Methodological considerations in measurement of bone mineral content

Introduction

The strength of bones depends on bone matrix volume, bone microarchitecture, and on the degree of mineralization of bone (DMB), and we have recently shown in patients with osteoporosis treated with alendronate that fracture risk and bone mineralization density (BMD) were changed without modifications of bone matrix volume or bone microarchitecture [1]. Thus, DMB must not be forgotten among the factors determining the mechanical competence of bone.

How to measure qualitative and quantitative differences in the mineral content of bone samples?

Bone mineral content depends on BMD and bone matrix volume. BMD can be quantified by noninvasive methods allowing for a global evaluation of mineral substance of the bone as an organ [dual x-ray absorptiometry (DXA), tomodensitometry, quantitative ultrasound]. Conversely, invasive methods quantify the mineral substance of the bone tissue alone. The 2 methods provide very different information; however, only the latter method will be discussed here. The first systematic study of bone mineral content was made on vertebral trabecular bone [2], and the values were derived by water displacement, using Archimedes' principle. Different chemical dosages can also be performed to determine the main composition of bone tissue. However, these methods destroy the bone samples, which cannot then be used for other measurements. Furthermore, such measurements are always global, and it is not possible to separate values corresponding to the different structures of the bone tissue. Neither BMD at organ level nor evaluation of bone volume gives information on the BMD at the tissue level or better at the intermediary level of organization of bone (cortical osteons or trabecular packets).

Among invasive methods (Table 1) preserving tissue structure, bone histomorphometry gives access to the direct measurement of the amount of bone matrix in a given volume of bone biopsy specimen, in particular through measurements of cancellous or total bone volume expressed in percentage of either spongy bone tissue or core volume. Bone histomorphometry is performed on undecalcified transiliac bone biopsy specimens embedded in methyl methacrylate. This method allows measurement of static parameters reflecting the bone structure and microarchitecture in cancellous and compact bone, the bone remodeling (resorption and formation), and dynamic parameters as the rate of osteoblast activity using double tetracycline labeling. It is the only method suited to evaluate tissue and cell changes at the level of the intermediary organization of bone, that is, the basic structural unit (BSU), which is either the osteon in cortical bone or the cancellous bone packet in spongy bone. Bone histomorphometry allows a good discrimination of completely unmineralized matrix; however, this method cannot provide information on the individual DMB of each BSU. DMB is commonly measured by x-ray attenuation experiments (Table 1). Most studies have used contact microradiography, which gives a high linear resolution. Quantitative microradiography using a computerized microdensitometric method allows measurement of the focal DMB of each BSU within the limits imposed by the thickness of the section [3, 4, 5]. Microscopic mineral variations and mineral density distributions have also been evaluated by quantitative backscattered electron imaging (qBEI) [6, 7, 8, 9, 10, 11, 12]. Recently, DMB has also been measured using synchrotron radiation microtomography [13]. Finally, the structure of the mineral substance at the nanometer level (shape, typical predominant orientation, and average size of crystals) can be investigated by small-angle x-ray scattering [14, 15] and by Fourier transform infrared microspectroscopy (FTIRM) [16, 17].

Table 1. Bone mineral content depends on both the bone mineral density at tissue level (DMB) and the amount of bone. It can be measured using different invasive methods (DMB degree of mineralized of bone, qBEI quantitative backscattered electron imaging, SRμCT synchrotron radiation microtomography)

How to interpret the measurement of the degree of mineralization of bone?

Bone is a composite material composed of an organic phase synthesized by the osteoblasts and of an inorganic phase composed mainly of a calcium phosphate crystallized as a nonstoichiometric apatite. The more protein solid that is present, the less degree of mineralization that can be reached; thus, a decreasing degree of mineralization is observed in various adult calcified tissues with enamel>dentine>calcified cartilage>woven bone>lamellar bone. Lamellar bone has the highest volume occupancy by collagen fibrils and is the least mineralized of the normal calcified tissues. During bone remodeling and after bone resorption, bone formation is a multistep process [4]. First, the organic matrix (osteoid) is synthesized and laid down at specific sites. Following this, the new matrix begins to mineralize after about 5–10 days from the time of deposition, and the rate of this 'primary mineral apposition' can be measured directly in vivo using double tetracycline labeling. After full completion of the BSU, a secondary mineralization begins (Fig. 1). This process consists of a slow and gradual maturation of the mineral component, including an increase in the number of crystals, a moderate augmentation of crystal size toward their maximum and, to a greater extent, changes in the internal order of the crystals reflecting their degree of perfection [18]. Such changes are illustrated on electron micrographs showing mineral substance at different steps of the mineralization process (Fig. 2) or in conditions known to reduce (primary hyperparathyroidism) or to more or less prolong (osteopetrosis, osteomalacia) secondary mineralization.

Fig. 1.

The different steps of the mineralization of bone tissue (DMB degree of mineralization of bone)

Fig. 2.

Electron micrographs illustrating the distribution of mineral substance during the mineralization of iliac bone tissue. ×10,000

What are the main methodologies used to measure DMB at tissue level?

Contact quantitative microradiography (Fig. 3) using a computerized microdensitometric method has been developed by our team and allows both the measurement of the mean DMB and the observation of the distribution of the different DMB [1, 3, 5]. DMB is quantitatively evaluated by first exposing an aluminum calibration step-wedge and a plane-parallel calcified tissue section (100±1 µm thick) simultaneously to the same beam of x-rays, then determining from the resulting microradiograph the thickness of aluminum that produces the same x-ray absorption as a given region of the bone tissue section. We used custom-developed software for the automatic analysis of gray levels of microradiographs with Visiolab 1000 (Biocom, France), a true color image processing workstation operating under Microsoft Windows (Redmond, Washington). Each field analyzed is divided into 4350 square measurement units (about 100 pixels each). Data are converted from gray-level values in DMB with the construction of a calibration curve based on the measurements obtained on the aluminum step-wedge. DMB is finally expressed in g mineral/cm³ bone, adjusted according to the precise thickness of each part of the section measured before microradiography. The coefficient of variation (intraobserver) for the quantitative microradiography technique is 3% and the interobserver variance is 7%. Irrespective of the bone structure analyzed (compact, cancellous, or total bone), interindividual variations for mean DMB were less than 10% in controls and 5% to 7% in placebo- and alendronate-treated osteoporotic women.

Fig. 3.

Microradiograph showing the diversity of the degrees of mineralization of iliac bone from a male control (age 86 years)

The mineral content of bone samples has also been evaluated [10, 11] by qBEI. Carbon and aluminum were used for gray-level references and osteoid and hydroxyapatite were employed as references to convert gray-level values into calcium weight percentage values. The intensity of the backscattered electron signal from the sample is directly proportional to the bone calcium concentration and can therefore be used for the generation of bone mineralization density distribution (BMDD). BMDDs display the frequency of certain calcium concentrations and are analyzed for the weighted mean calcium concentration (Ca mean), the most frequent calcium concentration (Ca peak), and the homogeneity of mineralization (Ca width). The BMDD of trabecular bone from healthy adult individuals was shown to be nearly constant over several biological factors (gender, age, ethnicity, skeletal site). Technical and biological variations (intra- and interassay) showed that it is a method highly sensitive and precise for subtle changes in mineralization. Interindividual variation was less than 3% for Ca mean. Using the same method [7, 8, 12], DMB expressed as weighted mean gray levels were strongly correlated with ash percentage measurements. Mineral content has also been grouped into 4 density fractions (low, medium, high, and very high), and the results were expressed as bone density profiles showing the distribution of bone within the fractions [6, 9].

Recently, we have tested a synchrotron radiation microtomography (SRµCT) method [13]. The availability of a three-dimensional (3D) measuring technique coupled to a specific image-processing method opens new possibilities. SRµCT may provide 3D images with spatial resolution as high as 1 µm. The acquisition of 3D bone sample images at high spatial resolution using SRµCT has proved to be very accurate for quantifying human bone microarchitecture. Moreover, the use of a monoenergetic synchrotron beam, which avoids beam-hardening effects, allows quantitative measurements of DMB on 3D images. SRµCT is a nondestructive, fast, and precise procedure to determine the DMB in 3D simultaneously to the microarchitecture. Results were compared from the same bone slice imaged by the 2 techniques (SRµCT and microradiography) to confirm this. Because x-rays are used in both cases, the major difference is that SRµCT is a reconstructed 3D image, whereas microradiography is directly acquired. The main limitation of the SRµCT is the access to SR facilities, although new SRµCT are in development in different countries. The DMB was evaluated from the distributions of the 2 different techniques, respectively, in compact, cancellous, and total bone. The 2 distributions of DMB are well overlaid and exhibit the same range of variation (0.5–1.6 g mineral/cm³) of the DMB values. The mean difference between the 2 techniques was around 4% and was slightly higher in cancellous than in compact bone.

Can the measurements of DMB at tissue level distinguish between bone conditions and treatments?

The main results obtained in different conditions and/or after the use of various therapeutic agents correspond to our measurements with quantitative microradiography. When known, results reported with the other methods will be mentioned. First, control values were measured [5] in 43 iliac bone samples taken at necropsy from 30 women and 13 men who died suddenly, showing no apparent bone disorder. In terms of mean values, distribution, and evolution with age, the DMB was not significantly different between the 2 sexes. The mean [± standard error of the mean (SEM)] DMB expressed in g mineral/cm³ was 1.08±0.02 in compact bone, 1.10±0.02 in cancellous bone, and 1.09±0.02 in total bone (compact + cancellous). In these 3 bone structures, DMB did not change significantly with age [5, 19]. The distributions of DMB revealed a small shift toward the high values in cancellous bone compared with compact bone (DMB maximums were 1.10 and 1.05 g/cm³, respectively, and the index of homogeneity readings were 0.28 and 0.34 g/cm³, respectively) [19]. This may reflect either the presence in cancellous bone of a higher proportion of interstitial bone than in compact bone or an "edge effect" (100-µm-thick sections), with suppression of the measurements in the lowest mineralized part of trabeculae.

In adult bone, the main biological determinant of mineralization is the rate of turnover. Thus, in our model (Fig. 4), any agent (e.g., parathyroid hormone, PTH) or event (menopause, ovariectomy) that provokes an augmentation in the "birth rate" or activation frequency of basic multicellular unit (BMU) induces a decrease of the "lifespan" of BSU; in other words, in the time available for the secondary mineralization. This leads to the fact that new BSUs are resorbed before they have fully completed their secondary mineralization, as proved by the presence of a large amount of incompletely mineralized BSU and a low mean DMB. In primary hyperparathyroidism (Figs. 5, 6), a shift of DMB toward the low values was observed (mean 0.92±0.07 g/cm³, DMB maximum 0.90) but the index of homogeneity was slightly decreased (0.30 g/cm³). In patients treated with PTH, the decreased DMB was confirmed but with an increased heterogeneity of the values [20].

Fig. 4.

Model illustrating the changes in the DMB due to the modifications of the bone remodeling rate (ALN alendronate, BP bisphosphonates, SERMs selective estrogen receptor modulators, OVX ovarectomized, PTH parathyroid hormone, BMU basic multicellular unit, BSU basic structural unit)

Fig. 5.

Microradiograph showing the increase in the number of BSU having a low degree of mineralization in a patient suffering from primary hyperparathyroidism (male, 42 years) (BSU basic structural unit)

Fig. 6.

Distribution of DMB in patients with primary hyperparathyroidism compared to a control group (HPT hyperparathyroidism)

Conversely, antiresorptive agents [bisphosphonates, estrogen, selective estrogen receptor modulators (SERMs)] that cause a marked reduction in the "birth rate" of BMU prolong the "lifespan" of the BSU, allowing a more complete secondary mineralization. This should finally provoke an increase of DMB. Recently, mean DMB was measured by quantitative microradiography [1] on transiliac bone biopsy specimens taken from 53 postmenopausal osteoporotic women who had been treated with alendronate (10 mg/day) for 2 or 3 years or with placebo. In the same patients, BMD values were obtained by DXA at the lumbar spine level. After 2 years of alendronate, mean DMB in total bone was 7.5% (P=0.0026) higher than with placebo treatment. After 3 years of alendronate, mean DMB in total bone was 10.7% (P=0.0001) higher than placebo. After alendronate, the distribution of the DMB in total bone clearly showed a shift toward the highest mineralization values concomitantly with a decrease in the number of bone structure units having low values of mineralization. The between-group differences in mean DMB were similar to those of BMD at the lumbar spine level (+8.7% after 2 years and +9.6% after 3 years, respectively), suggesting that mean DMB augmentation probably accounts for the major part of the increase in BMD seen with alendronate. After 2 and 3 years of alendronate, the index of homogeneity was smaller than after placebo in compact, cancellous, and total bone. This illustrates the fact that alendronate, by allowing the secondary mineralization to be more complete, aids the homogeneity of the mineralization. Once again, the increase in DMB and the homogenization of the mineralization do not mean that a hypermineralization occurs. These results have been confirmed using qBEI in a subset of our 53 patients [21]. Similar results have been reported in patients suffering from steroid-induced osteoporosis and treated with alendronate. Similar data have also been reported after etidronate treatment for 1 and 2 years. DMB (compared to baseline value) was increased in compact (+4%, then +12%) and in cancellous (+1.3%, then +8%) bone [22]. Treatments with other bisphosphonates, such as risedronate and zoledronate, also revealed decreases of vertebral fractures and augmentation of BMD without significant changes of bone mass and microarchitecture but marked reduction of activation frequency. Even if DMB at tissue level was not yet measured in the bone biopsy samples of the corresponding patients, the explanation could be similar to 1 of the effects of alendronate.

Another antiresorptive agent (the SERM raloxifene) has also been tested [23], even though its action on the decrease of remodeling rate is less potent than alendronate. Raloxifene induces a mild increase of BMD, a moderate decrease of the biochemical markers of bone turnover, and decreases the risk of vertebral fractures in postmenopausal women. In the absence of significant changes in bone mass and microarchitecture, the modifications of BMD could be the reflection of mild modifications in the DMB (Boivin et al., Contribution of raloxifene and calcium and vitamin D3 supplementation to the increase of the degree of mineralization of bone in postmenopausal women; submitted).

Finally, strontium ranelate, a new orally effective and safe treatment of vertebral osteoporosis, has been shown to decrease vertebral fractures and increase BMD. The unique mechanism of action of strontium ranelate is to decrease bone resorption and increase bone formation (decoupling agent). The effects at the bone tissue level are now being investigated in animals and in humans, and the preliminary observations show no modifications of the DMB, suggesting changes of the bone volume (Boivin et al., Strontium deposition and degree of mineralization of bone in post-menopausal osteoporotic women treated with S12911; submitted).

The measurement of the DMB offers insights on the relative effectiveness of different antiresorptive osteoporosis therapies. The greater the antiresorptive effect, the higher the DMB will be, due to the prolongation of the secondary mineralization. For example, alendronate appears more potent than etidronate and raloxifene. Concerning anabolic treatments, the expected mild decrease of DMB is under investigation using teriparatide and strontium ranelate.

The reduction of the fracture rate is not always due to an increase of DMB, as shown with antiresorptive agents. The amount of bone tissue and the microarchitecture must also be taken into account. When a bone-forming agent (i.e., teriparatide) induces an increase in the amount of bone matrix and better connectivity in the trabecular network, these changes may induce a reduction in the fracture risk in spite of a DMB decrease. The effects of these parameters on the biomechanical properties of bone tissue still need to be investigated.

The optimal change in bone turnover that allows an optimal modification of DMB and leads to the most effective decrease in fracture risk is not known. The combination of different therapies appears theoretically a good option, but to date we have no data to support such a treatment or knowledge of the types of drugs and sequence of therapies that would be optimal.

Conclusions

Changes in bone remodeling activity directly influence the degree of mineralization of bone. This may explain the modifications in fracture incidence, the increase in the BMD, and in bone strength without changes necessarily being seen in the bone matrix volume and bone microarchitecture. This new approach to measuring the determinants of bone strength and BMD and the results of our recent studies as well as others now in progress, using antiresorptive (estrogens, raloxifene) and decoupling (strontium ranelate) agents used in the treatment of osteoporosis, emphasize that bone mineral substance is an important factor to be taken into account in the pathophysiology of osteoporosis and other bone conditions. It is important to rediscover the mineral dimension of bone, which has been forgotten for many years.

Discussion

Dr. Bouxsein: The missing link seems to be that, whereas you both have presented observational studies demonstrating treatment effects, there hasn't been a direct link between changes either in mineral crystal distribution or size and mechanical behavior.

Dr. Schaffler: We have to introduce some other mechanical definitions besides stiffness and strength. Dr. Boskey alluded to brittleness and fracture resistance, but there is something else that governs the way bone breaks. For example, what happens after yield, the so-called postyield region or the fracture toughness? Both Drs. Boskey and Boivin have shown evidence that the mean degree of mineralization increases with bisphosphonates. In addition, because remodeling is shut off, you lose heterogeneity in both mineralization and crystal structure. From the standpoint not of stiffness or strength, but from the standpoint of resisting cracks propagating through material, this loss of heterogeneity is a bad thing. Heterogeneity, which is what nature puts into bone, is a good thing because interfaces stop the propagation of cracks.

Dr. Boskey: In fact, additional homogeneity also occurs with regard to collagen maturity. We see the same sort of narrowing of distribution with regard to not only the amount of collagen but also the comparative number of mature and immature cross-links and how the cross-links are distributed. Moreover, there is a trend for this in osteoporosis patients in general. With regard to the issues raised by Dr. Bouxsein and Dr. Schaffler, the challenge is to prove that a spatial distribution relates to a change in mechanical behavior. Currently, the best that we can do is to perform nano-indentation, which has its own limitations.

Dr. Schaffler: Nano-indentation is a stiffness-type technique and the things that we're talking about have to do with fracture and what happens after things start to break. Therefore, these testing techniques need to be a destructive test to assess these mechanisms.

Dr. Boskey: I agree. To test this, we're going to have to use analogs where you have the same mineral content but differences in crystal size within the same collagen matrix.

Dr. Boivin: I would like to make the point that we have never observed hypermineralization after bisphosphonates. However, I agree with you that too much homogeneity of the mineral is probably bad, but I have no data to know if it is a very big increase in the homogeneity. We have a small increase of a parameter that reflects homogeneity of the mineralization, but this response is expected if you prolong the secondary mineralization. Under these circumstances, you have more BSU with, if not a maximum, a high degree of mineralization. I do not think we have a pathological state of the bone after treatment; it is an expected consequence of treatment with bisphosphonates. The loss of heterogeneity is not so great that it would be perhaps negative in terms of biomechanical property.

Dr. Burr: Ultimately, the only way to determine whether these things are really bad or not is by relating the mean degree of mineralization to some mechanical property. A classical fracture mechanics approach would be appropriate. Take a small piece of bone, notch it, and you look at how easy it is to grow a crack in it and you can calculate fracture toughness. That will tell you whether that material is essentially more brittle or not, and whether it's likely that cracks will grow faster in it.

Dr. Ferrari: How much variance is there in this crystallinity and mineralization degree, and how does this variance relate to fracture risk in the population? Moreover, do we actually know about the variance of these parameters in men and women?

Dr. Dempster: Perhaps I can address that. We have recently completed a study in collaboration with Klaus Klaushofer and Paul Roschger from Vienna. We studied a number of skeletal sites, a wide range of age, men and women, and different races. We found that the mineralization density distribution was very similar across all of the samples. The degree of mineralization seems to be a biological constant. This suggests the subtle changes that Dr. Boivin has presented could be important, but I agree that how you prove this is very difficult.

Dr. Majumdar: I have 2 questions. Do the biopsies differ with regard to the distribution of rods and plates? And second, does the mean degree of mineralization and crystallinity distribution differ between rods and plates?

Dr. Boskey: This is a good question, but we do not have an answer at this time.

Dr. Jepsen: A question about the changes you're reporting in terms of either mineral crystallinity or distribution of mineral: How do those show up in BMC and BMD? For example, if you had 2 materials that had identical architectures but different crystallinity and/or differences in distribution, would their BMC be different?

Dr. Boskey: We can get different distributions for the same BMD.

Dr. Majumdar: Although BMC and BMD are good tools for the physicians, they would not be likely to detect these subtle changes.