Prevention of bone mineral density loss in patients with rheumatoid arthritis treated with anti-TNFα therapy

Abstract

This review focuses on recent advances in the effect of anti-TNFα therapy on bone metabolism and bone mineral density (BMD) in rheumatoid arthritis (RA). RA is a chronic disease characterized by inflammation of the synovial joint, cartilage degradation, and subsequent bone destruction. Bone damage is often manifested as erosions, localized juxta-articular bone loss, or generalized bone loss. Thus, blockade of TNFa not only serves to block inflammation, but also halts the erosive nature of RA and generalized/localized juxta-articular bone loss. Here, we review recent findings showing that anti-TNFa therapy is also effective on halting systemic bone loss. In vitro, TNFa reduces osteoblast activity and increases osteoclast activity through RANKL-RANK pathway. In arthritis animal models, an imbalance between bone formation and resorption is observed. In humans, this coupling of destruction is restored by anti-TNFα therapy early on, but only for a few months. Thus, anti-TNFα prevents the BMD loss in RA patients. In summary, TNFa blockade is not only able to prevent joint destruction, but it is also able to prevent bone loss in RA patients. Future studies are needed to address if TNFa blockers have an effect on bone fractures.

Introduction

Rheumatoid arthritis (RA) is a chronic disease characterized by inflammation of the synovial joint, cartilage degradation, and subsequent bone destruction. Bone damage often is manifested as erosions, localized juxta-articular bone loss, or generalized bone loss (Kroot et al 2001; Gough et al 1994). The later is likely attributed to a number of factors including disease itself, reduced exercise activity, steroids therapy (Laan et al 1993; Hall et al 1993; Cortet et al 1997; Haugeberg et al 2002; Mikuls 2003), or common postmenopausal osteoporosis. Accelerated generalized bone loss is often considered as one of the major complications of RA (Westhovens and Dequeker 2000; Green and Deodhar 2001; Sambrook et al 1987; Goldring and Gravallese 2000), as patients often have an increased risk of hip (Cooper et al 1995; Huusko et al 2001; van Staa et al 2006) or vertebral fractures (Spector et al 1993; Peel et al 1995; van Staa et al 2006). Blockade of mediators of joint inflammation is a promising strategy to halt or perhaps improve the severity of these bone related manifestations.

TNFα a key cytokine target

For almost a decade, anti-TNFα therapy has established a new standard in RA therapy, creating new objectives for disease remission on both the clinical and radiological level. Currently, three anti-TNFα drugs are commercially available and have been successfully used in halting both joint inflammation and destruction in RA patients (Charles et al 1999; Maini et al 1999; Weinblatt et al 1999; Genovese et al 2002; Lipsky et al 2000; Keystone et al 2004; Smolen et al 2005). Furthermore TNFα plays a central role not only in RA, but also in common osteoporosis. In fact, TNFα has been shown to increase bone resorption in systemic osteoporosis related to oestrogen deficiency (Pacifici and Avioli 1993). In addition, transgenic mice expressing soluble TNFα receptor to neutralize TNFα, were protected from oestrogen deficiency-related bone loss (Ammann et al 1997). Thus, blockade of TNFα not only serves to block inflammation, but also halts the erosive nature of RA and generalized/localized juxta-articular bone loss. Here, we review recent findings showing that anti-TNFα therapy is also effective on halting systemic bone loss.

In vitro effects of TNFα on osteoblasts and osteoclasts (Figure 1)

Figure 1.

Increasing the balance of receptor activator of nuclear factor κB ligand (RANKL)-receptor activator of nuclear factor κB (RANK) induced by tumor necrosis factor alpha (TNFα).

Abbreviations: OPG, osteoprotegerin; +, stimulation; −, inhibition.

Over the last two decades, there have been a number of findings on the effects of TNFα on osteoblast activity. These include the different down regulatory effects of TNFα on osteoblasts: differentiation, mineralization, and expression of the skeletal bone matrix (Bertolini et al 1986; Nguyen et al 1991; Taichman and Hauschka 1992; Kuroki et al 1994; Nakase et al 1997; Gilbert et al 2000, 2002). Furthermore, TNFα stimulation also induces apoptosis in osteoblasts (Hill et al 1997; Jilka et al 1998).

Research on the development of bone erosions has rapidly increased over recent years. First was the detection of osteoclasts at sites of arthritic bone erosion (Ishikawa et al 1984). Second was the discovery of the balance between TNF superfamily molecules: osteoprotegerin (OPG), receptor activator of nuclear factor κB ligand (RANKL), receptor activator of nuclear factor κB (RANK), and TNF-related apoptosis inducing ligand (Gravallese et al 2000; Kong et al 1999, 2000; Hofbauer and Heufelder 2001). RANKL is a transmembrane receptor protein and its inhibition by osteoprotogerin (OPG), a decoy receptor, has important implications for inflammation research as well as for bone physiology (Kong et al 1999; Emery et al 1998).

Osteoblasts express RANKL constitutively on their cell surface. RANKL interacts with its cognate receptor RANK, which is expressed on osteoclast precursors and promotes osteoclast differentiation. Interaction of RANKL with RANK on mature osteoclasts results in their activation and prolonged survival. Importantly, osteoclasts are clearly the principal instruments of this destruction. As shown by studies using osteoclast deficient animal models of disease, treatment with antiresorptive agents such as OPG, delays bone damage in vivo (Joosten et al 1999; Kong et al 1999; Pettit et al 2001; Gravallese 2002; Redlich et al 2002b; Saidenberg-Kermanac’h et al 2004).

RANKL also plays a key role in the regulation of osteoclastogenesis, osteoclast activation, dendritic cell survival, lymphocyte development, and lymph node organogenesis (Kong et al 1999). Studies showing that the RANKL/RANK ratio regulates osteoclastogenesis have profoundly influenced our current understanding of inflammation mediated bone loss (Kong et al 1999; Gravallese et al 1998, 2000; Shigeyama et al 2000; Takayanagi et al 2000; Redlich et al 2002a; Romas et al 2002; Pettit et al 2001). OPG deficient mice exhibit massive osteoporosis and also develop calcified lesions in the aorta and renal arteries (Bucay et al 1998; Min et al 2000). In contrast, the overexpression of OPG in transgenic mice results in severe osteopetrosis and splenomegaly (Min et al 2000). The inhibition of bone resorption by the administration of OPG has been demonstrated in postmenopausal women (Bekker et al 2001) and in mice challenged with interleukin-1β (IL-1β) or TNFα (Morony et al 1999). Treatment with anti-RANKL antibody has also resulted in an increased amount of bone mineral density (BMD) (McClung et al 2006).

TNFα elevates RANKL expression in numerous cell types: stroma cells, synoviocytes, osteoblasts, endothelial cells, human microvascular endothelial cells, T cells, and B cells (Hofbauer et al 1998; Cenci et al 2000; Kanematsu et al 2000 Nakashima et al 2000; Page and Miossec 2005; Collin-Osdoby et al 2001). TNFα also has differential effects on OPG expression depending on cell type: a decrease is observed in osteoblasts/stromal cells (Nakashima et al 2000), while an increase is observed in endothelial cells (Collin-Osdoby et al 2001), osteoblast cells (Hofbauer et al 1998), and in RA synoviocytes (Marotte et al 2005). In this later, an inhibition of this stimulation is observed when infliximab is added (Marotte et al 2005).

RA and psoriatic arthritis have a marked increase in the number of pre-osteoclasts in their peripheral blood mononuclear cell populations compared to normal and osteoarthritis controls (Ritchlin et al 2003). Importantly, this increase also appears to be reversible with anti-TNFα therapy, and may be a predominant mechanism by which this treatment inhibits erosions in these patients.

Unbalance between bone formation and resorption in arthritis model

Much less is known about bone formation in animal models of immune-mediated arthritis. Given the insidious erosive course of these conditions, it is obvious that bone resorption outweighs bone formation. However, bone formation is strongly down-regulated during chronic arthritis. For example, studies in animal models, which addressed systemic bone changes, showed reduction in bone formation in femurs by histomorphometry as well as low serum levels of the bone formation marker, osteocalcin (OC) (Bonnet et al 1993; Hoshino et al 2001). Furthermore, blocking TNFα prevents bone loss in animal models and increases trabecular thickness (Joosten et al 1999; Saidenberg-Kermanac’h et al 2004).

Effect of anti-TNFα on bone markers

As shown above, bone remodeling is the result of two opposing activities, the production of new bone matrix by osteoblasts and the destruction of old bone by osteoclasts. Biochemical markers of bone turnover are substances in the blood or urine that are produced or released during bone remodeling. They provide semi-quantitative information on bone remodeling. At present, the most specific markers for bone formation are serum OC, bone-specific alkaline phosphate (BAP), and procollagen type I N-terminal propeptide (PINP). OC is down regulated by steroid therapy (Vergely et al 2002), whereas BAP and PINP changes are not linked to this therapy (Cortet et al 2001). For the evaluation of bone resorption, numerous serum markers of type I collagen telopeptide epitopes (CTX-I, NTX-I, and ICTP) and deoxypyridinoline are currently used. Among them, C-telopeptides (CTX-I and ICTP) are the most sensitive markers of bone resorption (Garnero and Delmas 2004). By two different enzymatic pathways, two C-telopeptide fragments are released from bone type I collagen (Sassi et al 2000; Garnero et al 2003). ICTP is generated by matrix metalloproteases whose activity plays an important role in collagen degradation associated with RA (Brennan et al 1997). In contrast, CTX-I is generated by cathepsin K, which is the key osteoclastic enzyme for systemic bone resorption (Inui et al 1997).

Several cross-sectional studies in human RA support these experimental findings. RA patients have significantly lower osteocalcin levels than healthy controls (Al-Awadhi et al 1999; Garnero et al 1999; Seriolo et al 2002). Indeed, in one of our previous studies (Garnero et al 1999), uncoupling between bone destruction and formation was observed in destructive RA, whereas such uncoupling was not observed in benign non-destructive RA. It now appears that TNFα inhibition may restore this coupling of destruction with formation. Initial clinical studies on the effect of TNFα blockers on bone focused primarily on bone markers. Bone formation markers (OC and PINP) showed an early increase at 2 and 6 weeks after initiation of infliximab treatment in a study conducted in 68 RA patients (Vis et al 2003), whereas only ICTP was down regulated at 6 weeks despite having negligible effects on CTX-I. The long-term effect of infliximab at 1 year was described in a small population of 26 patients with RA (Lange et al 2005), with a persistent increase in OC and a persistent decrease in CTX-I. However, in another study of 70 patients with RA treated with infliximab plus methotrexate, a decrease was observed only in bone resorption markers (urinary excretion of NTX-I and deoxypyridinoline), whereas BAP did not increase (Torikai et al 2006). The same pattern with a rapid decrease of CTX-I levels followed by a return to baseline levels at 1 year was observed in several studies in RA (Lange et al 2005; Vis et al 2005, 2006; Chopin et al 2007), and also in spondylarthropathy (Briot et al 2005). Only one study explored the changes of CTX-I and ICTP under infliximab over one year (Chopin et al 2007). ICTP level showed a later decrease compared to CTX-I, suggesting different specific resorption processes. In another study, infliximab therapy had a less significant effect on serum OPG and sRANKL levels in RA patients (Ziolkowska et al 2002). Recently, decreases in CTX-I and sRANKL on serum were observed over 1 year of infliximab treatment (Vis et al 2006). In summary, early changes in bone remodeling under infliximab was characterized by an uncoupled response with TNFα inhibition inducing a decrease in bone resorption. This was followed by a secondary steady-state with higher resorption level to match bone formation. Finally, the ratio between markers of bone formation and bone resorption appeared to be increased.

Effect of anti-TNFα on BMD

Bone mineral density (BMD) can be assessed by numerous methods, but dual x ray absorptiometry (DXA) is considered the “gold standard” (Genant et al 1996). In practice, lumbar spine (L1-4) and total hip are measured after a daily calibration using an external phantom to detect a machine drift. For longitudinal studies, several assessments need to be performed in the same bone densitometer.

High BMD loss in RA patients was associated with joint damage progression, disease activity, functional disability and immobility in previous longitudinal studies, even in early RA (Gough et al 1994; Shenstone et al 1994; Cortet et al 2000; Forslind et al 2003; Jensen et al 2004). However, several observational studies were performed without enrolled control groups. The first studies enrolled a small number of patients, and showed an increase of BMD in numerous inflammatory disorders. In patients with RA (n = 26) (Lange et al 2005), with spondylarthropathy (n = 29) (Allali et al 2003) and with Crohn’s disease (n = 46) (Bernstein et al 2005), a significant increase of BMD values was observed. Another study enrolled 102 RA patients in an open prospective cohort. After 1 year of infliximab therapy, no bone loss was observed in the spine and hip, while a decrease of BMD was observed at the hands (Vis et al 2006). Unfortunately, no control group was used in the study.

The lack of a control group is a major limitation for predicting a true estimation of the effect of anti-TNFα treatment on bone. The gold standard for this type of study would be a double blind randomized vs placebo trial. However, since TNFα blockers are now on the market, ethical issues would prevent such a randomized vs placebo trial. In order to have a control population for such a study, we used another option. We performed a historical control study, composed of active RA patients followed before the time of TNFα blockers and treated with methotrexate alone (Marotte et al 2007). This historical control group is of critical interest because it was not influenced by the use of TNFα inhibitors. In the control group (n = 99), we found a significant decrease of BMD (–3.4% at the femoral neck and −3.9% at lumbar spine, p < 0.001), while no decrease was observed in the group treated by methotrexate and infliximab. Similar results were observed when putative cofactors are considered (sex status, age, menopause status, steroids or biphosphonate therapy, and RA activity). Furthermore, there was no significant difference in BMD changes between responders and non responders. For the non responders, the change in lumbar spine BMD was +0.4% vs −0.8% for the responders. Same trend was observed in femoral neck with BMD change at +2.0% in the non-responders vs −0.4% in the responders. This protective effect observed in clinical non responders to infliximab is in accordance with the findings of a recent study (Lipsky et al 2000; Genovese et al 2002; Keystone et al 2004; Smolen et al 2005) that demonstrated protective effects on joint destruction in the absence of clinical response.

Since our study, new data is available coming from the BeSt study, which is a multicenter, randomized clinical trial designed to compare the effectiveness of four different treatment strategies in patients with recently diagnosed, active RA (Guler-Yuksel et al 2008). The main finding of this study was that there were no differences between the treatment groups. Again in this study, no appropriate control group was used according to ethical issues. Furthermore, the lack of bone loss in the group with steroids can be explained in part by the large prescription of biphosphonate therapy in this subgroup.

Conclusion

In summary, TNFα blockade is not only able to prevent joint destruction, but it is also able to prevent bone loss in RA patients. Future studies are needed to address if TNFα blockers have an effect on bone fractures.

Table 1.

Effect of TNFα and anti-TNFα on bone cells, bone biomarkers, X-rays, and bone mineral density

		TNFα	Anti-TNFα
In vitro	Osteoblasts	Inhibition of differentiation (Gilbert et al 2000; Gilbert et al 2002)	?
		Inhibition of mineralization (Taichman and Hauschka; 1992; Kuroki et al 1994; Nakase et al 1997)
		Skeletal bone matrix synthesis:inhibition (Taichman and Hauschka 1992; Kuroki et al 1994; Nakase et al 1997; Yudoh et al 2000)
		Induction of apoptosis (Hill et al 1997; Jilka et al 1998)
		Stimulation of RANKL expression (Nakashima et al 2000; Hofbauer et al 1998)
	Osteoclasts	Activation via RANK/RANKL	?
		Increasing of OPG production (Collin-Osdoby et al 2001; Hofbauer et al 1998; Marotte et al 2005) Decreasing of OPG production (Nakashima et al 2000)	Decrease of OPG production (Marotte et al 2005)
In vivo	Biomarkers of bone remodeling	Decreasing of biomarkers of bone formation (Bonnet et al 1993) Increasing of biomarkers of bone resorption (Saidenberg-Kermanac’H et al 2004; Joosten et al 1999)	Early restoration of the balance between biomarkers of bone formation and resorption (Lange et al 2005; Vis et al 2005; Vis et al 2006; Chopin et al 2007)
	X-rays	Erosion progression	No erosion progression (Smolen et al 2005; Genovese et al 2002; Keystone et al 2004; Lipsky et al 2000)
	Bone mineral density	Bone loss	Prevention of bone loss (Marotte et al 2007)

Abbreviations: TNFα, tumor necrosis factor alpha; RANKL, receptor activator of nuclear factor κB ligand; RANK, receptor activator of nuclear factor κB; OPG, osteoprotegerin.