Bone remodelling in osteoarthritis David B. Burr and Maxime A. Gallant Abstract | The classical view of the pathogenesis of osteoarthritis (OA) is that subchondral sclerosis is associated with, and perhaps causes, age-related joint degeneration. Recent observations have demonstrated that OA is associated with early loss of bone owing to increased bone remodelling, followed by slow turnover leading to densification of the subchondral plate and complete loss of cartilage. Subchondral densification is a late event in OA that involves only the subchondral plate and calcified cartilage; the subchondral cancellous bone beneath the subchondral plate may remain osteopenic. In experimental models, inducing subchondral sclerosis without allowing the prior stage of increased bone remodelling to occur does not lead to progressive OA. Therefore, both early-stage increased remodelling and bone loss, and the late-stage slow remodelling and subchondral densification are important components of the pathogenetic process that leads to OA. The apparent paradoxical observations that OA is associated with both increased remodelling and osteopenia, as well as decreased remodelling and sclerosis, are consistent with the spatial and temporal separation of these processes during joint degeneration. This Review provides an overview of current knowledge on OA and discusses the role of subchondral bone in the initiation and progression of OA. A hypothetical model of OA pathogenesis is proposed. Burr, D. B. & Gallant, M. A. Nat. Rev. Rheumatol. 8, 665–673 (2012); published online 7 August 2012; doi:10.1038/nrrheum.2012.130
Introduction Understanding the role of bone remodelling in the initi ation and progression of osteoarthritis (OA) requires a clear definition of the boundaries and distinct physio logical differences in the subchondral bone ‘compart ments’. The term ‘subchondral bone’ has been used to refer to any bone found distal to calcified cartilage, but this definition confounds spatial, architectural, physiological and mechanical differences that may be critical to under standing the role of subchondral bone in OA (Figure 1). Immediately beneath the calcified cartilage is a 1–3 mm thick plate of corticalized bone1–3 that is physiologically and mechanically similar to cortical bone in other skeletal locations, but somewhat less stiff than diaphyseal cortical bone.4 Distal to this structure is subchondral cancellous bone that is more porous and metabolically active and has a lower volume, density and stiffness than the cortical plate. The term subchondral bone is often used to refer to both subchondral cancellous bone and the cortical plate, without making a proper distinction for their differences. It is important, however, to distinguish between the two regions because, at least in late-stage disease, the changes that occur in the subchondral cortical plate are different from those that occur in the cancellous bone.5,6 It is also important to recognize that calcified cartilage, which sepa rates subchondral cortical bone from non-mineralized articular cartilage at the tidemark, can contribute to sclerosis observed in advanced OA.
With OA progression, there is a well-known re- establishment of the process of endochondral ossification at the tidemark, which can be detected histologically by the presence of multiple tidemarks signifying tidemark advancement.7–9 As a result of this renewed developmental process, the calcified cartilage, which is more highly mineralized than bone10 (Figure 2), gets thicker. The thickening of the calcified cartilage makes the overlying layer of articular cartilage, which is not very dynamic and is not capable of making enough new cartilage to main tain its volume, thinner. The changes in the subchondral cortical and subchondral cancellous bone have distinct mechanical effects on articular cartilage. Implanting a stiff metal cylinder in the subchondral cancellous bone has no effect on stresses in the overlying cartilage unless it is close to the osteochondral junction.11 If the implant is within the subchondral cortical bone, however, stresses in the deep layers of cartilage increase by about 50%.11 This finding suggests that processes that occur in the cancellous bone beneath the subchondral plate are inconsequential to the mechanical function or health of the joint. Many years ago, trabecular micro fractures were identified as a potential pathophysiological mechanism accelerating joint failure.12 However, because these microfractures occur in cancellous bone distant from the articular cartilage, they cannot increase joint stiffness and do not have a role in progressive deterioration.13
Competing interests D. B. Burr declares associations with the following companies: Amgen Inc., Eli Lilly & Co., Merck and Wright Medical. See the article online for full details of relationships. M. A. Gallant declares no competing interests.
Initiation and progression of osteoarthritis To understand the pathophysiology of OA, it is important to acknowledge that the initiation of cartilage damage and the progressive deterioration of the articular cartilage
NATURE REVIEWS | RHEUMATOLOGY
Department of Anatomy and Cell Biology, MS 5035, Indiana University School of Medicine, Indianapolis, IN 46202, USA (D. B. Burr, M. A. Gallant). Correspondence to: D.B. Burr [email protected]
VOLUME 8 | NOVEMBER 2012 | 665 © 2012 Macmillan Publishers Limited. All rights reserved
REVIEWS Key points ■■ Subchondral cortical bone and subchondral cancellous bone are architecturally, physiologically and mechanically different, and respond differently in osteoarthritis (OA) ■■ The initiation and the progression of OA are distinct pathophysiological processes ■■ The early stages of OA are characterized by increased vascularity and reduced bone density ■■ Late-stage OA is characterized by decreased bone resorption without a decrease in bone formation, and by the development of subchondral sclerosis ■■ Increased bone remodelling is a necessary condition for OA; increased subchondral bone density alone does not lead to OA ■■ Antiresorptive agents that suppress bone remodelling during late stages of OA have not proven, and are not likely, to be effective treatments
that lead to complete cartilage loss and OA are driven by different conditions and represent distinct patho physiological processes. Every person sustains some car tilage damage by the time they reach 60 years of age—an inevitable result of aging; however, not everyone develops OA. In experimental studies, scarification of the articu lar cartilage does not inevitably lead to OA,14 suggesting that there are special conditions that promote progres sion of cartilage damage. Therefore, the most effective treatments for OA will be drugs that target the condi tions that accelerate progression of disease, rather than the chondroprotective agents that prevent the initiation of damage.
Subchondral density in osteoarthritis Subchondral sclerosis is a hallmark and indisputable sign of the progression of OA. However, its primary role in OA progression—as a driving force or as a consequence of the catabolism of the articular cartilage—has been disputed for four decades. One role of subchondral bone in the joint is to distribute forces, and to adapt in ways that maintain conformation of the joint and prevent stress concentrations. Radin et al.15 proposed in 1970 that sclerotic sub chondral bone creates an overly stiff substrate and internal stress concentrations in cartilage that lead to deterioration of cartilage. The original idea was that because articular cartilage and subchondral bone are viscoelastic, the joint ends deform when loaded. This viscoelasticity allows maximum contact area under the load and minimizes stresses. Subsequently, the idea was refined16 and we now have a better understanding of the mechanical performance of articular cartilage. Articular cartilage is designed for loadbearing, and the high water content of cartilage allows it to deform under compressive loads without failure. However, it is mechani cally less capable of withstanding tension or shear stresses that occur at the edges of the joint contact regions. High tensile and shear stresses at these margins predispose the cartilage to splitting or fibrillation. This predisposition could be exacerbated by inhomogeneities in the density and stiffness of the underlying subchondral bone, which would cause cartilage to deform more in regions over lying less dense parts of the subchondral plate than in regions over denser portions of the plate. The junction between the stiffer regions and the less stiff regions of 666 | NOVEMBER 2012 | VOLUME 8
the subchondral plate are sites of stress concentration at which the cartilage is more likely to fail in tension. Evidence from mathematical modelling and from experimental animal models provide a mixed and conflict ing picture about whether subchondral bone densification is associated with progressive OA in a cause-and-effect manner. Early studies using a rabbit impulsive loading model suggested that increases in bone volume and cata bolic changes in cartilage were concurrent.17 Nonhuman primates that develop OA spontaneously present with a thickened subchondral plate prior to the development of apparent cartilage damage,18,19 as does a well-established guinea pig model.20 Other models in which OA is experi mentally induced also demonstrate that bone changes occur earlier than cartilage deterioration.21 This finding suggests a temporal relationship between the changes in bone and cartilage during OA progression. Moreover, experimental studies in rabbit and mouse models of OA have shown that deterioration of articular cartilage occurs in regions where the subchondral plate is thick ened, but not in regions overlying subchondral bone that is not thickened.22,23
Bone remodelling in early-stage disease Rationalizing subchondral sclerosis in OA with observa tions of increased bone remodelling is difficult. 24–26 In contrast to sclerosis, increased bone remodelling is associ ated with a transient loss of bone, increased porosity in the subchondral region, and reduced density. It is possible, however, that sclerosis and accelerated remodelling in OA are not inconsistent observations, but simply reflect differ ent temporal stages of disease or different spatial locations, or both. This paradox can be partially resolved by answer ing the following questions: what are the early changes to subchondral bone in experimental conditions that lead to OA? What happens to articular cartilage when these early changes in bone are prevented? Does experimen tal induction of subchondral sclerosis lead to progressive cartilage deterioration?
Remodelling rate In early OA, the mineral apposition rate, normally around 0.70–1.00 μm/day, is increased by threefold– fivefold to an average of >3.50 μm/day.22,27 There is also an increase in the initiation of new remodelling sites within the subchondral bone. This increased rate of remodelling reduces the thickness of the subchondral plate.28 Thinning of the subchondral plate early in OA has been shown in two different canine models of OA in which OA was induced by either creating grooves on the femoral condyles or transecting the anterior cruciate ligament (ACL).29 20 weeks after OA induction, a con siderable reduction in the thickness of the subchondral plate was observed that was associated with increased articular cartilage destruction and reduced synthesis of glycosaminoglycans. Thinning of the subchondral plate was also associated with increased cartilage damage in a rabbit model.30 Rabbits in which accelerated bone remodelling and bone loss were induced prior to surgical induction of OA had considerably more severe cartilage
www.nature.com/nrrheum © 2012 Macmillan Publishers Limited. All rights reserved
BONE RESEARCH damage than those rabbits without prior induction of accelerated bone remodelling.30 Similar effects can be shown in humans with early-stage but progressive cartilage deterioration.31–33 Data from the Chingford study conducted in a cohort of women aged 45–64 years show that resorption markers (for example, C‑terminal and N‑terminal telopeptides: CTx and NTx) are considerably elevated in individuals with progres sive OA, but not in those with non-progressive forms of OA.24 Resorption markers are also elevated in young and middle-aged adults (aged 27–56 years) with early-stage OA, but without clinical symptoms.25 Interestingly, bone resorption and cartilage loss seem to be spatially associ ated within a single joint, with a sevenfold increase in risk of cartilage loss in regions with subchondral attri tion.26 Other researchers have also noted that changes in subchondral bone mineralization and volume occur only beneath the areas of significant cartilage destruc tion.34 The increased rates of bone remodelling in earlystage OA may cause alterations in joint shape and load transmission that predispose to progressive cartilage loss. These observations in both humans and animal models, however, are not sufficient to determine whether the association between subchondral bone loss and cartilage damage is part of a pathogenetic sequence, or simply rep resents non-causal and independently occurring changes in bone and cartilage during the development of OA.
Causes of increased remodelling The causes of increased bone remodelling in early OA are unknown. Several different mechanisms involved may include cellular signalling for microdamage repair, stimulation of vascular invasion by angiogenic factors, and bone–cartilage crosstalk via subchondral pores. Cellular signalling Elevations in transforming growth factor β, insulin-like growth factor, interleukin (IL)-1, IL-6 and prosta glandin E2 protein levels have been detected in deterio rating cartilage,35 but all these proteins are products, as well as stimulators, of bone remodelling. Cell culture experiments showed that osteoblasts taken from joints of patients with OA produce twofold–sixfold more IL‑6 and prostaglandin E2 than osteoblasts taken from indivi duals without OA.36 Moreover, upregulation of Wnt signal ling has been reported in knee joints of rats with OA.37 However, these findings beg the question, what stimulates the secretion of these proteins? There is good evidence to suggest that normal repetitive loading of joints can create microcracks in the subchondral plate, even in non- diseased joints.8,9 What was not fully recognized at the time of these initial observations, but is now well-known, is that microcracks are a nidus for the initiation of new remodelling events,8,38,39 and might stimulate osteocytes in the region of damage to produce RANKL (recep tor activator of nuclear factor κ‑B ligand; also known as TNF superfamily member 11 [TNFSF11]) and down regulate osteoprotegerin (OPG), an inactivating receptor for RANKL, to induce bone resorption.40,41 A decreased OPG:RANKL ratio has been observed in animal models
Tidemark Subchondral bone plate thickness
Tidemark Subchondral bone plate thickness
Loss of aggrecan Fibrillation Articular cartilage Calcified cartilage
Subchondral bone plate thickness
Figure 1 | Stages of progressive join degradation in OA. The mineralized tissues beneath the articular cartilage are sometimes referred to collectively as subchondral bone or subchondral mineralized tissues, but in reality include different kinds of tissues that vary compositionally, architecturally, physiologically and mechanically. The subchondral plate is dense cortical bone, but the subchondral cancellous bone beneath it is quite porous. The distinction between these tissues is important because they change in different ways and at different times during the development of OA. In early-stage OA, the subchondral plate becomes thinner as a consequence of an increased remodelling rate. At the same time, cancellous bone is lost as the trabecular plates become thinner and more rod-like. In late-stage disease, the subchondral plate thickens, but the subchondral cancellous bone remains osteopenic. The calcified cartilage begins to advance into the articular cartilage in late-stage disease, leaving a footprint of multiple tidemarks as the mineralization front advances. This creates an even thicker mineralized plate, and reduces the thickness of the non-mineralized articular cartilage which cannot replace itself. This is accompanied by a loss of aggrecan beginning superficially in the articular cartilage (shown by a change in stain color), and surface fibrillation. The sum result of these changes is subchondral sclerosis (that includes both the subchondral plate and calcified cartilage) and thinner, more fibrillated articular cartilage. Abbreviation: OA, osteoarthritis.
NATURE REVIEWS | RHEUMATOLOGY
VOLUME 8 | NOVEMBER 2012 | 667 © 2012 Macmillan Publishers Limited. All rights reserved