Bone 55 (2013) 377–383

Contents lists available at SciVerse ScienceDirect

Bone

j ourna l homepage: www.e lsev ie r .com/ locate /bone

Original Full Length Article

Fracture history of healthy premenopausal women is associatedwith a reduction of cortical microstructural components at thedistal radius☆,☆☆

T. Chevalley a,⁎, J.P. Bonjour a, B. van Rietbergen b, S. Ferrari a, R. Rizzoli a

a Division of Bone Diseases, University Hospitals and Faculty of Medicine, Geneva, Switzerlandb Department of Biomedical Engineering, Eindhoven University of Technology, Eindhoven, Netherlands

☆ Disclosure information's: Bert van Rietbergen is a co☆☆ Funding sources: The Swiss National Science Fou(Grant 3247B0-109799).

⁎ Corresponding author at: Division of Bone Diseaseand Faculty of Medicine, Rue Gabrielle Perret-Gentil 4, CHFax: +41 22 382 99 73.

E-mail addresses: Thierry.Chevalley@hcuge.ch (T. CJean-Philippe.Bonjour@unige.ch (J.P. Bonjour), B.v.Rietb(B. van Rietbergen), Serge.Ferrari@unige.ch (S. Ferrari),(R. Rizzoli).

8756-3282/$ – see front matter © 2013 Elsevier Inc. Allhttp://dx.doi.org/10.1016/j.bone.2013.04.025

a b s t r a c t

a r t i c l e i n f o

Article history:

Received 29 January 2013Revised 4 April 2013Accepted 24 April 2013Available online 6 May 2013

Edited by: R. Baron

Keywords:Healthy premenopausal womenFracture riskAreal and volumetric BMDMicrostructureHR-pQCTBone strength

Objectives: The objective of this study is to determine in healthy premenopausal women with a history of frac-turewhich bone structural components of the distal radius are themost closely associatedwith a risk of fracture.Methods and participants: The method was as follows: measurement of radial areal bone mineral density(aBMD) by DXA, microstructural components by high-resolution quantitative peripheral computerized tomog-raphy (HR-pQCT) and strength variables bymicro Finite Element Analysis (μFEA) in 196 healthy premenopausalwomen aged 45.9 ± 3.7 (±SD) years with (FX, n = 96) and without (NO-FX, n = 100) a history of fracture.We evaluated differences in T-scores between FX and NO-FX and risk of fracture by Odds ratios (OR with 95%confidence intervals, CI) per one SD decrease, using logistic regression analysis after adjustment for age, height,weight, menarcheal age, calcium and protein intakes, and physical activity.Results: In the whole group the mean radial metaphysis aBMD T-score was not significantly different from zero.In the FX as compared to the NO-FX group, the differences in T-scores were as follows: for radial metaphysis:aBMD, −0.24 (P = 0.005); for distal radius microstructure components: cortical volumetric BMD, −0.38(P = 0.0009); cortical thickness, −0.37 (P = 0.0001); cross-sectional area (CSA), +0.24 (P = 0.034); andendosteal perimeter, +0.28 (P = 0.032); and for strength estimates: stiffness, −0.15 (P = 0.030); failure

load, −0.14 (P = 0.044); and apparent modulus, −0.28 (P = 0.006). T-scores of trabecular volumetric BMDand thickness did not significantly differ between the FX and the NO-FX group. Accordingly, the risk of fracture(OR, 95% CI) for 1 SD decrease in radius bone parameters was as follows: radial metaphysis aBMD: 1.70 (1.18–2.44), P = 0.004; cortical volumetric BMD: 1.86 (1.28–2.71), P = 0.001; and cortical thickness: 2.36 (1.53–3.63), P = 0.0001. The corresponding fracture risk for the strength estimates was as follows: stiffness: 1.66(1.06–2.61), P = 0.028; failure load: 1.59 (1.02–2.47), P = 0.041; and apparent modulus: 1.76 (1.17–2.64),P = 0.006.Conclusions: In healthy premenopausal women, a history of fracture is associatedwith reduced T-scores in the dis-tal radius, with the cortical components showing the greatest deficit. A reduction of one SD in cortical thickness isassociated with a nearly three-fold increased risk of fracture. This finding strengthens the notion that, in healthywomen, a certain degree of bone structural fragility contributes to fractures before the menopause and thereforeshould be taken into consideration in the individual prevention strategy of postmenopausal osteoporosis.

© 2013 Elsevier Inc. All rights reserved.

nsultant for Scanco Medical AG.ndation supported this study

s, Geneva University Hospitals-1211 Geneva 14, Switzerland.

hevalley),ergen@tue.nlRene.Rizzoli@unige.ch

rights reserved.

Introduction

In postmenopausal women, low areal bonemineral density (aBMD)is a significant risk factor for fractures [1–4]. Likewise, it is establishedthat low trauma fracture occurring after themenopause is a risk for fur-ther fragility fractures [5–8]. There is also evidence that, in women aged65 years and older, fractures before menopause are independent riskfactors for subsequent fractures after menopause [9]. This greater riskof about 30% of fracture aftermenopause is similar to amaternal historyof fracture or a history of two or more falls in the previous year [9]. Therelation between fracture beforemenopause and deficit in bonemineralmass and microstructural alterations has been well documented in

378 T. Chevalley et al. / Bone 55 (2013) 377–383

women with idiopathic osteoporosis [10–12]. Besides the pathologicalcondition of premenopausal idiopathic osteoporosis, markedly reducedvolumetric (v) BMD, as measured by peripheral quantitative computedtomography (pQCT) at the distal radius, is an important risk factor forlow-energy or fragility fractures in premenopausal women [13]. Inpremenopausal women with fragility fracture, aBMD measured at thespine and proximal femur was markedly lower than that measured inpostmenopausal women without a fracture history [13]. However, inhealthy women without idiopathic osteoporosis, or substantially re-duced vBMD due to secondary osteoporosis [14], whether a fracturehistory recorded before themenopausewould be associatedwith subtlesigns of defective bone integrity remains to be documented.

In the context of parent–offspring investigation we assessedseveral bone variables in healthy premenopausal women (mean age~46 years). These women were mothers of two cohorts of healthygirls [15,16] and boys [17,18] whose bone acquisition characteristicswere studied over childhood and adolescence. Among them, about halfhad a history of fracture. In these healthy premenopausal women,we re-port which, among bone variables measured at the radial level bydual-energy X-ray absorptiometry (DXA), high-resolution peripheralquantitative computerized tomography (HR-pQCT) andmicro-finite ele-ment analysis (μFEA), differ in T-score and are significantly associatedwith an increased risk of fracture before the menopause.

Subjects and methods

Participants

Fracture history andbonemineral density, and cortical and trabecularmicrostructure including porosity as well as bone strength estimatesassessed by HR-pQCT and μFEA, respectively, were studied in 196healthy premenopausal women with a mean age of 45.9 ± 3.7(mean ± SD) years. All subjects lived within the Geneva area. Thepersistence of regular menstruation at the time of examination was aprerequisite for inclusion in this study. The Ethics Committee of theUniversity Hospitals of Geneva approved the protocol and written in-formed consent was obtained from the participants.

Clinical assessment

The participantswere examined over a period of 1.5 years, fromApril2006 to October 2007. Technical staff and instruments used for data ac-quisition did not change throughout the study. Weight, stadiometermeasured standing height and body mass index (BMI, kg/m2) weredetermined in all subjects. Menarcheal age, as defined by the age of thefirst menstruation, was assessed retrospectively. Fracture history sincebirth was retrospectively self-reported, including skeletal site and yearof event. After the age of 20 years, only low-trauma fractures definedas a fall from the standing height were recorded.

Calcium and protein intakesSpontaneous calcium intake, essentially assessed from intake of

dairy sources and protein intakeswas estimated by frequency question-naire [19]. The total animal protein intakewas expressed in either g/d org/kg BW·d−1. This included dairy, meat, fish and egg proteins.

Physical activityUsual physical activity and weight-bearing physical activity were

assessed by a frequency questionnaire [20]. It included organized sports,recreational activity and usual walking and cycling. Subsequently, thecollected data were converted and expressed as physical activity energyexpenditure (PAEE kcal/d) using established conversion formulas [21].

Measurement of bone variables

The aBMD was determined by DXA using a Hologic QDR 4500instrument (Waltham, MA) at radial metaphysis and diaphysis inantero-posterior view as previously reported [22]. The coefficient ofvariation (CV, %) of repeated measurements at these sites, as deter-mined in young healthy adults, varied from 1.0 to 1.6% for BMD.

Volumetric bone density and microstructure were determined atthe distal radius by high resolution peripheral computerized tomog-raphy (HR-pQCT) with a XtremeCT instrument (Scanco medicalAG®, Büttisellen, Switzerland) that acquires a stack of 110 parallelcomputerized tomography slices (9-mm length) with an isotropicvoxel size of 82 μm as previously described [22,23]. Unless therewas a history of fracture, DXA and HR-pQCT measurements wereperformed on the nondominant forearm. The following variableswere measured: total, cortical and trabecular volumetric vBMD bonedensities expressed as milligrams hydroxyapatite (HA) per cubiccentimeter, trabecular bone volume fraction (percent), trabecularnumber, thickness and spacing (micrometers), mean cortical thick-ness (micrometers) and cross-sectional area (square millimeters).

The HR-pQCT images were filtered and binarized using the standardmanufacturer's method [24]. The periosteal and endosteal boundarieswere definedusing an automated contouringmethod similar to the tech-nique described previously by Buie and colleagues [25] as implementedin Image Processing Language (IPL V5.07, Scanco Medical). The auto-mated segmentation used two threshold values and a series of morpho-logic operations (e.g., dilation and erosion operations) to extract theendosteal and periosteal surfaces of the cortex as well as the “apparent”cortical thickness after filling the pores, and was validated earlier[26,27]. Cortical porosity (Ct.Po, %) was calculated as the number ofvoid voxels in each binary cortex image divided by the total numberof voxels [27]. The in vivo short-term reproducibility of HR-pQCT atthe distal radius assessed in 15 subjects with repositioning variedfrom 0.6 to 1.0% and from 2.8 to 4.9% for bone density and for trabecularstructure, respectively. These reproducibility ranges are similar to thosepreviously published [28]. One technician per device performed all thescans, as well as daily quality control phantom, to check for possibledrifts in the X-ray sources.

Finite element analysis

A finite element model of the radius was created directly from thesegmented HR-pQCT images using a procedure similar to that used inearlier clinical studies [29–31]. In summary, a voxel-conversion pro-cedure was used to convert each voxel of bone tissue into an equallysized brick element [32] thus creating micro-finite element (μFE)models that can represent the actual trabecular architecture in detail.The models contained approximately 2 million elements for theradius and could be solved in approximately 3 h. Material propertieswere chosen isotropic and elastic. Both cortical and trabecular boneelements were assigned a Young's modulus of 10 {GPa} and aPoisson's ratio of 0.3 [30,33]. A compression test was simulated torepresent loading conditions during a fall from standing height [34].Bone failure load was calculated as the force for which 2% of thebone tissue would be loaded beyond 0.7% strain [33,35]. In additionto failure load {N}, μFEA-derived variables used in our study alsoincluded stiffness {kilo-Newton per millimeter = kN/mm} and ap-parent modulus {N/mm2}. This latter variable represents a measureof stiffness after correction for cross sectional area. All μFE analyseswere done using the FE solver integrated in the IPL software version1.15 (Scanco Medical AG).

Expression of the results and statistical analysis

The various anthropometric and osteodensitometric variables aregiven as mean ± SD. The differences in density, microstructure,

379T. Chevalley et al. / Bone 55 (2013) 377–383

mechanical parameters and clinical characteristics among participantswith orwithout a positive history of fracturewere assessed by unpairedStudent t-test or by Wilcoxon signed rank test whenever the variablewas not normally distributed. For the differences in density, microstruc-ture and mechanical parameters between fracture and non-fracturesubjects, an analysis of covariance was used to control the influence ofage, height, weight, menarcheal age, calcium and protein intakes, andphysical activity. In order to test whether the deficits in cortical vBMDor in cortical thickness between the FX and NO-FX groups are greaterthan the deficit in metaphysis aBMD, we analyzed whether there wasa significant interaction between the groups (FX and NO-FX) and themeasured variables (cortical vBMD vs. metaphysis aBMD or corticalthickness vs. metaphysis aBMD) using two methods: A) a two wayANOVAwith repeatedmeasure design; and B) amixed linear regressionmodel. The T-score values of radius aBMDbyDXA and ofmicrostructureand mechanical variables at the distal radius by HR-pQCT and μFEAwere calculated from a homogenous cohort of healthy young adultwomen in their early twenties [36]. Associations between density,microstructure, mechanical parameters and fracture status were evalu-ated by logistic regression analysis with adjustment for age, height,weight, menarcheal age, calcium and protein intakes, and physicalactivity. Theywere expressed asOdds ratio (ORwith 95% confidence in-tervals (CI) for a lower age-adjusted aBMD of 1 SD). The significancelevel for two-sided P-values was 0.05 for all tests. The data wereanalyzed using STATA software, version 9.2. (StataCorp LP, CollegeStation, TX, USA).

Results

Fracture history

One hundred and forty-six fractures were reported by 49% of the 196premenopausal women (Table 1). More than one fracture (2 to 5) wasrecorded in 34 of them, accounting for 84 fractures and 58% (84 of146) of all fractures. Fractures occurred in 60 (62.5%) and 36 (37.5%)women up to and after 20 years of age, respectively. Fractures were lo-calized in forearm/wrist (19.2%), hand/fingers (13%), arm/shoulder(10.3%), lower limb (29.5%) and 28.1% at other sites. Anthropometric,nutritional and physical activity did not significantly differ betweenwomen who fractured up to or later than 20 years of age (data notshown). Likewise, there was no significant difference between thesetwo subgroups of fractured women for all measured bone variables

Table 1Characteristics of healthymiddle-aged premenopausal women according to their fracturehistory.

All (n = 196) NO-FX (n = 100) FX (n = 96) P⁎

Age (years) 45.9 ± 3.7 45.9 ± 3.7 46.0 ± 3.7 0.850Height (cm) 164.6 ± 6.1 164.8 ± 6.4 164.6 ± 5.9 0.851Weight (kg) 65.6 ± 11.8 64.5 ± 10.3 66.8 ± 13.1 0.422Body mass index(kg/m2)

24.2 ± 4.2 23.8 ± 3.6 24.7 ± 4.7 0.218

Menarcheal age(years)

13.2 ± 1.6 13.4 ± 1.5 13.0 ± 1.8 0.198

Pregnancies (n)# 2.98 ± 1.23 2.96 ± 1.12 3.01 ± 1.33 0.774Calcium intake(mg/d)

869 ± 408 846 ± 442 892 ± 369 0.153

Protein intake(g/d)

48.0 ± 16.6 48.0 ± 18.4 48.0 ± 14.6 0.409

Protein intake(g/kg BW·d−1)

0.75 ± 0.28 0.76 ± 0.32 0.74 ± 0.23 0.981

Physical activity(kcal/d)

329 ± 195 302 ± 174 356 ± 212 0.051

Values are means ± SD.⁎ P difference between women with (FX) and without (NO-FX) fracture.# The corresponding number of life birth pregnancies was in all: 2.38 ± 0.73. In

NO-FX 2.38 ± 0.79 and FX 2.39 ± 0.67, P = 0.959.

adjusted or not for chronological andmenarcheal age, height andweight,calcium and protein intakes and physical activity (data not shown).

Participants' characteristics

Anthropometric data and reproductive life hallmarks did not differbetweenwomenwith (FX) and thosewithout (NO-FX) a history of frac-ture (Table 1). Likewise, dietary intakes of calcium and protein werevery similar in the two groups (Table 1). The daily energy expenditureresulting from physical activity was ~18% higher in the FX than in theNO-FX group, a difference at the limit of statistical significance(Table 1).

Bone variables

The aBMD T-score of the whole cohort in the radial metaphysis wasnot significantly different from zero (P = 0.418) (Table 2). It was aboutone quarter SD above zero for the radial diaphysis (P = 0.023)(Table 2). The simultaneously measured aBMD T-scores of thewhole cohort (mean ± SD) were, in femoral neck: −0.48 ± 0.98(P b 0.001); in total hip:−0.28 ± 1.07 (P = 0.018); and in L2–L4 ver-tebrae −0.34 ± 1.17 (P = 0.006). Based on aBMD values used in theclinical unit dedicated to the diagnosis of osteoporosis at the UniversityHospital in Geneva [16], 14.6% were osteopenic and 85.4% had normalBMD at the radial metaphysis in the FX group. At the femoral neck,these percentages were 35.4% and 64.6%, respectively. In the FX group,as compared to the NO-FX group, a small but significant reduction inaBMD by about 2.5% was recorded in the radial metaphysis but not inthe radial diaphysis (Table 2).

Among themicrostructural components, total volumetric bonemin-eral density (vBMD)was significantly lower in the FX than in theNO-FXgroup (Table 2). This reduced vBMD was essentially due to a thinnerstandard (−7.2%) as well as “apparent” (−7.0%) cortical thicknessand, but to a smaller extent, to a lower cortical density (−1.9%)(Table 2). By a two way ANOVA analysis as well as a mixed linear re-gression model, the deficits in the FX group, as compared to theNO-FX group, were greater in cortical vBMD (P = 0.004 and P =0.003, respectively) and cortical thickness (P = 0.030 and P = 0.005,respectively) than the deficits in radial metaphysis aBMD. In contrast,there was no significant difference between the FX and NO-FX groupsin the trabecular components: vBMD or BV/TV, Tb.N, Tb.Th, Tb.Spand Tb.Sp SD (Table 2). The thinner cortex was associatedwith a signif-icantly greater CSA and endosteal perimeter (Table 2). Cortical porositywas low and did not differ between the FX and NO-FX groups (Table 2).The strength estimates, stiffness (−2.9%), failure load (−2.6%) and ap-parent modulus (−6.0%) were significantly lower in the FX than in theNO-FX group (Table 2).

Whether the bone size (CSA) greater in the FX than in the NO-FXgroup could account for the statistically significant differences de-scribed above was also analyzed. After further adjustment to CSA,all bone variables (total vBMD, cortical vBMD, cortical thickness and“apparent” cortical thickness, cortical area, trabecular area, stiffness,estimated failure load and apparent modulus) remained significantlylower in the FX group as compared to those in the NO-FX group, withthe exception of the endosteal perimeter (Table 2).

The possibility that the differences in cortical microstructural com-ponents between the FX and NO-FX groups might entirely depend onthe difference in radial metaphysis aBMD was analyzed too. After ad-justment to usual confounding factors and further adjustment to radialmetaphysis aBMD, the deficits in cortical vBMD (P = 0.035) andcortical thickness (P = 0.003) remained significant in premenopausalwomen with a fracture history.

Forearm fractures were recorded in 26 out of the 96 women with afracture history. In this subgroup, cortical vBMD (P = 0.019) and corti-cal thickness (P = 0.014) were also significantly lower in the FX ascompared to the NO-FX group. After further adjustment to CSA, cortical

Table 2Areal BMD, microstructure, porosity and FEA of the distal radius in 196 middle-aged premenopausal women according to their fracture history.

All T-score NO-FX (n = 100) FX (n = 96) P P⁎ P#

aBMD (mg/cm2)Radial metaphysis 457 ± 46 0.09 ± 0.91 463 ± 42 451 ± 50 0.063 0.005 –

Radial diaphysis 723 ± 49 0.26 ± 0.97 727 ± 45 720 ± 54 0.313 0.144 –

vBMD (mg HA/cm3)Total 351 ± 61 0.40 ± 1.08 362 ± 63 339 ± 58 0.011 0.0011 0.014Cortical 951 ± 44 1.32 ± 0.91 960 ± 43 942 ± 42 0.004 0.0009 0.009Trabecular 157 ± 35 −0.17 ± 1.06 157 ± 33 157 ± 37 0.989 0.651 0.849

BV/TV (%) 13.1 ± 2.9 −0.14 ± 1.05 13.1 ± 2.8 13.1 ± 3.1 0.971 0.636 0.869Tb.N (mm−1) 1.91 ± 0.27 −0.31 ± 1.10 1.90 ± 0.26 1.93 ± 0.29 0.625 0.864 0.655Tb.Th (μm) 68.1 ± 9.9 0.01 ± 0.99 68.6 ± 9.6 67.6 ± 10.1 0.478 0.307 0.731Tb.Sp (μm) 467 ± 97 0.37 ± 1.47 470 ± 102 465 ± 93 0.659 0.964 0.812Tb.Sp SD (μm) 205 ± 86 0.69 ± 2.21 203 ± 82 206 ± 91 0.505 0.589 0.533Ct.Th (μm) 854 ± 143 0.26 ± 0.81 884 ± 144 820 ± 133 0.002 0.0001 0.0007Apparent Ct.Th (μm) 946 ± 150 NA 978 ± 159 910 ± 131 0.002 0.0003 0.0017CSA (mm2) 250 ± 42 −0.27 ± 0.94 245 ± 41 256 ± 42 0.069 0.034 –

Cortical area (mm2) 56.2 ± 8.6 NA 57.7 ± 8.7 54.6 ± 8.3 0.037 0.004 0.008Trabecular area (mm2) 192.0 ± 41.4 NA 183.9 ± 36.3 201.1 ± 45.1 0.028 0.001 0.009Cortical porosity (%) 1.22 ± 0.52 3.85 ± 2.46 1.19 ± 0.55 1.25 ± 0.49 0.148 0.401 0.451Endosteal perimeter (mm) 62.0 ± 6.7 −0.29 ± 0.93 61.0 ± 7.0 63.0 ± 6.3 0.047 0.032 0.240Periosteal perimeter (mm) 68.8 ± 6.4 −0.12 ± 0.95 68.2 ± 6.7 69.5 ± 6.1 0.182 0.226 0.343Stiffness (kN/mm) 78.5 ± 12.2 −0.13 ± 0.77 79.5 ± 11.8 77.2 ± 12.6 0.206 0.030 0.018Estimated failure load (N) 3747 ± 578 −0.15 ± 0.79 3793 ± 562 3695 ± 594 0.259 0.044 0.023Apparent modulus (N/mm2) 2003 ± 377 0.08 ± 0.85 2061 ± 370 1938 ± 377 0.029 0.006 0.026

Values are means ± SD. NA: not available. BV/TV, trabecular bone volume fraction; Tb.N, trabecular number; Tb.Sp, trabecular spacing; Tb.Th, trabecular thickness; Ct.Th, corticalthickness; CSA, cross sectional area.P difference between women with (FX) and without (NO-FX) fracture.Bold values indicate significance at p b 0.05.⁎ P after adjustment to age, menarcheal age, height, weight, calcium and protein intakes, and physical activity.# P after adjustment to age, menarcheal age, height, weight, calcium and protein intakes, physical activity and CSA.

380 T. Chevalley et al. / Bone 55 (2013) 377–383

thickness remained significantly lower (P = 0.042) and cortical vBMDwas at the limit of significance (P = 0.060).

The significant differences in the T-score bone variables betweenthe FX and NO-FX groups are illustrated in Fig. 1. The highest likeli-hood of fracture was associated with one standard deviation drop incortical thickness (Odds ratio: 2.36, 95% CI 1.53–3.63) (Fig. 2). This

-0.5

-0.4

-0.3

-0.2

-0.1

0.0

0.1 P = 0.0009

P = 0.034

P

0.2

0.3

0.4

Δ T

-sco

re F

X m

inus

NO

-FX

0.5

P = 0.005 P = 0.0011 P = 0.0001

-0.24 -0.30 -0.38 -0.37

+0.24

Fig. 1. Difference in bone variable T-scores between healthy premenopausal women witprobability (P) of statistical significance are indicated within each and above each column,calcium and protein intakes, and physical activity. The corresponding absolute values for e

probability was somewhat smaller for radial metaphysis aBMD andcortical vBMD (Fig. 2). In contrast, the trabecular variables of thedistal radius were not significantly associated with the risk of fracture(Fig. 2). A drop of one standard deviation in the apparent moduluswas the strength estimate most closely associated with an increasedfracture risk (Fig. 2).

= 0.004

P = 0.001

P = 0.032

P = 0.030 P = 0.044 P = 0.006

-0.36

+0.28

-0.15 -0.14 -0.28

+0.41

h (FX) and without (NO-FX) a history of fracture. The absolute differences and therespectively. These differences were adjusted for age, menarcheal age, height, weight,ach group are given in Table 2.

0.81.01.21.41.61.82.02.2

P = 0.001

P = 0.0001

P = 0.634P = 0.286

1.86

2.36

1.071.18

0.65

2.4

2.6

2.8

Fra

ctur

e R

isk

(OR

, 95%

CI)

P = 0.035

3.0

3.2

3.4

3.6

P = 0.028

1.59

P = 0.041

1.66

P = 0.006

1.76

0.6

1.70

P = 0.004

1.70

P = 0.001

0.4

3.8

Fig. 2. Risk of fracture in healthy premenopausal women for 1 SD decrease in radial aBMD or in microstructure and strength variables of the distal radius. Each column correspondsto Odds ratios ± 95% CI (horizontal line = mean) adjusted for age, menarcheal age, height, weight, calcium and protein intakes, and physical activity. The probability (P) ofstatistical significance is indicated above each column.

381T. Chevalley et al. / Bone 55 (2013) 377–383

Discussion

Relation between bone deficit and fracture risk

Our study shows that fractures occurring before the menopause inhealthy middle-aged women are associated with alterations in severalbone variablesmeasurable at the distal radius. As calculated from the ra-dial metaphysis aBMD, the increased risk of fractures (OR: 1.7, 95% CI1.18.–2.44) is very close to that obtained in postmenopausal women inthe important meta-analysis including eleven separate populationswith about 90,000 person years of observation time and over 2000 frac-tures by Marshall et al. [37]. For measurement at the level of the distalradius, the relative risk (OR, 95% CI) for a lower age-adjusted meanaBMD of 1 SD was 1.7 (1.4–2.0) for forearm fractures and 1.4 (1.3–1.6)for all fractures [37]. Note that this meta-analysis published about16 years ago remains the key reference for the World Health Organiza-tion [38] regarding the performance of bone mineral measurement forassessing the risk of fracture. The similarmagnitude of the risk of fracturebetween the classical study by Marshall et al. [37] in postmenopausalwomen and our results strongly suggests that fracture before the meno-pause should potentially be flagged for osteoporosis prevention at thetime of menopause. The ability of a technique to predict facture is tradi-tionally expressed as the increase in relative risk per standard deviationunit decrease in the selected bone measurement [38]. This gradient ofrisk appears to be similarwhenmeasuring distal forearmBMD in healthypremenopausal and in postmenopausal women. Furthermore, the possi-bility that a fracture occurring during childhood and adolescence [15,36],and now also during the third to the fifth decade, could be a marker ofbone fragility in postmenopausal life as previously proposed [9], is cor-roborated by the results above described.

Bone component contributing to the risk of fracture

The greatest deficit and therefore highest risk was for cortical thick-ness followed by cortical volumetric mineral density. The reducedcortical thickness of the distal radius was associated with higher riskof fracture not only in the forearm and wrist but also in other sites

such as upper arm and shoulder and also in lower limbs. This is in keep-ing with studies showing that microstructural and strength alterationsdetected by HR-pQCT and μFEA in the distal radius were associatedwith higher risk of all types of fragility fractures in both postmenopausalwomen [31,39,40] and men [41]. In our cohort of premenopausalwomen with a history of fracture, the reduced cortical thicknessobserved at the distal radius may have contributed to weaker bonesdespite a larger cross sectional area which is likely a compensatoryresponse to either less endocortical deposition in adolescent girls withlater menarche [16] and increased endocortical resorption with aging[42].

Origin of the fracture-associated distal radius deficits measured inpremenopausal women

About two thirds of the recorded fractures occurred ≤20 years ofage. This suggests that for a substantial number of fractures, the deficitwas already present during bone development. In a recent study wehave documented that in healthy young adult women (mean age:20.4 years), fractures occurring during childhood and adolescencewere also associated with alterations in some distal radius variables[36]. However, the microstructural characteristics associated with thehigher fracture risk differed from those reported here in healthypremenopausal women (mean age: 45.9 years). In the cohort in theirearly twenties [36], the higher risk of prior fractures was rather associ-ated with a deficit in the trabecular [36] than in the cortical compart-ment as reported here in the mid-forty women cohort. Despite thisdifference in the affected bone compartment of the distal radius, theexcess risk of fracture for 1 SD drop in the μFEA-estimated strengthvariables – stiffness, failure load and apparent modulus –was of similarmagnitude, varying from 1.79 to 2.02 in women in their early twenties[36], and from 1.59 to 1.76 in the mid-forty cohort of women reportedin the foregoing analysis. The reason why the structural deficit involvedmore cortical than trabecular components in the midforty than inwomen in their early twenties with a history of fracture can only be amatter of speculation. The HR-pQCT slices proximal to the endplate ofthe distal radius selected in order to define the measured volume of

382 T. Chevalley et al. / Bone 55 (2013) 377–383

interest might not have been localized at the same site. Related to thispossibility, the proportion of load carried by the cortical versus the tra-becular bone might change from the early twenties to the mid-forties,and thus might shift the fracture risk from one to the other compart-ment. Furthermore, the use of a newmethod of segmentation for quan-tifying porosity [43] might reveal a subtle difference in cortical porositybetween premenopausal women with or without a history of fracture.

Strengths of the study

The two groups of participantswith andwithout a history of fracturewere very homogenous in terms of demographic, reproductive life andnutritional variables. Their bone measurements were not made in rela-tion with knowledge of a fracture history. Therefore, the subjects of thisstudy cannot be considered as having been recruited for a typical case–control study, but rather as a randomized sample of healthy womenwith and without a fracture history.

Limitations

Fractures were not prospectively recorded, and based onself-questionnaire. There was a trend to higher physical activitywhich might have increased the number of participants included inthe fracture group. Nevertheless, the significance of the difference inthe bone variables and fractures risk was not attenuated after adjust-ment by the degree of physical activity. Our study cannot ascertainwhether the deficits observed were already present by the end ofthe growth period and thus resulted essentially from a lower peakbone mass, even in those women whose fractures occurred aftertwenty years of age. Alternatively, we cannot rule out that some met-abolic disturbance, acquired after the attainment of peak bone mass,would have been causally related to the greater bone cortical fragilitymeasured in the fracture group. However, a vitamin D insufficiencysevere enough to induce secondary hyperparathyroidism and corticalthinning appears to be unlikely in these healthy premenopausal,physically active and in good nutritional state women.

Conclusions

This study shows that, at the distal radius, a history of fracture inhealthy premenopausal women is associated with a significantlylower T-score with a predominant deficit in the cortical components.A reduction of one standard deviation in the distal radius cortical thick-ness nearly triples the risk of fracture occurring before the menopause.This finding strengthens the notion that in healthy women a subtle de-gree of bone structural fragility contributes to premenopausal fracturesand therefore should be taken into consideration in the individual pre-vention strategy of postmenopausal osteoporosis.

Acknowledgments

We thank Giulio Conicella and the team of the Division of NuclearMedicine for DXA and HR-pQCT measurements; Fanny Merminod,certified dietician, for the assessment of food intakes and her assistancein managing the study; Pierre Casez, MD, for the elaboration of thedatabase; François Herrmann, MD, MPH, for the help with statisticalanalysis. The Swiss National Science Foundation supported this study(Grant 3247BO-109799).

References

[1] Melton III LJ, Atkinson EJ, O'Fallon WM, Wahner HW, Riggs BL. Long-term fractureprediction by bone mineral assessed at different skeletal sites. J Bone Miner Res1993;8:1227–33.

[2] Black DM, Cooper C. Epidemiology of fractures and assessment of fracture risk.Clin Lab Med 2000;20:439–53.

[3] Cummings SR, Melton LJ. Epidemiology and outcomes of osteoporotic fractures.Lancet 2002;359:1761–7.

[4] Johnell O, Kanis JA, Oden A, Johansson H, De Laet C, Delmas P, et al. Predictivevalue of BMD for hip and other fractures. J Bone Miner Res 2005;20:1185–94.

[5] Lauritzen JB, Schwarz P, McNair P, Lund B, Transbol I. Radial and humeral fracturesas predictors of subsequent hip, radial or humeral fractures in women, and theirseasonal variation. Osteoporos Int 1993;3:133–7.

[6] Mallmin H, Ljunghall S, Persson I, Naessen T, Krusemo UB, Bergstrom R. Fracture ofthe distal forearm as a forecaster of subsequent hip fracture: a population-based co-hort study with 24 years of follow-up. Calcif Tissue Int 1993;52:269–72.

[7] CuddihyMT, Gabriel SE, Crowson CS, O'FallonWM,Melton III LJ. Forearm fractures aspredictors of subsequent osteoporotic fractures. Osteoporos Int 1999;9:469–75.

[8] Klotzbuecher CM, Ross PD, Landsman PB, Abbott III TA, Berger M. Patients withprior fractures have an increased risk of future fractures: a summary of the liter-ature and statistical synthesis. J Bone Miner Res 2000;15:721–39.

[9] Hosmer WD, Genant HK, Browner WS. Fractures before menopause: a red flag forphysicians. Osteoporos Int 2002;13:337–41.

[10] Cohen A, Liu XS, Stein EM, McMahon DJ, Rogers HF, Lemaster J, et al. Bonemicroarchitecture and stiffness in premenopausal women with idiopathic osteo-porosis. J Clin Endocrinol Metab 2009;94:4351–60.

[11] Liu XS, Cohen A, Shane E, Stein E, Rogers H, Kokolus SL, et al. Individual trabeculaesegmentation (ITS)-based morphological analysis of high-resolution peripheralquantitative computed tomography images detects abnormal trabecular plateand rod microarchitecture in premenopausal women with idiopathic osteoporo-sis. J Bone Miner Res 2010;25:1496–505.

[12] Misof BM, Gamsjaeger S, Cohen A, Hofstetter B, Roschger P, Stein E, et al. Bone ma-terial properties in premenopausal women with idiopathic osteoporosis. J BoneMiner Res 2012;27:2551–61.

[13] Hung LK, Wu HT, Leung PC, Qin L. Low BMD is a risk factor for low-energy Colles'fractures inwomenbefore and aftermenopause. ClinOrthop Relat Res 2005:219–25.

[14] Ferrari S, Bianchi ML, Eisman JA, Foldes AJ, Adami S, Wahl DA, et al. Osteoporosisin young adults: pathophysiology, diagnosis, and management. Osteoporos Int2012;23:2735–48.

[15] Ferrari SL, Chevalley T, Bonjour JP, Rizzoli R. Childhood fractures are associatedwith decreased bone mass gain during puberty: an early marker of persistentbone fragility? J Bone Miner Res 2006;21:501–7.

[16] Chevalley T, Bonjour JP, Ferrari S, Rizzoli R. Deleterious effect of late menarche ondistal tibia microstructure in healthy 20-year-old and premenopausalmiddle-aged women. J Bone Miner Res 2009;24:144–52.

[17] Chevalley T, Bonjour JP, Ferrari S, Rizzoli R. High-protein intake enhances thepositive impact of physical activity on BMC in prepubertal boys. J Bone MinerRes 2008;23:131–42.

[18] Chevalley T, Bonjour JP, van Rietbergen B, Ferrari S, Rizzoli R. Fractures duringchildhood and adolescence in healthy boys: relation with bone mass, microstruc-ture, and strength. J Clin Endocrinol Metab 2011;96:3134–42.

[19] Fardellone P, Sebert JL, Bouraya M, Bonidan O, Leclercq G, Doutrellot C, et al. Eval-uation of the calcium content of diet by frequential self-questionnaire. Rev RhumMal Osteoartic 1991;58:99–103.

[20] Boot AM, de Ridder MA, Pols HA, Krenning EP, de Muinck Keizer-Schrama SM.Bone mineral density in children and adolescents: relation to puberty, calciumintake, and physical activity. J Clin Endocrinol Metab 1997;82:57–62.

[21] Ainsworth BE, Haskell WL, Leon AS, Jacobs Jr DR, Montoye HJ, Sallis JF, et al.Compendium of physical activities: classification of energy costs of humanphysical activities. Med Sci Sports Exerc 1993;25:71–80.

[22] Chevalley T, Bonjour JP, Ferrari S, Rizzoli R. Influence of age at menarche on fore-arm bone microstructure in healthy young women. J Clin Endocrinol Metab2008;93:2594–601.

[23] Boutroy S, Bouxsein ML, Munoz F, Delmas PD. In vivo assessment of trabecularbone microarchitecture by high-resolution peripheral quantitative computedtomography. J Clin Endocrinol Metab 2005;90:6508–15.

[24] Laib A, Hauselmann HJ, Ruegsegger P. In vivo high resolution 3D-QCT of thehuman forearm. Technol Health Care 1998;6:329–37.

[25] Buie HR, Campbell GM, Klinck RJ, MacNeil JA, Boyd SK. Automatic segmentation ofcortical and trabecular compartments based on a dual threshold technique for invivo micro-CT bone analysis. Bone 2007;41:505–15.

[26] Burghardt AJ, Kazakia GJ, Ramachandran S, Link TM, Majumdar S. Age- andgender-related differences in the geometric properties and biomechanical signif-icance of intracortical porosity in the distal radius and tibia. J Bone Miner Res2010;25:983–93.

[27] Nishiyama KK, Macdonald HM, Buie HR, Hanley DA, Boyd SK. Postmenopausalwomen with osteopenia have higher cortical porosity and thinner cortices atthe distal radius and tibia than women with normal aBMD: an in vivo HR-pQCTstudy. J Bone Miner Res 2010;25:882–90.

[28] MacNeil JA, Boyd SK. Accuracy of high-resolution peripheral quantitative computedtomography for measurement of bone quality. Med Eng Phys 2007;29:1096–105.

[29] Boutroy S, Van Rietbergen B, Sornay-Rendu E, Munoz F, Bouxsein ML, Delmas PD.Finite element analysis based on in vivo HR-pQCT images of the distal radius isassociated with wrist fracture in postmenopausal women. J Bone Miner Res2008;23:392–9.

[30] Dalzell N, Kaptoge S, Morris N, Berthier A, Koller B, Braak L, et al. Bonemicro-architecture and determinants of strength in the radius and tibia:age-related changes in a population-based study of normal adults measuredwith high-resolution pQCT. Osteoporos Int 2009;20:1683–94.

[31] Vilayphiou N, Boutroy S, Sornay-Rendu E, Van Rietbergen B, Munoz F, Delmas PD,et al. Finite element analysis performed on radius and tibia HR-pQCT images andfragility fractures at all sites in postmenopausal women. Bone 2010;46:1030–7.

383T. Chevalley et al. / Bone 55 (2013) 377–383

[32] van Rietbergen B,WeinansH,Huiskes R, Odgaard A. A newmethod to determine tra-becular bone elastic properties and loading using micromechanical finite-elementmodels. J Biomech 1995;28:69–81.

[33] PistoiaW, van RietbergenB, Lochmuller EM, Lill CA, Eckstein F, Ruegsegger P. Estima-tion of distal radius failure load with micro-finite element analysis models based onthree-dimensional peripheral quantitative computed tomography images. Bone2002;30:842–8.

[34] Chiu J, Robinovitch SN. Prediction of upper extremity impact forces during falls onthe outstretched hand. J Biomech 1998;31:1169–76.

[35] Mueller TL, Christen D, Sandercott S, Boyd SK, van Rietbergen B, Eckstein F, et al.Computational finite element bone mechanics accurately predicts mechanicalcompetence in the human radius of an elderly population. Bone 2011;48:1232–8.

[36] Chevalley T, Bonjour JP, van Rietbergen B, Rizzoli R, Ferrari S. Fractures in healthyfemales followed from childhood to early adulthood are associated with latermenarcheal age and with impaired bone microstructure at peak bone mass. JClin Endocrinol Metab 2012;97:4174–81.

[37] Marshall D, Johnell O, Wedel H. Meta-analysis of how well measures of bone min-eral density predict occurrence of osteoporotic fractures. BMJ 1996;312:1254–9.

[38] World Health Organization. Assessment of osteoporosis at the primary healthcare level. Report of a WHO scientific group; 2007.

[39] Sornay-Rendu E, Boutroy S, Munoz F, Delmas PD. Alterations of cortical andtrabecular architecture are associated with fractures in postmenopausal women,partially independent of decreased BMD measured by DXA: the OFELY study. JBone Miner Res 2007;22:425–33.

[40] Vico L, Zouch M, Amirouche A, Frere D, Laroche N, Koller B, et al. High-resolutionpQCT analysis at the distal radius and tibia discriminates patients with recentwrist and femoral neck fractures. J Bone Miner Res 2008;23:1741–50.

[41] Vilayphiou N, Boutroy S, Szulc P, van Rietbergen B, Munoz F, Delmas PD, et al. Fi-nite element analysis performed on radius and tibia HR-pQCT images and fragilityfractures at all sites in men. J Bone Miner Res 2011;26:965–73.

[42] KimKM, Brown JK, KimKJ, ChoiHS, KimHN,RheeY, et al. Differences in femoral neckgeometry associated with age and ethnicity. Osteoporos Int 2011;22:2165–74.

[43] Zebaze R, Ghasem-Zadeh A, Mbala A, Seeman E. A newmethod of segmentation ofcompact-appearing, transitional and trabecular compartments and quantificationof cortical porosity from high resolution peripheral quantitative computed tomo-graphic images. Bone 2013;54:8–20.