Mechanisms of the Anabolic Effects of Teriparatide on Bone: Insight From the Treatment of a Patient With Pycnodysostosis

Authors


  • Drs Karsdal, Segovia-Silvestre, and Neutzsky-Wulff are employed by Nordic Bioscience A/S. All other authors state that they have no conflicts of interest

  • Published online on February 25, 2008

Abstract

Pycnodysostosis is an extremely rare genetic osteosclerosis caused by cathepsin K deficiency. We hypothesized that teriparatide, a potent anabolic agent used in the treatment of osteoporosis, might reduce skeletal fragility by activating bone turnover. We studied a typical case of pycnodysostosis in a 37-yr-old woman who exhibited short stature, skull and thorax deformities, and a history of severe fragility fractures. Cathepsin K gene sequencing was performed. Before and after 6 mo of 20 μg/d teriparatide, biochemical markers of bone turnover were measured, and 3D bone structure and microarchitecture was assessed in vivo by HR-pQCT. Qualitative and quantitative analysis of transiliac bone biopsies were performed, and the degree of mineralization was evaluated by quantitative microradiography. In vitro assessment of bone resorption was performed after separation and differentiation of CD14+ monocytes from peripheral blood. Bone structure assessed by HR-pQCT on the radius and tibia showed augmentation of cortical and trabecular density. Transiliac bone biopsy showed highly increased bone mass (+63% versus age- and sex-matched controls), a decrease in bone remodeling without evidence of active osteoblasts, and a severe decrease in the dynamic parameters of bone formation (mineralizing surfaces, −90% and bone formation rate, −93% versus age- and sex-matched controls). This depressed bone turnover probably explained the increased degree of mineralization. The presence of a novel missense mutation leading to an A141V amino acid substitution confirmed a genetic defect of cathepsin K as the cause of the disease. The deficiency of active osteoclasts was confirmed by an in vitro study that showed a decreased concentration of CD14+ monocytes (the precursor of osteoclasts) in blood. These osteoclasts had low resorptive activity when incubated on bone slices. After 6 mo of teriparatide, the structure, microarchitecture, and turnover of bone—assessed by HR-pQCT, histology, and bone turnover markers—remained unchanged. Our data strongly suggest that some features of the osteoclastic phenotype—that are absent in pycnodysostosis—are a prerequisite for the anabolic effect of PTH on osteoblasts.

INTRODUCTION

Pycnodysostosis is an extremely rare autosomal recessive bone disease (∼100 cases reported) caused by an inactivating mutation of cathepsin K, the key enzyme for the degradation of the bone matrix protein by the osteoclasts. Cathepsin K is localized in intracellular vacuoles in contact with the ruffled border and in the subosteoclastic space.(1) The impaired degradation of bone matrix is the fundamental pathogenic defect.(2) Mice deficient in cathepsin K have differentiated osteoclasts, which can demineralize bone but are unable to degrade the organic matrix.(3) To date, several cathepsin K (CTSK) mutations have been identified that always result in the absence of enzymatic activity. The deficiency of osteoclastic resorption leads to osteopetrosis. Generally, pycnodysostosis is diagnosed during childhood because of short stature and dysmorphic features of the skull, narrow thorax, short hands and fingers, and increased lumbar lordosis. Despite the increased bone mass, pycnodysostosis is characterized by skeletal fragility with recurrent fractures involving the lower limbs. Radiographic examinations show a generalized osteosclerosis with fractures and delayed cranial sutures and fontanels.(2) No treatment has been validated in pycnodysostosis. We hypothesized that teriparatide, the 1-34 N-terminal fragment of PTH, a potent anabolic agent used for the treatment of postmenopausal osteoporosis,(4) might reduce skeletal fragility. Although teriparatide is unlikely to overcome the genetic osteoclastic defect, it might enhance osteoblastic activity and therefore bone remodeling. Intermittent administration of PTH results in increased bone turnover with an augmentation of bone resorption and formation. Given to osteoporotic patients, teriparatide has an anabolic effect and decreases the incidence of new fractures.(4)

We studied a typical case of pycnodysostosis in a woman before and after treatment with teriparatide. We describe the histologic features of the bone tissue and evaluate the bone structure, microarchitecture, remodeling, and mineralization. The ability of the CD14+ monocytes, osteoclast precursors, to differentiate into active osteoclasts and resorb bone was studied. In addition, the CTSK mutation was described at the molecular level. Our observations after teriparatide treatment provide new insights into the potential mechanism of action of PTH on bone tissue.

MATERIALS AND METHODS

Patient

A 37-yr-old woman with short stature (131 cm; 34 kg) was diagnosed as having pycnodysostosis at the age of 8 yr. She had dysmorphic features with a short fourth metacarpal, skull and thorax deformities, and no endocrine abnormalities. She presented several low-trauma fractures including metatarsals and the shoulder blade at 16 yr of age, a tibia at 17 yr of age, a fracture of the left femur diaphysis and then the right femur diaphysis at 33 yr of age. BMD measured by DXA (QDR 4500; Hologic, Bedford, MA, USA) was significantly increased at all skeletal sites, with a T-score of +2.9 at the lumbar spine, +4.3 at the femoral neck, +4.8 at the total hip, and +4.9 at the distal radius.

The patient received daily 20-μg subcutaneous injections of teriparatide for 6 mo. The patient gave her informed consent to participate in the study.

Cathepsin K sequencing

Cathepsin K gene (CTSK) sequencing was performed in a sample of genomic DNA isolated from peripheral blood. All eight exons and the corresponding exon-intron boundaries of CTSK were amplified by PCR using specific primers and sequenced on an ABI 3730XL DNA Analyzer at MWG-Biotech (Ebersberg, Germany). PCR amplifications were performed using Thermo-Start DNA Polymerase Mastermix from Abgene (Epsom, UK) on 30–60 ng of genomic DNA according to the manufacturer's recommendations. The cycling conditions were as follows: 1× for 15 min at 95°C, 35× for 30 s at 95°C, 30 s at 55°C, 2 min at 72°C, 1× for 5 min at 72°C, and a final cooling step at 4°C. Before sequencing, PCR products were purified by means of a modified polyethylene glycol precipitation protocol. The dried products were dissolved in 5 mM Tris, pH 9.0. Approximately 30–60 ng of purified products were sequenced using the BigDye terminator chemistry on ABI 3730XL capillary sequencers (Applied Biosystems, Foster City, CA, USA).

Biochemical studies

Serum and urinary calcium and serum phosphorus levels were measured by a colorimetric method (Roche Modular analyzer; Roche Diagnostics, Mannheim, Germany). Serum intact PTH (sPTH) was measured by an immunochemiluminometric assay (ECLIA; Roche Diagnostics). Serum 25-hydroxyvitamin D level was determined after extraction by a radioimmunologic assay (DiaSorin, Stillwater, MN, USA).

Bone formation was evaluated with a human radioimmunoassay for serum total osteocalcin (sOC)(5) and intact procollagen type I N-terminal propeptide (sP1NP).(6) Bone resorption was assessed by measuring serum β isomerized C-terminal cross-linking telopeptide of type I collagen (βCTX) using the β-Crosslaps/serum reagents (Roche Diagnostics).(7) sPTH, sOC, sP1NP, and βCTX were measured with the Elecsys 2010 automated analyzer (Roche Diagnostics). Total urinary deoxypyridinoline (uTotalDPD) was measured by high-performance liquid chromatography according to a modification of a previously published technique.(8)

Urinary levels of helical peptide were determined by a specific competitive ELISA (Metra Helical Peptide assay; Quidel Corp., Mountain View, CA, USA).(9)

3D microarchitecture

The distal radius and tibia were scanned using a HR-pQCT system (XtremeCT; Scanco Medical, Basserdorf, Switzerland). This system enables the simultaneous acquisition of a stack of parallel CT slices with a nominal resolution (voxel size) of 82 μm. The details of the acquisition and analysis have been previously described.(10) Briefly, at each skeletal site, 110 CT slices were obtained, thus delivering a 3D representation of ∼9 mm in the axial direction. The volumetric total, trabecular, and cortical bone densities (mg HA/cm3) were obtained. Mean cortical thickness was derived from the mean cortical volume divided by the outer bone surface. The trabecular structural parameters were either directly measured as the number of trabeculae or derived according to the Parfitt's formulae(11) as the trabecular thickness and separation. The distribution of trabecular separation reflects the heterogeneity of the trabecular network.

Bone histomorphometry

Before and after treatment, a transiliac bone biopsy was obtained with a 7.5-mm inner diameter trephine after double tetracycline labeling. Seven-micrometer-thick undecalcified sections were stained with either Solochrome cyanin R or Goldner's trichrome. Unstained sections were used for UV observations of tetracycline labels.(12) The histomorphometric parameters were performed with an automatic (Osteolab; Explora Nova, La Rochelle, France) or semiautomatic (Tablet'Measure; Explora Nova) images analyzer. The entire cancellous tissue area of three sections was analyzed. All parameters were expressed according to the standard nomenclature recommended by the subcommittee on Bone Histomorphometry of the American Society of Bone and Mineral Research.(13) The bone structure and microarchitecture were assessed by measuring cortical thickness (μm) and porosity (%), cancellous bone volume (%), and wall thickness (W.Th) of cancellous packets measured under polarized light on Solochrome cyanin-stained sections. The following parameters were measured to evaluate the bone formation: osteoid volume/bone volume (%), osteoid surface/bone surface (%), osteoid thickness (μm), and mineral apposition rate (MAR, μm/d) measured on unstained sections under UV light. The extent of the mineralizing surface (MS, %) was calculated as the length of the double-labeled surface plus one half of the single-labeled surface. Bone formation rate was calculated as MS × MAR (μm3/μm2/d). The activation frequency, which represents the probability that a new cycle of remodeling will be initiated at any point of the bone surface, was calculated as (BFR/BS)/W.Th and expressed per year. The bone resorption was evaluated by the measurement of eroded surfaces including active (with presence of osteoclasts) and inactive (without osteoclasts) eroded surfaces and by the osteoclast number per unit of bone surface.

Degree of mineralization of bone tissue

The degree of mineralization was measured by computerized quantitative microradiography on 100-μm-thick sections. Microradiographs were performed with a Philips X-ray diffraction unit (PW 1830/40; Philips, Limeil Brevannes, France) operated at 25 kV and 25 mA and equipped with a PW 2272/20 diffraction tube. A monochromatic X-ray (nickel-filtered copper Kα radiation) was used with a focus-to-film distance of 30 cm. The mean degree of mineralization of bone (MDMB) measured on microradiographs was derived from the quantitative evaluation of the absorption of X-rays.(14)

In vitro assessment of bone resorption

Separation and differentiation of CD14+ monocytes:

CD14+ monocytes were isolated from peripheral blood from the pycnodysostosis patient and compared with that of three healthy sex- and aged-matched controls by CD14 magnetic cell sorting, as previously described.(15) Briefly, the blood was layered on top of Ficoll-Paque (Amersham Pharmacia, Burkinghamshine, UK) and centrifuged for 20 min. The lymphocyte fraction was collected from the interface and washed three times in PBS. The cells were incubated with CD14 magnetic beads (Dynal Biotech) for 20 min, with end-over-end rotation. Finally, the CD14+ cells were separated from the lymphocyte fraction by magnets. The isolated cells were seeded on cortical bovine bone slices (Nordic Bioscience, Herlev, Denmark) in a cell density of 71,000 cells/cm2 and grown in medium containing 25 ng/ml macrophage-colony stimulating factor (M-CSF; R&D Systems, Minneapolis, MN, USA) and 25 ng/ml RANKL (R&D Systems) to differentiate the monocytes into mature osteoclasts. Medium was changed every second or third day, and culture supernatants were saved for biochemical analysis.

Measurement of bone resorption:

Bone resorption was measured in supernatants from differentiating monocytes/osteoclasts by measurement of the amount of released C-terminal telopeptide fragment of type I collagen (CTX-I). The CrossLaps for culture ELISA assay (Nordic Bioscience) was used for measurement of CTX-I. TRACP was measured in the conditioned medium to study osteoclast formation.(16)

Statistical analysis

The comparison between two groups was performed by the nonparametric Mann-Whitney U-test because of unequal variance between the two groups.

RESULTS

Cathepsin K sequencing

A novel mutation associated with pycnodysostosis, a single base C→T transition at nucleotide 546 (GenBank record: NM_000396) leading to a A141V amino acid change in the protein, was discovered. The homozygosis of the putative mutation was confirmed by sequencing the primer binding sites.

Biochemical studies

Before treatment, levels of serum and urinary calcium and phosphorus, 25-hydroxyvitamin D, and PTH were normal. In contrast, the markers of bone formation, sOC and sP1NP, and the markers of bone resorption, sCTX, uTotalDPD, and helicoidal peptide, were low (Table 1). With teriparatide, bone formation markers (sOC and sP1NP) showed a transient increase at the first month of treatment and returned to baseline levels thereafter; resorption markers (sCTX, uTotalDPD, and helicoidal peptide) did not change (Table 1).

Table Table 1.. Effects of Teriparatide on Bone Markers in a Case of Pycnodysostosis
original image

3D microarchitecture

3D microarchitecture and volumetric BMD measured at the distal radius and tibia confirmed osteosclerosis (Fig. 1). There was a significant increase in trabecular and cortical density and a thickening of the trabeculae compared with age-matched controls. Teriparatide treatment did not significantly modify the bone architecture (Table 2).

Table Table 2.. Effects of Teriparatide on 3D Bone Structure and Microarchitecture Measured by HR-pQCT in a Case of Pycnodysostosis
original image
Figure Figure 1.

Microstructure of radius assessed by HR-pQCT in the patient with pycnodysostosis (A) and in healthy premenopausal woman (B) showing an increased cortical and trabecular density with a thickening of the trabeculae in the pycnodysostosis patient (scale bar = 1 mm).

Bone histomorphometry

Qualitative observations showed osteosclerosis (Fig. 2A) with a normal lamellar bone texture except in some area where the organization of the lamellae appeared disordered (Fig. 2B). Cartilage residuals were included in the bone matrix (Fig. 2C).

Figure Figure 2.

Undecalcified bone sections from a woman with pycnodysostosis. (A) Cortical (Ct) and trabecular osteosclerosis (Goldner's trichrome; bar: 1 mm). (B) The lamellar texture of bone is normal except in some areas where the bone matrix is disorganized (*) (Solochrome cyanin R; bar: 200 μm). (C) Cartilage residuals are included in bone tissue (arrowhead; Solochrome cyanin R; scale bar = 100 μm). (D) The extent of eroded surfaces is slightly increased but resorption lacunae are not deep, contain very few osteoclasts, and may be covered by an unmineralized bone matrix (arrow) (Solochrome cyanin R; scale bar = 50 μm). (E) Trabecular surfaces are boarded by a very thin layer of osteoid tissue without evidence of active osteoblasts (arrow) (Goldner's trichrome; scale bar = 100 μm). (F) The depressed active bone formation is confirmed by the absence of tetracycline label on most of trabecular surfaces (Tb) (unstained section observed under UV light; scale bar = 100 μm).

When compared with age- and sex-matched controls,(17,18) pycnodysostosis was characterized by a high cancellous bone mass with numerous thick trabeculae (Table 3). The extension of eroded surfaces was slightly increased, but resorption lacunae were sometimes covered by a demineralized matrix (Fig. 2D) and contained very few osteoclasts. These osteoclasts were small and contained two or three nuclei. Trabecular surfaces were covered by few and thin layer of osteoid tissue (Fig. 2E). Some trabecular surfaces were covered by a very thin and light tetracycline label. The absence of active formation (Fig. 2F) was confirmed by a severe decrease of the mineralizing surfaces and bone formation rate (Table 3).

Table Table 3.. Iliac Bone Histomorphometric Parameters of Structure and Remodeling in a Case of Pycnodysostosis Before and After 6 mo With 20 μg/d Teriparatide
original image

A posttreatment transiliac bone biopsy was performed on the controlateral side. The parameters of bone structure and microarchitecture were unchanged. The eroded surfaces decreased showing no sign of osteoclast activity. Static and dynamic parameters of bone formation remained unchanged (Table 3).

Degree of mineralization of bone tissue

The mean degree of mineralization of bone tissue was increased by 8% compared with premenopausal control women: 1.159 ± 0.112 versus 1.080 ± 0.086 g/cm3, respectively.

In vitro assessment of bone resorption

To study osteoclastogenesis and osteoclastic bone resorption, CD14+ monocytes were isolated from peripheral blood of the patient and three healthy controls. The concentration of osteoclast precursors, CD14+ cells, in the blood from the patient was very low compared with the concentration in blood from the controls (Table 4). The ratio between the patient's concentrations compared with that of controls was ∼1:150 (Table 4).

Table Table 4.. Concentration of CD14+ Monocytes in Blood From a Pycnodysostosis Patient
original image

Equal cell densities of monocytes from either patient or controls were seeded on bone slices, and differentiated into osteoclasts. The bone resorption marker, CTX-I, measured in cell culture supernatants showed a highly decreased ability of the patient-derived cells to resorb bone compared with differentiated monocytes from the controls (Fig. 3). Similar levels of TRACP were observed between patients and controls (data not shown).

Figure Figure 3.

Measurement of bone resorption from differentiated monocytes from pycnodysostosis patient and healthy controls. CD14+ monocytes were isolated from peripheral blood by CD14 MACS from one pycnodysostosis patient (▪) and three healthy, sex- and age-matched individuals (inline image). Cells were seeded on bone slices at a cell density of 71,000/cm2. Culture supernatants were collected on the indicated days, and CTX was measured. The results from the three controls were pooled. Mann-Whitney U-test: *p < 0.05, **p < 0.01.

DISCUSSION

Pycnodysostosis is an exceptional sclerotic bone disease associated with skeletal fragility, which results from a cathepsin K deficiency, and few studies have investigated its bone tissue characteristics.(19) In this study, we investigated the bone structure and remodeling, the osteoclasts properties, and the cathepsin K mutation in a 37-yr-old woman. The osteosclerosis was associated with a marked decrease of bone turnover. The extent of eroded surfaces was in the normal range but contained very few osteoclasts. The low concentration in monocytes, which are the precursors of osteoclasts, observed in the peripheral blood could potentially explain the low number of osteoclasts. The reason for the extreme low number of osteoclast precursors is not clear at this point and might be caused by different cell surface markers because we used a CD14 technique. In contrast, the presence of large multinucleated osteoclasts similar to those found in Paget's disease has also been described in pycnodysostotic patients,(20) suggesting that the potentiality of osteoclast precursors may differ from patient to patient. Specifically, genetic heterogeneity—our patient had a novel mutation—may explain those osteoclastic differences. Decreased osteoclastic activity, but not osteoclastogenesis, was shown both in vivo and in vitro by the diminution of bone resorption markers, clearly demonstrating that resorption of the organic part of the bone matrix is strongly inhibited. On bone sections, we observed crenated surfaces covered by a light unmineralized matrix partly bordered by flat cells, morphologically similar to bone-lining cells. It has been shown in pycnodysostosis and cathepsin K-deficient animals that osteoclasts may dissolve the mineral(3) and contain large cytoplasmic vacuoles filled with collagen fibrils.(21) In the absence of cathepsin K activity, a low rate of bone resorption suggests another pathway for the degradation of the bone matrix. The bone-lining cells contain matrix metalloprotease, which may contribute to the degradation of the demineralized collagen(22) by using different cleavage sites, resulting in the accumulation of larger fragments of collagen.(23) The possible contribution of other proteases in pycnodysostosis explains the high level of previously reported carboxy-terminal cross-linked telopeptide of collagen type I (ICTP), a marker detecting a more proximal portion of the C terminus of type I collagen.(23) Actually cathepsin K destroys ICTP, whereas matrix metalloproteases generate ICTP.(23,24)

The markedly reduced bone turnover resulted in a prolonged secondary mineralization and consequently in an increased mineral deposition as reflected by the elevated mean degree of mineralization.(25) The skeletal fragility observed in pycnodysostosis probably results mainly from the markedly depressed bone turnover that does not allow repair of microdamage(26) and possibly from the abnormal quality and quantity of the mineral.(27–29)

We found a new point mutation in exon 5 of the cathepsin K gene in our patient, which predicts the substitution of an alanine residue for a valine at position 141. This is the third mutation reported in the helix comprising residues 139–156, a highly conserved region of the protein, leading to a pycnodysostotic phenotype. The other two previously described mutations are F142L, which presumably renders the protein biologically inactive,(30) and G146R, which effectively abrogates enzyme activity in expression studies.(31) One would predict a similar outcome for A141V, particularly considering its proximity to Cys,139 an essential residue for catalytic activity.

No treatment has been validated in pycnodysostosis. We hypothesized that teriparatide, a potent anabolic agent used for the treatment of postmenopausal osteoporosis, might reduce skeletal fragility. Teriparatide therapy increases bone mass by increasing bone turnover with a stimulation of bone formation higher than that of resorption.(32–34) Although teriparatide is unlikely to overcome the genetic osteoclastic defect, we hypothesized that teriparatide might enhance other pathways of bone resorption as described above and increase osteoblastic activity. With 20 μg/d teriparatide, bone formation markers and to a lesser extent resorption markers increased during the first month of treatment and returned to low levels thereafter, reflecting a transient stimulation of the few existing osteoblasts, as has been suggested after 1 mo of PTH treatment(35) or an increased longevity of the osteoblasts by inhibiting the osteoblast apoptosis.(36) The bone formation marker P1NP was 2.5 times lower than in postmenopausal women after 6 mo of teriparatide.(37) This may also explain the decrease of eroded surfaces observed on the second bone biopsy, which have been partly filled by the small amount of matrix synthesized by these osteoblasts. At the end of the therapy, mineralizing surfaces and bone formation rate remained decreased by 88% and 90%, respectively, compared with control values.

Our observation showed that teriparatide was unable to stimulate bone turnover in pycnodysostosis, contrasting with its effect in postmenopausal and male osteoporosis, suggesting that some aspects of osteoclastic activity are required to induce the anabolic effect of teriparatide. It has been previously shown that, when given concomitantly with bisphosphonates, PTH or its 1-34 N-terminal fragment does not stimulate bone formation in sheep(38) and in humans.(39) In contrast, the concomitant administration of a weaker inhibitor of bone resorption, raloxifene, does not blunt the anabolic effect of teriparatide.(37) Whether the inhibition of parathyroid-induced bone formation is dependent on the type of antiresorptive agent, on the level of bone resorption, or on the osteoclast phenotype is unclear. Our observation suggests that the latter is more likely. In our case of pycnodysostosis, the rate of bone resorption assessed by histology and by biochemical markers was extremely low and was not activated by teriparatide because of the genetic defect in cathepsin K. Teriparatide is able to stimulate bone formation in remodeling units that were active before treatment by increasing both the production rate of osteoblasts and the recruitment of pre-osteoblasts.(35) However, de novo bone formation on quiescent surfaces as it has been hypothesized several years ago(40) is unlikely to result from the extension of bone formation beyond the limits of the remodeling units.(35) Our data strongly suggest that teriparatide is unable to induce directly the recruitment of new osteoblast teams or to initiate modeling-based formation. The mechanism by which bone formation rate is tightly coupled in time, space, and magnitude to the amount of bone resorbed within a remodeling unit is unclear. Animal models of osteopetrosis caused by osteoclast deficiencies support the hypothesis that osteoclasts may control the bone formation, not only through its resorptive activity.(41) When osteoclastogenesis is absent as in tl/tl rats, which lack M-CSF,(42) or in c-fos–deficient mice,(43) the bone formation is reduced. In contrast, mice deficient for ClC-7 or the α3-subunit of the V-ATPase show a normal bone formation despite the absence of resorption activity of the osteoclasts.(44,45) It has been suggested that the resorptive action of teriparatide is necessary for its sustained anabolic effects.(46) The anabolic effect of PTH is absent in c-fos−/− mice unable to develop osteoclasts(43) but is normal in c-src−/− mice, which have increased number of nonresorbing osteoclasts.(47) Furthermore, the resorbed bone surface may also be involved in the recruitment of osteoblasts. Of interest in this aspect are findings showing that osteoblasts preferentially form bone only when the correct 3D structures are present.(48,49) These studies showed that osteoblasts do not form bone on smooth surfaces but require either previous resorption or mechanical “grooving” to form bone. In absence of cathepsin K, the matrix degradation is impaired but the demineralization is normal.(3) In our pycnodysostosis patient, we confirmed the presence of nonmineralized matrix in erosion lacunae as previously reported.(19) Therefore, the collagen fibers contain RGD motifs, which have been suggested to induce osteoblast apoptosis,(41) contributing to the low bone formation and the absence of anabolic effect of teriparatide in pycnodysostosis.

In conclusion, teriparatide treatment was ineffective in a case of pycnodysostosis, which is characterized by deficient osteoclastic activity. These data strongly suggest that previous stimulation of osteoclast activity is a prerequisite for the sustained anabolic effect of teriparatide.

Acknowledgements

The authors thank Stephanie Boutroy for performing the HR-pQCT measurements; Evelyne Gineyts and Olivier Borel for performing biochemical measurements; and Dr John Bilezikian for helpful comments on the manuscript.

Ancillary