Dr Phipps is an employee of Procter & Gamble. Dr Klaushofer is a consultant for Amgen Inc., Procter & Gamble, Roche, Novartis, and Merck Sharp & Dohme Ltd. Dr Klaushofer is also an editorial board member of Osteo Update. Dr Paschalis receives funding from Procter & Gamble Pharmaceuticals, Eli Lilly & Co., Osteotech Inc., Servier, BioRad Laboratories Inc., and Novartis. Dr Roschger receives consultancy fees from Procter & Gamble and Merck Sharp & Dohme Ltd. as a member of advisory board meetings. All other authors state that they have no conflicts of interest.
Ludwig Boltzmann Institute of Osteology at the Hanusch Hospital of WGKK and AUVA Trauma Centre Meidling, and 4th Medical Department, Hanusch Hospital, Vienna, Austria
Long-term effects of risedronate on bone mineralization density distribution in triple transiliac crest biopsies of osteoporotic women were evaluated. In this double-blinded study, 3- and 5-year treatment with risedronate increased the degree and homogeneity of mineralization without producing hypermineralization. These changes at the material level of bone could contribute to risedronate's antifracture efficacy.
Introduction: Risedronate, a nitrogen-containing bisphosphonate, is widely used in the treatment of osteoporosis. It reduces bone turnover, increases BMD, and decreases fracture risk. To date, there are no data available on the long-term effects of risedronate on bone mineralization density distribution (BMDD) in humans.
Materials and Methods: Osteoporotic women enrolled in the VERT-NA trial received either risedronate (5 mg/day, orally) or placebo for up to 5 years. All subjects received calcium and vitamin D supplementation if deficient at baseline. Triple iliac crest biopsies were collected from a subset of these subjects at baseline and 3 and 5 years. BMDD was measured in these biopsies using quantitative backscattered electron imaging, and the data were also compared with a normal reference group.
Results: At baseline, both risedronate and placebo groups had a lower degree and a greater heterogeneity of mineralization as well as an increase in low mineralized bone compared with the normal reference group. The degree of mineralization increased significantly in the risedronate as well as in the placebo group after 3- and 5-year treatment compared with baseline. However, the degree of mineralization did not exceed that of normal. Three-year treatment with risedronate significantly increased the homogeneity of mineralization and slightly decreased low mineralized bone compared with placebo. Surprisingly with 5-year risedronate treatment, heterogeneity of mineralization increased compared with 3-year treatment, which might indicate an increase in newly formed bone.
Conclusions: Long-term treatment with risedronate affects the homogeneity and degree of mineralization without inducing hypermineralization of the bone matrix. These changes at the material level of the bone matrix may contribute to risedronate's antifracture efficacy in osteoporotic patients.
Bisphosphonates, synthetic analogs of pyrophosphate, are potent inhibitors of osteoclastic bone resorption, and as such, are valuable therapeutic agents in the treatment of osteoporosis. The nitrogen-containing bisphosphonates, risedronate and alendronate, significantly reduce the risk of vertebral and nonvertebral fractures in osteoporotic subjects.(1,2) However, the exact mechanisms for the antifracture benefit of bisphosphonates are not fully understood. Bisphosphonate-mediated changes in BMD are related to decreases in fracture risk in groups of patients, but the relationship is neither linear nor proportional.(3–5) Multiregression analysis of several clinical trials with antiresorptive treatments shows that the increases in BMD contributed <20% to the observed reductions in vertebral and nonvertebral fracture risk.(4–9) Thus, it is evident that, in addition to BMD, other factors such as bone quality are important when assessing bone strength and fracture risk. Bone quality can be defined as the sum of all factors mediating mechanical competence at constant bone mass. It includes factors such as bone geometry and mass distribution, microarchitecture, and bone mineral and matrix tissue properties including mineralization, turnover, and microdamage.(10) The degree of bone matrix mineralization is an important determinant of the stiffness and hardness(10) of the bone material and can be measured by quantitative backscattered electron imaging (qBEI).(11–13) This established and validated method(11–13) enables the description of the bone mineralization density distribution (BMDD) at the microscopic level. BMDD is the frequency distribution of the appearance of a certain mineral content found in the individual bone packets of varying age and mineral content comprising the bone structure.(14,15) It reflects the rate of bone modeling and remodeling, the dynamics of mineralization, the age distribution of the bone packets, and the influence of the collagen matrix on the mineralization process. Trabecular bone of normal skeletally mature individuals has a nearly constant BMDD, independent of skeletal site, age, sex, or ethnicity.(12) Therefore, already small deviations from normal BMDD caused by diseases and/or treatment can be detected and quantified. Bisphosphonates have been shown to increase the degree and homogeneity of mineralization in bone from human osteoporotic subjects(16) as well as in normal bone from animals.(17,18)
Despite the clinically proven antifracture effectiveness, there is still a debate about potential safety issues with the long-term treatment of bisphosphonates. One concern is the possibility of hypermineralization of the bone matrix in long-term treatment that could impair the intrinsic mechanical properties of bone and make it more brittle.(19)
Most previous studies looking at the effects of bisphosphonates on the degree of mineralization in bone of osteoporotic patients were limited in their sensitivity, because they only used endpoint biopsies, and not pre- and post-treatment biopsies from the same subjects.(16,20)
Here we report data from a double-blinded, prospective study in women with postmenopausal osteoporosis looking for the first time at the longitudinal effects of 3- and 5-year treatment with risedronate on BMDD in triple transiliac biopsies from the same subjects.
MATERIALS AND METHODS
Iliac crest bone biopsies used for this analysis were obtained from postmenopausal osteoporotic women enrolled in the Vertebral Efficacy with Risedronate Therapy-North America (VERT-NA) trial.(2,21,22) In this clinical trial, all subjects received 1000 mg elemental calcium per day. Subjects with low serum 25-hydroxyvitamin D levels at baseline (<40 nM) also received vitamin D (cholecalciferol) supplementation up to 500 IU/day. In the trial, subjects were randomized to receive risedronate 5 mg/day, orally, or matching placebo. Results from this trial on BMD, fracture incidence, bone turnover markers, histomorphometry, and trabecular microarchitecture have been reported previously.(2,21,22)
In this study, triple biopsies taken at baseline and 3 and 5 years from a subset of subjects in the VERT-NA trial were analyzed by qBEI for BMDD. Results were also compared with BMDD data from a normal reference group (n = 52).(12) This normal reference BMDD was derived from trabecular bone of iliac crest, vertebra, femoral neck, femoral head, and patella from 52 healthy women and men of white and black ethnicity, covering an age range from 25 to 90 years. For the triple biopsies study, biopsy sets for analysis were selected on basis of having baseline mineralizing surface (MS/BS; %) ≥ biopsy cohort median baseline MS/BS, as well as the biopsies having two intact cortices and being considered evaluable by histomorphometry. Baseline and 3-year biopsies were taken from iliac crest on opposite sides. The 5-year biopsy was taken from the same side as was the baseline biopsy, but at least 2 cm apart from the first biopsy site. The risedronate group consisted of 10 paired baseline and 3-year biopsies (n = 10), and a third biopsy was obtained from 8 of these 10 subjects after 5-year treatment with risedronate (n = 8). The placebo group consisted of eight sets of paired biopsies at baseline and 3 years (n = 8). A third biopsy taken at 5 years was available from only two of the eight placebo subjects. The data are presented, but no statistical analysis was performed. All biopsies had been fixed and dehydrated in alcohol, embedded in polymethylmethacrylate, and processed as described earlier for histomorphometry.(11,12,16)
BMDD was determined using qBEI by operators blinded to treatment groups. The details of qBEI have been published elsewhere.(11,13) Briefly, a digital electron microscope (DSM 962; Zeiss, Oberkochen, Germany) equipped with a four-quadrant semiconductor BE detector was used. The accelerating voltage of the electron beam was adjusted to 20 kV, the probe current to 110 pA, and the working distance to 15 mm. Settings were adjusted to a ×50 nominal magnification, corresponding to a pixel resolution of 4 μm/pixel. The digital backscattered (BE) image was generated by a single frame with a slow scan speed of 100 s/frame. At least five images of 2.5 mm width were taken from the cancellous bone area for each bone biopsy and were used for evaluation of the BMDD parameters as described below.
The intensity of the backscattered electrons is proportional to weight concentration of the calcium in bone.(11,13) The method uses digital (pixel) images acquired from the bone biopsies reflecting local calcium (Ca) content within the specified bone areas. From these images, gray-level histograms were generated indicating the percentage of mineralized bone area (y-axis) corresponding to the number of pixels with a certain gray level (x-axis; Fig. 1). The gray levels were transformed into weight percent (wt%) Ca values in two steps: first, the gray scale was calibrated using the “atomic number contrast” between carbon (C, Z = 6) and aluminum (Al, Z = 13) as reference materials. This was achieved by adjusting brightness and contrast of the BE detector before BE imaging of the bone areas was started. The gray scale of C was set to a gray-level index of 25 and that of Al to 225. Second, a calibration curve to convert the calibrated gray scale into wt% Ca was generated by measuring the gray value of osteoid (nonmineralized collagen) as 0 (<0.17) wt% Ca and pure hydroxyapatite (HA) reference as 39.86 wt% Ca. Because the gray-level value of HA is beyond that of Al (which is the upper limit of the gray scale calibration range), the HA value was subsequently corrected to the original brightness setting. Thus, the BE image gray-level distribution can be interpreted as a wt% Ca BMDD from the osteoid to HA gray level calibration line. Four parameters of BMDD were calculated to characterize the BMDD for statistical analysis.(12) (1) CaMean is the weighted average Ca concentration of the mineralized bone area. (2) CaPeak is the peak position of the BMDD histogram showing the most frequent wt% Ca of bone area. Differences between CaMean and CaPeak indicate the degree of asymmetry of the BMDD. Only in case of a perfect symmetric distribution CaMean and CaPeak would be equal. However, the BMDD is generally skewed to the low mineralization side. (3) CaWidth is the width at one-half-maximum of the BMDD histogram peak, indicating the heterogeneity of mineralization caused by the co-existence of bone packets of different ages and different degrees of mineralization. (4) CaLow, is the amount of bone area mineralized below 17.68 wt% Ca (primary mineralized bone(15) (Fig. 1).
Data were analyzed with one-way ANOVA followed by Fisher's protected least significant difference (PLSD) posthoc test for between-group comparisons. All calculations were performed using StatTView 4.5 (Abacus Concepts, Berkeley, CA, USA). Differences were considered statistically significant at p < 0.05.
The BMDD histograms and the BMDD-parameters CaMean, CaPeak, CaWidth, and CaLow determined from the BE images of the paired and triple biopsies are shown in Tables 1 and 2. Figure 1D shows typical BMDD plots of a triple biopsy from a risedronate-treated subject together with BMDD plots of a paired biopsy from a placebo-treated subject. Figure 2 shows the effect on BMDD of risedronate treatment after 3 and 5 years compared with baseline and the normal reference group.
Table Table 1.. Effects of 3-Year Treatment With Placebo and Risedronate (Paired Biopsies) on BMDD
Table Table 2.. Effects of 3- and 5-Year Treatment With Risedronate (Triple Biopsies) on BMDD
Average mineral content and most frequent Ca concentration
At baseline, CaMean of the risedronate group was reduced by 3.9% (p < 0.0001) and that of the placebo group by 5.0% (p < 0.0001) compared with the normal reference group. CaPeak of the risedronate group was reduced by 4.4% (p < 0.0001) and that of the placebo group by 5.3% (p < 0.0001). There were no statistical differences between the risedronate and placebo groups at baseline. At 3 years, there were significant increases in CaMean and CaPeak in the risedronate group (+4.2%, p < 0.001 and +3.2%, p < 0.001, respectively) and in the placebo group (+3.6%, p < 0.01 and +3.4%, p < 0.001, respectively) compared with baseline. CaMean and CaPeak increased toward values of normal. No significant differences in CaMean as well as CaPeak were detected between the risedronate and the placebo group. At 5 years, CaMean and CaPeak of the risedronate group were still significantly increased compared with baseline (+3.4%, p < 0.01 and +2.8%, p < 0.01, respectively), but were not significant different from 3-year placebo. At 5 years, CaMean and CaPeak of the placebo group with the small sample size (n = 2) appeared slightly lower compared with 3 years but was still higher than baseline.
Heterogeneity of mineralization
At baseline, CaWidth of the risedronate group was increased by 8.0% (p < 0.05) and that of the placebo group by 7.9% (p < 0.05) compared with normal controls. There was no statistical difference in CaWidth between the risedronate and placebo groups at baseline. At 3 years, (1) CaWidth decreased (−14.8%, p < 0.001) in the risedronate group compared with baseline, whereas the placebo group was unchanged; (2) CaWidth was significantly smaller (−15.2%, p < 0.001) in the risedronate group compared with placebo and was also lower than that of the normal group; and (3) CaWidth of the placebo group remained significantly increased compared with the normal group (+8.5%, p < 0.05). At 5 years, CaWidth of the risedronate group was significantly increased (+17.2%, p < 0.01) compared with 3 years but was not significantly different from baseline.
CaWidth of the placebo group with the small sample size (n = 2) seemed to be increased further after 5 years compared with 3 years as well as to baseline.
Low mineralized bone
At baseline, both the risedronate and the placebo groups had an increased amount of low mineralized bone areas (CaLow) compared with the normal group (+20.8%, p < 0.05 and +31.8%, p < 0.01, respectively). No statistical difference was detected in CaLow between the risedronate and the placebo groups at baseline. At 3 years, (1) CaLow decreased significantly in the risedronate group (−38.2%, p < 0.01) but not in the placebo group compared with baseline; (2) CaLow of the risedronate group was significantly reduced with respect to the placebo group (–28.0%, p < 0.05); and (3) CaLow of the risedronate group was lower than that of the normal group (–25.2%, p < 0.05). At 5 years, CaLow of the risedronate group showed no significant differences compared with 3 years. CaLow of the placebo group with the small sample size (n = 2) seemed to be increased at 5 years compared with 3 years but not to baseline.
In this study, we examined the effects of 3- and 5-year treatment with risedronate on the degree of mineralization and local variations/distribution of mineralization (BMDD) in iliac crest bone biopsies from women with postmenopausal osteoporosis. The BMDD values were compared with those from placebo-treated subjects and with those from the normal human reference group.(12) Interestingly, after 3 and 5 years of treatment, both the risedronate and the placebo group exhibited a significant increase in the average mineral content (CaMean), as well as in the most frequent calcium concentration (CaPeak), compared with baseline. However, the homogeneity of mineralization (CaWidth) was increased and the amount of low mineralized bone (CaLow) was reduced in the risedronate-treated group only. At baseline (i.e., untreated osteoporosis), there was a distinctly lower CaMean and CaPeak and an increased CaWidth and CaLow in risedronate- and placebo-treated subjects compared with normal values. This “hypomineralization” reflects relatively more newly formed bone matrix with a lower degree of mineralization and is in accordance with the higher bone turnover status of these subjects seen through baseline histomorphometry and biochemical markers.(21,22) Such a hypomineralization of bone in untreated osteoporotic subjects is also consistent with previous observations.(16,23)
The increase in the BMDD parameters CaMean and CaPeak after 3 years in both the placebo (calcium and vitamin D) and risedronate (combined with calcium and vitamin D) groups might indicate an inadequate intake of calcium and/or vitamin D in these subjects at baseline. The data suggest that increased intake of calcium and vitamin D to recommended levels was able to increase bone mineralization density toward normal values in these subjects. Moreover, risedronate treatment did not result in abnormal mineralization levels. It should be emphasized that, although potential “hypermineralization” caused by bisphosphonate treatment has been hypothesized, we never have observed above normal mineralization with any bisphosphonate in clinical or nonclinical studies.(16–19)
In contrast, the significant reductions in CaWidth and CaLow, which were only seen in the risedronate-treated subjects, are likely caused by the rapid and sustained antiresorptive action and suppressed bone turnover with risedronate. Already formed bone packets continue to mineralize, but fewer new bone packets are generated, resulting in less bone in the primary mineralization stage (lower mineral content) and more bone in the prolonged secondary mineralization stage (higher and more homogeneous mineral content). The reduction in CaWidth with 3-year treatment with risedronate was very similar to that found with 2- and 3-year treatment with alendronate in subjects with postmenopausal osteoporosis.(16) This seems remarkable, because at clinical dose levels, alendronate suppresses bone turnover more than does risedronate, indicated by a greater reduction in biochemical markers of bone turnover as well as a greater decrease in bone activation frequency.(22,24) Despite these differences, the two drugs produce similar reductions in fracture risk in nonhead-to-head studies.(1,2) A separate direct action of risedronate on the mineralization process itself could contribute to the observed effects and cannot be excluded.
This increase in the degree of mineralization as well as the reduction in the heterogeneity in mineralization density after 3-year risedronate treatment observed by qBEI is in agreement with the very recently reported data from quantitative synchrotron radiation μCT (SRμCT) of these same biopsies.(25) There were, however, some differences between the two analyses. Mineralization levels at baseline were not significantly different between placebo and risedronate groups with qBEI analysis, but seemed significantly higher in the placebo group with SRμCT analysis. This higher baseline level in the placebo group might be responsible for the apparent lack of effect with SRμCT analysis on mineralization with 3-year calcium and vitamin D (placebo) intake compared with the increase seen with qBEI analysis. The reason for this different detection of placebo baseline levels between the two methods of analysis remains an open question. It should be noted that SRμCT is a new and still developing method that is a completely different physically technique compared with qBEI. It is based on the absorption of a focused, monochromatic X-ray beam by blocks of bone. The information obtained from the 3D measuring technique is combined with an imaging processing method that derives the mineralization distribution from analysis of voxels (3D bone image× elements). In the discussed study, a 4-μm isotropic resolution was used, and a trabecular bone volume of 2 × 2 × 2 mm3 per biopsy was analyzed.
Beside the normalization of the mineralization density during treatment, the reduction in heterogeneity of mineralization seems to be an important and common effect of bisphosphonates on the material level of bone. Both risedronate and alendronate produced decreases in CaWidth in postmenopausal osteoporotic subjects, and it seems that this effect is a common mechanism contributing to antifracture efficacy. However, the real impact of increased homogeneity of mineralization on bone strength has not been studied directly. Experiments on crack propagation and fracture mechanism such as those recently reported(26,27) might be a promising way to determine the role of heterogeneously mineralized bone packets in bone strength.
Interestingly, changes in bone mineralization density were quite different from changes in BMD as obtained directly from the subjects of this study,(21) as well as from subjects of another risedronate study.(28) Whereas the mineralization density increased similarly both for placebo and risedronate groups, BMD increased significantly in the risedronate group only. It needs to be emphasized that the mineralization density and BMD measure physically distinct parameters. Indeed, BMD is influenced by both the mineralization density and bone volume, which may vary independently. These results taken together with the published BMD data suggest that risedronate and placebo groups differ mostly in the changes of bone volume, which are reflected in BMD but not in the degree of mineralization.
After 5-year treatment with risedronate, CaMean remained significantly elevated compared with baseline, but there was no significant difference compared with 3-year treatment and to normal controls. Interestingly, there was a significant increase in CaWidth compared with 3 years. This reversal of effects on variability of mineralization with 5-year risedronate treatment is a novel observation. It is unlikely to result from loss of the antiresorptive effect because degree of mineralization was maintained, and biochemical markers and histomorphometry data show sustained suppression of turnover.(21,22) The data taken together suggest that the effects of 5-year treatment with risedronate on bone tissue mineralization are both quantitative (increase of bone mass) and qualitative (effects at the material level including increase in primary mineralized areas). The increase in variability of mineralization may result from a slow systemic increase in newly formed bone. Whereas lumbar spine BMD continues to increase from 3 to 5 years, the degree of mineralization did not increase over this time period. The increase seen in BMD therefore could be caused by an increase in bone volume. However, histomorphometrical analysis did not show any evidence of an increase in bone formation in the biopsies.(22) Despite the fact that the third (5 year) biopsy was taken 2 cm apart from the site of the baseline biopsy (as per protocol), it is possible that there was a residual influence of bone healing processes on the third biopsy. However, no evidence of an ongoing repair process was detected with histomorphometrical analysis.(22) Further studies are required to discern the mechanism(s) responsible for this time dependent change in mineralization pattern with 5-year risedronate treatment.
In conclusion, in this first triple biopsy study in postmenopausal osteoporotic women, 3- and 5-year treatment with risedronate increased the degree and homogeneity of mineralization without producing hypermineralized bone. These changes at the material level of the bone matrix, together with previously published effects on BMD, bone turnover, and microarchitecture, may contribute to risedronate's antifracture efficacy in osteoporotic patients.
The authors thank Gerda Dinst, Sabrina Thon, and Daniela Gabriel for careful sample preparations and qBEI measurements at the Bone Material Laboratory of the Ludwig Boltzmann-Institute of Osteology, Vienna, Austria. This study was supported by the AUVA (Austrian Social Insurance for Occupational Risk), the WGKK (Social Health Insurance Vienna), and the FWF (The Austrian Science Fund) Project P16880-B13.