Article Text

Download PDFPDF

The coupling of bone and cartilage turnover in osteoarthritis: opportunities for bone antiresorptives and anabolics as potential treatments?
  1. M A Karsdal1,
  2. A C Bay-Jensen1,
  3. R J Lories2,3,
  4. S Abramson4,
  5. T Spector5,
  6. P Pastoureau6,
  7. C Christiansen1,
  8. M Attur4,
  9. K Henriksen1,
  10. S R Goldring7,
  11. V Kraus8
  1. 1Nordic Bioscience, Herlev, Denmark
  2. 2Laboratory of Tissue Homeostasis and Disease, Department of Development and Regeneration, Skeletal Biology and Engineering Research Center, KU Leuven, Belgium
  3. 3Division of Rheumatology, University Hospitals Leuven, Leuven, Belgium
  4. 4New York University School of Medicine, and Hospital for Joint Diseases at NYU Langone Medical Center, New York, New York, USA
  5. 5King's College, Genetics and Molecular Medicine, London, UK
  6. 6Servier Research Institute, Suresnes, France
  7. 7Hospital for Special Surgery, Weill Cornell Medical School, New York, USA
  8. 8Division of Rheumatology, Department of Medicine, Duke University School of Medicine, Durham, North Carolina, USA
  1. Correspondence to Dr Morten A Karsdal, Nordic Bioscience A/S, Herlev Hovedgade 207, Herlev DK-2730, Denmark; mk{at}


Osteoarthritis (OA) is the most common form of arthritic disease, and a major cause of disability and impaired quality of life in the elderly. OA is a complex disease of the entire joint, affecting bone, cartilage and synovium that thereby presents multiple targets for treatment. This manuscript will summarise emerging observations from cell biology, preclinical and preliminary clinical trials that elucidate interactions between the bone and cartilage components in particular. Bone and cartilage health are tightly associated. Ample evidence has been found for bone changes during progression of OA including, but not limited to, increased turnover in the subchondral bone, undermineralisation of the trabecular structure, osteophyte formation, bone marrow lesions and sclerosis of the subchondral plate. Meanwhile, a range of investigations has shown positive effects on cartilage health when bone resorption is suppressed, or deterioration of the cartilage when resorption is increased. Known bone therapies, namely oestrogens, selective oestrogen receptor modifiers (SERMs), bisphosphonates, strontium ranelate, calcitonin and parathyroid hormone, might prove useful for treating two critical tissue components of the OA joint, the bone and the cartilage. An optimal treatment for OA likely targets at least these two tissue components. The patient subgroups for whom these therapies are most appropriate have yet to be fully defined but would likely include, at a minimum, those with high bone turnover.

  • Osteoarthritis
  • Osteoporosis
  • Arthritis
  • DMARDs (Synthetic)
  • Pharmacokinetics

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Materials and Methods

Pubmed was searched with the following keywords: subchondral, bone, remodeling, osteoclasts, osteoblasts, osteocytes, phenotype, sclerosis, osteoarthritis, bone marrow edema, bone marrow lesion (BML), SERM, Tibolone, bisphosphonates, strontium ranelate, cathepsin K, PTH, PTHrp, DKK and calcitonin. The search was not limited to a particular timeframe. Only peer-reviewed papers and no abstracts from these searches were included. Additionally, each of the authors had specialised areas of focus resulting in additional references and viewpoints that were included. Additionally, selected key references from key opinion leaders were included and discussed. Subsequent to review from experts in the field, additional key references were included.


Osteoarthritis (OA) is the most common form of arthritis.1 ,2 A hallmark of the disease is progressive degeneration of articular cartilage and subsequent joint space narrowing. Animal and clinical observations suggest that the structural integrity of articular cartilage is dependent on normal subchondral bone turnover, intact chondrocyte function and physiological biomechanical stresses.3 ,4 An increasing line of evidence suggests that cartilage and bone metabolism and, in particular, subchondral bone turnover and its interaction with the articular cartilage, are even more important than traditionally thought.5–8 Thus, an ideal therapeutic agent may logically be directed at regulating the metabolic activity of bone and cartilage. As illustrated in figure 1,9 there are concomitant changes in bone and cartilage, supporting the notion of a coupling of these processes.

Figure 1

The macroscopic features involved in the development of osteoarthritis (OA). OA is characterised by a disorganisation of the articular cartilage, chondrocyte apoptosis, cloning, tidemark undulation and cartilage loss. Additionally, the subchondral cortical bone plate is thickened, with vascularisation of the calcified cartilage and advancement of the tidemark. This figure is reprinted with permission from Bay-Jensen AC, et al.9

For the majority of patients, the aetiology of OA is unknown. In addition to genetic risk factors,10 ,11 known risk factors for OA are age, significant trauma, obesity, altered gait, altered biomechanics (eg, a varus or valgus deformity), and excessive loading on the affected joint.12–18 Moreover, a subset of OA could result from a metabolic component, as evidenced by data showing that decreased sex hormone levels in animal models of OA and postmenopausal women are associated with the well-described increase in bone turnover and also increased cartilage degradation.8 ,19–22 A thickened subchondral cortical bone plate and increased bone turnover often associated with OA can be a consequence and further contribute to an altered composition and biomechanical properties of the OA joint pathology.23–29 Most of these studies were performed in humans and focused on late and end-stage disease, which reveals little about the early dysregulation of cartilage and bone turnover. Newer preclinical studies have provided key information on the early and pivotal role of bone turnover5–8 during the progression of disease. Based on current knowledge, we hypothesise that targeting two primary components of the joint organ, bone and cartilage, and the interactions between their cell types, most importantly between osteoclasts in bone and chondrocytes in cartilage, will provide a potential therapy for a selected subgroup of OA patients for a particular treatment option.

The role of subchondral bone turnover and structure in OA

The relationship between bone turnover and cartilage degradation should be considered a dynamic process depending on the origin and stage of disease. Figure 2 schematically outlines the differences between a normal and an OA knee joint, the latter displaying loss of articular cartilage, development of inflamed synovium, sclerosis of the subchondral plate, osteophyte formation, BMLs, undermineralised bone as a consequence of hyper-remodelling with apparent trabecular thinning in vertical versus horizontal trabeculae, development of chondroclasts and ‘compressed’ osteoblasts characteristic of the OA phenotype, and vascularisation of the subchondral bone with penetration into the calcified matrix, and presence of nerve endings in the osteophytes.30 All these represent key features of OA that need to be better understood to appreciate the OA joint phenotype. The goal of this state-of-the-art manuscript will be to review the involvement of the bone compartment. Consequent to this, we discuss whether an optimally selected antiresorptive treatment for the correctly selected patients may alleviate these pathological features.

Figure 2

Schematic representation of a healthy joint and an osteoarthritis (OA)-affected joint. Compared with the healthy joint, the OA joint organ is damaged in the following manner: complete cartilage loss (denudation), hypertrophic chondrocytes and cloning of chondrocytes (local proliferation), vascularisation of the subchondral bone and penetration into the calcified matrix, presence of nerve endings in the osteophytes, altered phenotypes of osteoblasts and osteoclasts (chondroclasts), subchondral bone sclerosis, undermineralised areas consequent to hyper-remodelling (woven bone), bone marrow lesions, trabecular thinning, synovial inflammation and invasion.

The relationship between bone and cartilage in OA has been a source of controversy for a long time, and originally it was proposed in clinical settings that OA was protective for the bone-loss disease, osteoporosis (OP), and vice versa.31–33 Subsequent larger studies found conflicting results.34–36 On balance, it appears most likely that OA is not protective for bone loss, and OP is not protective for OA. Whereas some risk factors were identified to predispose to either OA or OP, these seemed to have divergent effects.31 As an example, weight is one of the major risk factors for OA, thereby providing a plausible explanation for the fact that lean individuals are at a lower risk for OA and a higher risk for OP. In the large Framingham and Chingford studies,37–39 mean femoral bone mineral density (BMD) was higher in patients with knee OA than in those without OA, but in patients with an OA Kellgren severity score of grade IV, there was no difference in their BMD compared with healthy controls. In alignment, no association was found between BMD at the radius and OA severity score.40 Local versus generalised effects of bone on OA may, therefore, be very different, as associations with BMD varied by OA score and site. Importantly static observations, such as obtained by imaging, may be different in their prognostic nature compared to that of dynamic measures obtained by use of serological biochemical markers. Spector and colleagues demonstrated that bone turnover is important for progression of OA, as increased bone resorption markers, most likely reflecting systemic skeletal changes, predicted progression of OA.37–39

Over recent years, further insights into these local interactions have been obtained. Examinations of periarticular bone in the knees and hips of patients with OA have confirmed that the subchondral bone is abnormal in OA joints, with sclerosis of the subchondral plate and altered trabecular structure,41 ,42 exemplified by the thinning and fenestration of the vertical trabeculae and undermineralised bone consequent to the high remodelling rates as visualised in figure 2. Bone scintigraphy was the first technique used to demonstrate increased subchondral bone turnover.43 These pivotal studies by Dieppe and colleagues strongly suggested the importance of subchondral bone in the pathogenesis of OA and for progression of disease pathology.43–49 Cross-sectional studies have also established that women with advanced knee or hip OA have higher BMD near, or at the site of OA pathology.50 Impressively, turnover of subchondral bone in OA has been shown to be as much as 20-fold higher than that of normal bone.24 Insights into these apparent discrepant findings are now provided by studies of OA bone using fractal signature analysis that permits characterisation of the subchondral bone architecture during OA progression in which the horizontal trabeculae are thickened while the vertical trabeculae are thinned.51 This thinning of the vertical trabeculae is believed to be mediated by high trabecular turnover which results in decreased bone density and is likely a reflection of so-called ‘stress-shielding’ (unloading) of bone in this region. BMLs are receiving increased attention, as key findings have shown this feature to be associated with pain and radiographic progression.52 ,53 Taljanovic et al54 analysed the histopathologic findings associated with BML in hip OA at the time of joint replacement. This revealed the presence of microfractures of the trabecular bone, various stages of healing associated with fibrous replacement of the marrow, and increased activation of bone-remodelling units54; all histological parameters suggesting high bone remodelling. In alignment, others have concluded that BML from OA knees are characterised by sclerotic bone that is less well mineralised.55 With regards to symptoms and structure, quantitative assessment of BML size in osteoarthritic knees was shown to correlate with cartilage damage and predict longitudinal cartilage loss,56 and Felson et al57 demonstrated that BML on MRI were strongly associated with the presence of pain in knee OA. Consequently, pain, progression and BML are closely associated, and all part of the subchondral bone compartment.

In preclinical settings, studies using anterior cruciate ligament transection (ACLT) in dogs, as well as ACLT and meniscectomy (MNX) in rats, have been instrumental in characterising the role of subchondral bone changes in OA.58–60 In these two models of OA, a gradual loss of subchondral trabecular bone in the early postoperative period was observed. Many studies support the importance of osteoclasts in OA by showing, for example, accelerated bone turnover in traumatic MNX and ACLT, and in oestrogen-deficient models (ovariectomy (OVX)), in which increased bone resorption is associated with articular damage,61 and augmented articular cartilage erosion. The loss of trabecular bone in the subchondral regions is also associated with a decrease in the elastic modulus.58 ,60 ,62 By contrast, appositional endochondral and direct bone formation in the periphery of the articular cartilage plate, laterally and medially, results in osteophytes.3 These bone changes are all important hallmarks of the pathology of OA3 ,4 ,23–27 ,53 ,59 ,61 ,63–65 (illustrated in figure 2). Using, STR/ort mice, Stok et al,66 performed multimodel imaging, and documented the role of bone in OA progression, by careful longitudinal image analysis. Rabbit models of OA have further validated the role of bone in the early pathogenesis of OA.5–8 A bone densitometric evaluation in the meniscectomised guinea pig model of OA revealed alterations in bone metabolism seen at the onset of OA, with early bone loss at the subchondral bone followed by increased bone density.67 This finding corroborates previous extensive findings of altered subchondral turnover in the pathogenesis of OA,23–27 ,61 ,63 ,68 which is supported by the Dunkin–Hartley guinea pig model indicating that subchondral bone changes precede changes in the cartilage.69 ,70 In addition to the results obtained in the meniscectomised guinea pig model, other have noted age-related subchondral bone thickening.71 When more sensitive indicators of cartilage pathology are used, such as collagen birefringence, it is apparent that cartilage abnormalities in this model either coincide with bone pathology or precede it.72 Bone remodelling and joint destruction is further highlighted by the preclinical demonstration of dickkopf (DKK) as a master regulator of bone turnover, as inhibition almost abrogated bone erosion,73 and that DKK-1 expression in chondrocytes inhibited experimental osteoarthritic cartilage destruction in mice,74 which is further supported by other key observation on joint erosion and DKK levels.75–79 Studies in these animal models underscore the critical importance of bone remodelling in the development and progression of OA. Reviews by Goldring et al52 ,80–84 have further provided insights into the interactions between the cartilage and bone compartments in OA and have noted that bone may adapt more rapidly than cartilage in response to altered mechanical environments. It is thus readily understandable that early manifestations of OA would be first apparent, or be detectable, in the bone tissue component.

Taken together, bone turnover and osteoclast functions are likely to be important elements in the pathogenesis of OA, in which osteoclastic bone resorption and signals coming from osteoclasts may play important roles85 in the different sclerotic and osteoporotic alterations in subcompartments of bone.

The cells of the bone: osteoblasts, osteocytes and osteoclasts

Osteoblasts and osteocytes

The Pelletier group documented that OA osteoblasts have a different phenotype than normal human osteoblasts, which is manifested by the OA cells’ altered response to different stimuli.86–97 Compared to normal osteoblasts, they suggested that OA osteoblasts have different ratios of important bone factors (receptor activator of nuclear factor κ-B ligand (RANK-L) and osteoprotegerin (OPG)), and different responses to insulin-like growth factor (IGF), Prostaglandin E2 (PEG2) and PTH, with the alterations all being pro-bone-resorptive. In agreement with this, the group of Henrotin clearly documented the effect of compression on the OA osteoblast phenotype. They found compression induces a large increase in interleukin (IL)-6 and downregulation of OPG, both of which may possibly drive increased osteoclastogenesis and bone resorption,98–105 whereas compression downregulates aggrecan synthesis while upregulating matrix metalloproteinase (MMP) activity. In further support of this, a recent study examining the gene profiles of site-matched cartilage and bone samples from OA joints demonstrated a strong coordinated regulation of multiple genes in these two tissues.106 However, this study did not differentiate between the gene profiles in osteoblastic and other bone cells. Further studies of OA-derived osteoblasts have shown increased expression of dickkopf-related protein 2 (DKK2), which was attributed to the autocrine effect of transforming growth factor (TGF)-β.107 This was speculated to be a component of the dysregulated osteoblast function in vivo.108 Interestingly, OA osteoblasts were also shown to overexpress vascular endothelial growth factor (VEGF), a finding which could provide some insight into the increased vascularity and osteoclast activity observed in the subchondral plate in OA.109 ,110 Finally, a recent study showed marked alterations in the osteocyte phenotype in OA patients compared with controls; however, in this case, it was unclear whether these changes were caused by OA or contributed to OA.111

Osteoblasts have been documented to control cell types in other joint tissues. Cocultures of human OA osteoblasts and human cartilage led to increased proteoglycan degradation in cartilage compared with coculture with osteoblasts from healthy controls.112 The proteoglycan degradation was suggested to be a result of increased MMP-2 expression.63 This may indicate that altered osteoblast phenotypes during disease development may result in the above-described alterations in extracellular cartilage matrix function and composition, which is supported by others.113 In particular, loss of the articular chondrocyte's stable phenotype with induction of chondrocyte hypertrophy could be orchestrated at least partially by osteoblasts.104 The same group demonstrated that the osteoblasts from sclerotic regions of subchondral bone were capable of switching the profile of chondrocytes towards a catabolic and antianabolic phenotype, as illustrated by a reduction in aggrecan production, but also by an upregulation of MMP production.104 Finally, a study showed that the bone remodelling rates in healthy individuals correlated with RANK-L expression levels, whereas in OA patients, this correlation no longer existed.114 As osteoblast lineage cells are a source of RANK-L,115 ,116 this further supports the view that changes in osteoblasts are involved in OA pathology and progression.

A role for osteocytes in OA is newly emerging. Osteocytes are critical mechanosensing cells and, therefore, of key interest in the complex pathogenesis of a load-associated disorder such as OA. For example, recent studies have demonstrated altered expression of RANK-L, OPG and sclerostin in osteocytes, providing further evidence for a role of these cells in modulating subchondral bone remodelling.117 ,118


Osteoclasts have been shown to play important roles in joint damage, in particular, bone erosion in rheumatoid arthritis (RA),73 ,119–122 whereas, the role of osteoclasts in the pathogenesis of OA is still unclear. Increased expression of osteoclast differentiation factors, such as RANK-L, are known to occur in the synovial membrane of chronically diseased OA joints, and could contribute to increased bone resorption.119 ,123 Logar et al124 investigated the relationship between the expression levels of RANK-L and OPG and several osteoclast-associated genes (cathepsin K, MMP-9 and (Tartrate-resistant acid phosphatase (TRAP)) mRNA) in the proximal femurs in patients with OA or osteoporotic (OP) hip fractures. In the OA group, cathepsin K, MMP-9 and TRAP mRNA were present at higher levels for a given RANK-L expression level and RANKL/OPG ratio compared with the fracture group.124 The researchers suggested that the inhibition of RANK-L-dependent osteoclast formation may constitute a potential target to prevent the skeletal changes seen in OA.

With respect to the phenotype of osteoclasts in the pathogenesis of OA, unfortunately only sparse information is available, as osteoclast biology and bone turnover only recently received attention in the OA field.83 ,84 ,125 ,126 In an inflammatory model of arthritis by Pettit and colleagues, the involvement of osteoclasts in bone erosions was elegantly demonstrated by the absence of bone erosions in RANK-L knockout mice deficient of osteoclasts.127 These mice are deficient in calcitonin receptor (CTR)-positive cells representing mature osteoclasts that can adhere to the surface of mineralised tissues. This suggests that bone and/or calcified cartilage provide signals that are critical for haematopoietic osteoclast precursors to fully differentiate into osteoclasts.128 A recent study that isolated peripheral blood mononuclear cells (PBMCs) from patients with OA showed an increased osteoclastogenic potential of these cells compared with PBMCs from matched controls.129 This finding is consistent with the known increased osteoclastic activity in OA.129

It remains to be investigated whether the osteoclasts that are involved in the development of OA indeed are authentic osteoclasts or chondroclasts, that is, whether osteoclasts localised on calcified cartilage in the subchondral bone are different from those in the normal bone matrix130 ,131 and may define a specific subtype of calcified matrix resorbing cells.132 Interestingly, Ota and colleagues investigated chondroclasts and found processes reminiscent of the pathogenesis of OA in the bone growth plate during fracture repair. The local OPG produced by chondrocytes controlled cartilage resorption, thereby serving as a negative regulator for chondrocyte-dependent chondroclastogenesis.133 These results begin to elucidate the interactions of chondrocytes, cartilage matrix and osteoclasts.

Insights into OA and osteoclasts may be provided by further characterisation of the bone, cartilage and joint phenotype of several monogenic disorders. Such studies may provide molecular insights into whether specific osteoclast phenotypes, such as osteoclast-rich or osteoclast-poor bone phenotypes, are associated with the pathogenesis of OA.134 ,135 For instance, osteopetrosis in humans was reported to be associated with premature OA,136 ,137 and provides an opportunity to evaluate the pathogenesis of OA in the context of loss of osteoclasts or osteoclast activity.134 Taken together, it appears that osteoclastic bone resorption, bone formation by osteoblasts, and signals coming from osteoclasts, osteocytes, osteoblasts and chondrocytes may all play important roles in the altered metabolism of the subchondral bone during the pathogenesis of OA.23 ,85

Inflammation, bone resorption and OA

It is well appreciated that inflammatory mediators are produced by joint tissues in OA, and that these proinflammatory cytokines increase osteoclastogenesis and bone resorption.135 ,138 The mediators are particularly important in the initiation of post-traumatic OA, and contribute to the progression of established disease, as recently reviewed.139 ,140 Furthermore, preclinical studies using STR/ort mice as a model for spontaneous OA, demonstrated elevated levels of local and systemic inflammatory mediators.141 While not typically accompanied by the classical signs of inflammation, inflammatory processes, including the production of IL-1, tumour necrosis factor (TNF)-α and prostaglandins (PGE2), drive catabolic events in cartilage. Abramson and colleagues recently demonstrated that patients with symptomatic knee OA whose peripheral leukocytes exhibit increased expression of IL-1β are at higher risk for disease progression142 than those patients with normal levels of IL-1β. They also provided genetic evidence of a role of IL-1 and its antagonist in the pathogenesis of OA, by showing that the radiographic severity of knee OA is conditional on IL-1 receptor antagonist gene polymorphisms.143 Thus, inflammation is common in OA and characterised by immune cell infiltration and cytokine secretion by joint tissues, although the inflammation in OA seems to be quantitatively and qualitatively different from inflammation in RA.144 ,145

Many of the inflammatory mediators are detected at elevated levels in OA synovial fluids, such as IL-1β, macrophage migration inhibiting factor (MIF), IL-8, tenascin-C, complements, cartilage oligomeric matrix protein fragments, collagen and aggrecan fragments and PGE.140 These mediators have been shown to activate or amplify the inflammatory process in the joint as well as increase osteoclast activity. Scanzello et al146 recently presented evidence that cartilage matrix fragments can activate an innate immune system in OA, leading to synovial inflammation and progression of OA. An inflamed synovium or bone marrow may stimulate osteoclast-mediated bone resorption through a range of proinflammatory cytokines, such as IL-1, IL-6, TNF-α and TGF-β. As examples, the biomarker of bone resorption (CTX-I), was shown to be highly correlated to systemic inflammation as measured by high sensitive C-reactive protein (hsCRP).138 IL-6 is receiving increased attention in RA and OA, for effects on the joint and bone turnover.138 IL-6 has been implicated in the cross-talk between bone and cartilage, and studies have shown that IL-6 in combination with other cytokines can switch osteoblasts from a normal phenotype to a sclerotic phenotype,104 as well as altering expression of different chemokines, such as regulated and normal T cell expressed and secreted, and monocyte chemoattractant protein-1 (MCP-1).147

These complex inflammatory pathways may indicate an ‘inside-out’ and ‘outside-in’ roles in cartilage erosion and disease progression, a concept recently elegantly described by Schett and Firestein.148 ‘Outside’ refers to a primary inflammation of the synovial membrane (synovitis), which later spreads to adjacent structures resulting in penetration into the bone marrow (the outside-in concept). Alternatively, arthritis may start ‘inside’ in the bone marrow space, and later encroach upon the synovium (inside-out concept). These two processes are fundamentally different and provide interesting concepts regarding the disease-specific forms of arthritis.148

Targeting bone and cartilage in the treatment of OA

The presence and activation of osteoclasts seem to be involved in the pathogenesis of OA and other cartilage-destructive diseases, such as RA, psoriasis arthritis149 and spondyloarthropathies.120 ,121 ,150 Does this suggest that interventions that reduce bone turnover may have secondary positive effects on the progression of OA?

Due to the widespread nature of OA and the lack of effective therapies, there is an urgent need to develop safe and effective disease-modifying treatments. An optimal treatment for OA would entail bone and cartilage protective effects. However, options seem limited. This section briefly highlights the most important finding with different bone-targeting treatments and their potential as OA structure modifiers.


Doung and colleagues demonstrated that bisphosphonate treatment in surgical models of OA resulted in a 50% decrease in disease severity scores, and protection of bone and cartilage from pathological changes.4 ,59 ,151 ,152 In agreement, Zhu et al153 have also shown that early treatment of OVX rats with alendronate significantly attenuated cartilage erosion by inhibiting subchondral bone loss. Interestingly, Strassle et al154 demonstrated that when the bisphosphonate zolendronate was applied to the monoiodoacetate model of painful arthritis in rats, it protected against subchondral bone loss, cartilage degradation and, importantly, also pain. Bisphosphonates were subsequently demonstrated to be disease progression specific,155 underscoring further the role of bone turnover in early OA and need for selection of the right patient population and state of disease for a specific intervention. The chondroprotective effects of bisphosphonates have been validated further by other research groups.155–163

In clinical settings, the picture is somewhat more complicated. One bisphosphonate (risedronate) has been tested in phase II and phase III clinical settings, with positive effects on OA in phase II and negative effects in phase III.164–167 Bisphosphonate treatment was shown in a clinical setting to inhibit bone and cartilage degradation by assessment of biochemical markers, although joint space narrowing observed on X-rays failed to show attenuation of structural deterioration.166 This latter finding is complicated by the lack of progressors in the study, and thus the inability to conclude whether bisphosphonates could slow the progression of OA. Furthermore, the X-ray technology applied in the study, rather than the more advanced MRI methodology, may have reduced the ability to detect efficacy. In fact, a secondary analysis identified those with the highest levels of cartilage degradation as assessed by CTX-II levels to be those in whom OA progressed. CTX-II was the same marker that was influenced by bisphosphonate therapy.168 Recent discussions have indicated strong correlations between bone and cartilage degradation markers CTX-I and CTX-II, and a potential causal relationship.169 On the one hand, this correlation has suggested that CTX-II may not originate exclusively from cartilage, although, recent evidence by Catterall and colleagues documented that CTX-II cannot be generated by in vitro enzymatic treatment of bone by MMPs or cathepsin.170 To date, CTX-II may be the best validated marker for progression of OA.171 ,172 Importantly, different biochemical markers may have different interpretations depending on the pathology in question and treatment effect.


Human calcitonin exerts potent antiresorptive effects.173–175 It is a small, 32-amino-acid peptide, secreted in response to excess calcium in serum.176 Salmon calcitonin (sCT) is approved for the treatment of OP and other diseases involving accelerated bone turnover.176 ,177 Calcitonin was shown to be chondroprotective in the gold standard canine ACLT OA model. Calcitonin significantly affected trabecular structure and prevented subchondral bone resorption and trabecular thinning, which was speculated to be a major factor in the reduced level of cartilage degradation.178–180 The mode of action of calcitonin may be different from that of other antiresorptives, as calcitonin has been shown to have direct and indirect actions on articular cartilage on human OA chondrocytes.181–185 In other traumatic and non-surgical animal models of OA, calcitonin has been demonstrated by many independent research groups to have positive effects on articular cartilage surface erosion and bone structure.186–190

Although an oral formulation of sCT was recently developed,191 the administration of sCT in clinical practice has so far been limited to either the subcutaneous or intranasal route.192 Initially, Manicourt and colleagues reported clinically significant positive effects of calcitonin on joint pain and function, and positive effect on bone and cartilage biomarkers.193 ,194 In direct agreement, Karsdal and colleagues investigated the effect of the novel oral formulation of sCT on bone and cartilage degradation by using biochemical markers in multiple clinical settings.195–199 sCT inhibited bone and cartilage degradation, evidenced by the biochemical markers CTX-I and CTX-II, respectively. While these data may be positive, two large Phase III studies did not demonstrate similar positive effects. Additionally, the Food and Drug Administration (FDA) recently questioned the cost-benefit ratio for salmon calcitonin in OP. More pivotal phase III clinical trials and subsequent analysis are needed to investigate the potential positive effect of calcitonin on the pathogenesis of OA.

Cathepsin K

An interesting target in relation to treatment of OA, cathepsin K is the main osteolytic protease of osteoclasts. A Cathepsin K inhibitor has been suggested to have a special mode of action, namely inhibition of osteoclastic resorption without removing the osteoclasts, and a modest reduction of bone formation.200 In agreement with these results, a novel cathepsin K inhibitor was shown to protect against the progression of OA in a guinea pig model of spontaneous OA, and in the canine partial medial MNX model of OA.201–204 Furthermore, a recent study indicated that inhibition of cathepsin K prevented ACLT-induced OA in rabbits, with beneficial effects on subchondral bone, cartilage and osteophytes.152 Interestingly, when studying the development of inflammation-driven bone destruction in arthritis in cathepsin K-deficient mice, the prevention was only partial.205 This result was in contrast to a study employing the ACLT model in cathepsin K-deficient mice, where full protection was observed.152 These data may indicate that the inflammation status of OA may affect how well inhibition of cathepsin K will work, although this remains to be studied in clinical settings. Cathepsin K inhibitors have demonstrated promising effects in OP phase II studies, and one is currently being evaluated in phase III studies.200 ,206 ,207


Oestrogen is one of the most thoroughly investigated hormones affecting bone, cartilage, muscle and inflammation, and warrants further attention in the present context, consequent to the well-documented antiresorptive effects. Systemically increased turnover of cartilage is found more frequently in postmenopausal women than in the premenopausal population.19 ,208–211 An epidemiological study supports the concept that older people, and especially women, are at higher risk than younger women and men of developing the most common forms of OA.3 ,212 ,213 This supports the hypothesis that there is a relationship between female sex hormones and the incidence of OA.

Induction of OA by ACLT in 12 female rabbits, where half were ovariectomised prior to ACLT, showed that OVX aggravated the severity of OA compared with animals with no OVX.5 ,8 These data indicate that OVX, in which oestrogen is absent, results in high bone resorption, and hence, an OP-like phenotype and worsening of microscopic OA features, which is also seen in standard OVX models.214 In a corroborating study, OA was induced in mice by destabilisation of the medial meniscus (DMM). Half these animals were also orchidectomised (ORX) or underwent OVX.20 OA severity was markedly higher in male DMM mice than female DMM mice. However, OVX-DMM female mice developed more severe OA than controls, whereas ORX-DMM male mice developed significantly less severe OA than control mice. Also, the cartilage of ovariectomised cynomolgus monkeys and rats showed histopathological features of OA associated with oestrogen deficiency.21 ,214 ,215 Additionally, expression analysis of knee cartilage from OVX rats showed elevated levels of several tissue-degrading proteases and proinflammatory cytokines, compared with controls.216 This pattern is supported by earlier data from Claassen et al,217 who showed that oestradiol could suppress the expression of the same degrading proteinases in cultured osteoarthritic chondrocytes. Oestradiol was able to protect chondrocytes from oxidative damage.

One 3-year study examined the effect of antiresorptive oestrogen replacement therapy (ERT) for the prevention of OA in 180 female cynomolgus monkeys.61 ,64 Ovariectomised adult monkeys were divided into groups receiving ERT or receiving no treatment. By quantitative histology, significantly fewer cartilage lesions were seen in the ERT group compared with the control group. The authors suggested that high bone turnover contributed to a high risk of cartilage degradation, and that exogenous oestrogen may confer protection against the development of OA. Similar observation has been undertaken in rats,187 ,214 where oestrogen depletion results in increased bone turnover, and accelerated cartilage breakdown.

In alignment, the Women's Health Initiative (WHI) study documented that women taking unopposed oestrogen had an HR of 0.73 for total joints, and 0.55 for hip total joint replacement (TJR), that was analysed per-protocol, with follow-up time of participants censored at 6 months postnon-adherence, which was predefined as consuming 80% of assigned study pills, compared with those on placebo.218 The positive effect of oestrogen on cartilage health, is further supported by the fact that chondrocytes express oestrogen receptors,219–221 and respond to oestrogen intervention.214 However, it is not believed that this accounts for the entire detrimental effect of oestrogen loss on cartilage health,187 as elevated bone turnover and subchondral bone remodelling likely indirectly contribute to cartilage loss.

One SERM, levormeloxifene, was investigated in 301 postmenopausal women in a 12-month phase II trial for the prevention of bone loss and cartilage degradation.22 Levels of the biomarkers, CTX-II and CTX-I, indicators of cartilage and bone degradation, respectively, were decreased by approximately 50% in the treatment groups at 12 months compared with baseline. This was recently reproduced with another SERM, raloxifene, with reductions in bone and cartilage biochemical markers.222 Interestingly, after treatment cessation, CTX-II levels reverted back to baseline values while CTX-I remained strongly suppressed, indicating a short-term effect of the SERM on the cartilage, but a long-term effect on bone. These different pharmacodynamic effects may be important for understanding the coupling between bone and cartilage degradation. Consistent with these findings, CTX-II levels in 384 postmenopausal women were found to be significantly lower in those taking hormone replacement therapy (HRT) than in those not receiving HRT,19 indicating that HRT protects against cartilage degradation. By contrast, CTX-II levels were significantly higher in postmenopausal women compared with an age-matched group of premenopausal women.19

These lines of evidence suggest a positive effect of oestrogen on joint biology, and is most likely mediated by direct effects on bone, muscle, inflammation and synovium, which in combination, improve joint health. Additional evidence may be found in the increased incidence of erosive hand OA around the time of menopause, and in the important musculoskeletal side effects (arthralgia caused by oestrogen deprivation) that are associated with the use of aromatase inhibitors in the treatment and secondary prevention of breast cancer.223 Taken together, these studies indicate that sex hormone homeostasis plays a vital role in the regulation of skeletal health, and may have substantial effects on joint health.


Does any decrease in bone remodelling result in a secondary decrease in cartilage degradation? The effect of the steroid Tibolone was recently investigated for its effect on bone and cartilage turnover. Tibolone is a synthetic steroid ((7-α, l7-α &17-hydroxy-7-methyl-l9-norpregn-5(1O)-en-20-yn-3-one, Org OD 14, Livial, Organon, The Netherlands) with a combination of oestrogenic, androgenic and progestogenic properties that is capable of relieving climacteric symptoms, and used for treatment of postmenopausal symptoms.224–226 The oestrogenic characteristics of Tibolone imply an oestrogen-like effect on bone as well, which has been documented and characterised previously.227 In a current investigation, Tibolone significantly inhibited bone resorption by approximately 60% compared with placebo, whereas no effect was demonstrated on the cartilage degradation marker, CTX-II.228

This suggests that bone resorption can be strongly attenuated without a secondary effect on cartilage degradation, and consequently indicates that the tight coupling between bone and cartilage metabolism can be dissociated under certain circumstances. This important difference compared with other oestrogen-like molecules, which display protective effects on bone and cartilage, is interesting. This may, in part, result from the complicated mode of action of Tibolone with oestrogenic, androgenic and progestogenic properties. In fact, androgens have been shown to augment progression in OA models, and recently a comparison of male and female sex hormones in animal models of OA led to the conclusion that20 ‘sex hormones play a critical role in the progression of OA in the murine DMM surgical model, with males having more severe OA than females’. Further research is needed to understand these complicated effects of steroids and their metabolites, but the Tibolone study may suggest that some, but not all, steroids may have positive effects on the pathogenesis of OA, and that the androgenic component of the drug may be detrimental in females.


Strontium ranelate has been proposed as an antiosteoporotic drug capable of changing the balance between bone resorption and bone formation,229 ,230 protecting postmenopausal women from spinal and peripheral fractures.229 ,231–233 The mechanisms by which the drug exerts its effects remain relatively elusive, although some of these effects are mediated by the RANK-L-RANK pathway. Strontium ranelate stimulates OPG and inhibits RANK-L synthesis in osteoblasts,234 and activates the calcium-sensing receptor although with a lower affinity than calcium.235 Different observations suggested that this drug may also have beneficial effects in the management of patients with OA. In addition to the hypothetical effects on increased bone turnover in OA, based on the observations in OP and associated models, subanalysis of the OP trials suggested that strontium ranelate reduced levels of the cartilage turnover marker CTX-II in patients with and without OA.236 In a subanalysis of patients treated to prevent spinal fractures, lower spinal OA progression grades were found in those treated with strontium ranelate as opposed to healthy controls.237 At the molecular level, in vitro data suggest a direct effect of the drug on cartilage anabolism,238 and experiments in the dog ACLT model demonstrated a beneficial effect on cartilage damage which is most likely linked to a reduction in inflammation and protease expression.239 ,240 Most importantly, Reginster et al241 recently presented data from a large randomised clinical trial in patients with radiographic and clinical knee OA. The study shows that strontium ranelate provided a beneficial effect on radiographic progression of disease based on joint space narrowing after 3 years of treatment compared with placebo. Effects on pain appeared to be more modest and were only significant for the 2 g group, as assessed using total Western Ontario and McMaster Universities Arthritis Index (WOMAC) score and the pain subscore. These data are significant as the drug appears to have a true disease-modifying potential. Nevertheless, many questions remain, particularly with regards to subtypes or patient profiles, most likely to benefit from strontium ranelate.

PTH: an anabolic treatment for bone and cartilage?

If inhibition of bone turnover may be positive for cartilage health, does this mean that stimulation of bone turnover is bad as bone sclerosis, bone formation and osteophytes are integrated features of OA development242–246? Recently, Wnt signalling was shown to be essential for osteophyte generation73 and bone sclerosis in an animal model of cartilage destruction. Upregulation of genes in the Wnt family has also been observed247 in OA, suggesting that augmenting Wnt signalling may accelerate the development of OA. This highlights that anabolic pathways may exert differential effects on OA pathogenesis and, potentially, that additional attention should be directed to patient selection for such anabolic treatments.

PTH is presently the only bone anabolic treatment accepted by the FDA in the USA, and the European Medicines Agency (EMA). PTH stimulates osteoblasts to synthesise bone. Interestingly, chondrocytes and osteoblasts are from the same mesenchymal cell (MSC) lineage, which suggests that PTH might also affect chondrocytes anabolically. The most compelling evidence for a potential anabolic effect of PTHrP (or PTH) is the historical literature on PTHrp in the growth plate biology, where it has been shown to block chondrocyte hypertrophy which is important for cartilage and bone growth.248–257 Recently, the effect of PTH and PTHrp in OA settings has been further described.

A number of in vitro and in vivo studies have indicated that PTH influences articular cartilage homeostasis. Sampson et al showed that PTH1–32 protects against articular cartilage degradation in a mouse model.258 Furthermore, Chang et al showed that PTH1–32 inhibits terminal differentiation of human articular chondrocytes and OA progression in rats.259 Chondroprotective and chondrodestructive effects have been reported when articular cartilage is exposed to PTH. That may, in part, be mediated via the very interesting protective effect of cAMP on cartilage degradation and formation.260 ,261 Additionally, Wale et al showed that PTH1–32 inhibited expression of type X collagen in MSCs from OA patients in a time-dependent manner. In parallel, PTH1–32 stimulated expression of type II collagen, a marker of chondrogenic differentiation and cartilage repair. In other studies, PTH1–32 also inhibited the phosphorylation of p38 and Akt protein kinase signalling pathways in chondrocytes.262 These results indicate that PTH may be chondroprotective by inhibiting hypertrophy and cartilage calcification. PTH1–32 has also been shown to enhance bone repair in animal models of aging, inflammatory arthritis and glucocorticoid-induced bone loss.263

In summary, there are emerging lines of evidence suggesting that PTH could be chondroprotective. One major consideration is the limited allowed exposure period to PTH and analogues, but the potential positive effects on cartilage may be important for these bone anabolic treatments.

Taken together, this increasing line of evidence points toward some antiresorptive treatments, such as bisphosphonates, calcitonin, cathepsin K inhibitors, oestrogen or SERMs, having positive effects on cartilage and bone degradation, possibly due to the intimate interaction and interplay between bone remodelling and cartilage homeostasis. Some of these possible interventions may, in addition to the effects on osteoclasts and bone turnover, provide additional benefits by targeting chondrocytes directly (PTH and PTHrp). Combined, these different preclinical and clinical studies with various bone drugs indicate that a carefully selected antiresorptive or anabolic treatment could provide clinical benefits in OA for a selected patient population.

Past, present and future


One bisphosphonate (risedronate) has been tested in phase II and phase III clinical settings, with positive effects on OA in phase II and negative in phase III.164–167 While it is outside the scope of this review to discuss whether these outcomes were due to failures in clinical trial design or the hypothesis, it is important to note that there were very few progressors in the phase III study, virtually eliminating the possibility of a comparison and positive efficacy outcome. The result contrasts with the many positive preclinical data on bisphosphonates in OA, and suggests that further strong consideration is warranted of some antiresorptive treatments as potential OA therapies.


Very recently, another antiresorptive agent, strontium ranelate, has produced positive data in a pivotal phase III clinical study for OA. While the mode of action needs to be further understood, significant structural modification was achieved over the long term. This positive outcome was presaged by results from studies based on analysis of the urinary CTX-II biomarker indicative of joint turnover in OA.236 These data, and a better understanding of the patients who might respond best to antiresorptive therapies, may enable the testing of similar treatments in more selectively chosen OA populations. Whether the strongest antiresorptive treatment, anti-RANKL (Denosumab),264 would be beneficial for a selected OA population is highly interesting.


The OA field is becoming increasingly aware of the heterogeneity of OA patients. Not all OA patients should be treated with the same intervention. Some patients with chronic inflammation may be more appropriately treated with anti-inflammatory therapies, while others with mainly a bone-cartilage interface imbalance may benefit from an antiresorptive treatment. Finally, injectable options which regenerate cartilage in selected patient populations may also prove very useful. Optimal therapy in future will surely require identification of specific patient phenotypes for targeted therapy, and by inference, likely be reflected in pharmaceutical agents with narrower labels.

The consequence of this thinking is illustrated in figure 3. Different interventions may be needed at different stages of OA and, most important, one intervention that may be beneficial at one stage, or in one patient subgroup, may not be successful at a different stage or in a different subgroup with the disease.

Figure 3

The pathogenesis of osteoarthritis consists of bone and cartilage degradation, as well as synovial inflammation. The compartments are affected to different degrees at different stages of the pathology. Thus, one targeted intervention which may be effective at one stage of the disease may be ineffective at another stage. For example, an anti-inflammatory intervention may be suited for some patients (TREATMENT 2), while other patients may be more suited to a different treatment targeting the bone-cartilage interaction and not inflammation (TREATMENT 1). This figure is modified from: Karsdal MA, Bay-Jensen AC, Henriksen K, et al. The pathogenesis of osteoarthritis involves bone, cartilage and synovial inflammation: may estrogen be a magic bullet? Menopause 2012;18.



  • Contributors MAK, ACB-J, VK and KH discussed the outline of the paper. MAK made the first initial draft and outline of the manuscript. All authors contributed to the whole paper by critically revising the intellectual content, and making. All authors approved the final version of the manuscript.

  • Funding The funding from the ‘Danish Research Foundation’ is greatly acknowledged for the unrestricted research grant to MK.

  • Competing interests MAK and CC own stock in Nordic Bioscience. PP is a full-time employee of Sevier Laboratories. All authors declare that with the relation to this manuscript they have no conflicts of interests.

  • Provenance and peer review Not commissioned; externally peer reviewed.