). Whether MM cells can promote their survival via this OPG-depleting mechanism is unclear because evidence suggests that OPG-TRAIL binding might have relatively low affinity compared with the affinity of TRAIL for its canonical receptors (355). Furthermore, the resistance of primary human myeloma cells and other cancer cells to TRAIL-induced apoptosis did not correlate with OPG expression by those cells (356, 357). Evidence for meaningful OPG-TRAIL interactions in vivo is minimal. If such interactions occurred, one might expect that TRAIL reduces OPG activity and vice versa. However, TRAIL-deficient knockout mice had no skeletal abnormalities, and TRAIL had no cytotoxic effects in OPG-deficient knockout mice (358).
Although there are few data from human studies on the potential for OPG-TRAIL interactions, flow cytometry studies conducted with bone marrow aspirates from postmenopausal subjects showed that saturating levels of OPG did not reduce the ability of anti-TRAIL antibodies to recognize TRAIL-positive cells (76). It is also interesting to note that RANK-Fc, which has no TRAIL-binding activity, suppresses myeloma tumor burden in animal models (328, 345) similar to the effects of OPG (353). Direct lentiviral transfer of the human OPG gene into human myeloma cells led to their secretion of therapeutic levels of OPG, which was shown to prevent bone destruction in immunodeficient mice without altering their in vitro growth properties (359). Whether endogenous TRAIL was a component of this model or other myeloma models is not known. But recombinant OPG treatment has not been shown to increase tumor burden in bone or in other organs in any experimental models of myeloma or bone metastasis, including the MDA-231 studies described above. MDA-231 cells have been shown to be TRAIL-sensitive because the injection of recombinant TRAIL into MDA-231 tumor-bearing mice led to significant antitumor effects and increased mouse survival (360). Despite this antitumor effect, 14 d of daily high-dose TRAIL injections in these mice had no effects on bone parameters. Thus, an interactive role of OPG-TRAIL with regard to the tumor immunosurveillance, tumor cell growth, or bone turnover in vivo remains speculative. It is also worth noting that the only recombinant RANKL inhibitor that is currently being tested in clinical trials is denosumab, a fully human anti-RANKL antibody that has no TRAIL-binding properties (361).
K. RANKL and RANKL inhibition in fracture and fracture repair
Fracture healing is an ordered process that gradually restores the structural integrity of bone, eventually returning it to approximately its original strength. Normal fracture repair after immobilization in casts is characterized by callus formation. The periosteal callus forms from the hematoma at the fracture site with the production of a collar of fibrous tissue, fibrocartilage, and hyaline cartilage around the fracture fragments. Subperiosteal new bone formation begins some distance from the fracture site and, through a process similar to endochondral ossification, advances toward the central and peripheral regions of the callus. A similar, less prolific response occurs at the endosteal surface. Over time, the callus is remodeled from randomly oriented woven bone to mature cortical bone.
The involvement of bone remodeling in fracture repair implicates the OPG/RANKL system in the repair process. OPG and RANKL mRNA expression was elevated in the fracture callus of mice during the repair of standard transverse long bone fractures (362). The biological implications of these changes are difficult to evaluate, but interventional studies with different RANKL inhibitory strategies suggest that RANKL plays a role in callus remodeling but not callus formation. The effects of recombinant OPG were studied in rats subjected to closed, transverse tibial fractures. In this study, OPG was delivered at a very high dose (10 mg/kg, three times per week) at the time of fracture and for 8 wk thereafter (363). This treatment regimen resulted in serum OPG levels (up to 35 µg/ml) that were at least 10,000-fold higher than levels found in normal rat serum. Histological assessment of fracture calluses at wk 3 and 8 showed greater than 90% reductions in osteoclast numbers in OPG-treated rats compared with time-matched vehicle controls. At wk 3, there were no reported differences in callus dimensions, BMC, or biomechanical strength of the callus between OPG- and vehicle-treated controls. These results indicate that continuous RANKL inhibition did not influence the formation of a normal facture callus. At wk 8, fracture calluses from OPG-treated rats were about 20% larger and had 30% greater BMC compared with vehicle controls. These findings are consistent with a delay in callus remodeling, a phenomenon that is typically associated with bisphosphonates and other antiresorptive therapies in similar models (364, 365, 366).
Biomechanical strength is perhaps the most relevant preclinical endpoint for fracture repair. Destructive mechanical tests, which are not feasible in humans, provide important information that integrates various aspects of the repair process. Although delayed callus remodeling in humans can indicate a risk for nonunion, it can also represent a benign and expected response to antiresorptive therapy. Data from fracture repair studies using bisphosphonates (364) or with the RANKL inhibitor RANK-Fc (367) have suggested that the delay in callus remodeling with antiresorptive therapy has a neutral or positive effect on biomechanical integrity by virtue of increasing callus dimensions. Consistent with this notion, the delayed callus remodeling associated with OPG treatment in rats resulted in larger calluses that were able to withstand the same amount of biomechanical force as did the calluses of untreated rats (363). Similar findings were obtained when OPG was delivered via a gene therapy vector in a similar rat model of fracture repair (368). OPG delivery was associated with a 3-fold reduction in osteoclast numbers in the fracture callus and a significant increase in callus BMC compared with vehicle controls. Despite these changes, the amount of force required to break the calluses was similar in OPG and vehicle controls (368).
Fracture repair studies were also performed in normal mice that were treated with the RANKL inhibitor RANK-Fc. In this study, RANK-Fc treatment was associated with the elimination of osteoclasts in long bone fracture calluses, but union occurred in all animals within 28 d (369). This study also provided the only available data on the effects of RANKL inhibitory treatment withdrawal on fracture calluses. Osteoclast numbers in fracture calluses were shown to return to normal levels within 15 d of stopping RANK-Fc treatment, and animals from this experimental group also achieved union. This finding highlights the potent yet reversible nature of osteoclast suppression with RANKL inhibitors. These preclinical fracture repair studies have provided experimental evidence to support the provocative yet supportable hypothesis that osteoclasts do not play an important role in the biomechanical aspects of fracture repair in mice or rats.
|
IV. Role and Regulation of OPG/RANKL in the Human Skeleton |
|---|
|
TopAbstractI. IntroductionII. Discovery and Validation...III. Pharmacological Effects of...IV. Role and Regulation...V. Clinical Therapeutics...VI. SummaryReferences |
|---|
A. Postmenopausal osteoporosis
The central role of estrogen deficiency in the pathogenesis of osteoporosis in postmenopausal women has been clearly established. Estrogen deficiency results in increased bone resorption over bone formation and net bone loss. Estrogen has been demonstrated to up-regulate gene expression and protein synthesis of OPG in human osteoblastic cells (104, 378) in vitro. It is logical from animal and in vitro studies that the RANKL/OPG system is involved in the pathogenesis of postmenopausal osteoporosis. However, studies designed to assess the relation between serum OPG and RANKL levels and bone metabolism/osteoporosis in postmenopausal women have yielded conflicting results. There is general agreement that OPG levels increase after the menopause (373, 379, 380, 381) and with GnRH agonist therapy (382), both conditions of low estrogen levels. Residual estrogen levels after the menopause correlate with bone density and fracture risk, yet the relationship of estrogen levels with serum OPG levels in postmenopausal women has been inconsistent across studies, with some supporting a weak relationship (375) and others no relationship (373, 379) or a negative correlation (381). No relationship was found between serum OPG and serum estradiol during GnRH agonist therapy in premenopausal women in one small study (382). Serum OPG levels have been found to be positively related to BMD (375, 379), negatively related to BMD (374, 383), or unrelated to BMD (372, 373) in postmenopausal women. Similarily, the evidence is conflicting about an association between markers of bone turnover and circulating OPG because some studies report an association (154, 372, 375) or an association only among osteoporotic women (384), and others report no association (373, 379). Results of studies examining the relationship between serum OPG levels and fracture are also inconsistent. Low OPG levels were reported to be associated with prevalent vertebral fractures in two studies (379, 384), and high serum OPG levels were associated with fracture in another study (383). The risk of subsequent fractures in postmenopausal women was not associated with OPG levels (154). Collectively, the data regarding serum OPG levels in postmenopausal women have been conflicting, and their potential clinical applicability as a marker of bone disease is not yet proven.
Serum RANKL levels have also been investigated in postmenopausal women and reported not to be associated with menopausal status or BMD (379, 385) but to be associated with nontraumatic fracture (385). However, in one of the studies over half of the women (54.9%) had concentrations of RANKL that were undetectable (379), and in the other study 40% of samples were below the detection limit of the assay (385). This raises the concern that the current assay for RANKL may not be ideal for the study of serum levels of this cytokine in postmenopausal women. Levels of RANKL in the bone microenvironement may be more relevant, but only a few studies have attempted to investigate this. The relationship between menopause, estrogen use, and PTH on RANKL and OPG mRNA in human bone tissue samples was examined in pre- and postmenopausal women (386). No difference was found in expression of RANKL, OPG, or the ratio between premenopausal women, postmenopausal women, and postmenopausal women taking estrogen. In a separate study of cadaveric proximal femur bone biopsies, RANKL/OPG mRNA ratio was strongly associated with histomorphometric indices of bone turnover in trabecular bone (87). Similarly, the RANKL/OPG mRNA ratio in transiliac bone biopsies of women sustaining a hip fracture was increased compared with controls with osteoarthritis (387). In addition, RANKL/OPG mRNA ratio was significantly greater in bone from the proximal femur of hip fractures than controls with osteoarthritis (388). These studies suggest that a relative increase in the expression of molecular promoters of osteoclast formation and activity in the bone microenvironment may underlie the susceptibility to fracture, but more detailed prospective studies are needed to prove this conclusively. However, the data on tissue levels, unlike that regarding serum levels, are consistent with the notion that increased bone turnover is an important factor in fracture risk.
Few studies have examined the response to osteoporosis therapy and change in the OPG/RANKL/RANK system in postmenopausal women. Cell culture studies show a positive effect of estrogen (104, 378) and raloxifene (30) on OPG production by human osteoblastic cells, but correlation in vivo between estrogen levels and serum OPG levels has been inconsistent. In postmenopausal women receiving hormone replacement therapy, OPG levels declined by 3 months and positively correlated with changes in bone turnover markers at 3 months (389). However, no association was seen between the change in OPG and BMD after 1 yr of estrogen therapy (389). Eghbali-Fatourechi et al. (76) examined the cell surface concentrations of RANKL in bone marrow cells (osteoblastic, T, and B cells) of premenopausal and postmenopausal women, with and without estrogen treatment. RANKL expression per cell correlated with bone resorption markers and inversely with serum estradiol levels (76). The findings suggested that up-regulation of RANKL on bone marrow cells is correlated with increased bone resorption induced by estrogen deficiency. In contrast, a study of oral bisphosphonates in postmenopausal women found a positive correlation between changes in serum OPG levels and BMD response (390). Serum RANKL levels in this study did not change during bisphosphonate treatment. The effects of bisphosphonates on serum OPG may reflect changes in osteoblast cell pool or OPG production. Similar effects of bisphosphonates on serum OPG levels have been reported in Paget’s disease of bone (PDB) (see Section IV.D). Treatment with human PTH (1-34) in women receiving chronic glucocorticoid and hormone replacement therapy resulted in rapid increase in serum RANKL by 1 month and a decrease in serum OPG levels at 6 and 12 months of therapy (93). The interpretation of these results is complicated by the concomitant use of glucocorticoids and hormone replacement therapy.
B. Men
There are fewer studies investigating the OPG/RANKL system in men, but they rely on similar methodology as those in women: measuring serum concentrations of these cytokines. The limitations to this approach have already been outlined. As with women, serum OPG levels increase with age in men (372, 373, 376, 380, 381, 391, 392). OPG levels in men have been correlated inversely with BMD (373, 392), borderline positively with BMD (372), or not at all (376). The relationship of bone turnover and OPG levels is also conflicting. An inverse correlation of OPG with osteocalcin has been reported (372, 380, 392). However, a positive correlation with bone resorption markers but no correlation with bone formation markers has been reported (373). Yet another study reported a negative correlation with bone resorption markers (376). The relationship of OPG levels and sex steroids in men has also been examined. A positive correlation with free testosterone index and free estradiol index has been reported by one group (376). A negative correlation with bioavailable testosterone and bioavailable estradiol (373) and also with free testosterone index (391) has been reported. Testosterone replacement appeared to lower OPG levels in men made acutely hypogonadal by GnRH agonist and who also received aromatase inhibitor blockade of estrogen production (106). Men with prostate cancer who receive androgen deprivation therapy experience bone loss. However, studies looking at serum OPG/RANKL have largely investigated these as a marker of bone metastasis progression and as such do not shed light on the independent contribution of androgen deprivation to serum levels. As in postmenopausal women, the data in men are conflicting regarding the utility of serum levels of OPG to reflect bone disease states.
C. Glucocorticoid treatment
Glucocorticoids are used as treatment for a variety of conditions because of their modulation of inflammation and the immune system. The detrimental skeletal effects of glucocorticoid therapy include suppression of bone formation by increasing osteoblast apoptosis and, to a lesser extent, and perhaps indirectly, increase in bone resorption. However, glucocorticoids also reduce intestinal calcium absorption, increase renal calcium excretion, and disturb vitamin D metabolism. All of these actions have implications to adversely affect the skeleton. Investigations of glucocorticoid effects on bone in humans are complicated by the underlying disease state for which they are given, and drawing common conclusions about skeletal effects across disease states for which glucocorticoids are given can be problematic. Nonetheless, the effects of glucocorticoids on bone remodeling have been extensively studied in animal models, and a role of the OPG/RANKL/RANK system in their effects has been discussed earlier. As with most of the human studies of the OPG/RANKL/RANK system, those investigating these cytokines during glucocorticoid therapy have focused on evaluating serum levels. In a study of patients with Crohn’s disease, serum OPG levels decreased, whereas RANKL tended to increase with glucocorticoid therapy compared with baseline (244). Similarily, in a group of patients with glomerulonephritis, serum OPG levels were decreased with glucocorticoids compared with baseline, whereas markers of bone formation were transiently decreased and a marker of bone resorption was increased (267). There also was a positive association between OPG and bone density in this group of patients. A cross-sectional study of cardiac transplant recipients found a positive correlation between serum OPG levels and BMD, and a lower average serum OPG level was seen in those with prevalent vertebral fractures (393). In the same study, serum OPG levels fell after transplant, and bone loss based on BMD measurement was positively associated with the decrease in OPG level (393). Transplant recipients received not only glucocorticoids but also other immunosuppressive agents that may have independent effects on bone remodeling (especially cyclosporine). Although it is difficult to draw firm conclusions from the few human studies to date, there does appear to be some consistency in the effect of glucocorticoids to lower serum OPG levels in vivo that may partially explain the adverse skeletal effects of these drugs.
D. Paget’s disease of bone
PDB is a focal disorder of accelerated and disordered bone remodeling. Osteoclast-mediated bone resorption is increased with a compensatory increase in new bone formation. However, the resulting bone at affected sites lacks normal organization and is a mosaic of woven and lamellar bone. Osteoclasts in pagetic bone are increased in number and size and have abnormal appearance with substantially more nuclei than normal and with nuclear and cytoplasmic inclusions. Osteoblasts in pagetic bone are also numerous and active, rapidly synthesizing new bone. The increase in both bone formation and resorption at pagetic sites can be detected in the peripheral circulation where increased levels of both markers of formation and resorption are found. Although the exact etiology of Paget’s disease remains to be determined, there is abundant evidence of the involvement of the OPG/RANKL/RANK system and the NF-B signaling pathway in the pathogenesis of the disorder. RANKL expression is abundant in cultures of stromal cells or marrow cells derived from pagetic bone compared with nonaffected bone of the same individuals (394, 395). Inactivating mutations in the OPG gene have been demonstrated in juvenile PDB (396, 397, 398). Constitutive activation of RANK has been demonstrated in familial expansile osteolysis, another disorder with abnormal bone remodeling, and in familial early onset PDB (398). Most recently, mutations in the sequestosome 1 gene, a scaffold protein in the NF-B signaling pathway, have been demonstrated in some cases of familial and sporadic PDB (399, 400, 401, 402, 403, 404).
Complimenting the in vitro cell system and genetic evidence are several studies examining serum levels of OPG and RANKL in PDB and their response to treatment with bisphosphonates. OPG levels are consistently increased in patients with PDB compared with controls (405, 406, 407). This lends support to the concept of increased OPG circulating in conditions of elevated turnover potentially as a protective mechanism, although the relationship between OPG levels and bone turnover in unaffected individuals is less clear as discussed earlier. RANKL levels are elevated in PDB in some studies (406) but not in others (407). The response of OPG and RANKL levels in PDB to treatment with bisphosphonates has also been investigated. The primary mechanism of action of bisphosphonates is to inhibit bone resorption by impairing key metabolic pathways of osteoclasts, although some in vitro cell studies and animal studies suggest potential effects on osteoblasts (408, 409, 410, 411). OPG levels in PDB have been reported to increase modestly after treatment with pamidronate (406) and risedronate (407) and to decrease after treatment with tiludronate (405). These differing results may be due to differences in the mechanism of action of amino bisphosphonates compared with non-amino bisphosphonates on osteoclasts and osteoblasts, or due to differences in study populations. The increase in OPG levels after bisphosphonate treatment that lowers bone turnover does not support the general concept that serum OPG levels correlate with bone turnover in all situations.
E. Rheumatoid arthritis
RA is characterized by chronic inflammation of the synovium, the predominance of proerosive mediators, and subsequent destruction of bone and cartilage. Many investigations have implicated inflammatory cytokines (IL-1, IL-1β, IL-6, IL-11, TNF-) as important mediators stimulating bone lysis in RA. However, the critical role of the OPG/RANKL/RANK system in bone remodeling and osteoclast development has led to the concept of "osteoimmunology" to more accurately encompass the complex interactions between the immune system and the skeletal system in RA and other inflammatory arthritis conditions. In vitro cell system models and animal models discussed earlier support this concept, and the human in vivo data will be reviewed here.
The production of inflammatory cytokines in the inflamed joint is likely to up-regulate RANKL and RANK (90, 412), which promotes osteoclast development and bone destruction. Synovial cells isolated from rheumatoid joints have been found to express RANKL, RANK, and OPG (90, 412, 413, 414, 415). Activated T lymphocytes within arthritic joints have also been shown to express RANKL and therefore might contribute to the promotion of bone erosions (413, 416). Higher levels of RANKL expression have been observed in synovial tissue from active RA patients vs. those with quiescent disease (413). Also, in synovial tissue from active RA, RANKL expression was seen but OPG was not (33), and the ratio of OPG:RANKL in synovial fluid has been correlated positively with radiological scores (417). Both T cells and non-T cells have been observed to be the source of RANKL in the synovium (413, 418). These results suggest that the deficient production of OPG or the ratio of RANKL:OPG in the joint may underlie the development of erosions.
Peripheral measurements of serum or cellular levels of OPG and RANKL have also been investigated in RA. The ratio of circulating OPG/RANKL levels was shown to negatively correlate with progression of radiographic joint damage in newly diagnosed active RA patients (419). However, another group reported that whereas OPG and RANKL serum levels were higher in RA patients compared with controls, the ratio of OPG/RANKL was similar between the two groups, and no correlation was observed between disease score and serum levels in RA patients (53). RANKL and OPG mRNA levels have been found to be elevated in peripheral blood mononuclear cells in RA patients compared with controls, and the ratio of OPG/RANKL tended to be higher in the synovium than in the peripheral blood mononuclear cells in early RA patients (418). The use of serum levels of OPG or RANKL to understand disease progression requires additional study.
In addition to focal bone erosions and periarticular bone loss, systemic bone loss and increased fracture risk are important components of RA (224, 239, 420, 421, 422). The mechanisms responsible for bone loss at skeletal sites such as the lumbar spine and hip, in the absence of obvious inflammation, are not well understood. Certainly the high circulating levels of TNF- and IL-1 that are common in RA have the potential to promote osteoclastogenesis throughout the skeleton (223, 237, 423, 424, 425). Systemic stimulation of bone resorption in RA patients is suggested by increases in circulating biochemical markers of bone resorption (426) and increased osteoclastogenic potential of their bone marrow (426, 427). Recently, serum RANKL levels were reported to be increased in RA patients, and serum RANKL was inversely correlated with BMD of the hip and positively correlated with circulating bone resorption markers (239). Furthermore, the RANKL:OPG ratio was significantly elevated in serum of RA patients, and this increase was predictive of joint destruction (419). These results are consistent with the notion that RANKL might play a role in systemic bone loss in RA patients, in addition to its putative role in focal bone loss.
Several studies have reported that treatment of RA is associated with changes in OPG and RANKL in the peripheral circulation. Ziolkowska et al. (53) reported normalization of serum OPG and RANKL by 14–22 wk of therapy with infliximab, although the RANKL:OPG ratio did not change during therapy. Interestingly, higher baseline levels of OPG or RANKL were associated with larger decreases with therapy. Vis et al. (428) found that infliximab treatment in RA caused a decline in serum RANKL and the OPG/RANKL ratio but no change in serum OPG. Infliximab treatment was also associated with stabilization of BMD at the hip and spine but continued bone loss in the hand (428). This agrees with in vitro cell culture studies showing a decrease in RANK expression in peripheral blood mononuclear cells treated with infliximab and inhibition of RANKL expression and increase in OPG secretion in cultured synovial cells treated with infliximab, methotrexate, sulfasalazine, or hydroxychloroquine (429). Collectively, these investigations suggest that disease-modifying antirheumatic drugs exert some of their effects through the OPG/RANKL/RANK system.
|
V. Clinical Therapeutics Targeting the RANK/RANKL/OPG System |
|---|
|
TopAbstractI. IntroductionII. Discovery and Validation...III. Pharmacological Effects of...IV. Role and Regulation...V. Clinical Therapeutics...VI. SummaryReferences |
|---|
The second OPG construct to enter the clinic was OPG-Fc (also known as AMGN-0007; Fig. 2). With OPG-Fc, the Fc fragment of IgG1 was fused to the amino terminus of the active domain of OPG, which conferred a longer half-life compared with Fc-OPG. OPG-Fc was evaluated in a phase 1, active-controlled, dose escalation study in patients with MM or breast cancer-related bone metastases (23). A dose-dependent reduction in uNTX occurred within 1 d of dosing with OPG-Fc, and uNTX reached a nadir of approximately –60 and –84% in the MM and breast cancer strata, respectively, at the higher doses (1.0 and 3.0 mg/kg). Suppression of bone resorption was maintained throughout the duration of the study (56 d). The pharmacodynamic profile of OPG-Fc was generally similar to that observed with the open-label active control, pamidronate.
The results of these clinical trials suggested therapeutic potential for RANKL inhibition in the treatment of bone loss. The clinical development of these OPG constructs was stopped in favor of the development of denosumab, a fully human monoclonal antibody that inhibited RANKL. An important practical consideration was that denosumab had a significantly longer circulating half-life compared with OPG-Fc (430), which would allow for reduced dosing frequency.
Another potential advantage of a fully human anti-RANKL antibody is the low potential for immune responses (431), a theoretical attribute that has been borne out in clinical trials with denosumab. Denosumab and numerous other fully human therapeutic antibodies have been made in transgenic mice, wherein their murine Ig genes have been deleted and replaced with human orthologs (431, 432). Denosumab was obtained using Abgenix’s transgenic XenoMouse system (Abgenix, Fremont, CA). These mice were immunized with human RANKL, leading to an immune response that resulted in antigen-specific fully human IgGs that bind and inhibit RANKL (433). One such IgG (AMG 162) was cloned and expressed in Chinese hamster ovary cells and purified for large-scale production of clinical-grade material.
A. Mechanism of action of denosumab
Denosumab is a fully human monoclonal antibody with a high affinity (Kd 3 x 10–12 M) and specificity for RANKL and can thus bind and neutralize the activity of human RANKL in a similar fashion to the action of OPG. Denosumab does not cross-react with TNF-, TNF-β, TRAIL, or CD40 ligand (361).
Because the activity of denosumab is primate specific, it cannot be tested in mouse or rat models of bone disease. However, because OPG and denosumab have similar mechanisms of action, OPG has been used as a surrogate for denosumab to examine the effects of RANKL inhibition in numerous mouse and rat models of bone disease. Human OPG-Fc and denosumab were shown to have similar effects on suppressing bone resorption and increasing BMD in knockin mice that were genetically engineered to exclusively express a chimeric (human/murine) form of RANKL that is recognized by both OPG and denosumab (434). One potential advantage of denosumab is that it has a significantly longer half-life compared with human OPG-Fc (430), which promotes more sustained suppression of bone resorption and thereby permits infrequent dosing.
Ongoing phase 2 and 3 trials are evaluating denosumab in the treatment and prevention of postmenopausal osteoporosis, the treatment of bone loss associated with sex hormone ablative therapy in subjects with cancer, the treatment of bone erosions associated with RA, the treatment of bone metastases to delay or prevent skeletal-related events in subjects with advanced cancer, and the prolongation of bone metastasis-free survival in subjects with hormone-refractory prostate cancer.
In these studies, denosumab is administered sc, at intervals ranging from every 4 wk to every 6 months, depending on the therapeutic area being evaluated. This long dosing interval holds the potential to facilitate treatment compliance, which is increasingly being recognized as a major obstacle in therapy for these conditions.
B. Clinical studies of denosumab in postmenopausal women with osteoporosis
Denosumab administration in postmenopausal women with low BMD resulted in gains in BMD at the total hip, femoral neck, and distal radius at 12 and 24 months (230). Those gains in BMD were associated with decreases in bone resorption. This ongoing phase 2 study is evaluating the effects of denosumab (6, 14, or 30 mg every 3 months or 14, 60, 100, or 210 mg every 6 months), open-label oral alendronate 70 mg once weekly, and placebo on BMD of the lumbar spine, total hip, femoral neck, and distal radius at 1, 3, 6, and 12 months. Changes in bone turnover are also being regularly assessed by measurement of serum and urine telopeptides and bone-specific alkaline phosphatase.
Treatment with denosumab resulted in a mean increase in BMD at the lumbar spine of 3.0 to 6.7% (vs. an increase of 4.6% with alendronate and a loss of 0.8% with placebo), at the total hip of 1.9 to 3.6% (vs. an increase of 2% with alendronate and a loss of 0.6% with placebo), and at the distal third of the radius of 0.4 to 1.3% (vs. decreases of 0.5% with alendronate and 2.0% with placebo).
In a planned extension of this study, denosumab treatment for 2 yr resulted in persistent reductions in bone resorption and continued, significant increases in BMD from baseline in the 337 women (82%) who completed 2 yr of study (38 placebo, 259 denosumab, 40 alendronate) (435). Across dose groups, women treated with denosumab for 2 yr had significantly greater improvements in BMD compared with placebo (P < 0.001) at the lumbar spine, total hip, and distal radius; these improvements were similar to or greater than those observed in the open-label alendronate group. During yr 2, in response to denosumab, serum C-telopeptide (CTx) and uNTX/creatinine demonstrated continued suppression. The duration of the suppression of bone turnover appeared to be dose dependent. Optimal increases in BMD appeared to occur at denosumab doses of 30 and 60 mg, administered at 3 and 6 months, respectively.
C. Clinical studies of denosumab in rheumatoid arthritis
Based on the implication of RANKL involvement in the bone and cartilage changes characteristic of RA and the positive impact of RANKL inhibition in rat models of RA (26), the effects of denosumab (60 or 180 mg every 6 months) are being investigated in subjects with RA (436, 437, 438). In addition to studying BMD and bone turnover marker changes, a study by Cohen et al. (436) is assessing bone erosions using magnetic resonance imaging (MRI). Significant reductions in the progression of bone erosion were reflected in MRI erosion scores from this population of 227 patients with mild to moderately active RA (for at least 6 months) who had received a single 180-mg dose of denosumab compared with placebo (P < 0.019).
Results from a 6-month interim analysis were published by Dore et al. (437) showing that treatment with denosumab resulted in significant increases in BMD at the lumbar spine, total hip, and femoral neck, as well as suppression of bone turnover markers serum CTx type I (CTx-I) and serum procollagen type I N-propeptide at 3 and 6 months compared with placebo. In addition, significant suppression of the cartilage turnover marker, CTx-II/creatinine, was observed at 3 months. Of particular note, approximately 40% of patients in each treatment group had baseline CTx-II levels below quantifiable limits. However, at 3 months, this percentage increased to 81 and 89% in the 60- and 180-mg denosumab groups, respectively, and then returned to approximately 40% in all groups at 6 months, before the next denosumab dose (438).
In addition, a larger proportion of patients in the placebo group had an increase in MRI erosion score from baseline to 6 months than either denosumab dose group [53, 42, and 30% in the placebo, 60-mg, and 180-mg groups, respectively (P = 0.01 for 180-mg group vs. placebo)]. Adjustment for baseline erosion score did not affect the results (436).
D. Clinical studies of denosumab in patients with advanced cancer with bone metastases or multiple myeloma
In advanced cancer, bone is a common site of metastasis. Bone metastases lead to skeletal complications (such as vertebral or nonvertebral bone fractures, spinal cord compression, surgery, or radiation therapy to bone) that are associated with increased bone turnover. Studies in patients with metastatic bone disease (439, 440, 441) suggest that denosumab suppresses bone resorption and normalizes bone turnover, as measured by serum and urinary telopeptides. Although these studies provide intriguing results, the relationship between bone turnover suppression and clinical outcomes of skeletal metastasis with denosumab therapy remains to be better defined by data from larger prospective trials with sufficient statistical power. Large prospective trials comparing the effects of denosumab or bisphosphonates on skeletal-related events in metastatic cancer are currently under way.
A recent double-blind phase 1 study of denosumab published by Body et al. (442) in patients with bone metastases from breast cancer (n = 29 women) and patients with MM (n = 25; 14 men and 11 women) showed that denosumab was well tolerated, with an adverse event profile that was not significantly different from placebo patients. Denosumab therapy was associated with sustained decreases (
84 d) in uNTX and serum NTX after a single sc dose (0.1, 0.3, 1.0, or 3.0 mg/kg). These decreases were similar in magnitude, but of numerically longer duration, vs. those observed in a comparator group receiving an iv dose of pamidronate 90 mg. At the 3.0 mg/kg dose, denosumab demonstrated mean half-lives of 46.3 and 33.3 d in patients with breast cancer and MM, respectively. Population pharmacokinetic (PK) and pharmacodynamic (PD) analyses reported by Peterson et al. (440) suggest that denosumab PK and PD in breast cancer patients with bone metastases are consistent with those observed in other patient populations, and treatment with denosumab caused rapid and substantial suppression (>60% from baseline) of uNTX.
Lipton et al. (439) reported interim (13-wk) data from a phase 2 study of 255 patients with breast cancer and bone metastases who had not been previously treated with iv bisphosphonates (IV BP). Subjects were stratified by chemotherapy or hormonal therapy and randomized to receive either denosumab or IV BP in one of six cohorts [five denosumab (double blind), one IV BP (open label)]. Based on PK and PD assessments demonstrating decreased uNTX in cancer patients in response to treatment with denosumab (442), the primary endpoint of this study was the percentage change from baseline to wk 13 in uNTX, corrected for creatinine. The denosumab 120-mg every 4 wk dose resulted in the greatest percentage decrease from baseline in uNTX. Treatment with denosumab (n = 212) resulted in rapid and sustained suppression of bone turnover and was similar to IV BP at reducing the risk of skeletal-related events (439).
Suarez et al. (441) compared 49 (of a planned 135) cancer patients with bone metastases on established IV BP therapy who were randomized to receive denosumab (33 of 49) or continue with an established IV BP treatment (16 of 49; most of these patients received ZOL). Subjects had prostate (n = 24), breast (n = 20), or other (n = 5) cancers (including MM). The proportion of subjects with uNTX less than 50 nM bone collagen equivalent/mM creatinine at wk 13 was greater with denosumab (pooled arms) treatment than with continued IV BP treatment: 76% (95% confidence interval 60.3–91.2) vs. 38% (95% confidence interval 18.5–61.4; P = 0.015), respectively. Based on this interim assessment, denosumab appears to normalize uNTX more frequently than IV BP in patients with elevated uNTX, despite established IV BP therapy, across tumor types (441). Thus, it appears that these patients had become refractory to BP treatment, because they showed elevated uNTX despite receiving their established BP regimens. In contrast, patients who were switched to denosumab subsequently showed an overall normalization of uNTX levels. This study showed a safety profile similar to what might be typically observed in patients receiving standard chemotherapy regimens for the cancers studied.
E. Evaluation of adverse events profiles in clinical studies of denosumab
The largest and most recent single-study sample in cancer patients receiving denosumab is that of the 255 breast cancer patients reported by Lipton et al. (439) (discussed in Section IV.D). These authors report a denosumab safety profile similar to that seen in advanced breast cancer patients receiving systemic cancer treatment.
The largest sample size published to date is the 24-month report by Lewiecki et al. (435) of data from osteopenic or osteoporotic women comparing denosumab (n = 314) with alendronate (n = 46) and placebo (n = 46). Few significant differences were observed between the profiles of adverse events in the denosumab groups and those in the placebo group and the alendronate group. Hypertension and urinary tract infections were reported more often in the denosumab group than the placebo group (P = 0.01), whereas dyspepsia and osteoarthritis appeared to occur with significantly greater frequency in the open-label alendronate group (P = 0.01).
Indeed, the percentages of patients experiencing adverse events (92.0, 93.5, and 93.5%, respectively) and serious adverse events (13.4, 13.0, and 8.7%, respectively) in the denosumab, alendronate, and placebo groups were similar (435). The incidence and nature of adverse events continue to be examined in ongoing studies of denosumab, including assessments for unusual immune responses, infections, and neoplasms.
|
VI. Summary |
|---|
|
TopAbstractI. IntroductionII. Discovery and Validation...III. Pharmacological Effects of...IV. Role and Regulation...V. Clinical Therapeutics...VI. SummaryReferences |
|---|
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Conflicts of Interest: A.E.K. is involved in a clinical trial with Amgen and has received lecture fees from Amgen Inc. S.K. has nothing to declare. P.J.K. is an employee and shareholder in Amgen Inc.
First Published Online December 5, 2007
Abbreviations: BMC, Bone mineral content; BMD, bone mineral density; BMSC, bone marrow stromal cells; CaBr, breast cancer; CaP, prostate cancer; CTX, C-telopeptide; DDH, death domain homologous; DEX, dexamethasone; DXA, dual energy x-ray absorptiometry; IBA, ibandronate; IV BP, iv bisphosphonates; M-CSF, macrophage colony-stimulated factor; microCT, microcomputed tomography; MRI, magnetic resonance imaging; NFAT, nuclear factor of activated T cells; NF-B, nuclear factor-B; OB, osteoblast; OCIF, osteoclastogenesis inhibitory factor; ODF, osteoclast differentiation factor; OPG, osteoprotegerin; OPGL, OPG ligand; ORX, orchiectomy; OVX, ovariectomy; PD, pharmacodynamic; PDB, Paget’s disease of bone; PK, pharmacokinetic; RA, rheumatoid arthritis; RANK, receptor activator of NF-B; RANKL, RANK ligand; SRE, skeletal-related clinical event; TRAF, TNF receptor associated factor; TRAIL, TNF-related apoptosis-inducing ligand; TRANCE, TNF-related activation-induced cytokine; uNTX, urinary N-telopeptide; ZOL, zoledronic acid.
Received for publication May 25, 2007. Accepted for publication November 15, 2007.
|
References |
|---|
|
TopAbstractI. IntroductionII. Discovery and Validation...III. Pharmacological Effects of...IV. Role and Regulation...V. Clinical Therapeutics...VI. SummaryReferences |
|---|
B ligand. J Biol Chem 280:41155–41164, PTH, PTHrP, and 1,25(OH)2D3. J Bone Miner Res 14:1478–1485[CrossRef][Medline]B and osteoprotegerin expression by human microvascular endothelial cells, regulation by inflammatory cytokines, and role in human osteoclastogenesis. J Biol Chem 276:20659–20672. Diabetologia 48:561–568[CrossRef][Medline]B ligand and their differential expression in bone and thymus. Endocrinology 142:1419–1426B ligand. J Biol Chem 281:36846–36855B ligand (RANKL) in T lymphocytes: a potential role in multiple myeloma bone disease. Blood 100:4615–4621B ligand derived from osteoblasts. J Immunol 178:192–200B ligand: modulation of the expression by osteotropic factors and cytokines. Biochem Biophys Res Comm 275:768–775[CrossRef][Medline]B ligand in serum of rheumatoid arthritis patients and their normalization after anti-tumor necrosis factor treatment. Arthritis Rheum 46:1744–1753[CrossRef][Medline]B in osteoclasts. Biochem Biophys Res Comm 305:211–214[CrossRef][Medline]Vβ3-induced, NF-B-dependent survival factor for endothelial cells. J Biol Chem 275:20959–20962B (RANK) in endothelial cells. J Biol Chem 278:39548–39557B ligand) mRNA and protein in skeletal and extraskeletal tissues. Bone 25:525–534[Medline]B ligand and osteoprotegerin production by human osteoblastic and coronary artery smooth muscle cells. Biochem Biophys Res Comm 280:334–339[CrossRef][Medline]B ligand (RANKL) and osteoprotegerin regulation and partial rescue of osteoclastogenesis in Runx2–/– mice by RANKL transgene. J Biol Chem 278:23971–23977B ligand and inhibits osteoprotegerin expression via protein kinase A activation of cAMP-response element-binding protein. J Biol Chem 277:48868–48875B ligand to osteoprotegerin correlates with bone remodeling in normal human cancellous bone but not in osteoarthritis. J Bone Miner Res 16:1015–1027[CrossRef][Medline]B ligand (RANKL) regulate osteoclast formation by cells in the human rheumatoid arthritic joint. Rheumatology 40:623–630B ligand (RANKL)/osteoprotegerin (OPG) ratio is increased in severe osteolysis. Am J Pathol 163:2021–2031B ligand, osteoprotegerin, and interleukin-6 levels in patients with glucocorticoid-induced osteoporosis treated with human parathyroid hormone (1-34). J Clin Endocrinol Metab 89:3332–3336B ligand-osteoprotegerin ratio predicts survival in multiple myeloma: proposal for a novel prognostic index. Blood 102:1064–1069, but not interleukin 6, stimulate osteoprotegerin ligand gene expression in human osteoblastic cells. Bone 25:255–259[Medline]B and tumor necrosis factor from T cells. Proc Natl Acad Sci USA 100:125–130,25 dihydroxyvitamin D3 rapidly regulates the mouse osteoprotegerin gene through dual pathways. J Bone Miner Res 19:1411–1419[CrossRef][Medline]B ligand and support osteoclast differentiation. J Bone Miner Res 15:1459–1466[CrossRef][Medline]-mediated increase in CD11bhi osteoclast precursors but is essential for mature osteoclast formation in TNF -mediated inflammatory arthritis. J Bone Miner Res 19:207–213[CrossRef][Medline] induces differentiation of and bone resorption by osteoclasts. J Biol Chem 275:4858–4864 potently activates osteoclasts, through a direct action independent of and strongly synergistic with RANKL. Endocrinology 143:1108–1118 stimulates osteoclast differentiation by a mechanism independent of the ODF/RANKL-RANK interaction. J Exp Med 191:275–285 and IL-1β induced osteoclast formation and bone resorption is dependent on RANK signal transduction. J Bone Miner Res 16:S379 induces osteoclastogenesis by direct stimulation of macrophages exposed to permissive levels of RANK ligand. J Clin Invest 106:1481–1488[Medline]B ligand, receptor activator of nuclear factor B, and osteoprotegerin messenger mRNA in aged ovariectomized rat bones. J Bone Miner Res 16:1416–1425[CrossRef][Medline]-mediated joint destruction is inhibited by targeting osteoclasts with osteoprotegerin. Arthritis Rheum 46:785–792[CrossRef][Medline]B ligand in Monckeberg’s sclerosis and atherosclerosis. J Clin Endocrinol Metab 89:4104–4112. Endocrinology 142:2205–2212B ligand-mediated osteoclastogenesis. Endocrinology 145:2384–2391B, and its ligand, osteoprotegerin, bone-type alkaline phosphatase, and tartrate-resistant acid phosphatase in ovariectomized rats. J Cell Biochem 93:503–512[CrossRef][Medline]B (RANK) receptor complex by novel osteoprotegerin-like peptidomimetics restores bone loss in vivo. J Biol Chem 279:8269–8277-dihydroxytestosterone. J Bone Miner Res 17:2080–2086[CrossRef][Medline] antibodies and osteoprotegerin decrease systemic bone loss associated with inflammation through distinct mechanisms in collagen-induced arthritis. Bone 35:1200–1207[Medline]B/receptor activator of nuclear factor B ligand system in cartilage. Arthritis Rheum 44:2768–2776[CrossRef][Medline]-B ligand (RANKL) in serum is related to bone density, cartilage breakdown and inflammation in early active rheumatoid arthritis. J Bone Miner Res 17:S386B (RANK) receptors. J Exp Med 190:1741–1754B (RANK) ligand and osteoprotegerin in bone marrow stromal cells via cyclic AMP/protein kinase A pathway. Bone 30:553–558[Medline]B Fc diminishes prostate cancer progression to bone. Cancer Res 63:7883–7890B ligand expression by human myeloma cells mediates osteoclast formation in vitro and correlates with bone destruction in vivo. Cancer Res 63:5438–5445B) and RANKL expression in multiple myeloma. Brit J Haematol 117:86–92[CrossRef][Medline]B: new insights in the prostate cancer bone metastasis process. Bone 40:981–990[Medline]-related apoptosis-inducing ligand, TRAIL/Apo2L, using TRAIL/Apo2L gene-deficient mice. Eur J Immunol 32:2246–2254[CrossRef][Medline]B ligand that blocks osteoclast formation in vitro and in vivo. Eur J Cancer Suppl 4:62B ligand (osteoprotegerin ligand) and related proinflammatory cytokines during fracture healing. J Bone Miner Res 16:1004–1014[CrossRef][Medline]B (RANK) signaling blockade on fracture healing. J Orthop Res 21:676–684[CrossRef][Medline]B ligand: clinical utility in metabolic bone disease assessment. J Clin Endorinol Metab 90:6323–6331B ligand/osteoprotegerin/macrophage-colony stimulating factor in a presumably healthy human population. Eur J Endocrinol 150:305–311[Abstract]B ligand system are associated with bone metabolism in middle-aged males. Clin Endocrinol (Oxf) 62:92–98[CrossRef][Medline]B signaling. Ann NY Acad Sci 1068:143–164[CrossRef][Medline] converting enzyme (TACE). J Bone Miner Res 19:147–154[CrossRef][Medline]B ligand (RANKL) expression in synovial tissue from patients with rheumatoid arthritis, spondyloarthropathy, osteoarthritis, and from normal patients: semiquantitative and quantitative analysis. Ann Rheum Dis 61:1047–1054. Nature 408:600–605[CrossRef][Medline]B ligand in the serum and synovial fluid: a comparison of patients with longstanding rheumatoid arthritis and osteoarthritis. Rheumatol Int 26:63–69[CrossRef][Medline]B ligand mRNA expression in patients with rheumatoid arthritis and healthy controls. J Rheumatol 31:1483–1490, interleukin 1, and receptor activator of nuclear factor B ligand (RANKL) regulate osteoclasts. Ann Rheum Dis 63:354–359-mediated joint destruction. J Clin Invest 110:1419–1427[CrossRef][Medline]B ligand serum levels during treatment with infliximab in patients with rheumatoid arthritis. Ann Rheum Dis 65:1595–1599B, osteoprotegerin, and receptor activator of nuclear factor B ligand. Arthritis Rheum 50:3831–3843[CrossRef][Medline]B ligant inhibitor, Denosumab, in patients with multiple myeloma or bone metastases from breast cancer. Clin Cancer Res 12:1221–1228 (TNF-)-converting enzyme-like protease in shedding of TRANCE, a TNF family member involved in osteoclastogenesis and dendritic cell survival. J Biol Chem 274:13613–13618This article has been cited by other articles:
 |
|
 |
|
  M. Ohishi, R. Chiusaroli, M. Ominsky, F. Asuncion, C. Thomas, R. Khatri, P. Kostenuik, and E. Schipani Osteoprotegerin Abrogated Cortical Porosity and Bone Marrow Fibrosis in a Mouse Model of Constitutive Activation of the PTH/PTHrP Receptor Am. J. Pathol., June 1, 2009; 174(6): 2160 - 2171. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Kim, K. Kim, S. H. Lee, H.-S. So, J. Lee, N. Kim, and Y. Choi Identification of LRRc17 as a Negative Regulator of Receptor Activator of NF-{kappa}B Ligand (RANKL)-induced Osteoclast Differentiation J. Biol. Chem., May 29, 2009; 284(22): 15308 - 15316. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. Galli, Q. Fu, W. Wang, B. R. Olsen, S. C. Manolagas, R. L. Jilka, and C. A. O'Brien Commitment to the Osteoblast Lineage Is Not Required for RANKL Gene Expression J. Biol. Chem., May 8, 2009; 284(19): 12654 - 12662. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Vignot and D. Khayat A Step in the Journey of Denosumab From Bone-Targeted Therapy to Seed- and Soil-Targeted Therapy J. Clin. Oncol., April 1, 2009; 27(10): 1534 - 1536. [Full Text] [PDF]
|
|
|
|
 |
|
 S. Ferrari-Lacraz and S. Ferrari Effects of RANKL Inhibition on Inflammation and Immunity IBMS BoneKEy, March 1, 2009; 6(3): 116 - 126. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. N. Hirschhorn and L. Gennari Bona Fide Genetic Associations with Bone Mineral Density N. Engl. J. Med., May 29, 2008; 358(22): 2403 - 2405. [Full Text] [PDF]
|
|
|
|
|
|
U. Styrkarsdottir, B. V. Halldorsson, S. Gretarsdottir, D. F. Gudbjartsson, G. B. Walters, T. Ingvarsson, T. Jonsdottir, J. Saemundsdottir, J. R. Center, T. V. Nguyen, et al. Multiple Genetic Loci for Bone Mineral Density and Fractures N. Engl. J. Med., May 29, 2008; 358(22): 2355 - 2365. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
, significantly different from respective Pred + Veh control; P < 0.05 by ANOVA.
