Fracture, aging, and disease in bone - Berkeley · PDF fileFracture, aging, and disease in bone J.W. Ager III Materials Sciences Division, Lawrence Berkeley National Laboratory, Berkeley,

Fracture, aging, and disease in bone

J.W. Ager IIIMaterials Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, California 94720

G. BaloochMaterials Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, California 94720;and Department of Preventive & Restorative Dental Sciences, University of California,San Francisco, California 94143

R.O. Ritchiea)

Materials Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, California 94720;and Department of Materials Science and Engineering, University of California,Berkeley, California 94720

(Received 2 February 2006; accepted 2 March 2006)

From a public health perspective, developing a detailed mechanistic understanding ofthe well-known increase with age in fracture risk of human bone is essential. Thisalso represents a challenge from materials science and fracture mechanics viewpoints.Bone has a complex, hierarchical structure with characteristic features ranging fromnanometer to macroscopic dimensions; it is therefore significantly more complex thanmost engineering materials. Nevertheless, by examining the micro-/nanostructuralchanges accompanying the process of aging using appropriate multiscale experimentalmethods and relating them to fracture mechanics data, it is possible to obtain aquantitative picture of how bone resists fracture. As human cortical bone exhibitsrising ex vivo crack-growth resistance with crack extension, its fracture toughnessmust be evaluated in terms of resistance-curve (R-curve) behavior. While the crackinitiation toughness declines with age, the more striking finding is that thecrack-growth toughness declines even more significantly and is essentially absentin bone from donors exceeding 85 years in age. To explain such an age-induceddeterioration in the toughness of bone, we evaluate its fracture properties at multiplelength scales, specifically at the molecular and nano dimensions using vibrationalspectroscopies, at the microscale using electron microscopy and hard/soft x-raycomputed tomography, and at the macroscale using R-curve measurements. We showthat the reduction in crack-growth toughness is associated primarily with a degradationin the degree of extrinsic toughening, in particular involving crack bridging, and thatthis occurs at relatively coarse size scales in the range of tens to hundreds ofmicrometers. Finally, we briefly describe how specific clinical treatments, e.g., withsteroid hormones to treat various inflammatory conditions, can prematurely damagebone, thereby reducing its fracture resistance, whereas regulating the level of thecytokine Transforming Growth Factor-� can offer significant improvements in thestiffness, strength, and toughness of bone and as such may be considered a therapeutictarget to treat increased bone fragility induced by aging, drugs, and disease.

I. INTRODUCTION

The structural integrity of mineralized tissues such ashuman bone and teeth is clearly of particular clinicalimportance, especially in the case of bone, which formsthe body’s protective load-bearing skeletal framework.Bone is unique when compared to structural engineering

materials due to its well-known capacity for self-repairand adaptation to changes in mechanical usage patterns.Unfortunately, aging and disease are known to increasethe susceptibility of bone fracture, which in the case ofthe very elderly can lead to significant mortality.1 Al-though bone mineral density (BMD) has been routinelyused by clinicians as a predictor of fracture risk, particu-larly for the elderly, there is mounting evidence that thismeasure of bone quantity is not adequate as the solepredictor of bone fracture, and other factors pertainingprincipally to bone quality must be considered.2,3 For

a)Address all correspondence to this author.e-mail: [email protected]

DOI: 10.1557/JMR.2006.0242

J. Mater. Res., Vol. 21, No. 8, Aug 2006 © 2006 Materials Research Society1878

example, Hui et al.4 reported that while BMD was a goodpredictor of forearm fracture, age was a better predictorof hip fracture, suggesting that these two factors haveindependent effects; similarly, Aspray et al.5 concludedthat while the bone mineral content (BMC) of rural Gam-bian women is 10–40% lower than that of Europeanwomen of similar age, height, and weight, osteoporosicfractures are rare. Both these studies strongly challengethe concept of bone density as the primary determinant offracture risk. In addition, BMD has been relied upon toassess the therapeutic benefits of antiresorptive agents intreating osteoporosis, but this may be incomplete as well.

The importance of developing a more complete under-standing of fracture risk factors is thus evident. Numer-ous studies have established that there is a significantdeterioration in the toughness of bone with age (e.g.,6–10);nevertheless, a mechanistic framework for describinghow the microstructure affects the failure of bone is stilllacking. What is needed is an understanding of (i) howaging, disease, or therapeutic treatment can affect thestructure of bone, defined broadly from nano- throughmicro- to macroscopic size scales, and (ii) how this spe-cifically affects the mechanisms responsible for the de-formation and fracture of bone.

Bone has a complex hierarchical structure, as illus-trated schematically in Fig. 1, so to develop this under-standing, it is necessary to examine the bone-matrix

structure and how this affects the mechanistic aspects ofdamage and fracture over multiple (nano to macro) di-mensions.11,12 We first describe the characteristic fea-tures of the bone-matrix structure.

II. STRUCTURE OF BONE

The basic nanostructural building blocks of all miner-alized tissues in the human body, including bone, are anorganic matrix of roughly 90% collagen and mineralphase consisting of calcium phosphate-based apatitemineral. These units are organized into a hierarchy ofstructures ranging in dimension from molecular to themacroscopic size scales, and, accordingly, are far morecomplex than traditional engineering materials (Fig. 1).

Type-I collagen typically comprises ∼90% of the or-ganic matrix of human bone with a structure that is alsohierarchical in nature. Specifically, the collagen mol-ecule is composed of a triple helix of peptide chains,specifically two �1 chains and one �2 chain, each ofwhich are ∼1000 residues long. The molecules arestaggered by 67 nm and covalently cross-linked betweenlysine residues, the intra- and intermolecular cross-links providing for the tensile strength.13,14 They self-assemble into thin (10–300 nm in diameter) collagenfibrils and are impregnated with inorganic carbonatedapatite nanocrystals (tens of nanometers in length and

FIG. 1. Schematic illustration of the complex and hierarchical structure of cortical bone.38 The length scales relevant in our discussion of extrinsictoughening mechanisms range from tens to hundred of nanometers (diameter of collagen fibrils) to hundreds of micrometers (osteocyte lacuna andosteons).

J.W. Ager III, et al.: Fracture, aging, and disease in bone

J. Mater. Res., Vol. 21, No. 8, Aug 2006 1879

width, 2–3 nm in thickness). The fibrils are then aggre-gated into larger fibers that are micrometers in diam-eter.15 At the macrostructural level, bone is distinguishedinto cortical (compact) and cancellous (trabecular) bone;most long bones are composed of a cortical shell with acancellous interior. In this review, we will concentrate oncortical bone. At microstructural length-scales, corticalbone is organized into 200–300-�m-diameter secondaryosteons,16 which are composed of large vascular chan-nels called Haversian canals (50–90 �m diameter) sur-rounded by circumferential lamellar rings (3–7 �mthick), with so-called “cement lines” at the outer bound-ary; these secondary osteons are the end result of theremodeling process that repairs damage in vivo. On av-erage, the organic/mineral ratio in human cortical bone isroughly 1:1 by volume and 1:3 by weight.15 In additionto its hierarchical complexity, the composition and thestructure of bone varies with factors such as skeletal site,age, sex, physiological function, and mechanical loading,making bone a very heterogeneous structure, with theneed for vascularization adding to the complexity of thetissue.

One critical question in the understanding of fracturerisk in bone is discerning which of the characteristicdimensions in this hierarchical structure is most impor-tant in controlling the fracture properties of bone. Webelieve that specific toughening mechanisms in bone ac-tually exist over most of these dimensions; however, it isour contention that structure at the scale of hundreds ofmicrometers, i.e., at the level of the Haversian osteons, ismost critical to controlling the fracture toughness.17 Thisis not to say that molecular and nanostructure is unim-portant; rather it is simply the recognition that fractureinvariably involves collective phenomena that occur overdimensions far larger than atomistic or molecular size-scales. In this respect, bone is no exception.

Accordingly, in this paper, we attempt to characterizestructure, damage, and fracture properties in bone, asinfluenced by aging, disease, and clinical therapy, usinga multitude of techniques pertaining to this range of di-mensions. These will include measurements at the mo-lecular and nanoscales using pico-force atomic forcemicroscopy (AFM), nanoindentation, and vibrationalspectroscopies; at the microscale using electron micros-copy and hard/soft x-ray computed tomography; and atthe macroscale using fracture mechanics and fatigue test-ing. We begin by examining the mechanisms by whichbone derives its resistance to fracture.

III. ORIGINS OF TOUGHNESS IN BONE

A. Macroscopic quantification

Whereas the “strength” of a component such as a boneis often thought of as a measure of its fracture resistance,this is actually not the most appropriate descriptor as it

takes no account of the inevitable presence of flaws (i.e.,cracks) in the material which are known to have a sig-nificant effect on fracture. The same can be said formeasurements of the “work of fracture,” determinedfrom the area under the load–displacement curve in sucha test. The key to understanding whether a componentwill break or not is that it depends not only on the levelof applied stress but also on the presence of cracks(which may have been created during use). This is par-ticularly relevant to bone as an increasing volume frac-tion of microcracks (“micro-damage”) is known to formas bone ages.18 Fracture mechanics provides a viablemethod for quantifying the relationships between thestresses and strains applied to a body, the crack or flawsizes within it, and the resistance to fracture of the un-derlying material. Coupled with an examination of themechanisms of fracture and their relation to the nano/microstructure, it provides a framework for understand-ing the failure of materials under a variety of loadingconditions (e.g., tension, bending, compression, multi-axial loading, cyclic fatigue, etc.), and as such is usedextensively to quantify the onset of cracking in tradi-tional engineering.

In general, significant effort has been made over thepast ten years or so to elucidate how bone is toughened.There is now a large body of results in the literatureinvolving determinations of the fracture toughness ofcortical bone using the linear-elastic fracture mechanics(LEFM) approach. This has to a large extent involvedsingle-parameter characterization of the toughness usingeither the critical value of the mode I linear-elastic stressintensity KIc

19–24 or the related strain-energy release rateGc. [The stress intensity Ki can be defined for threemodes of loading: i � I (mode I tensile-opening), II(mode II shear), and III (mode III anti-plane shear). Foreach of these modes, a corresponding fracture toughnessKic may be defined as the critical value of Ki at fractureinstability, i.e., when Ki � Y�app(�a)1/2 � Kic, where�app is the applied stress, a is the crack length, and Y isa function (of order unity) of crack size and geometry.Alternatively, the toughness can be expressed as a criticalvalue of the strain-energy release rate Gc defined as thechange in potential energy per unit increase in crack area.For an isotropic material, Gc � KIc

2/ E� + KIIc2/ E� +

KIIIc2/ 2µ, where E� � E in plane stress and E� �

E/(1 − �2) in plane stress with E as Young’s modulus and� as Poisson’s ratio, and � is the engineering shearmodulus.] In terms of KIc, toughness values in corticalbone range from 2 to 7 MPa m1/2, with the fracturetoughness, in human humeri for example, typically beingup to twice as high in the transverse orientation comparedto the longitudinal (medial-lateral and proximal-distal)orientations.24,25

Whereas such a fracture mechanics approach repre-sents a significant improvement over measurements of

J.W. Ager III, et al.: Fracture, aging, and disease in bone

J. Mater. Res., Vol. 21, No. 8, Aug 20061880

toughness involving parameters such as the work of frac-ture, there are still problems with this simple single-parameter characterization when it is applied to bone.The first of these is that crack propagation in generalrepresents a mutual competition between two classes ofmechanisms: intrinsic damage mechanisms that operateahead of the crack tip and act to promote crack advance,and extrinsic toughening mechanisms that principally op-erate in the wake of the crack tip and act to impede crackadvance by “shielding” the crack from the full applieddriving force.26–28 Toughening can thus be achieved in-trinsically by enhancing the material’s resistance to mi-crostructural damage or extrinsically by promotingcrack-tip shielding. Whereas intrinsic mechanisms pri-marily govern the crack-initiation toughness, extrinsicmechanisms operate behind the crack tip along the crackflank and govern the crack-growth toughness. As thelatter effect is dependent on the size of the crack, thepresence of significant extrinsic toughening results inrising crack-resistance (or R-curve) behavior, where thevalue of K or G to “drive” a crack rises with crack ex-tension. We will show that akin to many ceramic mate-rials,29–31 toughening in cortical bone is predominantlyextrinsic and is associated with shielding mechanismssuch as crack deflection and bridging. Because this ne-cessitates an R-curve evaluation, single-value character-izations of the toughness are generally insufficient. De-spite this, R-curves have been utilized in only relativelyfew studies to characterize human bone fracture.10,17,32,33

A second problem pertains to the fact that, as dis-cussed below, such shielding in bone is due primarily tocrack bridging and can extend over quite large dimen-sions, approaching hundreds of micrometers to a fewmillimeters. Because such “process zones” are not nec-essarily small compared to the macroscopic size of thebone, a simple single-parameter LEFM characterizationusing KIc can be problematic as the resulting toughnessvalue will likely be size and geometry dependent. Thisproblem can be satisfactorily addressed using cohesive-zone modeling approaches34 but is beyond the scope ofthe present paper. As we are focusing on mechanisms inthe present work, we will use a simple R-curve approachto provide a macroscopic quantification of the toughnessof bone.

B. Mechanistic considerations

Bone principally derives its resistance to fracture fromextrinsic phenomena. Mechanistically, the tougheningcan arise from several sources often acting in concert,with their relative contribution typically depending onsuch factors as orientation and the size-scale. Severalsalient toughening mechanisms have been identified forhuman cortical bone, including (in decreasing order ofimportance) macroscopic crack deflection, crack bridging,

and constrained microcracking.35 Essentially all thesemechanisms result from the nature of the crack path, theover-riding feature being that certain features in the micro-structure provide microstructurally “weak” or preferredpaths for cracking; in bone, these are invariably the cementlines, which are the interfaces between the bone matrix andosteon structures. There are several implications from thisthat are critical for the toughness of cortical bone.

First, because the cement lines are oriented nominallyalong the longitudinal axis of the bone, preferred crack-ing paths will be tend to be along this direction. This canlead to the deflection of cracks attempting to propagate inthe transverse direction.

Second, cracking in the cement lines can lead to thegeneral formation of microcracks, i.e., microdamage,particularly as bone ages and the osteon density increaseswith remodeling. Such microcracking may predominatein the region ahead of a growing (macro) crack where thelocal stresses are highest.

Third, the nature of the coalescence of such micro-cracks to the growing crack can lead to the formation ofuncracked regions along the crack length which act to“bridge” the crack and increase fracture resistance.

Each of the mechanisms is described in more detailbelow.

1. Macroscopic crack deflection

For bones subjected to bending forces where the frac-ture should occur across the bone in the transverse di-rection, i.e., along a path of maximum tensile stress, thecrack will often macroscopically deflect along the longi-tudinal direction to follow a “weaker” path along thecement lines,21,23,24,35 as can be seen for human corticalbone in Fig. 2.35 The effect of this deflection, which isoften as much as 90°, is to increase the toughness sub-stantially by reducing the local driving force for crackadvance. This can be explained by the following fracturemechanics analysis. Assuming for the sake of simplicitythat the deflections are in-plane tilts through an angle �to the crack plane, the local mode-I and mode-II stressintensities, k1 and k2, at the deflected crack tip are givenby36,37:

k1(�) � c11(�)KI + c12(�)KII (1)

and

k2(�) � c21(�)KI + c22(�)KII , (2)

where KI (∼5.3 MPa m1/2, i.e., the fracture toughness ofbone) and KII (�0) are, respectively, the mode I andmode II far-field (applied) stress intensities for a maincrack, and the coefficients cij(�) are mathematical func-tions of the deflection angle � (∼90°).36,37 The effectivestress intensity at the tip of the deflected crack tip Kd can

J.W. Ager III, et al.: Fracture, aging, and disease in bone

J. Mater. Res., Vol. 21, No. 8, Aug 2006 1881

then be calculated by summing the mode-I and mode-IIcontributions in terms of the strain-energy release rate,namely,

Kd � (k12 + k2

2)1/2 , (3)

which suggests that the value of the stress intensity at thecrack tip is reduced locally by ∼50% due to such deflec-tion to ∼2.7 MPa m1/2, as compared to that for an unde-flected crack; the applied stress intensity must then beraised by a factor of two to achieve fracture. This quan-tification of the toughening effect from macroscopiccrack deflection for cracks attempting to propagate in thetransverse orientation, where the crack must try and cutthe osteons, provides the explanation for the observationthat the resistance to fracture in this orientation is ap-proximately twice as high as that in the longitudinal ori-entations.

2. Crack bridging

Crack bridging is a common toughening mechanismin ceramics and composites,26–31 which involves regions

of unbroken material behind a crack front (e.g., fibersin a composite) holding the crack faces together,thereby sustaining load that would otherwise be usedto advance the crack. Such “bridges” reduce the driv-ing force experienced at the tip and can result from avariety of crack/structure interactions. In bone, twoprincipal bridging mechanisms can be identified, whichoperate at very different dimensions (Figs. 3 and 4).

FIG. 2. (a) Schematic depiction of macroscopic crack deflection. Forbones subjected to bending forces where the fracture should occuracross the bone in the transverse direction, i.e., along a path of maxi-mum tensile stress, the crack will often macroscopically deflect alongthe longitudinal direction to follow a “weaker” path along the cementlines. This is shown in the SEM image in (b); cracking ahead of thenotch shows macroscopic crack deflection as the preferred crack pathis along the cement lines of the osteons.35

FIG. 3. (a) Uncracked ligament bridging created by microcrack-ing ahead of the main crack with “mother” and “daughter” cracks.(b) Typical optical micrograph of stable crack growth in 34-year-oldhuman cortical bone clearly shows the presence of uncracked liga-ments on the size scale of tens of microns (indicated by white arrows)in the crack wake. (c) Subsurface crack bridging is shown in three-dimensions in human cortical bone in the x-ray tomographic recon-struction. Uncracked ligaments are indicated, and the white arrow isthe direction of nominal crack growth.17,35

J.W. Ager III, et al.: Fracture, aging, and disease in bone

J. Mater. Res., Vol. 21, No. 8, Aug 20061882

3. Uncracked ligament bridging

Uncracked ligament bridging is the most potent bridg-ing mechanism in bone where tracts of material, oftenseveral hundred micrometers in dimension, compose thebridges (Fig. 3). [In the context of fracture, “ligament”here refers to any unbroken material—i.e., a crack“bridge”, of any type, shape, and size—that spans thecrack, and not a ligament in the anatomical sense.] Thismechanism results from microcracks forming ahead ofthe main crack tip, primarily at the cement lines, and theirimperfect linkage back to the main crack tip (it can alsoresult from nonuniform crack advance giving the appear-ance in any two-dimensional section of cracking aheadof the main crack tip). Such a configuration of “mother”and “daughter” cracks is shown in Figs. 3(a) and 3(b).The consequent bridging mechanism provides a markedcontribution to the macroscopic fracture toughness ofhuman bone3,10,17,35; moreover, its degradation with ag-ing has been identified as a major reason why the frac-ture resistance of bone deteriorates in the aged.10,38

[Uncracked ligament bridging also provides a majorcontribution to the toughness of dentin in teeth, al-though in this mineralized tissue, the microcracks that

form ahead of the main crack tip are initiated at thetubules.]

An estimate of the contribution from this mechanismto the toughness of cortical bone in the longitudinal ori-entation can be seen from the following simple analysis.Theoretical estimates of ligament bridging based on alimiting crack-opening approach39 give this tough-ness contribution in terms of the area fraction of bridg-ing ligaments ful on the crack plane (∼0.45, from crackpath observations), the total length lul of the bridgingzone (∼5 mm), and the applied (far-field) stress intensity(KI ∼4.5 MPa m1/2), namely,

Kbr � −fulKI [(1 + lul/rb)1/2 − 1]/[1 − ful + ful(1 + lul/rb)1/2] , (4)

where r is a rotational factor (0.20–0.47) and b is thelength of the remaining uncracked region ahead of thecrack. Substituting typical values for these parameters,the contribution to the toughness of bone due to thismechanism is of the order of Kbr ∼1 – 1.6 MPa m1/2;these predictions are comparable to those measured ex-perimentally.17

4. Collagen fiber bridging

Collagen fiber bridging (Fig. 4) is another bridgingmechanism where individual collagen fibers span thecrack.35,40 Here, the uniform traction Dugdale zonemodel of Evans and McMeeking41 can be used to esti-mate the contribution to the toughness, namely,

Kbf � 2 �b ff (2 lf / �)1/2 , (5)

where �b is the normal bridging stress on the fibrils(assumed to be ∼100 MPa), ff is the effective area frac-tion of the collagen fibrils active on the crack plane(∼0.15), and lf is the bridging zone length (∼10 �m). Thisanalysis implies a value of Kb

f ∼ 0.08 MPa m1/2 for thecontribution of collagen fiber bridging to the overallbone-matrix toughness. While this is a relatively minorcontribution to the overall toughness, collagen fiberbridging represents a toughening mechanism that oper-ates over far smaller dimensions; i.e., in the submicronrange, and this may well be important for promotingresistance to the extension of micron-scale microcracks,which often grow under the influence of much lowerdriving forces.40 Also, it has been reported recently thatthe angle between the collagen fibril alignment and crackpropagation directions can determine which extrinsictoughening mechanism is dominant, suggesting a morefundamental role in toughening than fiber bridgingalone.42 Finally, in addition to collagen fibers, there issome observational evidence that non-collagenous pro-teins form such small-scale bridges,43 although it is un-clear if these bridges actually are effective by carryingany load.

FIG. 4. Collagen fibril bridging is illustrated schematically in (a) andis shown in a scanning electron microscope image of human corticalbone in (b).35 The horizontal arrow in (b) indicates the direction ofcrack growth.

J.W. Ager III, et al.: Fracture, aging, and disease in bone

J. Mater. Res., Vol. 21, No. 8, Aug 2006 1883

5. Microcracking

Microcracking (Fig. 5) in bone represents a mecha-nism of inelastic deformation; however, in terms of re-sistance to fracture, the formation of microcracks in thedamage zone ahead of a crack tip acts to lower the in-trinsic toughness. Structural analysis techniques for ob-serving microcracking have been reviewed recently.44

However, several authors32,45,46 have suggested that mi-crocracking may extrinsically toughen bone because azone of microcracks surrounding a crack would be sub-jected to some degree of dilation. This dilation, if con-strained by surrounding rigid material, coupled with thefact that the microcracked region would be of lowermodulus, can thus act to shield the crack tip and hence,extrinsically toughen the material.47,48 However, recentwork has shown that the contributions solely from thismechanism in bone are relatively minor49; indeed, webelieve that the main significance that microcracks mayhave for toughening is that they result in the formation ofuncracked ligament bridges.

These considerations lead us to conclude that tough-ening in bone is primarily associated with extrinsictoughening mechanisms, most prominently crack deflectionand bridging, which result from the interaction of the

crack path with the underlying bone-matrix structure.The nature and direction of the crack path, as affected bythis structure; orientation; anatomical nature of the bone;and, as described below, aging, disease, and clinicaltherapy, is the crucial factor that determines the potencyof these mechanisms. Preferred cracking paths appear tobe along the cement lines between the osteon structures,which can cause significant toughening by macroscopiccrack deflection (for fractures in the transverse orienta-tion) and by uncracked ligament bridging. The relativecontributions of these mechanisms to the overall tough-ness provide an explanation for the anisotropy of tough-ness in bone, and (as described below) their progressivedegradation with age provides one reason why the risk offracture is significantly higher in the elderly.

IV. DETERIORATION IN BONE FROM AGING

A. Macroscopic scale: Fracturetoughness behavior

To first quantify how aging may affect the toughnessof bone, we examine macroscopic fracture toughnesstests on human cortical bone. The sample set for thisstudy was composed of human cortical bone taken fromthe humerii of nine cadavers (donor age: 34–99 years).10

Seventeen (N � 17) compact-tension C(T) specimens,were tested with samples divided into three age groups,arbitrarily named Young (age 34, N � 1; age 37, N � 4;and age 41, N � 2), Middle-Aged (age 61, N � 1; age69, N � 2; and 69, N � 2), and Aged (age 85, N � 1;age 85, N � 2; and age 99, N � 2). The samples were alloriented with the starter notch and the nominal crack-growth direction along the proximal–distal direction ofthe humerus (in the longitudinal–radial plane), i.e., par-allel to the long axis of the osteons and hence, long axisof the humerus. The crack-initiation toughness Ko wasobtained by extrapolating a linear fit of the data for eachsample to a crack extension of �a � 0 while the (linear)slope of the R-curve gave a measure of the crack-growthtoughness.

The resulting toughness R-curves are shown in Fig. 6with statistical summaries given in Fig. 7. Statisticalanalysis (non-parametric Kruskal–Wallis test) indicatedthat, for the three age groups, variation among groupmedians was significant (P < 0.05 and P < 0.01 for theinitiation and the growth toughness, respectively). Whilethe initiation toughness decreases by ∼40% over six de-cades from 40 to 100 years, the effect of aging is muchmore striking on the growth toughness, which is essen-tially eliminated in the Aged group. The decrease in theinitiation toughness is consistent with the trend observedin studies that report single-value toughnesses (e.g., Refs.23, 50–53). However, the value of R-curve measure-ments10 is that they clearly show that not only the intrinsicresistance to fracture (as reflected by the crack-initiation

FIG. 5. Microcracking surrounding a larger crack is shown (a) sche-matically and (b) in a scanning electron microscope image of dentin,a mineralized tissue similar to bone. The horizontal dark arrow in(b) shows the direction of crack growth. Dentin is chosen here to providea clear representation of the process, which in this case involves theformation of microcracks (white arrows) at the tubules.49 The struc-tural analysis of microcracks in bone has been reviewed in Ref. 44.

J.W. Ager III, et al.: Fracture, aging, and disease in bone

J. Mater. Res., Vol. 21, No. 8, Aug 20061884

toughness), but also the resistance to crack propagation(as reflected by the crack-growth toughness), decreaseswith age (Figs. 6 and 7). Indeed, the age-related deterio-ration in the crack-growth toughness appears to be thedominant effect.

B. Micrometer to millimeter scale: Osteons andcrack bridging

Reflective of this marked deterioration in the resis-tance to fracture of human cortical bone with increasing

age, there are significant changes in the bone-matrixstructure. One important change is the statistically sig-nificant (P < 0.001) increase in density of secondary os-teons with age; indeed, the osteonal density almostdoubles for 35–100-year-old bone.38 Regressions of theosteonal density against the crack-initiation and growthtoughnesses both show this effect (Fig. 8) and clearlyindicate that the significant age-related reduction in bothmeasures of the toughness (P < 0.01) with increase inosteonal density.38

By examining the surface of cracked specimens withoptical and scanning electron microscopy,38 the increasein osteonal density with age can be seen to be accompa-nied by a decrease in the extent of crack bridging. This isparticularly evident from three-dimensional x-ray com-puted tomography (XRT) imaging, which has provided

FIG. 7. Variation in crack-initiation toughness (left-hand scale) andcrack-growth toughness (right-hand scale) obtained from R-curvemeasurements on human cortical bone as a function of age (Young �34 to 41 years, Middle-aged � 61 to 69 years, Aged � 85 to99 years).

FIG. 6. Crack-resistance curves for stable ex vivo crack extensionin human cortical bone as a function of age from 34–41 years to 85–99 years (tested in Hanks’ Balanced Salt Solution at 25 °C). Thecrack-initiation toughness Ko is represented by the intercept of theR-curve on the stress intensity axis at �a → 0; the crack-growth tough-ness is given by the slope of the R-curve.10

FIG. 8. Variation in the (a) crack-initiation toughness Ko and the(b) crack-growth toughness (slope of the R-curve) with osteon densityfor human cortical bone. The density of (secondary) osteons increaseswith age. Each data point represents the average for all measure-ments from one donor. A linear regression of the data is shown in eachcase (fit equation and coefficient of determination R2 is also in-cluded).38

J.W. Ager III, et al.: Fracture, aging, and disease in bone

J. Mater. Res., Vol. 21, No. 8, Aug 2006 1885

the most quantitative evidence for the role of crack bridg-ing as a toughening mechanism in bone. [XRT was per-formed at the Stanford Synchrotron Radiation Labora-tory (25 keV), Menlo Park, CA, and at the AdvancedLight Source (18 keV), Berkeley, CA, with a typicalvoxel size (equivalent to the spatial resolution) of∼5 �m.] Shown in Figs. 9(a) and 9(b) are tomographicslices obtained from the Young and Aged groups as afunction of distance behind the crack tip. Crack bridgingfrom uncracked ligaments can be seen in both cases, asevidenced by intact regions, often tens of micrometersin size, observed along the crack path (the three-dimensional nature of the bridges can be seen in Fig. 3).

However, there are fewer bridges, which are smaller insize, in the Aged sample. Specifically, the bridging-zonefraction remains roughly constant up to the crack tip inthe Young sample [Fig. 9(c)] whereas in the Agedsample, it is initially comparable but then falls to nearlyzero within a few millimeters of the crack tip.

Worthy of note here are the large dimensions, on theorder of hundreds of micrometers or more, over whichthe phenomena that influence fracture occur. However, itis interesting to speculate how these toughening mecha-nisms, specifically crack bridging, degrade with age. Thecorrelation of the decay in toughness with an increase inosteonal density with age provides a clue here. As noted

FIG. 9. Two-dimensional computed x-ray tomographic reconstruction slices showing typical cracks in specimens taken from the (a) Young(34 years) and (b) Aged (85 years) human cortical bone groups. The numbers at the top indicate the distance from the (nominal) crack tip, andthe black arrows indicate uncracked-ligament “crack bridges.” (c) Fraction of such bridges with distance from the crack tip demonstrating thesmaller area fractions and bridging-zone size in the older bone.38

J.W. Ager III, et al.: Fracture, aging, and disease in bone

J. Mater. Res., Vol. 21, No. 8, Aug 20061886

above, uncracked ligament bridges are formed when mi-crocracks open up ahead of the main crack tip and thentemporarily fail to link perfectly with the main crack. Asthe microcracks preferentially form at the cement lines,the size of the resulting bridges is comparable with theosteon spacing. In older bone, excessive remodelingleads to a higher density of secondary osteons, whichnecessarily are closer spaced, and the consequences ofthis are smaller bridges, less toughening, and thus anincreased risk of fracture.

It is interesting to note here that in a recent study,54 ithas been reported that the fatigue resistance of equinebone may be enhanced with increase in osteon density, asthe osteons can provide local barriers to crack propaga-tion. Clearly, because the osteon density is affected byseveral other factors in addition to age, such as anatomi-cal location, gender, and physical activity, the mechanis-tic relationship between the fatigue and fracture tough-ness properties of bone—in particular the role of theosteon and its cement line in both initiating microcracksand perhaps acting as a local barrier to their further mo-tion—demands further in-depth study.

C. Molecular scale: Raman studies

At molecular, i.e., ultrastructural, dimensions, vibra-tional spectroscopy can be used to probe for structuraldegradation, and in this regard both Fourier transforminfrared (FTIR)55,56 and visible and near-infrared (IR)Raman spectroscopies9,57 have been used. It should benoted, however, that a specific challenge of applyingvisible Raman spectroscopy to solid biological tissue isfluorescence interference58; this has led to the use ofnear-IR excitation59–61 and highly sophisticated back-ground subtraction and data analysis techniques9,62 tomitigate this problem. To help resolve this issue, werecently reported the first in situ deep ultraviolet Ramanspectroscopy measurements on human cortical bone,63

for which the use of 244 nm excitation both completelyeliminates the fluorescence interference and increases thesignal strength of some features from the organic (colla-gen) phase due to resonance effects.

In our studies of bone63 and also dentin,64 we havefound that most of the variation occurs in the spectralregion from 1400 to 1800 cm−1, as shown by the datafrom the Young, Middle-Aged, and Aged cortical bonesample sets in Fig. 10. Spectral features due to both theamide backbone (amide I and II) and resonance-enhanced side chain vibrations (e.g., Y8a tyrosine ringstretching) were observed and are indicated in Fig. 10.Quantitative analysis was performed by using nonlinearleast-squares fitting to determine the heights of the fouroverlapping features with the indicated approximate fre-quencies, specifically CH2 wag, 1460 cm−1; amide II,1550 cm−1; Y8a, 1620 cm−1; and amide I, 1660 cm−1.The trend suggested by Fig. 10 of an increase in the

amide I peak height with increasing age could not beconsidered significant for the small data set studied(Pearson correlation, 0.05 < P < 0.10).63 Our workinghypothesis is that the changes in amide I peak height aredue to broadening of the resonance profile for the amide� → �* transition caused by changes in the intrafibrallarenvironment of the collagen molecules. It is well estab-lished that the density of bone on the macroscopic scaledecreases with age due to natural processes and to bonemetabolic diseases such as osteoporosis; however, thedensity of mineral per unit volume of bone, excludingporosities, actually increases.65 It is possible that the in-creased bone density is responsible for the age-relatedchanged in the amide I resonance Raman peak height, butclearly more study is required.

Statistical analysis (Pearson correlation) of the Ramanand R-curve data obtained from the same samplesshowed a significant relationship between the amide Ipeak height and the crack-initiation toughness (P < 0.05);the corresponding relationship with the crack-growthtoughness was not significant (P > 0.10). We considerthis to be further evidence to our notion that the reductionof crack bridging, and hence fracture toughness, in agedbone has its origins primarily at microscopic, as opposedto ultrastructural, length scales.

D. Mechanistic summary

These results show clear evidence of a marked degra-dation in bone quality, specifically from a significantdecrease in the fracture toughness (for both initiating andgrowing cracks) for human cortical bone between theages of 34 and 99 years. Such macroscopic changes in

FIG. 10. Deep ultraviolet Raman spectra for human cortical bonefrom the Young, Middle-Aged, and Aged data sets. Spectral assign-ments are indicated. The height of the resonance-enhanced amide Ifeature increases with increasing age.

J.W. Ager III, et al.: Fracture, aging, and disease in bone

J. Mater. Res., Vol. 21, No. 8, Aug 2006 1887

fracture resistance can be related to substantial age-related changes in structure over dimensions from mo-lecular to those in the hundreds of micrometers. Whereasthere are definitive changes in the collagen environmentat molecular levels, and an apparent degradation in col-lagen fibril structure and corresponding mechanicalproperties at sub-micrometer levels, an important factoris the presence of fewer and smaller crack bridges withage. In human bone, we believe that this is associatedwith an increased volume fraction of microcracks, mostprobably due to an increase in osteonal density, whichcan be caused by excessive remodeling with age; this inturn results in lower toughness and hence a greater risk ofbone fracture in older bone. This explanation is consis-tent with the observation that it is the crack-growthtoughness, i.e., slope of the R-curve (Fig. 6), that is mostaffected by age and that this effect occurs over larger(near-millimeter) dimensions.

V. DETERIORATION IN BONE FROM DISEASEAND CLINICAL TREATMENT

Skeletal development and homeostasis are regulatedby growth factors and hormones, which regulate celldifferentiation, cell function, and matrix deposition.

Glucocorticoids, which are steroid hormones widely usedfor the treatment of inflammatory conditions such as ar-thritis and dermatitis, have been associated in clinicalstudies with an increase in the risk of bone fracture,especially in the spinal vertebrae and the femoralhead.66–68 It is known that glucocorticoids alter bonemetabolism in such a way as to decrease bone densityand change trabecular bone architecture. However, thesechanges do not explain the observed increase in fracturerisk in patients treated with glucocorticoids. For ex-ample, the fracture risk in patients treated with gluco-corticoids is higher than that in women with postmeno-pausal osteoporosis who have lower bone mineral den-sities.69,70 It is therefore important to assess specificmechanical property changes in bone, in addition to bonemass and architecture, that occur in response to gluco-corticoid treatments; in this regard, our recent studieshave focused on steroid effects in mouse bone.71

Specifically, changes in the fifth lumbar vertebralbody were assessed for the trabecular bone structure(using microCT and histomorphometry), the elasticmodulus of individual lumbar vertebrae trabecula, andthe mineral-to-matrix ratio (Raman micro-spectroscopy)in glucocorticoid-treated mice and placebo-treated con-trols for comparison to estrogen-deficient mice and

FIG. 11. (a) Representative elastic modulus map from an individual trabecula from glucorticoid (GC) treated mice. The mean elastic modulus isE � 24.2 ± 1.9 GPa. At the remodeling surface, shown in the magnified image at the upper left by white dotted closed-loop lines, a significantreduction in E (30% below the mean values) was observed. Reductions in E were also observed around the osteocyte lacunae within the trabeculae(shown with black dotted closed-loop lines in the magnified images to the right). (b) Raman micro-spectroscopic imaging of glucocorticoid-treatedmouse trabecula. The dark spot in the right-hand images indicates a local deficit in the mineral phase and demonstrates that GC treatment reducesthe mineralized tissue around the osteocyte lacunae.71

J.W. Ager III, et al.: Fracture, aging, and disease in bone

J. Mater. Res., Vol. 21, No. 8, Aug 20061888

sham-operated controls. Elastic modulus mapping(EMM), a modified AFM-based nanoindentation tech-nique that assesses elastic modulus without plasticallydeforming the tissue, was used to map 256 × 256 pointsof elastic modulus with high spatial resolution (∼15 nm).Using EMM, the average elastic modulus over the entiresurface of each trabecula was found to be similar in allthe experimental groups; however, localized changeswithin the trabeculae in areas surrounding the osteocytelacunae were observed in the glucorticoid-treated mice[Fig. 11(a)]. In particular, the size of the osteocyte lacu-nae was increased, but more importantly, a zone of re-duced elastic modulus, concomitant with a “halo” ofhypo-mineralized bone [Fig. 11(b)], was observedaround these lacunae. From these results, it appears thatglucocorticoids may have direct effects on osteocytes,resulting in a modification of their microenvironmentfrom localized changes in elastic modulus and bonemineral-to-matrix ratio. Consistent with the fact thatsteroid treatment is known to increase bone fragility,we anticipate that these local changes in bone-matrixstructure will likely lead to reduced toughness levels inglucocorticoid-treated bone. However, it is by no meansclear that the mechanism(s) by which steroid treatmentdegrades bone quality will be similar to those caused byaging, particularly given that the elastic property changesappear to be localized. This subject will be addressed infuture work.

Bone architectural properties such as cortical bonethickness and trabecular bone volume and organizationare regulated by a variety of cytokines and hormones. Acytokine known to be important in bone formation isTransforming Growth Factor-� (TGF-�). The complexbiological role of TGF-� signaling on osteoblast prolif-eration and differentiation has been studied in vitro andin vivo. However, a direct connection between TGF-�signaling and the mechanical properties of the bone ma-trix itself has only recently been discovered.72 Measure-ments of the local mechanical properties, matrix compo-sition, and fracture toughness were performed on micewith different levels of TGF-� signaling. Compared towild-type mice, D4 and D5 mice, expressing 16- and2.5-fold increased levels of active TGF-� in bone,showed a reduction in elastic modulus, measured byAFM-nanoindentation [Fig. 12(a)]; bone mineral con-centration, assessed by synchrotron x-ray tomography[Fig. 12(b)]; and fracture toughness (by ∼30%). In con-trast, partial reduction in TGF-� signaling in transgenicDominant Negative TGF-6 Receptor II (DNT�RII) miceexpressing a dominant negative version of the type IITGF-� receptor in osteoblasts had elevated elastic modu-lus [Fig. 12(a)], bone mineral concentration [Fig. 12(b)],and fracture toughness (by ∼50%). This was also ob-served in mice with partial reduction in TGF-� signalingthrough heterozygote loss of Smad3, the intracellulartarget of TGF-� signaling in osteoblasts. Such results

FIG. 12. (a) Individual elastic modulus and hardness values obtained from consecutive nanoindentations in bone in six genotypes of mice withvarying levels of TGF-�. Columns are arranged left to right in order of decreasing TGF-� expression. Bones from two-month-old animals withelevated TGF-� signaling (D4 and D5 mice) had decreased elastic modulus and hardness. Two-month bones with impaired TGF-� signaling(DNT�RII, Smad3+/−, and Smad3−/− mice) had increased elastic modulus, hardness, and fracture toughness. Error bars show standard deviation,and stars indicate values that are significantly different from wild-type values (P < 0.001). (b) X-ray computed tomography (XRT) cross sectionsof two-month-old tibia showing effect of TGF-� on bone mineral concentration. Quantitative analysis of these data demonstrated regulation ofmineral concentration by TGF-�; (*)P < 0.05.72

J.W. Ager III, et al.: Fracture, aging, and disease in bone

J. Mater. Res., Vol. 21, No. 8, Aug 2006 1889

indicate that TGF-� can regulate mechanical propertiesand mineral concentration of bone matrix; in essence,reduced TGF-� signaling increases functional param-eters of bone quality and thereby contributes to thebone’s ability to resist fracture. However, the precisemechanism by which the over-expression of TGF-� canreduce the fracture toughness so significantly is as yetunknown and is currently under study.

In view of the fact that most bone disorder drugs todayprimarily treat the problem of bone quantity, which isnow known to be only a relatively small part of the issueof increased fragility of bone with age, these results sug-gest possible clinical treatments for the more importantproblem of bone quality. Specifically, a reduction ofTGF-� signaling should perhaps be considered as atherapeutic target for treating bone disorders. This is par-ticularly pertinent as numerous TGF-� inhibitors are cur-rently in preclinical or clinical trials for treatment ofcancer metastases.

VI. CONCLUDING REMARKS

In this work, we have examined in detail tougheningmechanisms in bone and then how this resistance to frac-ture can be degraded by processes of aging, disease, andclinical treatments. However, it must be noted that alimitation associated with the discussion of the fracturemechanism and characterization of human bone is therelatively small sample size; we used about 20 specimens(depending on the test) taken from 9 separate donors.More detailed investigations that look at the specific ef-fects of variables such as gender, anatomical location,microstructural variations, and bone mineral density arerequired to obtain a fuller understanding and are cur-rently being undertaken. In addition, our approach can beconsidered “worst-case” in that we have treated bone asa structural material and not considered that damage canbe repaired in vivo through remodeling; we have also notconsidered the specific role of fatigue loading. However,studies over a wide range of length scales lead us tobelieve that whereas bone is toughened at multiple di-mensions, it is phenomena at the scale of tens to hun-dreds of micrometers that dominate its fracture proper-ties. Specifically, microcracking, preferentially at thecement lines of the osteon structures promotes the twoprimary toughening mechanisms in bone, namely, crackbridging (principally due to uncracked ligaments) andmacroscopic crack deflection (for cracks propagating inthe transverse orientation). Indeed, similar behavior canbe seen in tooth dentin, although the microcracking thatpromotes the uncracked ligament bridging is now asso-ciated with the dentinal tubules at the micrometer scaleand macroscopic crack deflection is generally not an is-sue.73,74

The effects of aging on the structure and properties

of bone can also be identified through nano-/micro-mechanisms at multiple dimensions. Factors such as in-creased mineralization,9,75 increased microdamage,40,76

lowered collagen quality,52 and increased bone turn-over77 have all been implicated, although a consistentpicture of how all this precisely affects the bone-matrixtoughness is still uncertain. Again our premise is thatalthough we can directly measure molecular changesin the collagen environment and deterioration in the na-ture and properties of individual collagen fibrils at sub-micrometer dimensions, it is phenomena at the tens tohundreds of micrometers that are most pertinent to howaging degrades the toughness. That is, our results show aclear reduction in the fraction and size of crack bridges inolder bone, which we believe is associated with a higherdensity of (secondary) osteons from excessive remodel-ing.10,38

Our studies on disease and clinical treatment in theireffect on the structure and toughness of bone are still intheir infancy; however, the challenge is again to iden-tify and quantify the specific fracture mechanisms af-fected by each biological process in light of changesthat occur in the bone-matrix nano-/micro-structure. It isthrough such a multi-dimensional mechanistic approach,which combines biology, materials science, and frac-ture mechanics, that a clearer understanding of whattoughens and embrittles bone can be achieved. We be-lieve that such information can be used as basis for thedesign of improved drug therapies to reduce the risk ofbone fracture in the elderly and in other at-risk popula-tions.

ACKNOWLEDGMENTS

This work was supported by the Director, Office ofScience, Office of Basic Energy Science (BES), Divisionof Materials Sciences and Engineering, Department ofEnergy (authors J.W.A. and R.O.R.) and the LaboratoryDirected Research and Development Program ofLawrence Berkeley National Laboratory (for authorG.B.) under Contract No. DE-AC02-05CH11231. Weacknowledge the support of x-ray wiggler beamline (BL10-2) at the Stanford Synchrotron Radiation Laboratory(SSRL), supported by Department of Energy, ContractNo. DE-AC03-76SF00515, and the dedicated tomogra-phy beamline (BL 8.3.2) at the Advanced Light Source(ALS), supported by the Department of Energy (BES)under Contract No. DE-AC03-76SF00098. Specialthanks are due to the many individuals who were in-volved in the work reviewed in this paper, includingDr. R.K. Nalla, Dr. J.J. Kruzic, Dr. J.H. Kinney, Dr. A.E.Porter, Dr. T. Alliston, Dr. M. Balooch, Dr. A.P. Tomsia,Prof. N.E. Lane, and Prof. R. Derynck. We also thankDr. C. Puttlitz and Dr. Z. Xu for supplying the humancortical bone.

J.W. Ager III, et al.: Fracture, aging, and disease in bone

J. Mater. Res., Vol. 21, No. 8, Aug 20061890

REFERENCES

1. A.G. Jennings and P. de Boer: Should we operate on nonagenar-ians with hip fractures? Injury 30, 169 (1999).

2. R. Heaney: Is the paradigm shifting? Bone 33, 457 (2003).3. P.D. Miller, M.C. Hochberg, L.E. Wehren, P.D. Ross, and

R.D. Wasnich: How useful are measures of BMD and bone turn-over? Curr. Med. Res. Opin. 21, 545 (2005).

4. S.L. Hui, C.W. Slemenda, and C.C. Johnston: Age and bone massas predictors of fracture in a prospective study. J. Clin. Invest. 81,1804 (1988).

5. T.J. Aspray, A. Prentice, T.J. Cole, Y. Sawo, J. Reeve, andR.M. Francis: Low bone mineral content is common but osteopo-rotic fractures are rare in elderly rural Gambian women. J. BoneMiner. Res. 11, 1019 (1996).

6. A. Burstein, D. Reilly, and M. Martens: Aging of bone tissue:Mechanical properties. J. Bone Joint Surg. 58A, 82 (1976).

7. P. Zioupos and J.D. Currey: Changes in the stiffness, strength, andtoughness of human cortical bone with age—An underexploredfrontier. Bone 22, 57 (1998).

8. Y.N. Yeni and T.L. Norman: Fracture toughness of human femo-ral neck: Effect of microstructure, composition, and age. Bone 26,499 (2000).

9. O. Akkus, F. Adar, and M.B. Schaffler: Age-related changes inphysicochemical properties of mineral crystals are related to im-paired mechanical function of cortical bone. Bone 34, 443 (2004).

10. R.K. Nalla, J.J. Kruzic, J.H. Kinney, and R.O. Ritchie: Effect ofaging on the toughness of human cortical bone: Evaluation byR-curves. Bone 35, 1240 (2004).

11. J-Y. Rho, L. Kuhn-Spearing, and P. Zioupos: Mechanical prop-erties and the hierarchical structure of bone. Med. Eng. Phys. 20,92 (1998).

12. S. Weiner and H.D. Wagner: The material bone: Structure-mechanical function relations. Annu. Rev. Mater. Sci. 28, 271(1998).

13. D.R. Eyre, M.A. Paz, and P.M. Gallop: Cross-linking in collagenand elastin. Annu. Rev. Biochem. 53, 717 (1984).

14. L. Knott and A.J. Bailey: Collagen cross-links in mineralizingtissues: A review of their chemistry, function, and clinical rel-evance. Bone 22, 181 (1998).

15. A. Veis: Collagen fibrallar structure in mineralized and nonmin-eralized tissues. Curr. Opin. Solid State Mater. Sci. 2, 370 (1997).

16. J.D. Currey: Osteons in biomechanical literature. J. Biomech. 15,717 (1982).

17. R.K. Nalla, J.J. Kruzic, J.H. Kinney, and R.O. Ritchie: Mecha-nistic aspects of fracture and R-curve behavior in human corticalbone. Biomaterials 26, 217 (2005).

18. T.C. Lee, A. Staines, and D. Taylor: Bone adaptation to load:Microdamage as a stimulus for bone remodeling. J. Anat. 201,437 (2002).

19. W. Bonfield: Advances in the fracture mechanics of cortical bone.J. Biomech. 20, 1071 (1987).

20. T.L. Norman, D. Vashishth, and D.B. Burr: Fracture toughness ofhuman bone under tension. J. Biomech. 28, 309 (1995).

21. X.D. Wang, N.S. Masilamani, J.D. Mabrey, M.E. Alder, andC.M. Agrawal: Changes in the fracture toughness of bone may notbe reflected in its mineral density, porosity, and tensile proper-ties—Effects of sampling sites and crack orientations. Bone 23,67 (1998).

22. C.U. Brown, Y.N. Yeni, and T.L. Norman: Fracture toughness isdependent on bone location- A study of the femoral neck, femoralshaft, and the tibial shaft. J. Biomed. Mater. Res. 49, 380 (2000).

23. J.B. Phelps, G.B. Hubbard, X. Wang, and C.M. Agrawal: Micro-structural heterogeneity and the fracture toughness of bone.J. Biomed. Mater. Res. 51, 735 (2000).

24. P. Lucksanambool, W.A.J. Higgs, R.J.E.D. Higgs, andM.W. Swain: Fracture toughness of bovine bone: Influence oforientation and storage media. Biomaterials 22, 3127 (2001).

25. J.C. Behiri and W. Bonfield: Fracture mechanics of bone—Theeffects of density, specimen thickness, and crack velocity on lon-gitudinal fracture. J. Biomech. 22, 863 (1989).

26. R.O. Ritchie: Mechanisms of fatigue-crack propagation in metals,ceramics and composites: Role of crack tip shielding. Mater. Sci.Eng. 103, 15 (1988).

27. A.G. Evans: Perspective on the development of high-toughnessceramics. J. Am. Ceram. Soc. 73, 187 (1990).

28. R.O. Ritchie: Mechanisms of fatigue-crack propagation in ductileand brittle solids. Int. J. Fract. 100, 55 (1999).

29. R. Steinbrech, R. Knehans, and W. Schaawächter: Increase ofcrack resistance during slow crack growth in Al2O3 bend speci-mens. J. Mater. Sci. 18, 265 (1983).

30. P.L. Swanson, C.J. Fairbanks, B.R. Lawn, Y-W. Mai, andB.J. Hockey: Crack-interface grain bridging as a fracture resis-tance mechanism in ceramics: I. Experimental study on alumina.J. Am. Ceram. Soc. 70, 279 (1987).

31. Y-W. Mai and B.R. Lawn: Crack–interface grain bridging as afracture resistance mechanism in ceramics: II. Theoretical fracturemechanics model. J. Am. Ceram. Soc. 70, 289 (1987).

32. D. Vashishth, J.C. Behiri, and W. Bonfield: Crack growth resis-tance in cortical bone: Concept of microcrack toughening. J. Bio-mech. 30, 763 (1997).

33. P-C. Wu and D. Vashishth: Age-related changes in cortical bonetoughness: Initiation vs. propagation, in Proceedings of the 2ndJoint EMBS/BMES Conference Vol. 1, edited by J.W. Clark andL.V. McIntire (IEEE, 2002), p. 425

34. Q.D. Yang, B.N. Cox, R.K. Nalla, and R.O. Ritchie: Fracturelength scales in human cortical bone: The necessity of nonlinearfracture models. Biomaterials 27, 2095 (2006).

35. R.K. Nalla, J.H. Kinney, and R.O. Ritchie: Mechanistic fracturecriteria for the failure of human cortical bone. Nat. Mater. 2,164 (2003).

36. B.A. Bilby, G.E. Cardew, and I.C. Howard: Stress intensity fac-tors at the tips of kinked and forked cracks, in Fracture 1977, Vol.3, edited by D.M.R. Taplin (Pergamon Press, Oxford, UK, 1978),pp. 197–200.

37. B. Cotterell and J.R. Rice: Slightly curved or kinked cracks. Int. J.Fract. 16, 155 (1980).

38. R. K. Nalla, J.J. Kruzic, J.H. Kinney, M. Balooch, J.W. Ager, III,and R.O. Ritchie: Role of microstructure in the aging-related de-terioration of the toughness of human cortical bone. Mater. Sci.Eng., C 26 (2006, in press).

39. J-K. Shang and R.O. Ritchie: Crack bridging by uncracked liga-ments during fatigue-crack growth in SiC-reinforced aluminum-alloy composites. Metall. Trans. A 20A, 897 (1989).

40. Y.N. Yeni and D.P. Fyhrie: Fatigue damage-fracture mechanicsinteraction in cortical bone. Bone 30, 509 (2002).

41. A.G. Evans and R.M. McMeeking: On the toughening of ceramicsby strong reinforcements. Acta Metall. 34, 2435 (1986).

42. H. Peterlik, P. Roschger, K. Klaushofer, and P. Fratzl: From brittleto ductile fracture of bone. Nat. Mater. 5, 53 (2006).

43. G.E. Fantner, T. Hassenkam, J.H. Kindt, J.C. Weaver, H. Birkedal,L. Pechenik, J.A. Cutroni, G.A.G. Cidade, G.D. Stucky, D.E. Morse,and P.K. Hansma: Sacrificial bonds and hidden length dissipateenergy as mineralized fibrils separate during bone fracture. Nat.Mater. 4, 612 (2005).

44. N.D. Sahar, S-I. Hong, and D.K. Koln: Micro- and nano-structuralanalysis of damage in bone. Micron 36, 617 (2005).

45. D. Vashishth, K.E. Tanner, and W. Bonfield: Contribution, devel-opment and morphology of microcracking in cortical bone duringcrack propagation. J. Biomech. 33, 1169 (2000).

J.W. Ager III, et al.: Fracture, aging, and disease in bone

J. Mater. Res., Vol. 21, No. 8, Aug 2006 1891

46. D. Vashishth, K.E. Tanner, and W. Bonfield: Experimental vali-dation of a microcracking-based toughening mechanism for cor-tical bone. J. Biomech. 36, 121 (2003).

47. A.G. Evans and K.T. Faber: Crack-growth resistance of micro-cracking brittle materials. J. Am. Ceram. Soc. 67, 255 (1984).

48. J.W. Hutchinson: Crack tip shielding by micro-cracking in brittlesolids. Acta Metall. 35, 1605 (1987).

49. R.K. Nalla, J.J. Kruzic, and R.O. Ritchie: On the origin of thetoughness of mineralized tissue: Microcracking or crack bridging.Bone 34, 790 (2004).

50. C.U. Brown, Y.N. Yeni, and T.L. Norman: Fracture toughness isdependent on bone location—A study of the femoral neck, femo-ral shaft, and the tibial shaft. J. Biomed. Mater. Res. 49,380 (2000).

51. P. Zioupos, J.D. Currey, and A.J. Hamer: The role of collagen inthe declining mechanical properties of aging human cortical bone.J. Biomed. Mater. Res. 2, 108 (1999).

52. X. Wang, X. Shen, X. Li, and C.M. Agrawal: Age-related changesin the collagen network and toughness of bone. Bone 31, 1 (2002).

53. J.D. Currey, K. Brear, and P. Zioupos: The effects of ageing andchanges in mineral content in degrading the toughness of humanfemora. J. Biomech. 29, 257 (1996).

54. V.A. Gibson, S.M. Stover, J.C. Gibeling, S.J. Hazelwood, andR.B. Martin: Osteonal effects on elastic modulus and fatigue lifein equine bone. J. Biomech. 39, 217 (2006).

55. E.P. Paschalis, E. Shane, G. Lyritis, G. Skarantavos, R. Mendelsohn,and A.L. Boskey: Bone fragility and collagen cross-links. J. BoneMiner. Res. 19, 2000 (2004).

56. A. Boskey and R. Mendelsohn: Infrared analysis of bone in healthand disease. J. Biomed. Opt. 10, 031102 (2005).

57. A. Carden and M.D. Morris: Application of vibrational spectros-copy to the study of mineralized tissues (review). J. Biomed. Opt.5, 259 (2000).

58. J.J. Freeman and M.J. Silva: Separation of the Raman spectralsignatures of bioapatite and collagen in compact mouse bonebleached with hydrogen peroxide. Appl. Spectrosc. 56, 770(2002).

59. J.A. Timlin, A. Carden, M.D. Morris, J.F. Bonadio, C.E. Hoffler II,K.M. Kozloff, and S.A. Goldstein: Spatial distribution of phos-phate species in mature and newly generated mammalian bone byhyperspectral Raman imaging. J. Biomed. Opt. 4, 8 (1999).

60. C.G. Kontoyannis and N.V. Vagenas: FT-Raman spectroscopy:A tool for monitoring the demineralization of bones. Appl. Spec-trosc. 54, 1605 (2000).

61. R.J. Lakshmi, M. Alexander, J. Kurien, K.K. Mahato, andV.B. Kartha: Osteoradionecrosis (ORN) of the mandible: A laserRaman spectroscopic study. Appl. Spectrosc. 57, 1100 (2003).

62. A. Carden, R.M. Rajachar, M.D. Morris, and D.H. Kohn: Ultra-structural changes accompanying the mechanical deformation ofbone tissue: A Raman imaging study. Calcif. Tissue Int. 72,166 (2003).

63. J.W. Ager III, R.K. Nalla, K.L. Breeden, and R.O. Ritchie: Deep-ultraviolet Raman spectroscopy study of the effect of aging onhuman cortical bone. J. Biomed. Opt. 10, 034012 (2005).

64. R.K. Nalla, M. Balooch, J.W. Ager III, J.J. Kruzic, J.H. Kinney,and R.O. Ritchie: Effects of polar solvents on the fracture resis-tance of dentin: Role of water hydration. Acta Biomater. 1,31 (2005).

65. G.M. Kiebzak: Age-related bone changes. Exp. Gerontol. 26,171 (1991).

66. C. Cooper, C. Coupland, and M. Mitchell: Rheumatoid arthritis,corticosteroid therapy, and hip fracture. Ann. Rheum. Dis. 54,49 (1995).

67. N.E. Lane: An update on glucocorticoid-induced osteoporosis.Rheum. Disease Clin. N. Am. 27, 235 (2001).

68. K.G. Saag: Glucocorticoid-induced osteoporosis. Endocrinol.Metab. Clin. N. Am. 32, 135 (2003).

69. T.P. Van Staa, H.S. Leufkens, and C. Cooper: The epidemiologyof cortico-steroid osteoporosis. A meta-analysis. Osteoporos. Int.13, 777 (2002).

70. T.P. Van Staa, R.F. Laan, I.P. Barton, S. Cohen, D.M. Reid, andC. Cooper: Bone density threshold and other predictors of verte-bral fractures in patients receiving oral Glucocorticoid therapy.Arthritis Rheum. 48, 3224 (2003).

71. N.E. Lane, W. Yao, M. Balooch, R.K. Nalla, G. Balooch,S. Habelitz, J.H. Kinney, and L. Bonewald: Glucocorticoid treatedmice have localized changes in trabecular bone material propertiesand osteocyte lacunar size that are not observed in placebo treatedor estrogen deficient mice. J. Bone Miner. Res. 21, 466 (2006).

72. G. Balooch, M. Balooch, R.K. Nalla, S. Schilling, E.H. Filvaroff,G.W. Marshall, S.J. Marshall, R.O. Ritchie, R. Derynck, andT. Alliston: TGF-� regulates the mechanical properties and com-position of bone matrix. Proc. Natl. Acad. Sci. USA 102,18813 (2005).

73. J.J. Kruzic, R.K. Nalla, J.H. Kinney, and R.O. Ritchie: Crackblunting, crack bridging and resistance-curve fracture mechanicsof dentin: Effect of hydration. Biomaterials 24, 5209 (2003).

74. J.H. Kinney, R.K. Nalla, J.A. Pople, T.M. Breunig, and R.O. Ritchie:Age-related transparent root dentin: Mineral concentration, crys-tallite size, and mechanical properties. Biomaterials 26, 3363(2005).

75. O. Akkus, A. Polyakova-Akkus, F. Adar, and M.B. Schaffler:Aging of microstructural compartments in human compact bone.J. Bone Miner. Res. 18, 1012 (2003).

76. M.B. Schaffler, K. Choi, and C. Milgrom: Aging and matrixmicrodamage accumulation in human compact bone. Bone 17,521 (1995).

77. R.W. McCalden, J.A. McGeough, M.B. Barker, and C.M. Court-Brown: Age-related changes in the tensile properties of corticalbone. The relative importance of changes in porosity, mineraliza-tion, and microstructure. J. Bone Joint Surg. Am. 75, 1193 (1993).

J.W. Ager III, et al.: Fracture, aging, and disease in bone

J. Mater. Res., Vol. 21, No. 8, Aug 20061892