Overview, Vol 11, Issue 2

Epidemiology – oops, looking in the wrong place

The meeting of minds of the WHO committee many moons ago to establish a definition of osteoporosis as a T-score of <-2.5 SD gave voice to the silent epidemic of bone fragility which predisposes to fractures. That was a good thing but it is not without its limitations. One of the most important problems is the idea that fractures occur mainly in persons with osteoporosis. Over 50% of fractures in the community arise among persons without osteoporosis. Screening programs using a threshold of -2.5 SD will reassure and discharge these individuals. This is the screening paradox, described many years ago by Geoffrey Rose, whose books and papers on the topic are essential reading. Measurement of bone density is not sensitive; nor is it specific, many people with “osteoporosis” do not fracture and do not necessarily have fragile bones.

This point is reinforced by Oyen et al who reported that 40% of 263 women with distal radial fractures did not have osteoporosis. Osteoporos Int 2010;21:1247-55  Oyen et al also reported that of 218 men and 1576 women with distal radial fractures, T-scores <-2.0 and <-2.5 SD at femoral neck was found in 37.7% and 19.6% of men and 51.1% and 31.2% of women, respectively. The risk of hip fracture was for patients with T-score <-2.0 SD were 11.6% and 14.5% and for T-score <-2.5 SD 16.3% and 18.2%, respectively. Osteoporos Int 2010;21:1257-67

The point is made again in a study Pulkkinen et al who reported that in postmenopausal females with cervical (n=39) or trochanteric (n=18) hip fracture and 40 age-matched controls, a BMD T-score of -2.5 discriminated trochanteric fractures from controls (p<0.001); here 17 of 18 occurred in individuals with T-score ≤-2.5. However, only 20 cervical fractures (51.3%) occurred in individuals with BMD in osteoporotic range. Within the nonosteoporotic cervical fracture patients (N=19) and nonosteoporotic controls (N=35), 83.3% were classified correctly based on a model including NSA and FNC, area under the receiver operating characteristics curve being 0.85 for the model, while it was only 0.56 for BMD alone. Osteoporos Int 2010;21:1269-76

Morbidity, mortality and cost – note the nonvertebral fracture burden

There is a perception that vertebral fractures are the flagship of osteoporosis and bone fragility. This is not the case, 80% of all fractures in the community are nonvertebral and these fractures are costly, and more so as a group than hip fractures. Pike et al studied patients with a nonvertebral fracture (NVF). Although hip fractures had highest excess costs, nonvertebral, nonhip fracture patients made up a larger proportion of the sample and were more costly than controls. Comorbidity rates and resource use were higher among NVF patients during second year following an NVF compared with controls. Mean direct excess costs for NVF patients fell from $5267 in the first year to $2072 in the second year after a fracture. Patients with fractures of the pelvis, hip, and femur had the highest excess costs in the second year ($5121, $3930, and $3828, respectively). Osteoporos Int 2011;22:47-56

Song et al studied patients with incident fracture and >12-month pre-period and follow-up period from the incident fracture. Nationally projected annual cost of second fracture was $834 million for patients with commercial insurance and $1.13 billion for Medicare patients. For privately insured patients with hip, vertebral, or NHNV fracture, the 1-year second fracture rate was 8.0%, 5.1%, and 4.0%, and 1-year incremental costs were $47,351, $43,238, and $23,852, respectively; for Medicare patients, the corresponding rates and costs were 8.8%, 9.2%, and 8.2%, and $18,645, $19,702, and $19,697. Bone 2011;48:828-36

Huntjens et al report of 1921 consecutive patients 50+ years having a NVF, nearly one in five patients sustained a subsequent NVF and one in three died within 5 years after an initial NVF. One third of subsequent NVFs and mortality occurred within one year. The risk for a subsequent NVF was 17.6% and was related to age (HR per decade, 1.44). The risk for mortality was 32.3% and was related to age (HR per decade, 2.59), male sex (HR, 1.74), fracture at baseline (HR, 5.56) and subsequent fracture (HR, 1.65). The highest risks were found within the first year (NVFs, 6.4%; mortality, 12.2%) and were related to age and baseline fracture location for mortality. Osteoporos Int 2010;21:2075-82

Heterogeneity in morphology in patients with fractures

The prevailing view is that the large number of persons with BMD in the osteopenia range reflects the fact that they come from the larger pool of individuals in the bell of the bell shaped distribution, rather than from the smaller tail of the distribution. The question is whether these patients have structural abnormalities responsible for bone fragility. Some evidence is available that indeed, these patients have fragile bones but this is not detected by the bone densitometer. These patients have reduced trabecular BV/TV, reduced trabecular number and connectedness. They also have higher intracortical porosity (Seeman, unpublished data).

Persons with fractures are not all the same, the pathogenesis of bone loss and bone fragility differs markedly between individuals with “one or more minimal trauma fractures”. Until the structural basis is defined we will be left with the poor sensitivity and specificity of bone densitometry which fails to identify over 50% of all fractures in the community.

An example of this heterogeneity is reported by Parfitt et al, who studied the behaviour of the cellular activity upon the endocortical and trabecular envelopes that determines bone structure in three dimensional space and so its strength. The investigators reported that a combination of defective osteoblast recruitment and premature osteoblast apoptosis account for differing deficits in bone surfaces, the reductions in mean wall thickness and bone formation rate on the cancellous surface, and osteocyte deficiency in cancellous bone. Patients with vertebral fractures had reduced surface extent of osteoblasts, markedly on the cancellous surface, where 75% of the deficit was in intermediate (type III) cells, suggesting earlier transition to type IV (flat) cells and about 25% of the deficit was in cuboidal (type II) osteoblasts, suggesting impaired recruitment. On the endocortical and intracortical surfaces, the deficit was exclusively in type III cells. Mean bone formation rate was reduced by about 18% on the cancellous surface but not on the other two components of the internal surface and was due to a reduction in wall thickness. Bone formation rate was skewed left in the fracture patients. Osteoclast surface was lower on each subdivision. The data emphasise the histologic heterogeneity of postmenopausal osteoporosis. J Bone Miner Res 2011;26:475-85

The purpose of bone remodelling is to remove old or damaged bone. However, what precisely is ‘damage’; a crack, dead osteocytes, highly crosslinked bone, abnormalities in mineral content? If damage differs in nature, are the signals to repair it also different? Herman et al reported that fatigue microdamage occurs as either linear microcracks or diffuse sublamellar damage. Activation of resorption correlated to the number of linear microcracks, no activation of resorption occurred in response to diffuse microdamage and there was no change in osteocyte viability in response to diffuse microdamage. Bone 2010;47:766-72

Is cortical but not trabecular bone estrogen sensitive?

Some studies suggest that there is also heterogeneity on bone tissue regarding the response to estrogen and estrogen deficiency. The basis of this notion is that there appears to be little cortical bone loss before menopause; while trabecular bone loss appears to be independent of menopause, decreasing linearly with age and without an apparent acceleration at the time of menopause, at least in cross-sectional studies.

Syed et al studied ovariectomized, or ovariectomized and estrogen-replaced female C57/BL6 mice at 6 months of age. Compared with young mice, aged/sham-operated mice had a 42% reduction in spine BV/TV. Maintaining constant estrogen levels over life in ovariectomized/estrogen-treated mice did not prevent age-related trabecular bone loss. By contrast, lifelong estrogen in ovariectomized mice prevented the age-related reduction in tibial cortical vBMD and thickness. Moreover, as compared with cells from young mice, hematopoietic lineage negative cells from aged/sham-operated mice expressed higher mRNA for osteoblast differentiation and proliferation marker genes. J Bone Miner Res 2010;25:2438-46

Bone is not the centre of the universe

Gerard Karsenty has shifted the position of bone from the centre of the universe to its perhaps rightful position as one of the satellites rotating around the real centre of the universe, energy metabolism. In one of several studies confirming his ideas, Kanazawa et al reported that serum osteocalcin was negatively associated with fasting plasma glucose, HbA(1c), %trunk fat, and homeostasis model assessment (HOMA) for insulin resistance, and positively with HOMA for β-cell function in 101 postmenopausal women and 152 men with type 2 diabetes. In addition, positive association of serum osteocalcin with serum adiponectin was found in postmenopausal women. In the OGTT examinations, subjects were divided into tertiles by their serum osteocalcin levels. Postmenopausal women in the lowest tertile showed hyperglycemia and hyperinsulinemia compared to those in the highest tertile after oral glucose loading. Men in the lowest tertile also exhibited hyperinsulinemia, while hyperglycemia was not found. Serum osteocalcin is positively associated with insulin sensitivity and secretion in Japanese patients with type 2 diabetes. Bone 2011;48:720-5

Schulte et al presented fascinating use of noninvasive direct techniques to derive dynamic bone morphometry. Dynamic measures of bone formation correlated with MAR (R=0.68) and MS (R=0.78). After 4 weeks of loading, newly extracted trabecular BFR and MS were higher in the loading compared to the control group, ES was decreased. This indicates that modelling induced by mechanical loading takes place primarily by increased area, not width of formation packets. Bone 2011;48:433-42


Isaksson et al investigated the humeri from female New Zealand white rabbits of varying age. In mid-diaphyseal cortical bone samples the elastic stiffness increased with age, from 11 days to 6 months, based on the reduced modulus from the indentation probing, the storage modulus from the semidynamic test, and the first elastic parameter from the creep test. These elastic parameters correlated (R=0.88-0.94, p<0.01). Viscoelastic parameters, the phase-shift and time creep constant, decreased with age. For the viscous properties determined by the creep and the semidynamic testing correlated (R=0.90, p<0.01), however, no correlation was found between the phase-shift and the creep time constant. The elastic modulus of bone tissue increased by approximately 60%, whereas the viscoelastic parameters decreased by 10% to 25% during the first 6 months of the rabbit's life. Bone 2010;47:1030-8

Serum sclerostin

Modder et al reported that estrogen (E) treatment of postmenopausal women (n=17) for 4 weeks led to a 27% decrease in serum sclerostin. Similarly, in 59 elderly men, E, not testosterone (T), prevented increases in sclerostin following induction of sex steroid deficiency. In both sexes, changes in sclerostin correlated with changes in resorption, not formation, markers (r=0.62, p<0.001, and r=0.33, p=0.009, for correlations with changes in serum C-terminal telopeptide of type I collagen in the women and men, respectively). Sclerostin levels are reduced by E not T. Changes in sclerostin production may contribute to effects of E on bone resorption. J Bone Miner Res 2011;26:27-34


Macdonald et al reported that in 442 women and 202 men aged 20-99 years, total bone area, which was 33% larger in young men, changed similarly with age in women and men at the DT but increased 17% more in men than in women at the DR. Trabecular number and Tb.Th were higher in young men than in young women at both sites, and with the exception of Tb.Th at the DR, which declined more with age in men (-16%) than in women (-2%, p<0.01), the age-related decline in these outcomes was similar in women and in men. In the cortex, porosity (Ct.Po) was 31% to 44% lower in young women than in young men but increased 92% to 176% more with age in women than in men. The DR cortex carried 14% more load in young women than in young men, and the percentage of load carried by the DR cortex did not change with age in women but declined by 17% in men. FEA-estimated bone strength was 34% to 47% greater in young men, but the predicted change with age was similar by sex. In contrast, the load-to-strength ratio increased 27% more in women than in men with age. J Bone Miner Res 2011;26:50-62

Glucocorticoid-induced autophagy in osteocytes

Xia et al reported that glucocorticoid (GC) induced autophagy preserves osteocyte viability. GC resulted in an increase in autophagy markers and the accumulation of autophagosome vacuoles in vitro and in vivo promoted the onset of the osteocyte autophagy, as determined by expression of autophagy markers in an animal model of GC-induced osteoporosis. An autophagy inhibitor reversed the protective effects of GCs. In contrast, TNFα induced apoptosis but not autophagy. J Bone Miner Res 2010;25:2479-88

Drug therapy - is one plus one more than two or less than two?

Prior or concurrent use of remodelling suppressants may blunt the rise in remodelling markers and BMD produced by PTH1-34. The reasons for this are not understood. The anabolic effects of PTH1-34 are both modelling and remodelling based, but 70% of the effect is reported to be remodelling based and this might explain why there is blunting. When suppression of remodelling occurs with a bisphosphonate, the remodelling based component of the anabolic effect may be lost.

Not all pieces of the puzzle fit. Cosman et al reported the opposite. PTH 1-34 plus zoledronic acid initially produced a greater increase in BMD than PTH1-34. The observations raise many questions. One explanation of may be that the modelling based anabolic effect will occur and produce a net greater benefit than either agent alone. This doesn’t quite fit either. If correct, why isn’t the same observed with alendronate which should also allow the modelling dependent anabolic effect to proceed? One explanation to reconcile the differing observations is that zoledronic acid is a more potent remodelling suppressant than alendronate – initially. When first administered, the greater remodelling suppression allows porosity produced by remodelling at the time of treatment to go to completion with appearance of fewer new remodelling sites with zoledronic acid than with alendronate. Under these circumstances the rise in BMD may be greater with zoledronic acid plus PTH1-34 than PTH alone because the filling phase occurs without appearance of new remodelling sites. Alendronate suppresses remodelling less than zoledronic acid so porosity decreases less because any decrease in porosity produced by allowing filling of sites excavated at the time of treatment is now offset by more new pores as alendronate suppresses remodelling by only 50%. What is needed to sort this out is a study randomizing patients to alendronate plus PTH vs. zoledronic plus PTH, with a PTH alone arm and measurements of bone microarchitecture and tissue density to examine the effects of remodelling on any continued remodelling despite treatment on both porosity and tissue mineralization density. Densitometry is not going to help here. J Bone Miner Res 2011;26:503-11

The real question this study raises is why do these studies at all. The justification for studies comparing the BMD response of single vs. combined therapy is that if a greater BMD change is observed with one regimen over another, this will mean that the fracture risk reduction is greater. There is no evidence for this. A change in BMD provides little, if any, insight into the effects of combined therapy on the material composition and structure of bone over single therapy. Nor does it provide insight regarding potentially superior antifracture efficacy over single therapy.

Suppression or blunting of response to teriparatide (TPTD) is not always seen. Stepan et al reported the response to TPTD was not blunted by prior alendronate therapy. Paired biopsies at baseline and after 24 months of TPTD from 29 ALN-pretreated (~5 yrs) and 16 untreated patients revealed the increase in activation frequency was 0.11-0.34 cycles per year in ALN and 0.19-0.33 with TPTD, while osteoid surfaces were 3.96-9.8% in the ALN-pretreated group and 6.2-11.3% in the untreated group. The increased level of formation is similar in patients who were either untreated or had lower bone turnover due to previous ALN. Elevated bone formation in postmenopausal women with osteoporosis was sustained over a 24-month period by TPTD. Osteoporos Int 2010;21:2027-36

Misof et al paired transiliac bone biopsy before and after one year teriparatide in 16 women treated with alendronate (priorALN) or risedronate (priorRIS) for at least two years and then treated for 12 months with teriparatide. BMDD response to one year of teriparatide did not depend on the type of bisphosphonate. For priorALN patients increases in low mineralized bone areas (Cn.Ca(Low) +25.9%, Ct.Ca(Low) +62.0%, and Ct. heterogeneity of mineralization (Ct.Ca(Width) +22.8%). PriorRIS patients showed increased mineralization heterogeneity (Cn.Ca(Width) +14.8% and Ct.Ca(Width) +15.8%). J Bone Miner Res 2010;25:2297-303

Understanding the structural basis of a rise or lack of rise in BMD following antiresorptive therapy

Remodelling suppression with antiresorptive agents is thought of in terms of potency in suppressing remodelling, but the effects also depend on the remodelling that continues despite therapy. Calcium supplements suppress remodelling by about 10-20% so about 80% of the remodelling present before treatment continues. At the other extreme, denosumab reduces remodelling by about 80%, at least as measured using remodelling markers. When remodelling is suppressed, remodelling sites initiated before treatment complete remodelling with bone formation which reduces the existing intracortical porosity and focally fills excavated sites upon trabeculae and the endocortical surface (a change below the detection of most noninvasive methods). In addition, osteons that would be removed are not and continue to undergo secondary mineralization. So cortical vBMD increases for two reasons: the reduction in porosity and the increase in tissue mineral density of the osteons. The net effect also depends on the residual remodelling.

In the case of calcium supplementation 80% of the remodelling continues, so as the pores fill at the time of initiation of calcium, new porosity appears because the drug is a weak remodelling suppressant. Bone loss continues after a modest increase in BMD produced by the filling phase of resorption cavities present before treatment. This BMD increase is modest because of the appearance of almost as many new resorption sites as were present before treatment. With denosumab, only 20% of the remodelling sites present before treatment reappear during treatment so the filling phase of resorption sites present at the time of treatment occurs without being offset by new sites. The net effect of these changes on bone structure and tissue mineralization density varies depending on the drug used and on the baseline remodelling intensity in an individual.

Seeman et al reported that in 247 postmenopausal women randomized to denosumab, alendronate or placebo for 12 months, denosumab decreased serum C-telopeptide more rapidly and markedly than alendronate. Alendronate prevented the decline in total vBMD observed with placebo whereas denosumab prevented the decline or improved total and cortical vBMD. The polar moment of inertia also increased more in the denosumab than alendronate or placebo groups (p<0.001). J Bone Miner Res 2010;25:1886-94  The reason why denosumab improved cortical vBMD while alendronate only prevented the decrease can be understood from the above. With alendronate, remodelling suppression is about 50%. so the increase in tissue density produced by the remodelling suppression would increase cortical vBMD, but the continued remodelling reduces cortical density by removing osteons and leaving porosity.

Similar observations with alendronate were reported by Burghardt et al. During 24 months, alendronate prevented the decrease in distal tibia cortical vBMD seen in placebo but did not change cortical vBMD relative to baseline because the net effect of filling of resorption sites present before treatment plus secondary mineralization of osteons not removed (both of which increase cortical vBMD) what offset by the continued appearance of about 50% of remodelling sites which produce porosity. J Bone Miner Res 2010;25:2282-95

Zoledronate – is annual administration needed?

Grey et al report that the antiresorptive effects of a single 5 mg dose of zoledronate are sustained for 3 years in 50 postmenopausal women with osteopenia. Mean serum β-CTX and P1NP were 44% and 40% lower in the zoledronate group vs. placebo. BMD was higher in the zoledronate group than in the placebo group by 6.8% at the spine, 4.0% at the total hip, and 2.0% at the total body. Between-group differences in markers of bone turnover and BMD were stable from 12 to 36 months. J Bone Miner Res 2010;25:2251-5

Gamsjaeger et al analysed the biopsies (152 patients, 82 ZOL and 70 placebo) by Raman microspectroscopy. ZOL increased the mineral/matrix ratio. Mineral crystallites had carbonate content and maturity/crystallinity characteristics of younger bone, suggesting that ZOL may be exerting an effect on bone matrix formation in addition to its antiresorptive effect. J Bone Miner Res 2011;26:12-8


O'Neal et al reported that changes in architecture and matrix properties associated with alendronate reduce trabecular bone's ability to resist the formation of loading-induced severe and linear microcracks, both of which dissipate less energy prior to fracture than does diffuse damage. Trabecular cores from the distal femur of beagle dogs treated for one year with alendronate subjected to uniaxial compression had 25% lower von Mises stress for trabeculae exhibiting severe and linear microcrack patterns, whereas there was no reduction for diffuse microdamage formation. Severely damaged trabeculae were thinner, more aligned with the loading axis, and less mineralized than undamaged trabeculae. Similar relationships with damage morphology were found only with trabecular orientation in vehicle-treated control dogs. Bone 2010;47:241-7

Kusamori et al developed a transdermal patch of alendronate. The maximum permeation fluxes through rat and human skin were 1.9 and 0.3 µg/cm2 per hour, respectively. The bioavailability (BA) in rats was 8.3% (patch) and 1.7% (oral) administration. The plasma calcium was reduced in 1α-hydroxyvitamin D3-induced hypercalcemia in rats. The alendronate patch also suppressed the decrease in bone mass in rats with osteoporosis. Modest alendronate-induced erythema of rat skin was observed. Incorporation of butylhydroxytoluene in the alendronate patch suppressed this alendronate-induced skin damage maintaining permeation and pharmacologic effects. J Bone Miner Res 2010;25:2582-91

Adverse events of bisphosphonates

Shane et al provide recommendations of a task force that the definition of atypical fracture requires the location in the subtrochanteric region and femoral shaft, transverse or short oblique orientation, minimal or no trauma, a medial spike when the fracture is complete, and absence of comminution. Minor features include their association with cortical thickening, a periosteal reaction of the lateral cortex, prodromal pain, bilaterality, delayed healing, comorbid conditions, and concomitant drug exposures, including bisphosphonates, other antiresorptive agents, glucocorticoids, and proton pump inhibitors. J Bone Miner Res 2010;25:2267-94

Vestergaard et al studied users of bisphosphonates and other drugs (n=103,562) and three age- and gender-matched controls from the general population (n=310,683). After initiation of therapy, an increased risk of subtrochanteric fractures was seen for alendronate (HR=2.41, 95% CI 1.78-3.27), etidronate (HR=1.96, 95% CI 1.62-2.36), and clodronate (HR=20.0, 95% CI 1.94-205), not raloxifene (HR=1.06, 95% CI 0.34-3.32). An increased risk of subtrochanteric fractures was also present before alendronate (OR=2.36, 95% CI 2.05-2.72), etidronate (OR=3.05, 95% CI 2.59-3.58), clodronate (OR=10.8, 95% CI 1.14-103), raloxifene (OR=1.90, 95% CI 1.07-3.40), and strontium ranelate (OR=2.97, 95% CI 1.07-8.27). The authors suggest that not all subtrochanteric fractures are the result of treatment. Osteoporos Int 2011;22:993-1001

While prolonged therapy may increase the risk for atypical fractures, this risk is small and is probably much smaller than the risk of failure to treat or poor compliance with treatment. Imaz et al reported in a meta-analysis of five articles totalling 236,540 patients, followed for one year, had a pooled persistence of 184.09 days. The meta-analysis of five articles, totalling 234,737 patients, also followed for one year, provided a pooled medication possession ratio mean of 66.93%. The meta-analysis of six articles, totalling 171,063 patients, followed 1 and 2.5 years, provided a pooled 46% increased fracture risk in noncompliant patients vs. compliant patients. The increased fracture risk was 16% for nonvertebral, 28% for hip and 43% for clinical vertebral fractures. Osteoporos Int 2010;21:1943-51

Bisphosphonate therapy is associated with adverse events including ONJ. While no known treatment is available, in an experimental model in mice Kikuiri et al reported that zoledronate plus dexamethasone causes BRONJ-like in part by suppressing the adaptive regulatory T-cells and activating the inflammatory T-helper-producing interleukin 17 cells. Systemic infusion with mesenchymal stem cells prevented and cured BRONJ-like disease possibly via induction of peripheral tolerance. The suppressed Tregs/Th17 ratio in zoledronate- and dexamethasone-treated mice is restored in mice undergoing salvage therapy with Tregs. J Bone Miner Res 2010;25:1668-79

Allen et al studied female untreated beagle dogs (CON) or treated with intravenous zoledronic acid (ZOL). Following the extraction of the 4th premolars, initial infilling of the socket was not affected by ZOL, but subsequent removal of this bone was suppressed. After 8 weeks of healing, the alveolar cortical bone adjacent to the extraction socket had a remodelling rate 50% per year in CON animals while ZOL-treated animals had a rate of <1% per year. One ZOL-treated animal developed exposed bone post-extraction which eventually led to the formation of a sequestrum. Assessment of the sequestrum with µCT and histology showed that it had features consistent with those reported in humans with ONJ. Osteoporos Int 2011;22:693-702

DVT and drug therapy

Lamberg et al studied 4193 cased and 41,197 controls. 149 cases (3.6%) and 1078 controls (2.6%) were current bisphosphonate users. The adjusted OR for VTE among the current bisphosphonate users compared with nonusers was 1.03 (0.84-1.26), and when restricted to cases of unprovoked thromboembolism, the adjusted OR was 1.08 (95% CI 0.82-1.42). Osteoporos Int 2010;21:1911-7

Breart et al reported a greater prevalence of VTE in osteoporotic compared to nonosteoporotic patients, but not between persons with osteoporosis receiving treatment than those not receiving treatment, showing any greater association in treated patients with strontium ranelate or alendronate compared to untreated osteoporotic patients. In this study the authors reported there was an increased risk for VTE in untreated osteoporotic women vs. nonosteoporotic women (annual incidence 5.6 and 3.2 per 1000 patient-years, respectively; relative risk 1.75). The annual incidences of VTE in osteoporotic patients treated with strontium ranelate and alendronate were 7.0 and 7.2 per 1000 patient-years, respectively, with no significant difference between untreated and treated patients whatever the treatment. The sample size was large; 1546 untreated women, 20,084 treated with alendronate, 2408 patients treated with strontium ranelate and 115,009 nonosteoporotic women. Osteoporos Int 2010;21:1181-7 Well, large sample sizes in cohort studies do not solve the problem of potential bias in sampling. The point of a randomized trial is to control for covariates that might influence outcomes. In cohort studies such as this, it is not possible, even with the large sample sizes to exclude the possibility of confounding. The risk of VTE was reported in randomized trials of strontium ranelate and other drugs and this should be borne in mind.

Does taking a bisphosphonate prolong survival?

There are several studies that suggest this is so. Of course the study by Kyle et al of zoledronic acid following hip fracture demonstrated a 28% reduction in mortality unrelated to the prevention of fractures. Beaupre et al studied 220 hip fracture patients randomized to case manager or usual care. During 3 years, 101 (46%) patients started oral bisphosphonates and 65 (64%) remained on treatment. Overall, 24 (11%) patients died, 19 (9%) had new fractures, and 42 (20%) reached the composite outcome of death or fracture. Compared to no treatment, bisphosphonate exposure was associated with reduced mortality (17[16%] vs. 7[7%]; aHR=0.92 per month treated; 95%CI 0.88-0.97) and composite endpoints (28[26%] vs. 5[15%]; aHR=0.94 per month treated; 95%CI 0.91-0.97). Osteoporos Int 2011;22:983-91  The reasons for this association are not known. Whether this is a healthy user effect is one explanation.


Reid et al reported the results of iliac crest bone biopsies at 24 and/or 36 months from 45 women receiving placebo and 47 denosumab in the FREEDOM trial and at 12 months from women treated with alendronate in the STAND study (21 continuing alendronate and 15 changed to denosumab at trial entry). In the FREEDOM study, median eroded surface was reduced by >80% and osteoclasts were absent from >50% of biopsies. Double labelling in trabecular bone was observed in 94% of placebo and in 19% of denosumab treated subjects. Median bone formation rate was reduced by 97%. In the STAND trial, indices of bone turnover tended to be lower in the denosumab than alendronate group. Double labelling in trabecular bone was seen in 20% of the denosumab biopsies and in 90% of the alendronate samples. These findings reflect suppression of tissue level remodelling. What is needed is biopsies after stopping therapy or at the end of a treatment cycle. J Bone Miner Res 2010;25:2256-65

Genant et al reported that in 332 women with spine areal BMD T-scores between -1.0 and -2.5, denosumab 60 mg or placebo every 6 months during the 24-month study, increases in BMD were seen at the ultradistal region (4.7% [95% CI 3.6-5.7]; P<0.001) and total BMC (5.7% [95% CI 4.8-6.6]; P<0.001) over placebo. Cortical bone had increases in BMD, BMC, and thickness, and trabecular bone had an increase in BMD relative to placebo. Bone strength, estimated by density-weighted PMI, increased compared with placebo after 6 months, with the largest increase occurring at 24 months in the ultradistal region (6.6% [95% CI 5.6-7.6]; P<0.0001). Bone 2010;47:131-9


When teriparatide is administered, bone remodelling markers increase and then decrease despite continued administration. The significance of this observation is far from clear. Yu et al administered teriparatide either as a constant or an escalating subcutaneous dose (30 µg daily for 18 months or 20 µg daily for 6 months, then 30 µg daily for 6 months, and then 40 µg daily for 6 months). All markers differed between the two treatment groups. During the final six months, bone markers declined in the constant dose group but remained stable or increased in the escalating dose group. However, mean area under the curve did not differ between treatments for any marker, and BMD increases were equivalent. The mystery remains. Bone 2011;48:713-9

Sclerostin antibody

There are many unmet needs in therapeutics. One is the need for an anabolic agent that influences the periosteal and all three components of the endosteal envelope. Li et al reported that Scl-AbII increased areal BMD, improved trabecular and cortical architecture at L5, femoral diaphysis (FD), and femoral neck (FN), associated with a higher maximal load of L5, FD, and FN in the high-dose group. Mineralizing surface, mineral apposition rate and bone formation rate at the proximal tibial metaphysis and tibial shaft were greater on trabecular, periosteal, and endocortical surfaces in both Scl-AbII dose groups. What is interesting is that the drug, which inhibits sclerostin, an inhibitor of bone formation, also reduces bone resorption by mechanisms that are not understood. J Bone Miner Res 2010;25:2647-56

Tian et al reported that in rats treated with Scl-AbIII at 5 or 25 mg/kg, twice per week by s.c. for 4 weeks, both doses increased bone formation and trabecular bone volume and thickness and decreased resorption in the trabecular bone. Scl-AbIII decreased resorption and increased formation and bone volume in a red (hematopoietic) marrow site, the 4th lumber vertebral body. Bone 2010;47:529-33

Tian et al also reported that antibody-mediated blockade of sclerostin is a promising anabolic treatment for immobilization-induced osteopenia. Rats were normal weight-bearing (normal-loaded, NL) or right hindlimb-immobilized (under-loaded, UL were treated with Scl-Ab at 5 or 25 mg/kg for 4 weeks. UL rats had reduced body and muscle weights, increased marrow fat, increased trabecular bone resorption and periosteal mineral apposition rate (MAR) and decreased trabecular MAR and bone formation rate (BFR/BS). In NL bones, Scl-Ab increased bone formation and decreased bone resorption, resulting in increased trabecular and cortical bone mass. In UL trabecular bone, Scl-Ab at 5 or 25 mg/kg induced dose-dependent increases in trabecular bone volume and thickness, mineralized surfaces (MS/BS), MAR and BFR/BS, and a decrease in eroded surface (Er.S/BS) compared with UL controls. In UL cortical bone, Scl-Ab induced increases in cortical width, periosteal and endocortical MS/BS, MAR and BFR/BS, and decreases in endocortical Er.S/BS compared with UL controls. Bone 2011;48:197-201

Padhi et al repoted a randomized, double-blind, placebo-controlled, ascending, single-dose study, 72 healthy subjects received AMG 785 or placebo (3:1) subcutaneously (0.1, 0.3, 1, 3, 5, or 10 mg/kg) or intravenously (1 or 5 mg/kg). One treatment-related serious adverse event of nonspecific hepatitis was reported and resolved. AMG 785 pharmacokinetics were nonlinear with dose. Dose-related increases in formation markers and decrease in sCTx occurred. In addition, increases in BMD of up to 5.3% at the spine and 2.8% at the total hip compared with placebo were observed on day 85. Six subjects in the higher-dose groups developed anti-AMG 785 antibodies, two of which were neutralizing, with no effect on the pharmacokinetics or pharmacodynamics. This drug is a promising agent in osteoporosis treatment. J Bone Miner Res 2011;26:19-26

Cat on a hot tin roof

Cathepsin K inhibitors prevent resorption of bone by blocking the degradation of collagen. Unlike denosumab, the osteoclast numbers are not reduced, so that if there are products of the osteoclast that are important in promoting bone formation, the great hope is coupling with reduction in resorption but continued bone formation at the cellular level. Eisman et al studied postmenopausal women with BMD T-scores between -2.0 and -3.5 who received placebo or ODN 3, 10, 25, or 50 mg weekly. After 2 years, patients (n=189) were rerandomized to ODN 50 mg weekly or placebo for an additional year. Continued treatment with 50 mg of ODN for 3 years produced increases from baseline and from year 2 in BMD at the spine (7.9% and 2.3%) and total hip (5.8% and 2.4%). Urine NTx remained suppressed at year 3 (-50.5%), but bone-specific alkaline phosphatase (BSAP) was unchanged from baseline. Treatment discontinuation resulted in bone loss at all sites, but BMD remained at or above baseline. After ODN discontinuation at month 24, turnover markers increased above baseline but resolved by month 36. J Bone Miner Res 2011;26:242-51

Pennypacker et al compared two cathepsin K inhibitors (CatKIs) with ALN in OVX rabbits. OVX demonstrated loss (9.8% to 12.8%) in lumbar vertebral (LV) BMD at 13-weeks prevented by estrogen or ALN. L-006235 (L-235), at 10 mg/kg per day for 27 weeks, decreased LV BMD loss. ALN reduced spine cancellous mineralizing surface by 70%, L-235 had no effect. Endocortical bone formation rate and the number of double-labelled Haversian canals in the femoral diaphysis were not affected by L-235. ODN 4 or 9 µM/day for 27 weeks prevented LV BMD loss vs. OVX-vehicle control comparable with sham or ALN. ODN also dose-dependently increased proximal femur, femoral neck, and trochanter BMD. ODN did not reduce bone formation at any sites. Although CatKIs had similar efficacy to ALN in preventing bone loss in adult OVX rabbits, this novel class of antiresorptives differs from ALN by sparing bone formation. J Bone Miner Res 2011;26:252-62

Strontium ranelate

Strontium ranelate has established antifracture efficacy in human subjects. In this study, Geoffroy et al examined the effects of this drug on fracture incidence in mice overexpressing Runx2, a model of osteopenia associated with vertebral fractures. Treatment resulted in fewer fractures (-60%, p<0.05). In lumbar vertebrae, strontium ranelate improved resistance to compressive loading (ultimate force, +120%, p<0.05). What is a mystery with this drug is the mode of action. The authors report higher trabecular bone volume and trabecular number, lower trabecular separation and higher cortical thickness (+17%, p<0.05) and reduced marrow cavity cross section. These observations are based on histomorphometry and do support the notion of an anabolic effect. What is missing in the data is more detail about the growth rate differences in the animals and whether this was no different. If so, this would be the first histological evidence to support the claims of an anabolic effect. I reserve my interpretation for now and still favour the notion that the strength improvement is due to an effect on the material properties of bone. Osteoporos Int 2011;22:289-97

TGFβ antibody

Edwards et al report that an antibody (1D11) directed at all three isoforms of TGFβ increased BMD, trabecular thickness, and bone volume by up to 54%, accompanied by elevated osteoblast numbers and decreased osteoclasts. Biomechanical properties of bone also were enhanced with increased bending strength and tissue-level modulus with an 11% increase in the mineral-to-collagen ratio of trabecular bone. J Bone Miner Res 2010;25:2419-26

The tyrosine kinase inhibitor

Vandyke et al report that dasatinib is a tyrosine kinase inhibitor used to treat chronic myeloid leukemia. It inhibits M-CSF receptor c-fms, osteoclast formation and activity in vitro. Sprague Dawley rats administered dasatinib (5 mg/kg/day) or vehicle by gavage or ZOL (100 µg/kg/6 weeks) subcutaneously were followed 12 weeks. Whole body BMD and tibial cortical thickness where unchanged in the dasatinib- or ZOL-treated animals relative to controls. However, µCT analysis of cancellous bone at the proximal tibias showed that trabecular BV/TV and Tb.Th increased comparable with ZOL-treated group. These changes were associated with a decrease in osteoclast numbers (N.Oc/B.Pm) and surface (Oc.S/BS) and decreased serum levels of the CTX-1. MAR, BFR, osteocalcin and P1NP were not altered. J Bone Miner Res 2010;25:1759-70

What do vitamin D, snake oil, and bloodletting have in common?

Lack of evidence, that’s what. If randomized trials were properly designed and executed, we would not need meta-analyses. It is almost without exception, and I can’t think of an exception, all studies of calcium alone, vitamin D alone or combinations of both are plagued by serious issues in design and execution. The design flaw is that most, if not all the studies, were carried out in persons who were neither calcium nor vitamin D deficient. The two flaws in execution are dropouts and poor compliance, usually about 50% for each. If patients are not deficient in calcium and vitamin D. how can the effect of deficiency on fracture rates be determined and how can the effect of correction on whatever the consequence may be ascertained?

The answer submitted is subgroup analysis. This is problematic also if we adhere to the notion that the randomized trial is the best level of evidence we can make inferences from because covariates influencing outcomes are likely to be evenly distributed in both groups provided the sample size is large enough. Once a subgroup analysis is done, whether this is a subgroup of ‘low’ calcium intake group or compliers to therapy, randomization may be violated – the operative word is “may be” and that should be enough if we are rigorous. Without this, the independent effects of both known and unknown covariates influencing fracture rates cannot be excluded; that’s it! If we deviate from this requirement we may as well ask the opinion of the next door neighbour.

Healthy users of placebo have better outcomes than noncompliers to placebo, so it is possible that any benefit of calcium or vitamin D in compliers reflects this effect rather than the effect of treatment. This problem exists with dose finding as well. The claim that a dose of 800 U vitamin D, but not 400 U vitamin D, is the correct dose has never been tested in a randomized trial of patients selected on being vitamin D deficient randomized to placebo, 400 U, 800 U with fracture outcomes assessed prospectively. The argument that it is unethical to give placebo begs the question, it may be argued that it is unethical not to give placebo if we don’t know and the sample size requirements to detect a better fracture risk reduction with 800 U over 400 U is likely to be formidable.

Bischoff-Ferrari et al report the results of a meta-analysis of double-blind randomized control trials, 8 examining falls (n=2426) and 12 examining nonvertebral fractures (n=42,279). Benefits were observed at 700 to 1000 IU vitamin D per day or mean 25(OH)D between 75 and 110 nmol/l (30-44 ng/ml). Prospective cohort data on cardiovascular health and colorectal cancer prevention suggested benefits with the highest 25(OH)D evaluated (median between 75 and 110 nmol/l). Mean levels of 75 to 110 nmol/l were reached in most RCTs with 1800 to 4000 IU vitamin D per day without risk. Well, should we then proceed with these doses? Osteoporos Int 2010;21:1121-32.

Salovaara et al reported that among women aged 65-71 years, 1586 in the intervention group given 800 U vitamin D plus 1 g calcium, and 1609 in the control group. None of the effects reached statistical significance. During 3 years the risk of any fracture decreased by 17% [aHR=0.83; 0.61-1.12], nonvertebral fracture decreased by 13% (aHR=0.87; 0.63-1.19), distal forearm fractures decreased by 30% (aHR=0.70; 0.41-1.20), any upper extremity fractures decreased by 25% (aHR=0.75; 95% CI 0.49-1.16), lower extremity fractures (aHR=1.02; 0.58-1.80). J Bone Miner Res 2010;25:1487-95


Cummings et al reported that in women aged 60-85 years with osteoporosis or a vertebral fracture, and women with low bone mass were assigned to arzoxifene 20 mg or placebo daily. After 3 years, the incidence of vertebral fractures in patients with osteoporosis was 2.3% lower in the arzoxifene group than in the placebo group, a 41% relative risk reduction (0.45-0.77). In the overall population, the cumulative incidence of invasive breast cancer over 4 years was reduced by 1.3%, with a 56% relative reduction in risk (HR 0.44, 0.26-0.76); there was no significant decrease in nonvertebral fracture risk. Arzoxifene increased the cumulative incidence of venous thromboembolic events by 0.7%, with a 2.3-fold relative increase (95% CI 1.5-3.7). J Bone Miner Res 2011;26:397-404

Ivergard et al estimated costs and benefits of 5-year raloxifene based on 10-year fracture probability estimated with FRAX® below 20% and 5-year invasive breast cancer risk of 1-5%. The cost per QALY gained ranged from US $22,000 in women age 55 with 5% invasive breast cancer risk and 15-19.9% fracture probability, to $110,000 in women age 55 with 1% invasive breast cancer risk and 5-9.9% fracture probability. At lower fracture risk with lower cancer risk, the cost-effectiveness of raloxifene worsened. At fracture risk of 15-19.9% raloxifene was cost-effective also in women at lower invasive breast cancer risk. The cost-effectiveness is contingent on their 5-year invasive breast cancer risk. Bone 2010;47:966-74

Note from the Editor

The purpose of Progress in Osteoporosis is to provide the reader with a summary of the most important literature published in the preceding three to four months in the field of osteoporosis. Most reviews and original research are cited. In addition, summaries and figures are provided for readers who may not have easy access to all the specialist literature. The summaries are based on the contents of abstracts, which have been abbreviated to concisely convey the main theme. The contents of the abstracts and figures should be used only as a means of directing the reader to the original literature and should not be quoted verbatim or cited as a reference. The opinions expressed in the Overview are my own and do not necessarily reflect those of the International Osteoporosis Foundation.