The RANK–RANKL axis: an opportunity for drug repurposing in cancer?

Drug repurposing offers advantages over traditional drug development in terms of cost, speed and improved patient outcomes. The receptor activator of nuclear factor kappa B (RANK) ligand (RANKL) inhibitor denosumab is approved for the prevention of skeletal-related events in patients with advanced malignancies involving bone, including solid tumours and multiple myeloma. Following improved understanding of the role of RANK/RANKL in cancer biology, denosumab has already been repurposed as a treatment for giant cell tumour of bone. Here, we review the role of RANK/RANKL in tumourigenesis, including effects on tumour initiation, progression and metastasis and consider the impact of RANK/RANKL on tumour immunology and immune evasion. Finally, we look briefly at ongoing trials and future opportunities for therapeutic synergy when combining denosumab with anti-cancer agents such as immune checkpoint inhibitors.


Drug repurposing as a therapeutic strategy in cancer
Drug discovery and development requires many years of research and financial investment [1]. Repurposing an approved therapeutic agent is generally faster and more cost efective than novel drug development because the largest and most ineicient part of the development process has been bypassed [1]. Repurposing may take 3-12 years, compared with up to 17 years to develop a new drug, and repurposed drugs have the beneit of established safety proiles, which contribute to lower implementation failure rates than seen with novel agents [1]. Drugs with a history of safe use and a favourable toxicology proile may be candidates for repurposing if they have both a plausible mechanism of action and evidence of eicacy in the new indication [1]. Drug repurposing is one of the oldest development pathways for anti-cancer medicines; for example, anthracyclines, which are used extensively in the management of breast cancer and lymphoma, were tested as antibiotics but proved to be too toxic [2]. Thalidomide is another example of a drug successfully repurposed as a cancer treatment. Used to alleviate morning sickness in pregnant women, thalidomide was withdrawn after causing severe teratogenicity, but later demonstrated response rates of 25-35% in relapsed/refractory multiple myeloma and was subsequently approved for treatment of this cancer [3]. This inding prompted development of thalidomide analogues, such as lenalidomide and pomalidomide, which are now typically preferred in clinical practice over the original agent [3].
Recently, there has been renewed interest in repurposing in the oncology ield: drugs approved in other indications have been shown to induce cancer cell death, suppress aspects of cancer cell behaviour, or even prevent cancer development. Examples of repurposed drugs that have shown anti-cancer activity in at least one clinical trial include cimetidine, clarithromycin, verapamil and pravastatin [3]. Aspirin is also a potential candidate for repurposing as an anti-cancer agent. There is some evidence, from 1 3 clinical trials investigating prevention of cardiovascular disease and from observational studies, that it may reduce cancer incidence or death [4]. Similar indings have emerged from interventional studies evaluating the efect of aspirin on mortality and disease recurrence in cancer patients [4]. Potential antineoplastic properties of the oral antidiabetic agent metformin are also being investigated following observations that it reduced cancer incidence by 37% in patients with diabetes [5].
A further avenue open to exploration is repurposing drugs that target the receptor activator of nuclear factor kappa-B (RANK) ligand (RANKL) signalling system [6], which has an established role in regulating bone remodelling [7,8]. The RANKL inhibitor denosumab was irst approved in 2010 and is now indicated for the prevention of skeletalrelated events (SREs) in adults with advanced malignancies involving bone [9]. The denosumab label has recently been expanded to include the prevention of SREs in patients with multiple myeloma [9,10]. Observations that some stromal cells in giant cell tumour of bone (GCTB) express RANKL, possibly leading to the recruitment of RANK-expressing osteoclast-like giant cells, led to increased understanding of the role of RANK-RANKL signalling in the aetiology of GCTB and then to investigations of the therapeutic potential of denosumab in patients with GCTB [11,12]. Positive clinical trial results led to the approval of denosumab in 2013 for use in adults and skeletally mature adolescents with GCTB that is unresectable or where surgical resection may result in severe morbidity [13]. Accumulating data on the RANK-RANKL pathway raise the possibility that denosumab could be repurposed for use in other types of cancer. Evidence suggests that RANK-RANKL signalling may have osteoclast-independent efects on tumour biology, with RANK/RANKL activity demonstrating a positive correlation with tumour progression and advanced disease [14]. Moreover, trials including denosumab have demonstrated positive efects on survival outcomes in patients with cancer, including increased overall survival (OS) in lung cancer and increased disease-free survival in non-metastatic breast cancer [15,16]. Recent data also suggest that denosumab and immune checkpoint inhibitors may act synergistically to provide additional beneit [17]. This review explores the RANK-RANKL axis in cancer biology and the potential for repurposing denosumab.
The RANK-RANKL axis: more than bone turnover RANKL, its transmembrane signalling receptor RANK, and the decoy receptor osteoprotegerin (OPG) belong to the tumour necrosis factor superfamily [7]. Competition between OPG and RANK to bind RANKL regulates RANK activation and, therefore, its signalling function [7]. Binding of RANKL to RANK initiates various signal-transduction pathways, including production of nuclear factor kappa B [7], the eponymous member of a transcription factor family implicated in cancer development and progression, in inlammation and in crosstalk between these two processes [18,19]. While the RANK/RANKL/OPG system regulates osteoclast function and bone remodelling in bone homeostasis [7,8], it also has roles in immunity and development (Fig. 1a-c). Speciically, RANK-RANKL signalling sits at the interface between bone and the immune system, inluencing the pathology of diseases such as rheumatoid arthritis [20] and immune functions unrelated to bone including involvement in dendritic cell survival and function, M1 macrophage activation and T-cell diferentiation and activation [14,21,22]. Additionally, RANK-RANKL signalling functions in epithelial growth and diferentiation, including mammary physiology [7,23], in lymph node development [21], in skeletal muscle [24] and in the central nervous system (CNS) [8]. For example, RANK-RANKL signalling has been implicated in the formation and function of hair follicles, intestinal microfold cells, thymic medullary epithelium and mammary glands [7,25]. RANK protein is also expressed in bronchial (but not alveolar) epithelial cells [23]. Of note, in mammary gland diferentiation and the development of the lobuloalveolar breast tissue architecture required for lactation [26], RANK-RANKL signalling promotes the proliferation and survival of mammary epithelial cells [26] and constitutive signalling may promote the propagation of RANK-expressing stem cells at the expense of normal diferentiation [27]. RANKL expression is driven by progesterone-responsive cells in the breast and is required for progesterone-driven epithelial proliferation [28,29]. In skeletal muscle, the RANK-RANKL axis has been implicated in the regulation of calcium storage, inluencing denervated muscle function [24]. Furthermore, in the CNS, the RANK-RANKL axis plays a role in thermoregulation [8]. RANK, RANKL and OPG are, therefore, expressed across a variety of tissue types (Fig. 2a-c). Notably, RANK and OPG have relatively broad expression in normal tissues, while RANKL is largely restricted to osteoblasts and haematopoietic lineages (Fig. 2a), where it is expressed in T lymphocytes and basophils [30][31][32].

The RANK-RANKL axis in cancer biology
Given its varied roles in human physiology, interest has increased in the role of the RANK-RANKL axis in tumourigenesis, speciically in its functions beyond bone metabolism. Cells from various tumour types express RANK, RANKL and OPG (Fig. 3a-c) and RANK-RANKL signalling appears to be involved in tumour initiation, progression 1 3 and metastasis (Fig. 1d) [6,7,25,33]. With regard to bone metastasis, the efects of RANK/RANKL seem likely to be indirect, mediated by an osteoclast-dependent mechanism, or direct, via interaction of RANKL with RANK-expressing cells [6,7,25,33]. However, there is evidence for involvement of the RANK/RANKL/OPG system in the development and metastasis of melanoma, breast, hepatocellular, lung, prostate and renal cell carcinomas, as well as in primary malignant tumours of the bone (osteosarcoma, multiple myeloma and GCTB) [6,7,25,33]. The relative levels of RANK/RANKL/OPG expression may inluence prognosis in numerous cancer types including breast, lung, endometrial, renal cell and gastric cancers, osteosarcoma and multiple myeloma (Fig. 4a, b) [34][35][36][37][38][39][40][41]. In breast cancer, microarray analysis demonstrated that low RANK and high OPG expression were associated with longer OS (p = 0.0078 and p = 0.034, respectively) and disease-free survival (p = 0.059 and p = 0.040, respectively) [34]. Immunohistochemical analysis demonstrated an association between high RANK expression and bone metastasis (p = 0.023) and shorter survival without bone involvement (p = 0.037) [34]. In another study, low RANKL expression has been associated with increased risk for bone metastases (p = 0.018) and shorter disease-free survival (p = 0.018) in breast cancer [40]. In metastatic clear cell renal cell carcinoma, an elevated RANK/OPG ratio has been associated with a shorter median time to metastasis (p = 0.014), progression-free survival (PFS) (p = 0.001) and OS (p = 0.0001) [35]. Moreover, RANK expression in clear cell renal cell carcinoma has been identiied as an independent negative prognostic factor for both cancer-speciic survival and recurrence-free survival (p < 0.001) [42]. Additionally, high RANK expression has been linked to a shorter PFS (p = 0.01) and OS (p = 0.02) in endometrial cancer [36]. RANK-positive osteosarcomas Fig. 1 Involvement of the RANK-RANKL axis in bone remodelling, development, immunity and cancer. a In the mammary gland, following the binding of progesterone to its receptor, RANKL is produced. Increased RANKL stimulates mammary gland epithelial cell expansion via paracrine signalling, leading to lactation competence. Image modiied from [96] under the Creative Commons Attribution License (https :// creat iveco mmons .org/licen ses/by-nc-sa/3.0/). b Activated T cells and NK cells express RANKL, which stimulates various processes via monocytes, macrophages and dendritic cells including dendritic cell survival and maturation, T-cell activation and NK cell inhibition. c Increased levels of RANKL have been associated with increased tumour growth, stemness, EMT and metastases. Throughout these processes, RANKL acts as a chemoattractant for RANK-expressing cancer. d Increased RANKL produced by bone stromal cells, including cells of the osteoblast lineage, leads to an increased rate of bone remodelling [96]. EMT epithelial-mesenchymal transition, NK natural killer, PR progesterone, RANK receptor activator of nuclear factor kappa B, RANKL receptor activator of nuclear factor kappa B ligand are also associated with higher mortality than RANKnegative cases (p < 0.05) [41]. A high RANKL expression has been associated with higher stage disease (p = 0.035) and a shorter OS (p = 0.008) in gastric cancer [37]. In lung adenocarcinoma, both high RANK (p = 0.01) and RANKL (p = 0.02) expression have been linked to worse OS [38]. Furthermore, in multiple myeloma, the serum RANKL/OPG ratio was found to correlate with the extent of bone disease (p = 0.023), and a high RANKL/OPG ratio was associated with an increased risk of death (p < 0.001) [39]. Such correlation data should be interpreted with caution as the methods used to evaluate expression levels vary across studies, can include non-validated assays and require expertise to assess the results; therefore, further research is required to conirm these indings.

Indirect efects of RANK-RANKL signalling on tumour initiation, progression and metastasis: preclinical evidence
RANKL is well known to be involved in a cycle of osteoclast-mediated bone destruction and tumour activity that facilitates bone metastasis [7,8,25]. Both tumour cells and tumour-derived factors are involved in the production of RANKL and increase the RANKL/OPG ratio. Increased RANKL in the bone microenvironment upregulates osteoclastogenesis and the activation of mature osteoclasts (bone-resident macrophages), resulting in increased bone resorption. The subsequent release of growth factors and calcium into the bone microenvironment stimulates additional proliferation of tumour cells and release of tumour-derived Fig. 2 a RANK, RANKL and OPG tissue expression. Data from the Human Protein Atlas available at www.prote inatl as.org [97]. RANK expression is relatively ubiquitous, while RANKL expression is efectively restricted to haematopoietic cells. OPG expression is highest in mucosal epithelial tissues, consistent with its role in modulating immune responses. b RANKL positive immune cells within a lymph node. Image reproduced from the Human Protein Atlas [97] under the Creative Commons Attribution License (https :// creat iveco mmons .org/licen ses/ by-sa/3.0/); available at https :// www.prote inatl as.org/ENSG0 00001 20659 -TNFSF 11/tissu e/ lymph +node#img. c RANKL expression in normal breast tissue. Image reproduced from [98] under the Creative Commons Attribution License (http://creat iveco mmons .org/licen ses/by/4.0). OPG osteoprotegerin, RANK receptor activator of nuclear factor kappa B, RANKL receptor activator of nuclear factor kappa B ligand, TPM, transcripts per million factors, and thus further increases the RANKL/OPG ratio, thereby promoting continued RANKL-dependent osteoclastmediated bone destruction [43,44].
With respect to bone metastasis, preclinical evidence has shown that high RANK expression in breast cancer cells increases their ability to metastasise to bone [45]. RANKL inhibition promoted tumour apoptosis in a mouse model of bone metastasis from prostate cancer, and in combination with docetaxel, was synergistic in improving survival [46]. Moreover, preclinical data have indicated that the combination of anti-RANKL therapies and AT-406 (an antagonist of inhibitor of apoptosis proteins) may have a beneicial efect on bone metastases [47]. RANKL inhibition also suppressed bone invasion in mouse models of oral squamous cell carcinoma [48]. Furthermore, it has been shown that an elevated RANKL/OPG ratio increases the potential of human non-small cell lung cancer (NSCLC) to metastasise to bone [49]. In mouse models of multiple myeloma, RANKL has been shown to promote the release of myeloma cells from dormancy through osteoclast-mediated remodelling of the endosteal niche, thus promoting disease progression or relapse. In line with these observations, bone resorption has been linked to tumour burden in patients with multiple myeloma [50].
In addition to these well-known bone-related pathways, RANK-RANKL signalling may play a role at several other stages of cancer development, progression and metastasis [7,25,33]. RANKL may inluence carcinogenesis and Fig. 3 a Expression of RANK, RANKL and OPG in cells from various solid organ and haematological cancers. The results shown here are based upon data generated by the TCGA Research Network (http://cance rgeno me.nih.gov/) obtained using the cBioPortal for Cancer Genomic [95]. b RANKL expression in lung adenocarcinoma. c RANKL expression in pancreatic adenocarcinoma. Reproduced from the Human Protein Atlas [97] under the Creative Commons Attribution License (https ://creat iveco mmons .org/licen ses/by-sa/3.0/); image in b is available at https ://www.prote inatl as.org/ENSG0 00001 20659 -TNFSF 11/patho logy/tissu e/lung+cance r#img and image in c is available at https ://www.prote inatl as.org/ ENSG0 00001 20659 -TNFSF 11/patho logy/tissu e/pancr eatic +cance r#img. ACC adenoid cystic carcinoma, AML acute myeloid leukaemia, ccRCC clear cell renal cell carcinoma, chRCC chromophobe renal cell carcinoma, CS carcinosarcoma, DLBCL difuse large B-cell lymphoma, GBM glioblastoma multiforme, OPG osteoprotegerin, PCPG pheochromocytoma and paraganglioma, pRCC papillary renal cell carcinoma, RANK receptor activator of nuclear factor kappa B, RANKL receptor activator of nuclear factor kappa B ligand Fig. 4 a Association of RANK, RANKL and OPG expression with survival across diferent cancer types according to published studies. Each line represents diferent expression data as labelled; all comparisons are high vs low expression except for the breast cancer results, which are low vs high. b Association of RANK, RANKL and OPG expression (low or high) with 5-year overall survival rates across various solid organ and haematological cancers. OPG expression is signiicantly (p < 0.001) associated with survival in urothelial, renal and cervical cancers, but extreme gene expression for OPG, RANKL and RANK also show hypothesis-generating associations in other can-cers. Log rank p values for Kaplan-Meier plots showing results from analysis of correlation between mRNA expression levels and patient survival from TCGA datasets are plotted. Data extracted from the Human Protein Atlas available at www.prote inatl as.org [97]. a TCGA dataset. b Lung-1 dataset. + Indicates a better prognosis with high gene expression. − Indicates a worse prognosis with high gene expression. CI conidence interval, HR hazard ratio, mRNA messenger RNA, NR not reported, OPG osteoprotegerin, RANK receptor activator of nuclear factor kappa B, RANKL receptor activator of nuclear factor kappa B ligand, RCC renal cell carcinoma metastatic progression through efects on RANK-expressing tumour cells or cells from stem/progenitor compartments that express RANK [7,25,33]. Efects on tumour initiation may difer between cancer types and there has been particular interest in the role that RANKL may play in breast cancer given its physiological roles in mammary gland epithelia and the stem cell compartment [7,25,33]. Preclinical evidence suggests that RANK/RANKL mediate the pro-tumourigenic role of progesterone in the mammary gland [28,29,51]. In a mouse mammary tumour virus model, RANK overexpression promoted tumourigenesis in breast tissue after treatment with carcinogen and progesterone, and RANKL inhibition attenuated tumourigenesis in this and a spontaneous breast tumour model [51]. In another mouse model with breast tissue deleted for RANK, tumourigenesis, tumour growth and stem cell expansion driven by medroxyprogesterone acetate were also attenuated [52]. Furthermore, mammary carcinogenesis driven by BRCA1 mutations relies on autocrine or paracrine RANK-RANKL signalling in mammary progenitor cells [29]. Targeting the RANK-RANKL axis could conceivably, therefore, be beneicial in patients with BRCA1 mutation-positive breast cancer and may provide a rationale for designing clinical trials of cancer prevention [29].
Preclinical data have demonstrated that when expressed in lung epithelial cells or other cells of the tumour microenvironment, RANK may also promote lung cancer development. Signalling via RANK promoted expansion of cancer stem-like cells in a mechanism dependent upon mitochondrial respiration [23]. Female sex hormones are also able to promote lung cancer progression through the RANK-RANKL axis [23]. It has been hypothesised that the inluence of sex hormones on RANK/RANKL might in part explain gender-speciic diferences observed in epidemiological studies of human lung cancer [23,53]. RANKL expression has been associated with decreased survival in lung adenocarcinoma [23,38], and RANKL inhibition has reduced tumour progression in mouse models of lung adenocarcinoma [38].
The leucine-rich repeat-containing G-protein-coupled receptor 4 (LGR4) has recently been identiied as a RANKL receptor, has been shown to have an antagonistic efect on osteoclast diferentiation and to promote the proliferation of tumour cells when overexpressed [6]. Additionally, LGR4 expression has been associated with a poor prognosis and metastasis in breast, prostate and colorectal cancer [54][55][56]; however, LGR4 has been associated with improved survival in lung adenocarcinoma [23]. In multiple myeloma, LGR4 has been implicated in promoting Wnt/β-catenin signalling, which drives proliferation [57]. It could be speculated that patients with tumours expressing high levels of both LGR4 and RANKL might benefit from disruption of RANKL signalling.
Many of the mechanisms underlying the pro-metastatic activities of RANKL, including increased proliferation, migration and invasion of RANK-expressing tumour cells, could be relevant to numerous cancer types [7,14,25,33]. RANK overexpression induced transformation and stemness in untransformed mammary epithelial cells, induced markers of epithelial-mesenchymal transition (EMT), known to be important in acquiring these metastatic capabilities [58] and increased invasiveness, migration and anchorage-independent growth [59]. Additionally, RANK overexpression induced tumourigenesis and promoted metastasis of BRCA1deicient cells in immunodeicient mice [59]. Furthermore, disruption of RANK-RANKL signalling in mouse models of breast cancer reduced metastasis of RANK-expressing breast cancer cells to lung [7,60] and of RANK-expressing melanoma cells to bone [61]; conversely, exogenous RANKL promoted lung metastases in mice with RANKexpressing breast tumours [60]. In humans, high RANK/ RANKL messenger RNA expression has been signiicantly associated with metastatic breast adenocarcinomas [59], while RANKL induced markers of EMT in prostate cancer [62] and in head and neck squamous carcinoma cells [63]. RANKL has also been implicated both in the induction of angiogenesis and in increasing vascular permeability via RANK-expressing endothelial cells, which may impact intra/ extravasation and metastasis [25,64]. Vascular endothelial growth factor, which is commonly expressed by tumour cells, enhances endothelial cell responses to RANKL [25].

Potential immunomodulatory role of RANK-RANKL signalling
Immune cells may represent a subset of RANKL-expressing cells that mediate direct efects of RANK-RANKL signalling in cancer [7,14,25]. Such cells are commonly found in the tumour microenvironment but their precise roles are varied and not fully understood [7,14]. The complex role of RANK-RANKL signalling is evident from the fact that it has been implicated in both the generation and regulation of immune responses, suggesting that RANK-RANKL signalling can enhance and suppress immunity and may afect inlammation, T-cell activation and immune evasion [7,14,22]. Speciically, RANK-RANKL signalling supports dendritic cell survival and function, M1 macrophage activation and T-cell diferentiation and activation [14,22]. It is also involved in central and peripheral tolerance [7,14,25] and in the development of lymph nodes [21]. Additionally, RANKL contributes to regulation of early lymphocyte development [14,21]. With respect to suppression of immune responses, RANK-RANKL signalling is required for development of medullary thymic epithelial cells, which establish T-cell self-tolerance centrally through expression of the autoimmune regulator (Aire) gene [14,65]. Peripherally, 1 3 RANK-RANKL signalling mediates tolerance and immunosuppression in tissues such as pancreas, small intestine (in Peyer's patches) and skin; defective RANK-RANKL signalling is also implicated in the pathogenesis of autoimmune disease in these tissues [7,14]. RANK-RANKL signalling is required for the formation and function of regulatory T cells (Tregs) [66,67]. Speciically, RANKL is involved in the generation and activation of potent Tregs in the pancreas that are involved in preventing diabetes [66]. RANKL also mediates the induction of Tregs in studies of the immune environment in skin [67]. Modelling of the immune response in mice with colitis has shown that RANKL is expressed by Tregs and that blockade of RANK-RANKL signalling can impair Treg function and immune suppression [68].
Expression of RANK and RANKL is largely segregated among immune cells [7,25,69]. RANKL is primarily expressed in T cells and is upregulated upon T-cell stimulation, enhancing T-cell proliferation and function. In contrast, RANK is expressed in monocytes, macrophages and dendritic cells and is induced in T cells and natural killer (NK) cells by speciic cytokines [7,14,25,69]. With respect to the development of immune responses, RANK-RANKL signalling in dendritic cells enhances their survival and function, including their ability to prime and activate T cells, and to enhance T-cell memory [70,71], by inducing expression of activating cytokines [7,14,25,72]. Evidence suggests that RANK and the co-stimulatory molecule CD40 may act cooperatively in dendritic cell-mediated activation of T-cell responses [14,71]. Disruption of RANK-RANKL signalling can antagonise dendritic cell function and T-cell-mediated immune responses [73]. RANK-RANKL signalling in macrophages also enhances their survival and function, including antigen presentation [14,74]. However, the high levels of antigens and immunosuppressive factors typically found in tumour microenvironments can result in T-cell exhaustion, associated with the upregulation of markers, such as immune checkpoint proteins; therefore, although dendritic cells may initially mediate T-cell diferentiation and activation, extensive inlammation can induce immune suppression [75].
The importance of RANK-RANKL signalling in tumour immunology, and in stimulating or suppressing anti-tumour immune responses in cancer, is not yet fully understood [7,14,25]. Preclinical studies suggest that RANKL can promote the negative selection of tumour-speciic T cells by inducing the expression of self-antigens that are shared with tumours in the thymus, thereby promoting central tolerance to tumour antigens and tumour immune evasion [76]. RANKL inhibition can reversibly decrease self-antigen expression in the thymus and can spare from deletion T cells that target melanoma, thus enhancing the anti-tumour immune response and improving survival [76]. Furthermore, synergy has been observed between the suppression of central tolerance in Aire-deicient mice and peripheral immune checkpoint inhibition using anti-cytotoxic T-lymphocyte-associated antigen-4 (CTLA-4) antibodies with respect to slowing tumour growth and improving survival through increasing T cells that target melanoma [77]. Co-treatment with anti-RANKL and anti-CTLA-4 antibodies could not reproduce the synergistic efect of Aire-deiciency and anti-CTLA-4 antibodies on survival, although synergistic improvements in survival with co-treatment could be achieved if mice were also vaccinated with tumour cells producing granulocyte-macrophage colony-stimulating factor [77].
In addition to establishing central tolerance to tumours [7,14,76], RANK-RANKL signalling plays an important role within tumours where it may contribute to the development of a tolerogenic immune microenvironment [25,60,78,79]. RANK is highly expressed on tumour-associated macrophages (TAMs), and RANKL is expressed by tumouriniltrating T cells [7,25,69]. Tregs inhibit efector T-cell activity and can, therefore, play a role in tumour immune evasion [80], and as discussed, RANKL plays a crucial role in the expansion of Tregs in a variety of situations. Evidence suggests that RANK-RANKL signalling in M2 TAMs produces chemokines that recruit immunosuppressive Tregs to the tumour microenvironment [7,25,79]. Tregs have been shown to produce RANKL, which functions as a chemoattractant for RANK-expressing monocytes [81], suggesting that Tregs may perpetuate tumour progression [7,14,25,60]. Distant metastasis may also be inluenced by RANKL expression in tumour-iniltrating T lymphocytes [7,14,25]. Speciically, Tregs producing RANKL have been implicated in the development of lung metastases in a RANK-expressing breast cancer mouse model [60]. Although modulation of the tumour immune microenvironment could facilitate tumour progression, RANKL inhibition in mouse models of lung adenocarcinoma that responded to anti-RANKL therapy showed only modest changes in the tumour microenvironment, including non-speciic depletion of tumour T cells [38]. In addition to their suggested role in modulating the tumour immune microenvironment, M2 TAMs have also been implicated in angiogenesis and metastasis [25]. NK cells play an important role in tumour immunosurveillance [82]; RANKL can stimulate production of NK cell inhibitory factors by acute myeloid leukaemia cells, which also upregulate RANK expression by NK cells, promoting RANK-RANKL signalling that further contributes to NK cell inhibition [33].

Therapeutic potential of targeting the RANK-RANKL axis in cancer
Denosumab mimics the activity of endogenous OPG in binding to RANKL, although it does not inhibit the proapoptotic protein TRAIL [7,14,25]. Various trials have 1 3 recently completed or are in progress, particularly in breast and lung carcinoma, that are providing insight into the repurposing of denosumab to an anti-cancer therapy (Table 1). Anti-RANKL therapy may have the potential to help patients with primary bone tumours other than GCTB, or who have associated lesions, such as aneurysmal bone cysts [83]. Moreover, denosumab is hypothesised to have a number of anti-tumourigenic efects including increasing bone metastasis-free survival (BMFS), reducing proliferation and stemness in early tumourigenesis and augmenting checkpoint immunotherapy (Fig. 5a, b). In a phase 3 clinical trial of 1432 patients with castrationresistant prostate cancer at high risk of bone metastasis, denosumab signiicantly increased BMFS, but did not improve OS, compared with placebo [84]. A single-arm phase 2a study (D-BEYOND), conducted to determine whether a short course of neoadjuvant denosumab (two 120 mg doses a week apart) can decrease tumour proliferation rates (based on Ki67 immunohistochemistry) in newly diagnosed, early-stage breast cancer in pre-menopausal women, found that a short course of denosumab did not reduce proliferation rate, but did induce a signiicant increase in tumouriniltrating lymphocytes. These data support the hypothesis that denosumab may potentiate immunotherapy eicacy [85]. ABCSG-18, a phase 3 study using adjuvant denosumab (60 mg every 6 months) in patients with oestrogen and/or progesterone-positive non-metastatic breast cancer receiving aromatase inhibitors has demonstrated a 50% reduction in fractures alongside a clinically meaningful impact on disease-free survival (secondary endpoint) after a median follow-up of 72 months (hazard ratio = 0.823, 95% conidence interval 0.69-0.98, p = 0.026) [16]. However, the primary analysis of the phase 3 clinical trial D-CARE in non-metastatic high-risk early-stage breast cancer patients reported that the interventional scheme of denosumab (120 mg monthly for 6 months, then every 3 months for 5 years) failed to demonstrate an improvement over placebo with respect to the trial's unconventional composite primary Fig. 5 Targeting of the RANK-RANKL pathway may improve patient outcomes by a reducing self-renewal of cancer stem cells and b augmenting immunotherapy, tipping the balance in favour of the immune system. c Concomitant immune checkpoint inhibitor and denosumab therapy is associated with a higher response rate than has been reported for immune checkpoint inhibitors alone [17,99,100] a . Red colouring indicates studies of patients with melanoma; blue colouring indicates studies of patients with NSCLC; square = nivolumab; triangle = pembrolizumab; circle = atezolizumab, hexagon = avelumab; diamond = ICI + denosumab. a As the trials difer in study design and methodology, such crosstrial comparisons should be viewed as hypothesis-generating only. ICI immune checkpoint inhibitor, NSCLC non-small lung cell carcinoma, RANK receptor activator of nuclear factor kappa B, RANKL receptor activator of nuclear factor kappa B ligand endpoint of BMFS [86]. This was deined as the time from randomisation to the irst observation of bone metastasis or death and, at the primary analysis, death accounted for 42% of the BMFS endpoint [86], confounding the ability to observe an efect of denosumab in preventing the development of bone metastasis. Moreover, in contrast to ABCSG-18, there was a lack of disease-free survival beneit in D-CARE [16,86]. This may however be attributed to the heterogeneous high-risk early-stage breast cancer population and investigational dosing regimen used in D-CARE [86].
In a retrospective analysis of a phase 3 clinical trial of denosumab versus zoledronic acid in the treatment of bone metastases from solid tumours, denosumab was associated with increased OS in 811 patients with bone metastases from any lung cancer [15]. This outcome aligns with evidence from mouse models indicating that RANKL inhibition impairs the progression of lung adenocarcinomas [38]. However, recent prospective data from a phase 2 trial and the phase 3 SPLENDOUR trial showed that the addition of denosumab to standard irst-line therapy in patients with metastatic NSCLC (with or without bone metastases) did not improve OS or PFS [87,88]. Consistent with the above retrospective analysis [15], a slight improvement in OS was observed in the subgroup of patients with bone metastases receiving denosumab in the phase 2 study, but this efect was not signiicant [87,88].
Given the multifactorial roles of the RANK-RANKL axis in the immune system, there has been interest in the potential of RANKL inhibitors to improve response to immune checkpoint inhibitors in the treatment of cancer [6,89]. Inhibitors of immune checkpoint proteins such as CTLA-4 and anti-programmed death-1 (PD-1) or its ligand have revolutionised treatment for some patients [89]. These agents have been approved in several tumour types, including melanoma, head and neck squamous carcinoma, NSCLC, renal cell and urothelial carcinoma, gastric cancer and Hodgkin's lymphoma [89]. However, their efectiveness is limited by variable responses to treatment in individual patients; there is a need to improve their activity overall and to develop biomarkers that can predict which patients are likely to derive beneit from current strategies [89]. Preclinical data and emerging data from real-world practice suggest that denosumab and immune checkpoint inhibitors may act synergistically [69,90]. For example, in one case report, a patient with melanoma presented 13 months after initial diagnosis with an axillary nodal mass and widespread bone metastases and demonstrated an unexpected dramatic partial response following concomitant treatment with denosumab and single-agent ipilimumab (anti-CTLA-4) [90].
Preclinical studies suggest a synergistic efect of anti-RANKL in combination with anti-CTLA-4 and PD-1 antibodies, which have demonstrated tumour growth suppression in mouse models of melanoma, prostate and colon cancer [91]. In mouse models of melanoma, the addition of RANKL inhibition to anti-CTLA-4 antibodies did not deplete Tregs, but increased tumour-iniltrating CD4 + and CD8 + T cells compared with anti-PD-1 and anti-CTLA-4 therapy [69,91]. T-cell effector function (measured by cytokine production) was also enhanced in tumours treated with both agents compared with tumours treated with either single agent [69,91]. Interestingly, the synergistic efect of anti-RANKL and anti-PD-1 antibodies on lung metastasis was dependent upon NK cells, whereas the synergistic efect on subcutaneous tumour growth was dependent upon T cells [90,91]. Building on these data, real-world evidence using Flatiron Health's electronic health record database from ~ 255 cancer clinics across the US, including patients with advanced melanoma or NSCLC, found that a longer mean duration of concomitant immune checkpoint inhibitor and denosumab therapy was associated with improved overall response rate (p = 0.0172 and p < 0.0001 for melanoma and NSCLC, respectively) [17]. The response rate associated with the combination appears to be higher than reported historical response rates to immune checkpoint inhibitors alone (Fig. 5c). Additional survival analyses supported the association of concomitant therapy and improved survival outcomes, primarily in NSCLC (p < 0.0001) [17].
Prospective studies combining denosumab and checkpoint inhibitor immunotherapy are in progress. A phase 2 Australia and New Zealand Urogenital and Prostate Cancer Trials Group investigator-sponsored trial (KEYPAD) of denosumab and the PD-1 inhibitor pembrolizumab in patients with vascular endothelial growth factor receptor tyrosine kinase inhibitor-refractory clear cell renal cell carcinoma (NCT03280667), and an Australia and New Zealand Melanoma Trials Group phase 1b/2 trial (CHARLI) of ipilimumab-nivolumab-denosumab and nivolumab-denosumab in patients with unresectable stage III/IV melanoma (NCT03161756) are currently recruiting (Table 1). Of note, denosumab has an established safety proile in clinical practice, including when used in combination with novel anticancer agents, with no new detected safety signals to date [9,92].

Discussion and conclusions
Expanding knowledge of biological mechanisms as well as bioinformatic and combinatorial screening approaches are being used to identify existing agents that may be repurposed to beneit people with cancer. Improved knowledge relating to the role played by the RANK-RANKL axis in the cancer setting has highlighted that agents targeting this pathway may be able to modulate diferent stages of tumour progression from initiation to metastasis. In this respect, trials are investigating denosumab in diferent cancers with 1 3 promising results. Such trials also ofer the opportunity to identify potentially meaningful clinical synergy with checkpoint inhibitor immunotherapies through characterisation of the immune efects of denosumab on intratumoural cell populations.
A focus for further work will be to establish if there are groups of patients likely to derive beneit from denosumab as an anti-RANKL agent in cancer in combination with anticancer therapy. For example, the RANKL system is strongly inluenced by sex hormones [8,29], and RANKL activity is increased in post-menopausal women lacking oestrogen owing to decreased OPG expression [6,29]. Endogenous levels of OPG are lower in BRCA1 mutation carriers, who are at greater risk of breast cancer than the general population, compared with levels in women without the mutation [93]. A trial will assess the efect of denosumab on normal breast tissue in these individuals as well as BRCA2 mutation carriers and high-risk, non-BRCA carriers [94]. Moreover, RANK/RANKL/OPG and LGR4 expression data for primary tumours, such as renal, breast and lung, may assist in identifying patients most likely to beneit from denosumab therapy [95].
Overall, given the increased understanding of the pleiotropic roles of RANK/RANKL/OPG signalling in tumours and immune cells, the repurposing of denosumab, with its potential to enhance the eicacy of immuno-oncology drugs, could be a cost-efective component of such a combination approach for the treatmentof people with cancer.