Cortical remodeling during menopause, rheumatoid arthritis, glucocorticoid and bisphosphonate therapy

Bone mass, bone geometry and its changes are based on trabecular and cortical bone remodeling. Whereas the effects of estrogen loss, rheumatoid arthritis (RA), glucocorticoid (GC) and bisphosphonate (BP) on trabecular bone remodeling have been well described, the effects of these conditions on the cortical bone geometry are less known. The present review will report current knowledge on the effects of RA, GC and BP on cortical bone geometry and its clinical relevance. Estrogen deficiency, RA and systemic GC lead to enhanced endosteal bone resorption. While in estrogen deficiency and under GC therapy endosteal resorption is insufficiently compensated by periosteal apposition, RA is associated with some periosteal bone apposition resulting in a maintained load-bearing capacity and stiffness. In contrast, BP treatment leads to filling of endosteal bone cavities at the epiphysis; however, periosteal apposition at the bone shaft seems to be suppressed. In summary, estrogen loss, RA and GC show similar effects on endosteal bone remodeling with an increase in bone resorption, whereas their effect on periosteal bone remodeling may differ. Despite over 50 years of GC therapy and over 25 years of PB therapy, there is still need for better understanding of the skeletal effects of these drugs as well as of inflammatory disease such as RA on cortical bone remodeling.

(construction, growth) bone is deposited at the periosteal surface without prior resorption on the endosteal surface, leading to an increase in cortical diameter (Figure 1a) [9]. During bone remodeling (reconstruction, lifetime adaptation to load and age), osteoclasts resorb a small volume of bone at a distinct location at the inner site of the cortex adjacent to the marrow (Figure 1b) [10]. If periosteal appo sition increases (as an adaptive response) to compensate for the loss of strength (caused by endosteal bone loss), there will be no loss in bone strength (Figure 1c) [11][12][13]. Alternatively, if periosteal apposition is impaired, endosteal resorption will produce cortical thinning and loss of bone mass and strength, predisposing to fractures.
Various factors infl uence adaptation of bone geometry. Th ese factors include the menopause, GC therapy and treat ment with anti-resorptive drugs, as well as diseases such as RA, hyperparathyroidism, diabetes mellitus, hypophosphatasia and vitamin D defi ciency and mechanical factors such as weight. In the following paragraphs, current knowledge on the main factors causing bone loss and potentially leading to fractures in patients with RA will be summarized and resulting research questions highlighted.

Menopause
Estrogen loss during the menopause is the most common cause of fractures in older people. Trabecular bone loss occurs years before the menopause; at the onset of estrogen decline, a phase of rapid bone loss starts, leading to loss of an approximate mean of 10% of bone at the spine and about 5% at the hip over a period of 5 years [14,15]. Th is phase is characterized by bone resorption, leading to trabecular thinning and perforation, resulting in a loss of connectivity [16]. Th is initial acute phasethe high-turnover phase -is followed by a long-lasting period of slower bone loss where the dominant microarchitecture change is trabecular thinning and increased cortical porosity [17].
On a cellular level, the high-turnover phase is caused by an increased activity of osteoclasts, driven by the increased production of receptor activator of NF-κB ligand [18][19][20]. Th e second phase is caused by impaired osteoblastic activity due to increased osteoblast death and decreased function [21]. Estrogen therapy blunts bone resorption and stimulates bone formation. Th e dominant eff ect of estrogen is the inhibition of osteoclast diff erentiation and activation, and, to a lesser extent, impairment of apoptosis of the osteoblasts [19,20,22]. Studies on bone geometry and strength of the one-third distal radius in perimenopausal and postmenopausal women showed an accelerated endosteal resorption and decreased periosteal apposition at the radius shaft. Th is endosteal resorption during the postmenopausal years leads to cortical thinning -and because the outward apposition of bone is minimal, the cortical crosssectional area and bending strength decrease (Figure 2a) [23].
Th inning of the cortical shaft thickness of the radius and tibia is associated with an increased fracture risk [24]. In estrogen-treated women, loss of bone strength was partially prevented [24] and the risk of vertebral and hip fracture incidence in postmenopausal women declined [25]. In summary, estrogen defi ciency leads to thinning and loss of bony trabeculae as well as increased endosteal resorption and impaired periosteal bone apposition, resulting in a net loss of bone with diminished strength to compression and bending.

Rheumatoid arthritis
RA is a chronic infl ammatory joint disease characterized by destruction of periarticular bone and cartilage and soft-tissue damage, leading to impaired joint function [1]. Th e general consensus is that the pathogenic process of bone loss in RA patients is due to the abundance of osteoclasts [26]. Bone erosion, periarticular osteopenia and systemic bone loss are central features of RA [2]. Chronic infl ammatory diseases such as RA cause skeletal breakdown and enhance the risk of nontraumatic fractures [27,28]. Approximately 25% of patients with early RA show an osteopenic bone mineral density at the spine or hip before the onset of therapy, and 10% have osteoporosis [29][30][31][32]. A large case-control study based on the British General Practice Research Database has shown that RA doubles the risk of hip and vertebral fractures, supporting the concept that the eff ect of RA is independent and additive to GC [33].
With a newly established protocol using peripheral quantitative computed tomography (QCT) we investigated the diff erences in volumetric bone mineral density (BMD) and bone geometry at the metacarpal bone, tibia and radius in 50 female RA patients and compared it with 100 healthy female controls [3]. We found that RA patients had signifi cantly lower trabecular volumetric BMD at the distal epiphyses of the radius (-19%), tibia (-14%) and metacarpal bone (-12%). At the shafts of these bones RA patients had thinner cortices (-7% to -16%), and at the metacarpal shaft the outer diameter was increased by 5% (Figure 2b). Th e extent of bone erosions and the cumulative dose of GCs correlated negatively with trabecular and total volumetric BMD as well as with shaft cortical thickness [3]. In a subsequent study with 64 RA patients and 128 healthy controls, we assessed factors associated with diff erences in bone geometry between RA patients and healthy controls. Based on linear models with explanatory variables (muscle cross-sectional area, age, RA status and sex), we found that patients with RA showed a greater age-related decrease in cortical thickness with a concomitant increase in outer bone circumference consistent with an enhanced aging pattern [4].
Th ese observations suggest that the biological changes of endosteal resorption and periosteal bone apposition found in healthy populations [11,23,34] are accelerated in patients with RA. Possible explanations for this process may be infl ammation-driven endosteal resorption, which was also found in human juvenile idiopathic arthritis [35], experimentally induced infl ammatory knee arthritis or adjuvant-induced arthritis [36][37][38]. Similarly, inhibition of bone resorption by the anti-receptor activator of NF-κB ligand antibody denosumab prevented metacarpal shaft bone loss in patients with RA [39]. Periosteal bone formation has been found in human studies [35] and in animal studies [36][37][38], with one of these studies observing that periosteal bone formation occurred after the endosteal resorption, when infl ammation had already subsided [38]. Th is observation suggests that periosteal bone formation may be a compensatory mechanism to restore bone strength. In other words, the calculated load-bearing capacity of RA patients was not impaired and adaptation to muscular strength remained unchanged [4].
In summary, RA is associated with an increased endosteal bone resorption and compensatory periosteal bone formation of the aff ected shafts leading to a change in geometry with maintained strength. Longitudinal data for the eff ects of infl ammation on bone geometry and volumetric BMD in patients with RA have not yet been published.

Glucocorticoids
GC use, quantifi ed either by daily or cumulative dosage, is strongly related to an increased fracture risk [40,41]. Prednisone equivalent >5 mg daily is associated with vertebral fractures in postmenopausal women within the fi rst 6 months of treatment [42]. Th e relative risk for fractures in GC users in a general practice research database is 2.86 (95% confi dence interval = 2.31 to 3.16) for the spine and 2.01 (95% confi dence interval = 1.47 to 2.29) for the hip compared with subjects not using GCs [42]. Th e areal BMD at the lumbar spine and hip showed an up to 12% increased bone loss in GC users compared with in nonusers of similar age and sex [42]. A metaanalysis of the relationship between areal BMD changes and fractures showed that low areal BMD was associated with a relative risk of 1.48 for vertebral fractures and 1.41 for hip fractures [43]. Further, it has been shown that fractures occurred at a younger age in male GC users compared with nonusers at comparable areal BMD [44]. Th is fi nding indicates that GCs not only lead to decreased areal BMD, but fractures occur at an even higher rate than that expected based on areal BMD. One can therefore speculate that there is not only a loss in bone mass but also in bone quality in GC users. Th e eff ects of GC on bone cells include early and quick induction of osteoclastogenesis leading to in creased bone resorption as well as apoptosis of osteo blasts and osteocyts [45].
Histomorphometric analyses showed trabecular thinning and reduced osteoid thickness, a marker for osteoblastic activity in GC users [46,47]. At the iliac crest, cortical porosity and Haversian canal density were higher in patients treated with GCs [48]. Using QCT of the spine, rapid decline of vertebral trabecular volumetric BMD was found upon GC treatment with subsequent recovery following discontinuation [49]. Th inning of the cortical shell at the distal radius was found by peripheral QCT [50,51] and at the proximal femur by QCT [52]. Consistent with these results, GC administration to ovariectomized sheep showed that the cortical width and cortical bone area were reduced by 7 to 8% and the marrow area increased by 8% in GC-treated animals compared with controls ( Figure 2a) [53]. Trabecular bone formation in the GC-treated animals decreased by 68% after 2 months and by 90% after 4 months. Th e apparent trabecular BMD and compressive stiff ness were reduced by 34% and 55%, respectively. Th e results are consistent with a substantially reduced bone formation in the cortical and trabecular bone and reduced cortical width due to increased endosteal resorption. Impaired cortical bone remodeling together with increased cortical porosity bears the potential for increased fracture risk at appendicular bone sites such as the femur and tibia [48,53].
In summary, BMD and bone quality are reduced upon GC treatment due to increased endosteal resorption, cortical porosity and impaired periosteal apposition. Longitudinal data on bone shaft geometry changes in GC-treated patients, however, are as yet absent.

Bisphosphonates
BP therapy is the strategy of choice for prevention and treatment of bone loss and fractures associated with osteoporosis. Current clinical practice treats patients on high-dose long-term GC therapy with BP to prevent bone loss. BP therapy has been shown to reduce the relative risk for new vertebral and hip fractures in postmenopausal women by 70% and 50%, respectively, over 3 years [54]. Indications for treatment with BPs are a quantitative estimate of the 10-year probability of a major osteoporotic fracture (clinical spine, hip, forearm or shoulder) exceeding 15 to 30% [55][56][57], inadequate vertebral fracture or GC therapy depending on the risk category [58].
BPs accumulate in the bone's hydroxyapatite mineral phase, particularly at sites of active resorption [59,60] where the nitrogen-containing BPs enter osteoclasts and reduce resorption by initiating early osteoclast death [61]. Recent retro spective studies and case reports suggest that long-term BP therapy may result in the suppression of bone turn over and confer a predisposition to increased bone fragility, with an increased risk of atypical fractures [62]. Th e number of atypical fractures is low -corresponding to an incidence of subtrochanteric fractures of 3 per 10,000 person-years -compared with the overall incidence of hip fractures (103 per 10,000 person-years) [63]. Nevertheless, long-term BP use is more likely and the duration of BP use longer in patients with sub trochanteric and femoral shaft fractures compared with patients with typical osteoporotic hip fracture [63,64]. Hip fracture incidence declined by 12.8% from 1996 to 2007, since BPs were approved, whereas the number of subtrochanteric and femoral shaft fractures remained stable or even increased in the same period [65]. Th is coincidence made the US Food and Drug Administration issue an alert with a warning about the 'possible risk of rare atypical femur fractures associated with BP treatment in osteoporosis patients' [66].
A causal relationship between BP use and atypical femur fractures has not been established, but preclinical data evaluating the eff ects of BPs lend biologic plausibility to a potential association with long-term BP use. Of special interest was the fi nding that patients with femoral shaft fractures had a higher number of comorbid conditions -for example, diabetes mellitus, RA (odds ratio = 16.5) and concomitant use of additional antiresorptive agent (such as estrogen, raloxifene or calcitonin) or systemic GCs (odds ratio = 5.2) -than those with typical hip fractures [8]. Th is association is striking since guidelines suggest treating patients with RA undergoing GC treat ment with BPs, either preventa tive to alleviate GC-induced osteoporosis or as therapy of already established osteoporosis [58]. In a recent longitudinal study using high-resolution com puted tomography to measure the eff ect of alendronate treatment in postmenopausal women with low BMD, changes in trabecular and endosteal compartments at the epiphyses of tibia and radius were found [67]. Particularly at the weight-bearing tibia, a fi lling of endosteal bone cavities was found with a concomitant decrease in the medullary area [67]. At the tibia shaft, however, increased medullary area with unaltered periosteal circumference was seen in postmenopausal RA patients treated with alendronate [68] (Figure 2c).
In summary, the positive eff ects of BP therapy in reducing fracture risk are undoubted; in patients with comorbid conditions such as RA and GC use, however, BPs seem to be associated with atypical femoral fractures, for which the reason is not yet known. Th e recent reports of the potential association of BPs with atypical femur shaft fractures and the recommendations to treat RA patients on GCs with BP therapy are in stark contrast to the prevailing advice and therefore impede the clinician's decision-making in treating a large proportion of RA patients. Further research is needed to address these questions.

Conclusion
Current knowledge about the adaptation of bone shaft geometry to eff ects of estrogen defi ciency, RA and systemic GC therapy suggests an enhanced endosteal bone resorption. While in estrogen defi ciency and under GC therapy this endosteal resorption is insuffi ciently compensated by periosteal apposition (Figure 2a), in the RA it seems to be compensated by increased periosteal bone apposition (Figure 2b) resulting in a maintained load-bearing capacity and stiff ness. BP therapy results in fi lling of endosteal bone cavities and hence reduction of the endosteal diameter at the epiphysis. However, periosteal apposition upon BP treatment in patients with osteoporosis has not yet been seen (Figure 2c). More so, studies elucidating the eff ect of BP treatment on the bone geometry of patients with RA are missing. Further, the eff ects of concomitant use of GCs and BPs on the bone shaft's inner and outer bone envelope are unexplored. Th ere is a need for answering these questions, and studies that further clarify the eff ect of GCs and BPs (and their combination) on bone geometry and fragility will shed some light on these points.

Rheumatology key messages
• Estrogen loss, RA and GC show increased endosteal resorption of bone shaft. • Periosteal bone remodeling of the bone shaft may be increased under infl ammatory conditions.
• Studies are needed to elucidate the eff ect of BP on periosteal apposition under infl ammatory conditions.