Bone composition: relationship to bone fragility and antiosteoporotic drug effects

Abstract

The composition of a bone can be described in terms of the mineral phase, hydroxyapatite, the organic phase, which consists of collagen type I, noncollagenous proteins, other components and water. The relative proportions of these various components vary with age, site, gender, disease and treatment. Any drug therapy could change the composition of a bone. This review, however, will only address those pharmaceuticals used to treat or prevent diseases of bone: fragility fractures in particular, and the way they can alter the composition. As bone is a heterogeneous tissue, its composition must be discussed in terms of the chemical makeup, properties of its chemical constituents and their distributions in the ever-changing bone matrix. Emphasis, in this review, is placed on changes in composition as a function of age and various diseases of bone, particularly osteoporosis. It is suggested that while some of the antiosteoporotic drugs can and do modify composition, their positive effects on bone strength may be balanced by negative ones.

Introduction

Bone is a heterogeneous composite material consisting, in decreasing order, of a mineral phase, hydroxyapatite (Ca₁₀(PO₄)₆(OH)₂) (analogous to geologic ‘hydroxyapatite'),1 an organic phase (∼90% type I collagen, ∼5% noncollagenous proteins (NCPs), ∼2% lipids by weight)2 and water. Proteins in the extracellular matrix of bone can also be divided as follows: (a) structural proteins (collagen and fibronectin) and (b) proteins with specialized functions, such as those that (i) regulate collagen fibril diameter, (ii) serve as signaling molecules, (iii) serve as growth factors, (iv) serve as enzymes and (v) have other functions. The relative amount of each of these constituents present in a given bone varies with age,3 site,4 gender,5 ethnicity6 and health status.7 The amount, proper arrangement and characteristics of each of these components (quantity and quality) define the properties of bone. The tendency of bones to fracture depends on the quantity of mineralized tissue present (size and density) often measured by clinicians as bone mineral density or BMD8 and several other factors, grouped together as ‘bone quality'.8,9 ‘Bone quality' factors include composition (weight percent of each component), mineralization (organization of the mineral and its crystallite size and perfection), collagen content and collagen crosslinks, morphology,10 microarchitecture11 and the presence of microcracks.12 Each of these factors varies with health, disease and drug therapies. Their distribution in the heterogeneous tissue also varies with these perturbations. The focus of this review will be on the composition of bone and its site-specific variation. Materials present, their characteristics and their distribution will be discussed here. Readers are referred to the references above for more information on morphology, microarchitecture and the presence of microcracks, which will not be discussed.

Bone mineral

Hydroxyapatite is the principal component of the mineral phase of bone. This was demonstrated more than 60 years ago using X-ray diffraction, now viewed as the ‘gold standard' for such determinations.1 The quantity of mineral present in bone can be determined by a variety of techniques13 including gravimetric analyses (ash weight determination), analysis of calcium and phosphate contents, spectroscopic and densitometric analyses including bone mineral density distribution (BMDD), bone mineral density (BMD) and micro-computed tomography (micro-CT). Such methods show that the mineral content of bone ranges from ∼30%/dry weight (in the skate or ray appendicular skeletal element, the propterygium) to 98%/dry weight in the stapes of the human ear. Most bones have ∼60–70% mineral/dry weight, depending upon site, species and stage of development (Figure 1).13,14

Figure 1.

Bone composition. Ternary diagram illustrating the composition of mature bones in different species (adapted from Currey J.D. ‘Bones'. Princeton University Press: Princeton, NJ, 2002, p 436.) and reproduced with permission from Journal of Experimental Biologists, Figure 7B in ‘Comparison of structural, architectural and mechanical aspects of cellular and acellular bone in two teleost fish'.14 The speckled circle is the average of data from five iliac crest biopsies of females aged 69–75 years.

Variation in the distribution of mineral and its properties in bone can be illustrated by a variety of imaging techniques, discussed here, including BMDD, Raman and infrared spectroscopic imaging. It can also be determined by microprobe or synchrotron radiation-induced micro-X-ray fluorescence elemental analysis and mapping15 including trace elements such as strontium, aluminum, zinc or lead. In contrast, backscattered electron imaging in the scanning electron microscope is highly sensitive to the average atomic number of the bone material that is dominated by calcium. This technique is not a tool to identify specific elements in bone. Quantitative backscattered electron imaging is used for mapping the calcium concentrations and for the determination of bone mineralization density distribution (frequency distribution of Ca concentrations within the bone sample, BMDD; Figure 2).16 Parameters obtained from BMDD include the average and mode Ca content and the full-width at half-maximum of the BMDD peak, which is a measure of the heterogeneity of mineralization. Deviations from normal calcium distributions have been reported to date in: osteomalacia,17 osteoporosis18 and idiopathic osteoporosis19 (peak shifted to the left of normal), classical and new forms of osteogenesis imperfecta16,20 (peak shifted to the right of normal) and treatment with some but not all bisphosphonates examined by this technique.18,21,22

Figure 2.

BMDD distribution of bone. Measurement of bone mineralization density distribution (BMDD) using quantitative backscattered electron imaging (qBEI) in a transiliac bone biopsy sample (left insert) from a 39-year-old women with coeliac disease (PATIENT). The qBEI image shows a detail of the cancellous bone compartment. Bright gray levels mean high and dark gray levels mean low mineral content. The trabecular features display severe under mineralization and mineralization defects. Thus, the BMDD curve is extremely shifted to the left toward lower matrix mineralization (decreased CaMean, CaPeak, CaHigh and increased CaLow) compared with the normal reference BMDD (Ref.). In addition, the peak width of the BMDD (CaWidth) is distinctly enlarged, indicating an enormous heterogeneity in mineralization of the bone matrix. Examinations by histological staining techniques revealed a severe osteomalacia combined with signs of secondary hyperparathyroidism. Description of the individual BMDD indices derived from the BMDD curve: *CaMean, the weighted mean calcium concentration of the bone area obtained by the integrated area under the BMDD curve; *CaPeak, the peak position of the histogram, which indicates the most frequently occurring calcium concentration (mode calcium concentration); *CaWidth, the full-width at half-maximum of the distribution, describing the variation in mineralization density (heterogeneity); *CaLow, the percentage of bone area that is mineralized below the 5th percentile of the reference BMDD of normal adults, that is, below 17.68 weight percent calcium. This parameter corresponds normally to the amount of bone area undergoing primary mineralization. *CaHigh, the percentage of bone area that is mineralized above the 95th percentile of the reference BMDD of normal adults that is above 25.30 weight percent calcium. This parameter corresponds to bone matrix having achieved plateau level of normal mineralization. Figures courtesy of P Roschger.

Variation in phosphate distribution is visualized by both Fourier transform infrared microscopic imaging (FTIRI)23 and Raman microscopy and imaging (Raman).24,25 These types of vibrational spectroscopic imaging describe the distribution of any elemental pair or larger moiety that vibrates when excited by incident light. Relevant vibrations for bone are those in phosphate, protein and lipid groups. The precise location of the vibrations, often given in wave numbers or reciprocal wavelength, reflects the molecular environment in which the vibrating ions are found. In addition, as with BMDD, the line width at half-maximum of any of the broadened peaks indicates the heterogeneity (number of pixels with different values in the section analyzed) for that particular vibration. The spatial resolution of the FTIRI experiment, unless synchrotron radiation is used, is ∼7 μm. Raman spectroscopy, in contrast, has a spatial resolution of ∼1 μm. Raman spectral data is not affected by the presence of water, making the analysis of non-dehydrated samples, not possible for FTIRI, yet possible using Raman. Vibrations that are strong in the infrared spectra are weak in Raman spectra, and vice versa. These two are complimentary techniques, providing information on both the mineral phosphate and the organic matrix distribution in tissues (Figures 3, 4, 5). The data from these images can be presented as ‘chemical photographs' (as in Figures 3 and 5) or as numerical averages (Figure 6). Data in Figure 6 compares the composition of cancellous bone in patients treated with bisphosphonate followed by 1 year of teriparatide (PTH) treatment, as determined by Raman spectroscopy. The lower half of Figure 6 shows baseline values as determined by FTIRI in an on-going study of female patients with and without fragility fractures. The mean heterogeneity for each of these parameters is also shown. An important observation is that the fracture cases had reduced heterogeneities relative to unfractured controls. We believe (discussed later) that this loss of compositional heterogeneity may allow microcrack propagation resulting in a weakened, more brittle tissue as the microcracks spread. Whether this change in heterogeneity is due to oversuppression or other mechanisms is yet to be determined.

Figure 3.

FTIRI images. Typical FTIR images of mineral compositional parameters in normal and osteoporotic bone trabeculae from iliac crest biopsies of a woman with fractures and her age-matched control. Left side, normal; right side, osteoporotic (fractured). Images are shown for: mineral/matrix ratio, carbonate/phosphate ratio, crystallinity (1030/1020, cm⁻¹) and acid phosphate substitution. The arrow indicates the size scale for all images. The color scales for each parameter for both normal and fractured images are shown.

Figure 4.

Raman analysis. (a) Image as seen under the Raman microscope showing the double tetracycline labels used to define newly formed trabecular bone, the three areas examined are shown. (b) Typical Raman spectra of bone from a premenopausal woman and a woman with osteoporosis. (c) Mean values for the indicated parameters calculated from the Raman spectra of these two patients. Figure courtesy of Dr Lefteris Paschalis.

Figure 5.

FTIR image of the organic matrix in osteoporosis. Typical cortical bone image of collagen maturity (ratio of intensity of 1660/1690, cm⁻¹) in iliac crest biopsies from age-matched women without and with a fracture. The magnification is indicated by the arrow. The infrared spectra corresponding to the pixel at the tip of the arrow is shown. The major peaks used in the analysis of mineral and matrix are indicated in this figure.

Figure 6.

Compositional changes in bone as detected by Raman and FTIRI. Top row of figures show Raman compositional data indicating changes from baseline in newly formed trabecular bone caused by 1 year treatment with teriparatide (TP) following longer treatment with alendronate (ALN) or risedronate (RIS). Differences between baseline and TP treatment are indicated. Mineral-to-matrix, inverse of crystallinity and proteoglycan-to-mineral ratios are shown (adapted with permission from Figures 2a, 3 and 4 in Gamsjaeger et al.51. Lower row shows FTIRI lack of compositional changes in iliac crest biopsies from a group of 48 women, half with fractures and half without who had comparable hip BMD values. The first figure shows mean and standard deviation in cancellous bone. The second figure illustrates the mean and standard deviation of the heterogeneity of these parameters. *P<0.05.

Using FTIRI spectroscopy, biopsies from patients with low-energy (fragility) fractures were found to exhibit differences in their mineral composition, relative to the same tissue in fracture-free controls of similar age and sex. Using these FTIRI analyses of 54 iliac crest biopsies from patients ranging in age from 30 to 84 years, with and without fragility fractures,26 models were constructed to determine which FTIRI parameters were associated with fracture. The mineral parameters significantly associated with fracture in the constructed model were high cortical mineral/matrix ratio (increased fracture risk) and high cancellous crystallinity (increased fracture risk). Carbonate-to-phosphate ratio was increased in both areas, however, not significantly. Raman analysis performed on femoral cancellous bone near the fracture site of women with fractures similarly demonstrated a higher carbonate/amide I area ratio than in those without fractures. Iliac crest biopsies in those fractured patients also revealed a higher carbonate/phosphate ratio in cortical bone samples of women with fractures.27 These compositional changes and the loss of heterogeneity reflect the persistence of older more mature bone (increased mineral/matrix ratio and carbonate/phosphate ratio and a lower acid phosphate substitution) along with the absence of new bone formation.

Changes in mineral composition also occur in other bone diseases associated with increased fracture risk in addition to osteoporosis. In most types of OI (or brittle bone disease), as in osteoporosis, the mineral content (mineral/matrix ratio) is increased.28 In osteoporosis, this increase is attributable to the lack of osteoid associated with increased remodeling and decreased bone formation, whereas in OI there is a lesser amount of collagen because OI patients make an improper collagen matrix.28 In osteomalacia, the mineral/matrix ratio is unchanged when mineralized tissue alone is examined.29 Crystallinity (measured based on different vibrations in FTIRI and Raman30) generally shows parallel trends. These trends, reflecting the changes in mineral properties in different diseases, are summarized in Table 1 along with other compositional parameters measured by FTIRI and Raman. Carbonate/phosphate ratio, based on the same vibrations in both techniques varies similarly. The extent of acid phosphate substitution31 varies inversely with crystallinity. For example, in chronic kidney disease, abnormalities in phosphate transport and clearance lead to osteoporosis and increased crystallinity when bone turnover is high, with no significant changes occurring when bone turnover is low.32

Table 1. Mineral property distribution in diseased human bone revealed by FTIRI and Raman analysis.

Parameter/condition	Compared with healthy age-matched similar site
	Cortical				Cancellous
	Min/matrix	CO3/P	XST	HPO4	Min/matrix	CO3/P	XST	HPO4
Aging52 primates	Inc	Inc	Inc	Dec	Inc	Inc	Inc	Dec
Osteogenesis imperfecta53,54,a
Type I	NC	Dec	Dec	NA	NC	Dec	Dec	NA
Type III
Type IV
Type VII	Inc	NC	Dec	NA	Inc	NC	Dec	NA
Type VIII	Inc	NC	Dec	NA	Inc	NC	Dec	NA
Osteomalaciab	NC	NA	NC	NA	NC	NA	NC	NA
Osteoporosisc	Inc	NC	Inc	NA	NC	NC	Inc	NA
Renal osteodystrophy32,55	NC	NC	NC	NA	Dec	NC	NC/INC	NA

Abbreviations: CO₃/P, carbonate-to-phosphate ratio; Dec, decreased relative to appropriate control; HPO₄, acid phosphate substitution; Inc, increased relative to appropriate control; Min/mat, mineral-to-matrix ratio; NA, not measured; NC, unchanged relative to appropriate control. XST=crystallinity.

^aA Boskey, E Carter, CL Raggio, unpublished data,

^bSee Faibish et al.29

^cSee Gourion-Arsiquaud et al.26,35

Bone matrix

Protein composition of bone was classically determined following demineralization of the tissue and isolation and characterization of the component proteins. Collagen, predominantly type I, accounted for the majority of the matrix, but other proteins, the so-called NCPs, accounted for ∼5% of the total bone weight. The major components of the NCPs were identified as belonging to the SIBLING (small integrin-binding N-glycosylated), SLRP (small leucine-rich proteoglycans), GLA protein (γ-carboxyglutamic acid protein) and CCN protein (small secreted cysteine-rich protein) families. Today, using proteomics and gene expression profiling, it is known that there are thousands of proteins in the bone matrix, some of which (based on whole-genome analysis) have been associated with changes in BMD, but most are yet to be identified and their functions determined. Major structural proteins and NCPs will be reviewed here.

Structural proteins

Collagen

The most abundant protein in the bone matrix is type I collagen, a unique triple helical molecule consisting of two identical amino-acid chains and one that is different. The collagen molecules that consist of repeating glycine–X–Y residues are often hydroxylated and glycosylated. This gives rise to some of collagen's unique crosslinking ability, in turn, making the collagen lattice ideal for its functions. These include: providing elasticity to the tissues, stabilizing the extracellular matrix, supporting or templating initial mineral deposition and binding other macromolecules. In different types of OI, mutations in the collagen genes are reflected in the inability of the OI bones to mineralize properly. Reviewed elsewhere,28 the origin of the mineralization defect is unknown. This defect may be in the altered structure of the collagen itself, or in the inability of extracellular NCPs, which regulate the mineralization process, to bind to the defective collagen, and hence regulate mineralization.

Chemical analyses of the crosslinks in bone collagen have demonstrated two types of crosslinks, those formed enzymatically and those that occur by glycation.33 Both types of crosslinks (enzymatic and non-enzymatic) increase with age, are altered in disease (Table 2) and affect the mechanical strength of the collagenous matrix. Enzymatic crosslinks are believed to enhance mechanical strength, whereas the advanced glycosylation end-products, which are elevated in uncontrolled diabetics and in oxidative stress, make for a more brittle bone. Distribution of total crosslinks in the bone matrix can be visualized in FTIRI data (Figure 5). Such analyses agree with the high-performance liquid chromatography chemical analysis of crosslinks.34 The number of these crosslinks, which reflect collagen maturity, is increased in osteoporotic individuals, especially in the center of the remaining cancellous bone. The heterogeneity of this distribution is decreased in consequence of age, osteoporosis and treatment with bisphosphonates.35

Table 2. Variation in collagen crosslinking in human bones with age, disease and therapy (relative to control values).

	FTIRI collagen maturity	Enzymatic crosslinks	AGEs
Agea	Increasesb	Increases56	Increases in cortical bone56
Osteogenesis imperfectac	Increasedb	No change57	No change57
Osteomalaciac	No changed	ND	ND
Renal osteodystrophyc	No change55	Increased58 (rats)	Increased58 (rats)
Osteoporosisc	Increased59	Increasede	Increasede
+Alendronatef	No change60,61	No change (dogs)62	Decreased (dogs)62
+Risedronatef	No change (dogs)61	No change (dogs)62	Decreased (dogs)62
+Zoledronatef	No changeb	ND	ND
+PTHf	No changeb	Increased (monkeys)63	Decreased (monkeys)63
+Estrogenf	Increased64	ND	ND
+Raloxifene (SERM)f	No change	ND	Decreased

Abbreviations: AGE, advanced glycosylation end-products; FTIRI, Fourier transform infrared microscopic imaging; ND, not determined; PTH, parathyroid hormone; SERM, selective estrogen receptor modulator.

^avs Younger individuals,

^bA Boskey, E Carter, CL Raggio, unpublished data,

^cvs Age-matched control,

^dSee Faibish et al.29,

^eSee Vashishth.33,

^fvs Untreated osteoporotic.

Fibronectin

Fibronectin, a minor constituent of bone matrix, is one of the first proteins produced by osteoblasts, and directs the initial deposition of collagen fibrils.36 Continued presence of fibronectin is also required to maintain the integrity of the collagenous matrix.37 Studies with a variety of different conditional knockout animals demonstrated that while osteoblasts produce fibronectin, they are not responsible for the presence of the same in the bone extracellular matrix; rather, the bone matrix fibronectin is derived from circulating liver fibronectin.38 In primary biliary cirrhosis, the incidence of osteoporosis is markedly elevated. This higher incidence is due to an increased production of a fibronectin isoform that lessens osteoblastic bone formation.39

Noncollagenous proteins

There are several families of proteins that account for a small proportion of the extracellular matrix, which, as reviewed elsewhere,40 serve important functions in matrix organization, cell signaling, metabolism and mineralization. Other than early studies showing a reduced NCP content of osteoporotic bone,41 little has been written on how these proteins change in expression or distribution in osteoporosis or other bone diseases, with or without anabolic or antiresorptive therapies. Known changes in the expression of NCPs in health and disease are summarized in Table 3. As shown in Table 3, recent gene-wide association studies have identified several NCPs that may be related to fracture risk. In terms of actual measurement of protein content, findings to date are that osteocalcin and osteopontin are important for fracture resistance,42 their concentrations are reduced in older osteonal bone43 and osteopontin may retard crack propagation.42 As a number of NCPs can and do interact with collagen fibrils,40 they may function as ‘glue', enhancing bone's resistance to fracture.42 Recent studies have shown compositional differences between lamellae and interlamellar areas of cortical bone. The interlamellar areas have lower collagen content and increased concentration of NCPs.44 The significance of such differences is as yet unknown. No published studies to date have examined changes in the expression of enzymes and signaling factors in the matrix of patients with metabolic bone disease.

Table 3. Variation of noncollagenous bone protein concentrations in healthy and diseased human and animal bonesa.

Protein	Bone conc. in osteoporosis	Bone conc. in OI (types I–IV)	Bone conc. after drug treatment
Albumin	Reduced65	Increased	?
Alpha 2-HS Glycoprotein (fetuin)	Unchanged65	Increased66	?
Bone Gla protein (osteocalcin)	Reduced67	Increased66	?
Fibronectin	?	Increased68	?
Matrix Gla protein	Reduced69	?	+ALN not affected (mice)70
Large proteoglycans	?	Decreased67	+ZOL reduced71,72
SLRPS
Biglycan	Depleted72,73	Decreased68	?
Decorin	Depleted72	No change68	?
Osteoadherin	No change74	?	?
SIBLINGS
BSP	Associated with BMD75	No change76	+ALN (rats) reduced77+PTH (mice) reduced78+Sr ranelate no effect79
DMP1	Increased (mouse)80	No change76	+Ca supplement increased81
DPP	?	Decreased80	?
MEPE	Reduced82,83Associated with BMD82,83,84	No change76	+Ca supplement decreased81
Osteopontin	Reduced85	No change76	+PTH serum levels lowered86
Osteonectin	Reduced (mouse)87Associated with BMD in males88	Reduced68	+ALN (rats) reduced77
Thrombospondins	Associated with BMD89,90	TSP1 increased76	?
Matrix metalloproteases
MMP13	Reduced;81,84 associated with BMD82	?	+ALN (rats) reduced91
ADAMTS18	Associated with BMD in Japanese women82	?	?
Phosphatases
FAM210A	Reduced;82 associated with BMD84	?	?
Alkaline phosphatase	Associated with BMD90No change in staining (rats)91	Decreased staining91	+ZOL-enhanced staining92
Tartrate-resistant acid phosphatase	Increased93,94	Increased staining (mice)95	+Estrogen reduced92

Abbreviations: ADAMTS18, A Disintegrin And Metalloprotease with ThromboSpondin repeats; ‘Associations', refer to gene expression studies; ALN, alendronate; BMD, bone mineral density; BSP, bone sialoprotein; Conc., relative concentration; DMP1, dentin matrix protein 1; DPP, dentin phosphoprotein; ?, no data available; MEPE, matrix extracellular phosphorylated glycoprotein; MMP13, matrix metalloproteinase 13; OI, osteogenesis imperfecta; PTH, parathyroid hormone; SIBLING, small integrin-binding N-glycosylated; SLRPS, small leucine-rich proteoglycans; TSP1, thrombospondin-1; ZOL, zoledronate.

^aSee review by Boskey and Robey40 for the function of these proteins and animal models in which the diseases mentioned are noted. Items without reference numbers are discussed in that text.

Lipids

Less than 3% of the total bone matrix is fat soluble. Lipids are important for cell function, surrounding the cell body, regulating the flux of ions and signaling molecules into and out of the cell. The distribution of lipids in the matrix can be observed from histology, based on sudanophilia, from FTIR and Raman analysis or by nuclear magnetic resonance (NMR).45 There are no recent published studies on lipid composition associated with fragility fractures or other bone diseases in humans. Thirty years ago, we did analyze the lipid composition of femoral heads from patients with avascular necrosis, reporting increased cholesterol content.46

Water

The water content of bone may be demonstrated by proton NMR and can be assessed quantitatively by Raman spectroscopy47 and gravimetric methods.48 Water serves many functions, including filling the pores, interacting with collagen fibrils and binding to mineral crystals.49 Unfortunately, the precise role of water in determining the mechanical competence of bone has not been determined. Analysis of water content shows a direct relationship between water and cortical porosity, which occurs with aging and osteoporosis (Figure 7) and is a key feature of renal osteodystrophy50 and its associated osteoporosis. It is assumed, but not yet demonstrated, that decreases in porosity caused by bisphosphonate treatment will result in a lesser water content in both osteoporosis and renal osteodystrophy.

Figure 7.

Water distribution in midshaft tibia cortical bone is demonstrated by three-dimensional ultrashort echo-time magnetic resonance imaging. (a) A young 33-year-old woman without osteoporosis, and (b) is a 64-year-old woman. Note the marked increase in bone water content (BWC) shown by both the concentration images (on the top) and their histograms (c and d, respectively, on the bottom) (reproduced from Yoder et al.49 (Figure 5) with permission from John Wiley & Sons Ltd).

Composition changes in aging and disease

Bone is a dynamic as well as a heterogeneous tissue; therefore, it is not surprising to see changes in composition as a function of age.3 The types of compositional changes that have been reported are summarized in Table 4. These are distinct from the changes discussed above that are associated with bone disease, fragility fractures or treatments to prevent such fractures. Important is the observation that some age-dependent changes in composition are due to alterations in cell activity and protein expression as well as changes in the concentration and post-translational modification40 of those NCPs that regulate matrix composition and mineralization. Therapies to limit the extent of bone disease are directed against these very same cells.

Table 4. Age-related changes in healthy bone composition (cortical and cancellous considered together).

Bone mineral densitya and tissue mineral densityb	Increase with age
Mineral to organic matrix ratioc	Increase with age
Calcium-to-phosphate ratiod	Increase with age
Carbonate-to-phosphate ratioc	Increase with age
Crystal size and perfection (crystallinity)c,e	Increase with age
Acid phosphate substitutionc	Decrease with age
Matrix heterogeneityb,c	Decrease with age
Total collagen crosslinks (collagen maturity)c	Increase with age
Collagen enzymatic crosslinksd	Increase with age
Collagen AGEsd	Increase with age

Abbreviations: AGE, advanced glycation end-products; BMDDD, bone mineral density distribution; DXA, dual photon absorptiometry; FTIRI, Fourier transform infrared microscopic imaging; XRD, X-ray diffraction.

^aDetermined by DXA,

^bDetermined by microcomputed tomography,

^cDetermined by FTIRI, BMDD and Raman spectroscopy,

^dDetermined by chemical analyses,

^eDetermined by XRD.

Compositional changes induced by therapies for osteoporosis

The interesting and important observation is that treatments for osteoporosis while decreasing fracture incidence do not consistently correct the above compositional abnormalities. Therapies currently used (Table 5), other than supplements of calcium, vitamin D and phosphate, fall into two classes: anabolic agents and antiresorptive agents. These therapies all increase or maintain BMD. They each, however, have distinct effects on compositional properties and heterogeneity of their distribution. The most informative of the studies in Table 5 are those based on biopsies before and after treatment. Not every study examined the same tissue site, dose of drug or duration of treatment, and many did not compare the resulting data to the same tissue site in an untreated control. The few instances where this comparison was carried out, and the therapy returned the parameter in question to healthy control values are emphasized in Table 5. Where no human data exists, animal data is included, with the caveat that this does not mean the same alterations will occur in man. Bisphosphonates, alendronate in particular, decreased the heterogeneity of the tissue, rather than increasing it, as would be appropriate for a mechanically strong tissue. Parathyroid hormone, an anabolic agent, and strontium ranelate both increase heterogeneity, determined by FTIRI, Raman and micro-CT measurements. Calcitriol's effect on bone composition has not been determined, neither were the effects of sclerostin antibodies. Odanatacib, the cathepsin K inhibitor, has been shown to affect composition based on BMDD measurements.

Table 5. Effects of current therapies on cancellous bone compositional properties.

	Ca content (BMDD)	Min/mat	BV/TV	TbN	TbS	XLR	XST	Heterogeneity
Antiresorptives
Estrogen	N96	I64	N96	N96	N96	I64	D64	NA
Calcitonin	NA	D97	I97	I97	D97	NA	NA	NA
Ibandronate	NA	I98	N99	N99	N99	NA	I98	N,99 D98
Alendronate	I18,22,100	N60	I101	N101	N101	N60	N60	D,18,60,98,100N,22,99
Odanaticib	I102	I102	NA	NA	NA	NA	NA	D102
Risedronate	I102	D*103	NA	NA	NA	D*103	N*103	D104
Zoledronate	I103	I105,106	I21	I21	D21	NA	D*105,106	D71
Anabolics
PTH	D16	Da	N107	N107	N107	D	Da	I,16,a
Antisclerostin	I15	NA	I15	NA	NA	NA	NA	NA
Other
SrRAN	I108,109	N109,110	N111	N111	N111	N110	N111	I108

Abbreviations: BMDD, bone mineral density distribution; BV/TV, bone volume fraction; D, decreased; I, increased; Min/mat, mineral/matrix ratio; N, no change; NA, not measured; TbN, trabecular number; TbS, trabecular separation; XLR, collagen maturity; XST, crystallinity.

I, D, N and NA show changes relative to untreated osteoporotic patients. Differences in reported values may be because of site, duration of treatment or method of analysis. Where no human data was available, other species are shown in italics. Bold indicates treatments that were reported to normalize indicated property to that in healthy controls.

^aA Boskey, E Carter, CL Raggio, unpublished data.

Most of the antiresorptive therapies for osteoporosis (estrogen, bisphosphonates, calcitonin, cathepsin K inhibitors) increase mineral/matrix ratio, decrease crystallinity and return other FTIR and Raman parameters to less osteoporotic values without returning these values to normalcy. The anabolic agent PTH and strontium-ranelate, which may have both catabolic and anabolic properties, correct many of these FTIRI properties and increase tissue heterogeneity. Many of the therapies lead to retention of existing ‘older' bone. Older bone has increased collagen maturity and increased crystal size. Anabolic agents, in contrast, stimulate new bone formation and the tissue acquired has characteristics of younger bone.

Loss of material heterogeneity, in fracture mechanics, is associated with an increase in brittleness, hence a greater risk of fracture. The importance of heterogeneity is seen in lumber structure and in development of stronger cements and plastics. Bisphosphonate treatment usually results in an increase of bone mineral density and bone volume, or in its maintenance. Bisphosphonate treatment is often accompanied by a decrease in heterogeneity. The reason for this event is as yet uncertain. It may be that there is a balance in these two opposing effects. The use of bisphosphonates results in an increase in BMD and bone volume, hence an increase in bone stiffness and strength. The use of bisphosphonates also causes a decrease in bone heterogeneity, which in turn increases bone brittleness. If this balance is disturbed, as in ‘oversuppression' of bone turnover, failure may occur. Thus, the composition of bone in the healthy individual must be maintained and adjusted, similar to the structure described by Wolff's law, so as to optimize the function of bone.

Conclusions

This review of mineral and matrix properties in healthy and diseased bones demonstrates that these properties show both age- and disease-dependent changes. Bone disease and therapies for these diseases also affect the composition of bone. Bisphosphonates increase bone quantity but their effects on bone quality are variable. The effects of many other agents used in the treatment of osteoporosis are still under investigation. Some bisphosphonates decrease tissue heterogeneity, which may in turn increase brittleness. The origins of this effect and the significance of the alteration in bone quality remain to be determined.