Strontium is incorporated into mineral crystals only in newly formed bone during strontium ranelate treatment

Abstract

Strontium ranelate has been shown to increase bone mass in postmenopausal osteoporosis patients and to reduce fracture risk. The aim of this study was to investigate the potential influence of strontium ranelate (Protelos) treatment on human bone tissue characteristics and quality at the micro- and nanostructural levels. We investigated transiliac biopsies from patients treated for 36 months with strontium ranelate or placebo (n = 5 per group) using synchrotron radiation with a microbeam combining scanning small-angle scattering, X-ray diffraction, and fluorescence spectroscopy (SAXS/XRD/XRF) for a detailed characterization of the mineral crystals within the collagenous bone matrix. A scanning procedure allowed the simultaneous determination of maps of the chemical composition together with thickness, length, and lattice spacing of these mineral crystals within each of the 15- or 25-µm-wide pixels in a thin bone section. The fluorescence results show that only bone packets or osteons formed during the strontium ranelate treatment contain significant amounts of strontium and that up to 0.5 of 10 calcium atoms in the mineral crystals are replaced by strontium, as revealed by a corresponding shift in apatite lattice spacing. The thickness and length of the plate-shaped bone mineral crystals were not affected by the strontium ranelate treatment. As a consequence, there was no indication for a change in human bone tissue quality at the nanoscale after a 36-month treatment of postmenopausal osteoporotic women with strontium ranelate, except for a partial replacement of calcium by strontium ions in the hydroxyapatite crystals, only in newly formed bone. © 2010 American Society for Bone and Mineral Research

Introduction

Strontium ranelate is a treatment for osteoporosis with antifracture efficacy1, 2 regardless of the severity of osteoporosis. Phase III studies performed over 5 years have shown sustained efficacy in reducing vertebral, nonvertebral, and hip fractures.3 The mechanical performance of bone and its potential fragility depend to a large extent not only on bone volume and shape and microarchitecture but also on its intrinsic material properties.4 Bone is a heterogeneous material at all hierarchical levels of organization. In particular, bone material is a composite consisting of a collagen-rich organic matrix with small essentially plate-shaped mineral crystals embedded in it.5, 6 The thickness of these mineral crystals is on the order of 2 to 4 nm, and the length and width are on the order of 10 nm.6 The mineral density distribution, as well as the shape, size, and arrangement of the mineral crystals, has a major influence on the mechanical behavior of the bone material.7, 8 Osteoporosis treatments interfere with the activity of bone-forming or bone-resorbing cells, and an influence on the bone material properties may be expected and, indeed, has been reported for several antiresorptive as well as anabolic treatments.9–16

The chemical synthesis of nonbiologic strontium–calcium hydroxyapatites leads to hydroxyapatite with strontium substitutions on the calcium site, and the lattice spacing in the hexagonal c direction of the resulting crystals is known to increase linearly with strontium substitution from 0.688 nm in stochiometric calcium hydroxyapatite (containing 10 calcium atoms per unit) to 0.728 nm in strontium hydroxyapatite when all 10 calcium atom are replaced by strontium.17 Moreover, some studies have suggested that strontium incorporation may change the physicochemical properties of the mineral and potentially may interfere with apatite formation and crystal properties both in vitro and in vivo.18, 19 It is therefore extremely important to assess potential influences of strontium ranelate treatment on factors determining bone material quality, such as size, orientation, and composition of mineral crystals. The effects of strontium ranelate on bone mineral have been studied in animal models, and the increase in bone strontium content was not found to be followed by a significant increase in the average lattice spacing measured in powdered bone.19 From this it was concluded that most of the strontium ions are not incorporated into mineral crystals but rather are stored in a nonspecific way on their surfaces.20–22 An inhomogeneous incorporation of strontium preferentially into newly formed bone was observed in a monkey model.20, 21 Recent data are available from biopsies sampled during strontium ranelate clinical studies24; they show the absence of deleterious effects of strontium ranelate on the primary mineralization of bone, with no evidence of osteomalacia.

The goal of this study was to determine, on a subgroup of biopsies described previously,24 the size, shape, composition, and crystal structure of the hydroxyapatite crystals in iliac crest biopsies after long-term antiosteoporotic strontium ranelate treatment. We used a combination of small-angle X-ray scattering (SAXS), X-ray diffraction (XRD), and X-ray fluorescence spectroscopy (XRF) to obtain position-resolved information on the mineral characteristics in bone sections with a spatial resolution of 15 or 25 µm. The advantage of this scanning approach is that areas of newly formed bone can be distinguished from areas of old bone formed before treatment. This also increases the sensitivity of the crystal structure analyses used to determine whether the strontium ions enter the crystal lattice20 because strontium ions are mostly expected to be found in areas of newly formed bone.23 Finally, the SAXS/XRD analyses provide information on crystal size and arrangement, which are major determinants of the biomechanical properties of the bone material.6

Materials and Methods

Specimen preparation

Transiliac bone biopsies from osteoporotic women supplemented with calcium and vitamin D and treated with strontium ranelate or placebo for 3 years1, 2 were fixed in formaldehyde and embedded in polymethylmethacrylate. Then 15-µm-thick bone sections were obtained by cutting and polishing. They were then mounted into lead foils with circular windows of about 2 mm. After this, two specimens (one treated patient and one placebo) were sputtered with gold by a sputter coater for 240 seconds. The sputtered thin film was optically transparent and was intended as a diffraction reference for precise measurements of lattice constants. Altogether, 10 biopsies (5 treated and 5 controls) were studied with SAXS, XRD, and XRF to determine the size and arrangement of the mineral crystals.

Experimental setup

The experiments were performed at the microfocus beamline (µSpot) at BESSY II in Berlin, Germany.25 The general principles of the SAXS and XRD analysis are summarized in Fig. 1. An on-axis charge-coupled device (CCD)-based area detector (MarMosaic 225, Mar USA, Evanston, IL, USA) with an active area of 225 × 225 mm and a pixel size of 73.24 µm was used for simultaneous SAXS/XRD/XRF measurements. An energy-sensitive detector (ASAS-SDD KETEK, Munich, Germany) with a 100-mm² sensitive active area and 167.4 eV energy resolution was used for the XRF measurements. The angle between the X-ray primary beam direction and fluorescence detector was about 50 degrees, and the distance to the sample was approximately 20 mm.

Figure 1. Schematic representation of the experimental setup at the µSpot synchrotron beamline at BESSY II, Berlin. A 15- or 25-µm-wide X-ray beam is hitting a thin bone section (a), where it generates several types of signals. Part of the X-rays are scattered at small angles that gives rise to a signal around the primary beam (labeled SAXS) that can be analyzed to give the thickness and orientation distribution of bone mineral crystals (b). Some photons are also diffracted to larger angles (XRD), where diffraction lines are generated. The 002 line of hydroxyapatite (labeled HA 002) is used to determine the crystal lattice spacing (specifically the hexagonal c spacing) of the mineral (from the peak position) and the length of the mineral crystals in the c direction (from the peak width). The diffraction line labeled “Au” is due to a thin gold layer deposited on the bone section to serve as an internal standard for lattice spacing determination. Finally, some of the photons are absorbed and give rise to X-ray fluorescence, which is collected in a separate detector (c) and used to analyze the atomic composition, in particular the strontium content. An azimuthally averaged diffracted intensity for large angles is shown in panel d. Since all these parameters are determined from the identical volume of the specimen (determined by the beam cross section and the section thickness only), a scanning of the bone section in two directions perpendicular to the beam allows the simultaneous determination of parameter maps of different kinds.

A sample goniometer consisting of an xyz translation stage and a θ–2θ goniometer, together with an additional small xyz scanning stage with linear encoders (resolution 100 nm) on top of the goniometer, was used for microbeam scanning. In order to observe and define the regions of interest on the specimen, a long-distance optical microscope (Infinity Optical, Boulder, CO, USA) with a resolution of 2.2 µm and a field of view of 1.6 mm was mounted on the 2θ goniometer arm. For the position calibration of the X-ray beam at the sample position, a high-resolution CdWO₄ crystal on a 1-mm-thick glass substrate was employed. The reproducible accuracy for the absolute beam position is about 2 to 3 µm.

The X-ray microbeam was defined by a toroidal mirror in combination with a beam-defining pinhole close to each of the samples. Slightly different setups with respect to beam size and X-ray energy were used in three consecutive experimental sessions. Session 1: Simultaneous SAXS/XRD (sample-detector distance D = 220 mm) with a 10-µm beam-defining pinhole, a 20-µm guard pinhole, and a W/Si multilayer monochromator were used at an energy E of 15 keV (wavelength λ = 0.0827 nm). Session 2: To obtain a better XRD resolution, a Si 111 crystal double monochromator was used at E = 12.398 keV (λ = 0.100 nm) together with a 20-µm beam-defining pinhole and a 50-µm guard pinhole (D = 234 mm). Session 3: For combined SAXS/XRD/XRF finally, the same setup as in session 2 at E = 18 keV (λ = 0.0689 nm) and D = 292 mm was employed. Owing to beam divergence, the beam size and therefore the effective position resolution at the sample position was about 15 µm (10-µm beam-defining pinhole) and about 25 µm (20-µm beam-defining pinhole).

Measurement protocol and data reduction

Simultaneous SAXS/XRD measurements of two regions (one cortical and one trabecular) of interest (typically 200 µm × 200 µm each) were performed for each of the 10 biopsies (5 per group). In addition, SAXS/XRD/XRF measurements were performed for one of the strontium ranelate–treated samples within three selected regions and for one placebo sample. The regions were selected to contain newly formed bone areas as visualized by low calcium content in backscattered electron imaging. The thin specimen sections were aligned exactly perpendicular to the primary beam, and mesh scans in the yz sample plane with a step width of 15 µm (25 µm) were performed. Prior to the SAXS/XRD/XRF measurements, a transmission scan in the selected region of interest was performed for all samples using a photodiode. This supplied an absorption image of the specimen that was used in the further SAXS/XRD analysis for background correction.26 Then SAXS/XRD patterns were recorded simultaneously for each scan point in the same region of interest (Fig. 1). The typical exposure time was 15 seconds for the measurements with the W/Si multilayer and 60 seconds for measurements with the Si 111 crystal double monochromator.

The 2D SAXS/XRD data were corrected for dark current (CCD readout noise) and reduced to 1D scattering profiles I(q) by azimuthal averaging. The length of the scattering vector q is defined by q = 4πsin(θ)/λ, where 2θ is the scattering angle. Moreover, a radial averaging of the SAXS signal was performed to obtain the SAXS intensity as a function of azimuthal angle I(χ). All data were corrected for background scattering by taking the local sample transmission properly into account.

XRD analysis

Hydroxyapatite is a crystal with hexagonal structure. Its unit cell contains 10 calcium atoms for 6 phosphates and 2 hydroxyls. The 002 reflection of hydroxyapatite (HA), which gives the size c of the unit cell in hexagonal direction, was used for accurate lattice spacing determination (Fig. 1). The 002 Bragg peak in the azimuthally averaged intensity was fitted using a Voigt function, and the crystal lattice spacing c of HA then was calculated by c = 4π/q⁰, with q⁰ being the fitted peak maximum. An internal standard (thin gold film sputtered on the specimen surface) was used for the accurate absolute determination of q⁰ by also fitting the (111) Au diffraction peak (Fig. 1) for each diffraction profile. The width of the 002 peak of apatite (after correction for instrumental resolution using a measured diffraction pattern of synthetic hydroxyapatite) was used to estimate the length L of the mineral crystals in the c direction via the Scherrer equation.27–29

SAXS analysis

The central part of the SAXS/XRD pattern, corresponding to the small angle scattering, was analyzed in a standard way from the azimuthally averaged intensity I(q) described previously.26, 30–32 In particular, we determined the typical thickness T of the mineral crystals, defined as T = 4φ(1 – φ)/σ, where φ is the volume fraction and σ is the specific surface of the mineral phase within the bone matrix.33 Moreover, we determined the typical orientation and the degree of alignment ρ of the mineral crystals at each measuring position on the specimen for the radially averaged intensity I(χ). For a perfect alignment of mineral crystals, ρ = 1, whereas for a random orientation, ρ = 0.26, 32 This parameter is known to reflect changes in tissue organisation, for example, in murine models of osteogenesis imperfecta34 or of alkaline phosphatase deficiency.35

X-ray fluorescence

The X-ray fluorescence signal was collected simultaneously with the SAXS/XRD data in the energy range from 0 to 18 keV (Fig. 1). This energy range is equally distributed to 2048 channel numbers, from which the energies of the different fluorescent lines can be determined. The peaks indicated in Fig. 1 are P Kα (2.01 keV), Ca Kα (3.69 keV), Au Lα (9.71 keV), Sr Kα (14.17 keV), Sr Kβ (15.84 keV). The two red peaks near the Au Lα were from the guard pinhole (Pt Lα 9.44 keV and Lβ 11.07 keV). For relative determination of strontium content, X-ray fluorescence intensities were used after normalizing to the Au Lα peak. No absolute calibration to element concentrations was performed.

The measurements for these data used a beam-defining pinhole with 20 µm diameter and Si 111 crystal double monochromator, the scan step size was 25 µm, and the wavelength was 0.0689 nm (E = 18 keV).

Statistical analyses were carried out using the software package SigmaStat (Systat Software, San José, CA, USA).

Results

Figure 2 shows a typical result from the SAXS data collected from a trabecula in the biopsy of a patient treated for 36 months with placebo. The area scanned with a resolution of 15 µm is shown in the light microscopic image on the left (Fig. 2b). Three parameters obtained from an analysis of SAXS data collected at each point on the specimen are plotted on the right. The typical thickness T of the mineral crystals is roughly constant throughout the area of investigation. The degree of alignment ρ as well as the particle orientation (Fig. 2d, 2e) clearly show that the trabecula is composed of several bone packets with different fiber orientation of the collagen-mineral composite. These packets probably were deposited at different points in time.

Figure 2. Example of parameter maps derived from the analysis of SAXS data alone for a bone section from a placebo-treated patient for 36 months. (a) An overview of the specimen holder. (b) The light microscopy image obtained in situ at the synchrotron radiation beamline. Scanning SAXS measurements were done in the area indicated by the white rectangle. The right-hand pictures show (c) the thickness T, (d) the degree of alignment ρ, and (e) the typical orientation of the elongated mineral crystals within the area indicated by the white rectangle in panel b. The bars for χ have a length proportional to ρ. The wavelength of this measurement was 0.0827 nm (15 keV) supplied by a W/Si multilayer monochromator, scan step size was 15 µm, exposure time was 15 seconds, and the beam-defining pinhole had a 10 µm diameter.

Orientation of mineral crystals

The typical picture, as shown in Fig. 2, is conserved throughout the placebo as well as the specimens from strontium ranelate–treated patients. The mineral crystal orientation of the bone material varies according to the distribution of bone packets in the lamellar bone structure.5, 6 This is visible in the maps of both ρ and of χ (where the bars are aligned with the direction χ and have a length proportional to ρ). For example, a bone packet on the right of the trabecula is blue in the ρ map (Fig. 2d), revealing little alignment within the plane of section. Accordingly, the bars in the orientation map are very short and do not show a special direction (Fig. 2e). This means that the collagen fibril direction is most likely perpendicular to the plane of section in this area. In the central part of the trabecula, the ρ map is orange, indicating a high degree of alignment within the plane of section. The direction of the bars in the χ map indicates that the orientation is predominantly vertical.

Mineral particle thickness from small-angle X-ray scattering

Most interestingly, the mineral particle thickness, as estimated by the parameter T, does not show any statistically significant variation between bone packets. T is defined as T = 4φ(1 – φ)/σ, where φ and σ are the volume fraction of the mineral and its total surface per unit volume, respectively. As is quite obvious from Fig. 3, there is no statistically significant difference between placebo and strontium ranelate–treated patients, as verified by a one-way ANOVA test. For this quantitative comparison between different patients, only newly formed cancellous bone identified previously in backscattered electron images was included. Additional pairwise comparison showed no statistically significant difference between any of the patients.

Figure 3. Summary (mean ± SD) of mineral crystal thickness T in biopsies from patients treated with strontium ranelate (white circles) and placebo (black circles). The T parameter was collected in about 100 or more data points separated by 25 µm in newly formed cancellous bone (based on areas with lower mineral content according to previous observation by backscattered electron imaging) for each patient biopsy. No significant difference is found between the two patient groups.

Mineral lattice constant and crystal length from X-ray diffraction

A more complete picture of the characteristics of mineral crystals is obtained when SAXS and XRD results obtained simultaneously in the experiment are combined. Figure 4 shows an area of osteonal bone in a strontium ranelate–treated patient. Thickness (as determined from the SAXS analysis) and length (as determined from the width of the 002 peak of hydoxyapatite) of the mineral crystals vary little throughout the area, including several osteons. It should be noted that all parameters represent average values within a bone volume of 15 × 15 × 15 µm³ (for Fig. 2) or 25 × 25 × 15 µm³ (for the other figures), as defined by the specimen thickness and the X-ray beam cross section. A somewhat different picture is obtained when the length of the c axis in the apatite structure is plotted as a function position on the bone section. It is clear from the map in Fig. 4d that the lattice spacing c is different in different areas of the section. Comparatively large values of c are obtained in one of the osteons in the figure (region labeled 1 in Fig. 4a). These values of c are larger than what would be expected from calcium hydroxyapatite and, therefore, indicate the incorporation of strontium into the lattice. Indeed, the average value for the hexagonal c axis of the calcium hydroxyapatite crystal is 0.687 ± 0.001 nm,17, 19 which corresponds well with the values of c found with placebo treatment and in some areas of biopsies after strontium ranelate treatment. An example is the osteon to the right of osteon 1 in Fig. 4, which is obviously older because osteon 1 penetrates into its area (Fig. 4a). This shows hydroxyapatite with a lattice spacing c in the usual range. From chemical studies, it is known that strontium may replace calcium atoms in the hydroxyapatite lattice.17 For a synthetic apatite with composition (Sr_xCa_1-x)₁₀(PO₄)₆(OH)₂, the lattice constant c in the direction of the hexagonal axis depends on the replacement of calcium atoms by strontium atoms.17 That is,

• 

where x is the Sr/(Sr + Ca) atomic ratio.

Figure 4. Parameter maps of some mineral crystal characteristics in osteonal bone from a postmenopausal women treated for 36 months with strontium ranelate (a). The mean thickness T (b) and length L (c) of the mineral crystals are derived from SAXS and XRD analysis, respectively. The parameter c (d) is the lattice spacing in the hexagonal direction of apatite, which is known to depend linearly on strontium content in synthetic hydroxyapatite.17 (e) The percentage of calcium ions substituted by strontium in the hydroxyapatite crystals, as derived from part d. The red lines in the microscopy image (a) indicate regions of increased strontium content in the mineral crystals (labeled 1, 2, 3, and 4). The newly formed osteon 1 (see arrows in panels a and d) has a higher strontium content than the old osteon, which it has partially replaced (see panel a). For these measurements, we used a beam-defining pinhole of 20 µm diameter and a Si 111 crystal double monochromator; the scan step size was 30 µm, and the wavelength was 0.1 nm (E = 12.398 keV).

Using this equation, we converted the c map into a strontium map (Fig. 4e). This map indicates the strontium content within the mineral crystals but ignores other types of noncrystalline strontium deposits. The atomic fraction of strontium replacing calcium on the apatite lattice goes up to 3% in the specific section illustrated in Fig. 5 and up to 5% maximum in certain areas but stays very small (at the same levels as in controls) in other areas. Interestingly, the areas exhibiting strontium correlate with structural features visible with the light microscope, such as osteons in the cortex or individual bone packets in trabeculae (Fig. 4a). For example, an obviously newly formed osteon that impedes on another older osteon exhibits strontium in the mineral crystals, whereas the older osteon has strontium basal levels (see arrows in Fig. 4a, d). This is pointing to the fact that strontium is incorporated into the mineral crystals of osteonal bone formed during treatment. This also suggests that there is essentially no strontium diffusion from “new” bone into older bone packets (as nicely visible for the two osteons in Fig. 4).

Figure 5. Strontium content in a newly formed bone packet on the surface of a trabecula for a patient treated for 36 months with strontium ranelate. XRD and XRF data were collected simultaneously in the area shown by the box in the light microscopic overview of the specimen (a). A beam-defining pinhole with 20 µm diameter and a Si 111 crystal double monochromator were used, the scan step size was 25 µm, and the wavelength was 0.0689 nm (E = 18 keV). The image in panel b shows a map of the strontium content in the mineral crystals (expressed as Sr/Ca atomic ratio in the mineral crystals) derived from XRD, and panel c shows a map of the Sr/Ca atomic ratio in the bone material as a whole determined from XRF. (d) A map of the calcium line intensity in the XRF spectrum.

Strontium content in tissue from X-ray fluorescence spectroscopy

Finally, we compare the levels of strontium found in the crystals to global strontium levels in the same tissue area by evaluating the XRF maps collected simultaneously with the SAXS and XRD maps. Figure 5 shows an example of a bone packet on the surface of a trabecula for a strontium ranelate–treated patient. Distributions of Sr/Ca ratios have been determined in two ways: First, XRD was used as described in the preceding paragraph to determine the atomic ratio Sr/Ca within the mineral crystals (Fig. 5b). Second XRF analysis was used to determine atomic ratio Sr/Ca within the tissue (including all calcium and strontium atoms in crystals and any other form). This analysis shows that this ratio is enhanced in the newly formed bone, where the calcium content is less compared with older bone (Fig. 5c) owing to a less complete secondary mineralization. Strontium is enhanced in the newly formed bone packet (recognizable by a lower calcium content; Fig. 5d) deposited on the surface of the trabecula, whereas the older bone below stays at basal levels (Fig. 5d). According to these analyses, the focal bone strontium content reaches at most about 5% in the affected areas, consistently throughout our measurements, in both trabecular and osteonal bone.

Discussion and Conclusion

The results of this study, employing a combinatorial analysis of iliac crest biopsies from osteoporotic patients treated with strontium ranelate, indicate that strontium is selectively uptaken in newly formed (during the course of the treatment) bone packets, incorporated into the apatite crystals, but does not alter any of the mineral crystal characteristics monitored.

Today it is widely accepted that bone strength depends on both the amount and quality, the latter an umbrella entity encompassing the structural and material properties of bone, both of which depend on bone turnover.7 Mineral crystal size, shape, and chemical composition are important contributors to the bone material properties.6 For instance, the alignment of mineral crystals has been found to be strongly modified in murine disease models of osteogenesis imperfecta34 and alkaline phosphatase deficiency.35 To date, the effects of various interventions such as alendronate,13 risedronate,10 and intermittent treatment with parathyroid hormone12 in humans and combination of osteoprotegerin and parathyroid hormone in a rat animal model11 on the mineral crystal characteristics have been investigated. In none of these treatments were the size and shape of the mineral crystals found to be significantly altered. This is in striking contrast to treatments with sodium fluoride, where an increase in mineral crystal size was found to accompany the increase in bone density. Moreover, the arrangement of mineral crystals within the collagenous matrix was found to be severely impaired with fluoride, which was linked to a reduced biomechanical competence of the bone material.14–16 The probable reason for this effect is the substitution of fluoride for hydroxide in the hydroxyapatite crystals, producing fluorapatite with radically different chemical properties and, in particular, decreased solubility.

Using a completely new experimental approach with a combination of scattering methods to analyze human iliac crest biopsies from patients treated with strontium ranelate over 3 years, we could show that some strontium ranelate is incorporated into the apatite crystals mainly within the new bone packets deposited during treatment with strontium ranelate, with a Sr/Ca atomic ratio not exceeding 5%. Furthermore, this hydroxyapatite composition modification does not interfere with other mineral crystal characteristics, such as their typical thickness or orientation. These results confirm earlier data on ranelate treatment at the level of the bone material, as reported in animal models.20–22 On the other hand, our new data do not support the idea put forward earlier that strontium is present in bone only loosely bound to crystal surfaces because some of the strontium present in bone is deposited into the hydroxyapatite crystal lattice with the mineral of the newly formed bone. This assessment was based on measurements with XRD on powdered bone specimens that did not reveal changes in crystal lattice spacing.20 Based on this study, we know that changes in crystal lattice spacing (owing to strontium incorporation) occur only in newly formed bone packets/osteons. When analyzing powdered bone samples, it was impossible to separate newly formed bone packets from older ones formed before treatment; thus the sensitivity of the techniques used in these studies was not sufficient to detect the relevant changes. With the new technology of microfocus scanning XRD, it is now possible to clearly distinguish effects in younger bone packets from older bone. This technical improvement compared with earlier studies may explain why these studies were not able to reveal strontium incorporation.21 The combination of scanning XRD with XRF analysis clearly shows that some strontium is incorporated into the crystals, based on the lattice constants determined for synthetic hydroxyapatite with different Sr/Ca ratios.17 However, this strontium uptake into the apatite crystal lattice does not directly influence bone material properties. This aspect remains an important issue for a long-term clinical observation, but the available 5- to 8-year efficacy and safety data3 suggest that this has no influence compared with the benefits of fracture prevention.

To conclude, our results show that only bone packets or osteons formed during strontium ranelate treatment contain significant amounts of strontium and that up to 0.5 of 10 calcium atoms in the mineral crystals are replaced by strontium, as revealed by a corresponding shift in apatite lattice spacing. However, the mean thickness and length of the plate-shaped bone mineral crystals were not affected by strontium ranelate treatment. As a consequence, there was no indication for a change in human bone tissue quality at the nanoscale after a 36-month treatment of postmenopausal osteoporotic women with strontium ranelate, except for a partial replacement of calcium by strontium ions in the hydroxyapatite crystals, only in newly formed bone.

Disclosures

All the authors state that they have no conflicts of interest.

Acknowledgements

We are most grateful to Dr Isabelle Dupin-Roger (Institut de Recherches Internationales SERVIER, Courbevoie, France) for many useful discussions and a critical reading of the manuscript. This article was supported in part by a research grant from servier.