- Email: [email protected]
Comparison of 2D and 3D bone microarchitecture evaluation at the femoral neck, among postmenopausal women with hip fracture or hip osteoarthritis
Bone 49 (2011) 1055–1061
Contents lists available at SciVerse ScienceDirect
Bone j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / b o n e
Original Full Length Article
Comparison of 2D and 3D bone microarchitecture evaluation at the femoral neck, among postmenopausal women with hip fracture or hip osteoarthritis☆ Stéphanie Boutroy a,⁎, Nicolas Vilayphiou a, Jean-Paul Roux a, Pierre D. Delmas a, Hubert Blain b, Roland D. Chapurlat a, Pascale Chavassieux a a b
INSERM, UMR 1033, Université de Lyon, Lyon, France Department of Internal Medicine and Geriatrics, University Hospital of Montpellier, University Montpellier 1, Montpellier, France
a r t i c l e
i n f o
Article history: Received 20 January 2011 Revised 30 June 2011 Accepted 25 July 2011 Available online 2 August 2011 Edited by: Rene Rizzoli Keywords: Femoral neck Microarchitecture Microcomputed tomography Hip osteoarthritis Hip fracture
a b s t r a c t Objectives: High resolution peripheral quantitative tomography (HR-pQCT) is used more widely to assess microarchitecture, but we are lacking comparisons between HR-pQCT and histomorphometry, which is considered the gold standard. They have only been assessed on different anatomical regions. The purpose of our study was to assess the microarchitecture and the relative contribution of cortical and trabecular bone in hip fracture with this 3D imaging technique, compared with the 2D histomorphometry. Material and methods: We compared the distribution of cortical and trabecular bone in the ultradistal femoral neck samples (~ 3 mm thick) obtained after total hip replacement in 21 hip osteoarthritis (HOA, 66 ± 8 yrs) and 20 hip fracture (HF, 79 ± 8 yrs) menopausal women by a direct 3D evaluation method (HR-pQCT: XtremeCT, Scanco Medical AG) and by histomorphometry, performed and averaged on three 10 μm-thick sections 800 μm apart. Results: Significant correlations were found between both techniques for trabecular bone volume, number, thickness, separation and cortical thickness (0.51 b r′ b 0.81, p b 0.01). The connectivity was also significantly correlated (r′ = 0.58, p b 0.001) between both techniques, as well as the trabecular bone pattern factor measured in 2D with the structural model index (SMI) measured in 3D (r′ = 0.62, p b 0.001). However HRpQCT overestimated the absolute value of most parameters, with higher values being even more overestimated. The agreement between the two techniques was weak for cortical porosity. With the 3D measurements we found that trabecular bone volume was 43% lower in HF than HOA (p b 0.01), associated with loss of trabecular connectivity (− 50%, p b 0.01) and a more rod-like structure (SMI, 22%, p b 0.01), mainly at the inferior (34%, p b 0.01) and posterior (22%, p b 0.05) quadrants. Cortical thickness was found to be lower in the posterior quadrants (− 22%, p b 0.05) and tended to be lower in HF than in HOA at the inferior quadrant (− 14%, p = 0.08), but it was still the highest at the inferior quadrant in both groups. In conclusion, 3D methods confirmed the alteration of trabecular and cortical bone found by histomorphometry in HF compared with HOA and the frequency of the rod-like structure in HF. © 2011 Elsevier Inc. All rights reserved.
Introduction The evaluation of bone microarchitecture has gained increased attention in the past years, with the recognition that the usual areal bone mineral density (aBMD) measurements failed to identify most fragility fractures [1–3]. Indeed, several limitations are associated with aBMD measurements: aBMD values are influenced by bone size as DXA does ☆ This work was supported in part by unrestricted grants from Eli Lilly and by the French Ministry of Health (Projet Hospitalier de Recherche Clinique Régional, Languedoc-Roussillon, 2003, UF-7755). ⁎ Corresponding author at: INSERM UMR 1033, Hôpital Edouard Herriot, Pavillon F, 69437 Lyon cedex 03, France. Fax: + 33 4 72 11 74 32. E-mail addresses: [email protected] (S. Boutroy), [email protected] (N. Vilayphiou), [email protected] (J.P. Roux), [email protected] (H. Blain), [email protected] (R.D. Chapurlat), [email protected] (P. Chavassieux). 8756-3282/$ – see front matter © 2011 Elsevier Inc. All rights reserved. doi:10.1016/j.bone.2011.07.037
not measure true volumetric BMD; DXA cannot distinguish cortical and trabecular bone compartments and moreover DXA cannot directly measure cortical and trabecular architecture. However structural changes are important determinants of bone strength and have been associated with hip fracture independently of aBMD [4]. Bone histomorphometry allows for assessment of bone microarchitecture and remains the gold standard for this purpose. It also permits a dynamic evaluation by estimating the level of bone remodeling. However, histomorphometry has inherent limitations relative to its 2D design. Measurements are based on the assumption of a plate-like structure of bone and they are limited to a few slices thus representing a very small fraction of the biopsy. Additionally, histomorphometry precludes the assessment of non-metric parameters able to provide insight into bone quality, such as the structure model index. By conventional 2D histomorphometric analysis of femoral neck biopsies, we have previously shown that in addition to cortical
1056
S. Boutroy et al. / Bone 49 (2011) 1055–1061
thinning, loss of trabecular bone mass and connectivity may distinguish women with hip fracture (HF) from women with hip osteoarthritis (HOA) [5]. To overcome the 2D histomorphometric limitations, those femoral neck samples have been scanned with a 3D device. Three-dimensional analysis of the femoral neck structure by high-resolution peripheral quantitative computed tomography will assess the microarchitecture and the relative contribution of cortical and trabecular bone in women with hip fracture when compared to women with hip HOA. Material and methods Subjects The subjects and bone biopsies procedures were described in detail in our earlier study, in which cortical and trabecular bone distribution at the femoral neck in fractured and hip osteoarthritis women was assessed by histomorphometry [5]. Briefly, femoral neck biopsies were obtained during arthroplasty for non-traumatic hip fracture (HF), i.e. fall from standing height or less, (HF; n = 20; aged 79.3 ± 8.0 years old) or hip osteoarthritis (HOA; n = 21; aged 65.9 ± 7.9 years old) in postmenopausal women. For both HOA and HF patients, femoral neck biopsies were taken at the ultradistal part of the femoral neck, beginning at the cut face prepared for the insertion of the prosthesis at the base of the neck, thus leading to a bone slice of 10 mm thick approximately. Among the 20 HF, there were 19 cervical fractures and 1 trochanteric fracture. Undecalcified femoral neck samples were fixed in 70% ethanol and embedded in methylmethacrylate [6]. The protocol was approved by an independent Ethics Committee, and all patients gave written informed consent before participation. Measurement of 3D bone microarchitecture Three-dimensional bone microarchitecture of embedded femoral neck biopsies was measured using a high-resolution peripheral quantitative computed tomography system (XtremeCT, Scanco Medical AG®, Brüttisellen, Switzerland), with a nominal isotropic voxel size of 82 μm, using the standard in vivo scanning protocol (60 kVp, 900 μA) [7]. According to the bone microarchitecture regional differences in femoral neck [5,8], each section was divided into four quadrants (superior, anterior, inferior and posterior). CT slices were visually matched with histomorphometric analysis to have consistent regions of interest (Fig. 1), leading to mean ROI of 33 ± 6 CT slices (2.7 ± 0.5 mm). Trabecular and cortical parameters were first assessed in their respective whole compartment and then on the 4 quadrants separately. Image processing included a Gaussian filter (support = 1, σ = 0.8). Cortical bone was segmented manually from trabecular bone on a slice-by-slice basis and both volumes were thresholded separately to delineate bone from non-bone voxels (Fig. 1) [9]. As shown in Fig. 1, the global threshold used for trabecular bone segmentation (337 mg.HA/cm 3) tended to slightly overestimate the trabecular structure in order to keep an intact connectivity with the relatively low resolution in regard of the structure measured, whereas the
global threshold for cortical bone segmentation (511 mg.HA/cm 3) was chosen to differentiate the cortical porosity [10]. To avoid potential errors in trabecular spacing measurement due to the relatively small ROI, mirror images of the sample were added at the 2 extremities of the sample, otherwise trabecular spacing may have been underestimated [11]. The trabecular bone volume ratio (BV/TV) was determined by dividing the number of voxels representing trabecular bone by the total number of voxels in the trabecular compartment. Methods used to process the trabecular microarchitectural data were based on direct measurements of structure. Trabecular number (Tb.N*, 1/mm), thickness (Tb.Th*, μm) and separation (Tb.Sp*, μm) measurements were based on the distance transformation method, where maximal spheres are filled into the segmented object [12], thus not relying on an assumed model (rod/plate) type. Non-metric indices for trabecular bone structure as SMI [13] and Conn.D [14] were also calculated from the segmented images. Mean cortical thickness (Ct.Th, μm) was calculated using an annular method [15] where mean cortical volume is divided by the periosteal surface area, thus being irrespective of any pores. Cortical porosity was calculated as the complement of the cortical bone volume ratio (1 −(BVcortical/TVcortical)) in the compact-appearing cortex (Ct.Po, %). Two HF biopsies were excluded because the material was insufficient to be measured in 3D, with a total of 21 HOA and 18 HF as a result. Moreover, in most HF biopsies, the cortex from the superior quadrant was impaired or missing. Thus, cortical thickness and porosity were measured in only 3 samples at this site. Histomorphometry Serial 10 μm-thick cross-sectional slices were cut at three different levels of the bone sample, 800 μm apart, and were stained with Goldner trichrome [16]. Measurements were done separately on the four quadrants, the mean value of each sample being the average of the 4 quadrants among the 3 analyzed slices. The trabecular ROI corresponded to a rectangular area of maximum 31 mm², centered in each quadrant and cortical measurements were evaluated in the area corresponding to the trabecular ROI, as shown in Fig. 1. Parameters of bone structure and micro-architecture were performed using Bone and MorphoExpert softwares (Explora Nova, La Rochelle, France) and standard abbreviations for bone histomorphometry were used [17]. Outcome parameters for trabecular bone were volume (BV/TV,%), thickness (Tb.Th; μm), separation (Tb.Sp; μm), number (Tb.N; 1/mm) based on Parfitt's formulae, i.e. derived from volume and surfaces measurements [18]. After skeletonization, we evaluated the number of nodes (N.Nd/TV, 1/mm²) and the trabecular bone pattern factor (TBPf; 1/mm), both reflecting the connectivity/topology of the network. Outcome parameters for cortical bone were thickness (Ct.Th; μm) and porosity (Ct.Po, %) representing the cortical area occupied by Haversian canals. Statistical analysis Descriptive statistics were summarized by means and standard deviations. Regarding the distribution of the variables, non-parametric
Fig. 1. Cross-section of a femoral neck. A — Histomorphometric slice showing the four quadrants and the regions of interest for trabecular and cortical bone (rectangle and its projection on the cortex respectively) B — HR-pQCT gray-level slice of the same sample C–E — 3D reconstruction of the whole trabecular (C) and cortical (porosity shown in gray) (D) bone and of each quadrants (E).
S. Boutroy et al. / Bone 49 (2011) 1055–1061
tests were performed. The relationships between 2D and 3D parameters were studied using Spearman correlation (r′) in the entire data set and within-group, and 2D and 3D values were compared by paired Wilcoxon Signed Rank test. Bland–Altman plots were provided for qualitative analysis of accuracy (Fig. 2). The differences among women with HF and HOA were assessed by Wilcoxon Signed Rank test. In each group, the regional comparisons between the 4 quadrants (3 quadrants for cortex because of impaired materiel) were performed by Friedman test, followed when significant by paired Wilcoxon Signed Rank test for pairwise comparisons. The differences among women with HF and HOA were also assessed in a subset of the population ranging from 63 to 81 years old to have groups comparable in age. Statistical analyses were performed using SPSS software (version 16.0). Results Comparison of 2D and 3D parameters Despite mean values differences between 2D and 3D measurements (Tables 1a and 1b), significant correlations (r′) were observed between both techniques, ranging from 0.51 (p = 0.001) for trabecular thickness to 0.81 (p b 0.001) for trabecular bone volume (Fig. 2).
1057
Moreover, 2D TBPf was correlated with the 3D SMI (r′ = 0.62, p b 0.001), and the number of nodes was associated with the connectivity density (r′ = 0.58, p b 0.001). 3D mean cortical thickness of the whole cortical compartment was almost similar to 2D cortical thickness (+2.4% in 3D vs 2D) with a coefficient of correlation of 0.72 (p = 0.004). Cortical porosity measured in 3D and 2D was not correlated for the full sample, yet measurements were only available in a subset of the population because of impaired cortical bone in the superior quadrant in HF. When analyzed by quadrants, cortical porosity measured in 2D and 3D was correlated for the inferior and anterior quadrants (r′ = 0.37–0.54, p b 0.05). These correlations were almost the same when studied only in the HF or hip HOA group (data not shown). Bland–Altman plots were provided to visualize the agreement between 2D and 3D measurements. Dotted lines represented the lower and higher limit of agreement, i.e. average difference ± 1.96 standard deviation of the difference (Fig. 2). For most parameters, differences between the 2 techniques were smaller for lower values and increased with higher values of the measured parameters (Figs. 2B–D and F). Tb.Th measured in 3D was systematically higher than measured in 2D. There was no systematic disagreement for cortical porosity (Fig. 2E). A few values in the upper range were out of the limit of agreement between the two techniques.
Fig. 2. Correlation and Bland–Altman plots between 2D and 3D methods. (A) — Correlation of BV/TV measured by 2D vs. 3D methods in the whole sample. (B–D) — Bland–Altman plot of BV/TV (B), Tb.N (C) and Tb.Th (D) assessed in the whole sample. (E–F) — Bland–Altman plot of Ct.Po (E) and Ct.Th (F) assessed in the inferior quadrant.
1058
S. Boutroy et al. / Bone 49 (2011) 1055–1061
Table 1a Mean values and correlations between 2D and 3D parameters in the entire sample.
BV/TV (%) Trabecular number (1/mm) Trabecular thickness (μm) Trabecular separation (μm) TBPf (2D)/SMI (3D) N.Nd/TV (2D)/Conn.D (3D) Cortical thickness (μm) a Cortical porosity (%) a
2D values
3D values
% Difference
Correlation
Mean ± SD
Mean ± SD
3D vs. 2D
r′
11.1 ± 7.1 0.51 ± 0.20 312 ± 44 2358 ± 1096 2.8 ± 0.6 0.33 ± 0.25 1902 ± 355 9.9 ± 4.7
43.2⁎⁎
0.81⁎⁎ 0.69⁎⁎ 0.51⁎⁎ 0.67⁎⁎ 0.62⁎⁎ 0.58⁎⁎ 0.72⁎
7.8 ± 4.2 0.78 ± 0.31 94 ± 19 2067 ± 1989 3.6 ± 4.0 0.80 ± 0.63 1857 ± 513 12.6 ± 4.2
− 35.0⁎⁎ 231.1⁎⁎ 14.1⁎ NA NA 2.4 − 21.7
0.06
NA: non-applicable. a Measurements were only available in a subset of the population because of impaired cortical bone in the superior quadrant in HF. ⁎ p b 0.01. ⁎⁎ p ≤ 0.001.
Within-group regional differences in 3D parameters When comparing the 4 quadrants in the HOA group, we observed significant regional differences in trabecular BV/TV (p = 0.025), thickness (p = 0.01) and SMI (p = 0.002) (Fig. 3). The same pattern was observed in the HF group but it reached significance only for trabecular thickness (p = 0.001). BV/TV and trabecular thickness were the highest in the inferior quadrant and were significantly associated with a more plate-like structure in the HOA group as shown by a lower value of SMI. In both HOA and HF groups, we observed strong regional differences in cortical thickness (p b 0.001 and p = 0.009 respectively) and porosity (p b 0.001 and p = 0.006, respectively) (Fig. 4). In both groups, cortical thickness was the highest in the inferior quadrant, where cortical porosity was the lowest. Differences in 3D parameters between HF and hip HOA samples Trabecular bone On the entire femoral neck (Table 2), HF samples had 43% lower trabecular bone volume than HOA samples (p = 0.005), due to loss of trabeculae, which were significantly thinner (−11%, p b 0.01). Similar trends were observed for each quadrant (Fig. 3). These results were also associated with a loss of the trabecular connectivity (−50%, p = 0.009), mainly in the superior and inferior quadrants (− 60% and −43% respectively, p b 0.05). This was characterized by a more rodlike structure as shown by the significant higher SMI (+ 22%, p = 0.002) in the HF group than in the HOA group, mainly in the inferior and posterior quadrants (34% and 22% respectively, p b 0.05) (Fig. 3).
Table 1b Mean values and correlations between 2D and 3D cortical parameters by quadrants.
Ct.Th — Post (μm) Ct.Th — Inf (μm) Ct.Th — Ant (μm) Ct.Po — Post (%) Ct.Po — Inf (%) Ct.Po — Ant (%)
2D values
3D values
% Difference
Correlation
Mean ± SD
Mean ± SD
3D vs. 2D
r′
1064 ± 413
1372 ± 434
3234 ± 1244
2956 ± 695
1206 ± 514
1479 ± 369
22.7⁎
0.44⁎
11.2 ± 5.5 9.8 ± 5.6 14.8 ± 9.2
11.0 ± 8.4 5.1 ± 4.1 15.4 ± 8.5
− 1.8 − 48.2⁎⁎ 4.1
0.02 0.37⁎ 0.54⁎⁎
29.0⁎⁎ − 8.6
0.71⁎⁎ 0.84⁎⁎
Ct.Th = cortical thickness, Ct.Po = cortical porosity, Post = posterior quadrant, Inf = inferior quadrant, Ant = anterior quadrant. ⁎ p b 0.05. ⁎⁎ p ≤ 0.001.
Cortical bone When compared to HOA, cortical thickness was significantly lower in the posterior quadrant (−22%, p = 0.025) and tended to be lower in the inferior quadrant (−14%, p = 0.08) in HF. Mean cortical porosity assessed in 3D was not significantly different between HF and HOA in any quadrants (posterior, inferior and anterior) but was lower in HF than in HOA (−39 to −42%, p b 0.05) when assessed by histomorphometry. Subset of HF and HOA with comparable age As HF were in average 13 years older than HOA, we determined a subset of biopsies ranging from 63 to 81 years old, including 10 HOA and 12 HF (respectively 71 ± 6 and 75 ± 5 years old, p = 0.13). With the 2D method, most of the trabecular parameters (−39%, −20%, +103% for BV/TV, Tb.Th and Tb.Sp respectively, p b 0.05) remained significantly impaired in HF compared to HOA, as well as cortical parameters (−30%, −25%, −44% for Ct.Th posterior, Ct.Th anterior and Ct.Po inferior, p b 0.05). With the 3D method, most of the trabecular and cortical parameters were also impaired in the HF group but the results were not significant (− 23%, −5%, −13%, −41%, for BV/TV, Tb.Th, Ct.Th posterior and Ct.Po inferior respectively), unless for SMI (2.7 ± 0.4 vs 3.0 ± 0.4, respectively HOA and HF, p = 0.03). Discussion In this study, the 3D method confirmed the cortical thinning and the loss of trabecular connectivity found by histomorphometry in postmenopausal women with a femoral neck fracture when compared to women with hip osteoarthritis [5]. In addition 3D analysis outlined the prevalence of a disconnected rod-like structure at the femoral neck in HF women compared to HOA women. Bone fragility relates to both cortical and trabecular bone, but their contributions to bone strength vary throughout the skeleton. At the proximal femur, a study of the distribution of maximal principal and effective strain by finite element analysis has found that the contributions of the trabecular and cortical bone to bone strength are quite similar [19]. However, the load distribution between cortical and trabecular bone varies strongly within the proximal femur, cortical bone carrying as low as 30% at the subcapital region, to approximately 50% at the mid-neck and up to 96% at the base of the neck [20]. These computational results have recently been supported by Holzer et al. [21]. Therefore, while focusing on femoral neck, a thick cortical shell is preferred over a dense trabecular core [4,19,22]. However, in an earlier study of younger individuals [23], the role of the trabecular bone was greater than in the analysis of Holzer et al. This may reflect trabecular bone loss with age. The axis of greatest strain is the inferoanterior to superoposterior direction in both stance and fall configuration, which results in a
S. Boutroy et al. / Bone 49 (2011) 1055–1061
1059
Fig. 3. 3D trabecular microarchitecture of femoral neck biopsies by quadrants from women with HF or HOA. Comparison HF vs. HOA (* p b 0.05), comparison between quadrants ( p b 0.05), mean ± SEM. Sup = superior quadrant, Post = posterior quadrant, Inf = inferior quadrant, Ant = anterior quadrant.
similar stress/strain distribution [20]. Several studies suggested that hip fractures may be initiated at the superior neck, by buckling of the thin superior cortex [4]. However, femoral neck is a complex structure and trabecular bone may have consequences for cortical bone resistance. Indeed, trabecular bone may be seen as buttressing the cortex, thus stiffening it and reducing the risk of buckling. Jordan et al. found that the lesser femoral neck fracture risk observed in HOA may be conferred principally through structural effects on trabecular bone, as they observed an increased trabecular bone mass and connectivity in HOA at this site [24]. In our study, we found not only a significant decrease in cortical thickness in patients with HF compared to HOA but also an impairment of trabecular bone. Other studies have focused on trabecular microarchitecture at the femoral head [25–27], or at the intertrochanteric region [28,29] and consistently with our study, Zhang et al. [27] reported, at the femoral head, higher BV/TV, trabecular thickness and a more plate-like structure in HOA compared to HF, but no difference in Conn.D. We observed correlations between histomorphometry and 3D direct measurements of the femoral neck biopsies. Most previous studies have shown a link between μCT measurements and bone histomorphometry measurements. However, comparisons between HR-pQCT and histomorphometry have only been assessed on different anatomical regions [30], thus reporting either non-significant or weak correlations [12]. Laib et al. [31], and more recently MacNeil et al. [32],
reported higher correlation between μCT measurements (with a resolution of 28 and 19 μm) and 3D-pQCT (165 μm) or HR-pQCT (82 μm) measurements on ex vivo femoral head and radii, respectively. In the latter study, they showed that application of direct measurement techniques from HR-pQCT scans gave good correlations for BV/TV (r = 0.84) and Tb.N* (r = 0.97) [32]. However HR-pQCT overestimated the absolute value of most parameters, as we also observed in our study, with higher value being even more overestimated. Similar to our study, for mean BV/TV values ranging from 5 to 20%, they observed differences between the two techniques ranging from 0 to almost 15% for higher values of BV/TV. This was particularly true for Tb.Th*, with trabeculae found to be 2 to 3 times thicker than measured by histomorphometry, but still correlated with trabecular width measured in 2D. This overestimation may be due to partial volume effect as the resolution of the device is 82 μm whereas the thickness of trabeculae is close to 100 μm. We also found a lower number of trabeculae with the 3D than the 2D methods. This can also be related to partial volume effect, with relatively small trabeculae not correctly depicted. On the other hand, the trabecular number may be overestimated on thin slices, in 2D: a single trabeculae can be cut twice and thus counted as two trabeculae by histomorphometry. The number of nodes and the trabecular bone pattern factor measured by histomorphometry reflect the connectivity/topology of the network and can correspond to density connectivity (Conn.D) and
Fig. 4. 3D cortical thickness and porosity of femoral neck biopsies by quadrants from women with HF or HOA. Comparison HF vs. HOA (* p b 0.05), comparison between quadrants ( p b 0.05), mean ± SEM. a measurements were only available in a subset (n = 3) of the HF because of impaired cortical bone in the superior quadrant. Sup = superior quadrant, Post = posterior quadrant, Inf = inferior quadrant, Ant = anterior quadrant.
1060
S. Boutroy et al. / Bone 49 (2011) 1055–1061
Table 2 3D trabecular microarchitecture of femoral neck biopsies from women with HF or HOA (mean values and percentage differences).
BV/TV (%) Trabecular number (1/mm) Trabecular thickness (μm) Trabecular separation (μm) SMI Conn.D
HF
HOA
% Difference
Mean ± SD
Mean ± SD
HF vs HOA
7.9 ± 4.3 0.48 ± 0.16 293 ± 31 2339 ± 785 3.1 ± 0.4 0.22 ± 0.16
13.9 ± 7.9 0.59 ± 0.17 329 ± 47 1878 ± 572 2.5 ± 0.7 0.43 ± 0.22
− 42.8⁎⁎ − 17.7 − 10.9⁎ 24.5 22.0⁎⁎ − 49.7⁎
⁎ p b 0.01. ⁎⁎ p ≤ 0.005.
Structure Model Index (SMI) assessed in 3D. Indeed, we found that SMI and Conn.D performed quite well with their homologous 2D parameters: trabecular bone pattern factor and number of nodes. Even better relationship (r = 0.95) was ascertained by Thomsen et al. between TBPf and SMI measured with an isotropic 20 μm resolution on proximal tibial metaphysis necropsies [33]. Sode et al. found that SMI and Conn.D measured by HR-pQCT at 82 μm resolution on ex vivo distal tibia strongly correlated with that measured at 16 μm by μCT [34]. MacNeil et al. reported similar relationship for Conn.D but not for SMI measured by HR-pQCT and μCT at 19 μm on cadaver radii [32]. One of the limitations of our study is the small sample size. The age difference between women with HF and HOA and the lack of healthy controls have already been addressed in details [5]. Indeed, most parameters obtained in 3D were altered in HF compared to HOA but were no more significant, unless for SMI (p = 0.03) when studied in a subset of HF and HOA of comparable age. Other limitations can explain some moderate relationships found in our study between HR-pQCT and histomorphometry. First, the measurements of histomorphometric parameters were based on plate model assumptions whereas 3D measurements were directly measured without model assumption. Second, histomorphometric parameters may not be representative of the whole sample and results may be biased by regional variation. At the iliac crest, several studies have shown an intra-individual difference of 15–30% in trabecular bone volume between contiguous biopsies [35–38]. In order to obtain the best representativeness, histomorphometry measurements were performed on 3 separated slices covering the whole tissue area measured in 3D (in the first, middle and last slices of the 3D ROI). Yet, differences observed in cortical thickness, especially in the inferior quadrant, can partly be attributed to the difference in ROI between 2D and 3D. Indeed, in 2D, Ct.Th was measured in a rectangle centered on the quadrant, i.e. where the cortex is thicker, whereas in 3D, Ct.Th was measured in the entire quadrant, thus including thinner regions near the posterior and anterior quadrants. Third, discrepancies might be explained by the resolution of the 3D device (82 μm) which was probably not sufficient to depict accurately some parameters such as small cortical pores and non-metric parameters. Sode et al. showed that non-metric parameters were dependent on the resolution and may be limited by partial volume effect; therefore they should be interpreted with caution [34]. The impact of partial volume effect on the measurements is related to the amount of tissue i.e., higher value of BV/TV results in an increased overestimation of itself, as shown in Fig. 2. The lack of agreement between histomorphometry and HRpQCT for cortical porosity (Fig. 2E) is likely related to the size of pores detectable by each technique. Indeed in 3D, only pores bigger than 82 μm × 82 μm × 82 μm can be measured whereas smaller pores, which can be observed at different densities in both low and high porosity samples, were depicted by histomorphometry. This is consistent with the higher porosity observed in HOA than in HF patients when measured in 2D but not in 3D. Moreover, cortical thinning concomitant to cortical porosity in HF, so-called trabecularized cortex, may partly contribute to their similar or lower cortical porosity
compared to HOA. Reproducibility of cortical porosity was not computed in this study but was reported in vivo by Burghardt et al. to be 12% at the radius and 4% at the tibia in women aged 61± 10 yrs [39]. We acknowledged that reproducibility may have been affected by the age of the population and the measurement site, yet as opposed to in vivo measurements, we were not influenced by movement artifacts, which play a major role in the reproducibility. The clinical device used in this study with an isotropic resolution of 82 μm is already used to measured bone microarchitecture in vivo, at peripheral sites. Recent studies have shown that microarchitecture measurements acquired using high resolution multi-detector CT (MDCT) imaging available in vivo correlate strongly with those assessed using either μCT or HR-pQCT [40,41]. Our results apply only to femoral neck fractures, but not to trochanteric fracture. There are some evidences showing that trochanteric fractures are more dependent on trabecular bone impairment than femoral neck fractures, which are more associated with cortical bone degradation. Altogether, these HR-pQCT results confirm those obtained by histomorphometry, showing that hip fragility results not only from cortical but also from trabecular bone deterioration. In addition, the HR-pQCT results show a more rod-like and disconnected structure of the femoral neck in postmenopausal women with hip fracture compared to women with hip osteoarthritis. According to recent progress provided by MDCT with a resolution close to HR-pQCT, the present results suggest that cortical and trabecular impairments involved in femoral neck fragility would be detected by non-invasive assessment of the 3D microarchitecture deterioration. Conflict of interest The authors have no conflict of interest. Acknowledgments The authors thank F. Bonnel, F. Canovas, M. Chammas, P. Maury, C. Belloc, R. Louahem and N. Portero-Muzy for their help in sample collection and preparation. This work was supported in part by unrestricted grants from Eli Lilly and by the French Ministry of Health (Projet Hospitalier de Recherche Clinique Régional, LanguedocRoussillon, 2003, UF-7755). References [1] Schuit SC, van der Klift M, Weel AE, de Laet CE, Burger H, Seeman E, et al. Fracture incidence and association with bone mineral density in elderly men and women: the Rotterdam Study. Bone 2004;34:195–202. [2] Siris ES, Miller PD, Barrett-Connor E, Faulkner KG, Wehren LE, Abbott TA, et al. Identification and fracture outcomes of undiagnosed low bone mineral density in postmenopausal women: results from the National Osteoporosis Risk Assessment. Jama 2001;286:2815–22. [3] Stone KL, Seeley DG, Lui LY, Cauley JA, Ensrud K, Browner WS, et al. BMD at multiple sites and risk of fracture of multiple types: long-term results from the Study of Osteoporotic Fractures. J Bone Miner Res 2003;18:1947–54. [4] Mayhew PM, Thomas CD, Clement JG, Loveridge N, Beck TJ, Bonfield W, et al. Relation between age, femoral neck cortical stability, and hip fracture risk. Lancet 2005;366:129–35. [5] Blain H, Chavassieux P, Portero-Muzy N, Bonnel F, Canovas F, Chammas M, et al. Cortical and trabecular bone distribution in the femoral neck in osteoporosis and osteoarthritis. Bone 2008;43:862–8. [6] Zanelli JM, Pearson J, Moyes ST, Green J, Reeve J, Garrahan NJ, et al. Methods for the histological study of femoral neck bone remodelling in patients with fractured neck of femur. Bone 1993;14:249–55. [7] Boutroy S, Bouxsein ML, Munoz F, Delmas PD. In vivo assessment of trabecular bone microarchitecture by high-resolution peripheral quantitative computed tomography. J Clin Endocrinol Metab 2005;90:6508–15. [8] Bell KL, Loveridge N, Power J, Garrahan N, Meggitt BF, Reeve J. Regional differences in cortical porosity in the fractured femoral neck. Bone 1999;24:57–64. [9] Tommasini SM, Hu B, Nadeau JH, Jepsen KJ. Phenotypic integration among trabecular and cortical bone traits establishes mechanical functionality of inbred mouse vertebrae. J Bone Miner Res 2009;24:606–20. [10] Bouxsein ML, Boyd SK, Christiansen BA, Guldberg RE, Jepsen KJ, Muller R. Guidelines for assessment of bone microstructure in rodents using microcomputed tomography. J Bone Miner Res 2010;25:1468–86.
S. Boutroy et al. / Bone 49 (2011) 1055–1061 [11] Harrigan TP, Jasty M, Mann RW, Harris WH. Limitations of the continuum assumption in cancellous bone. J Biomech 1988;21:269–75. [12] Hildebrand T, Laib A, Muller R, Dequeker J, Ruegsegger P. Direct three-dimensional morphometric analysis of human cancellous bone: microstructural data from spine, femur, iliac crest, and calcaneus. J Bone Miner Res 1999;14:1167–74. [13] Hildebrand T, Ruegsegger P. Quantification of bone microarchitecture with the structure model index. Comput Methods Biomech Biomed Engin 1997;1:15–23. [14] Odgaard A, Gundersen HJ. Quantification of connectivity in cancellous bone, with special emphasis on 3-D reconstructions. Bone 1993;14:173–82. [15] Laib A, Hauselmann HJ, Ruegsegger P. In vivo high resolution 3D-QCT of the human forearm. Technol Health Care 1998;6:329–37. [16] Chavassieux P, Arlot ME, Meunier PJ. Osteoporosis. 2nd ed. San Diego, CA: Academic Press; 2001. [17] Parfitt AM, Drezner MK, Glorieux FH, Kanis JA, Malluche H, Meunier PJ, et al. Bone histomorphometry: standardization of nomenclature, symbols, and units. Report of the ASBMR Histomorphometry Nomenclature Committee. J Bone Miner Res 1987;2:595–610. [18] Parfitt AM, Mathews CH, Villanueva AR, Kleerekoper M, Frame B, Rao DS. Relationships between surface, volume, and thickness of iliac trabecular bone in aging and in osteoporosis. Implications for the microanatomic and cellular mechanisms of bone loss. J Clin Invest 1983;72:1396–409. [19] Verhulp E, van Rietbergen B, Huiskes R. Load distribution in the healthy and osteoporotic human proximal femur during a fall to the side. Bone 2008;42:30–5. [20] Lotz JC, Cheal EJ, Hayes WC. Stress distributions within the proximal femur during gait and falls: implications for osteoporotic fracture. Osteoporos Int 1995;5: 252–61. [21] Holzer G, von Skrbensky G, Holzer LA, Pichl W. Hip fractures and the contribution of cortical versus trabecular bone to femoral neck strength. J Bone Miner Res 2009;24:468–74. [22] Van Rietbergen B, Huiskes R, Eckstein F, Ruegsegger P. Trabecular bone tissue strains in the healthy and osteoporotic human femur. J Bone Miner Res 2003;18: 1781–8. [23] Martens M, Van Audekercke R, Delport P, De Meester P, Mulier JC. The mechanical characteristics of cancellous bone at the upper femoral region. J Biomech 1983;16: 971–83. [24] Jordan GR, Loveridge N, Bell KL, Power J, Dickson GR, Vedi S, et al. Increased femoral neck cancellous bone and connectivity in coxarthrosis (hip osteoarthritis). Bone 2003;32:86–95. [25] Ciarelli TE, Fyhrie DP, Schaffler MB, Goldstein SA. Variations in three-dimensional cancellous bone architecture of the proximal femur in female hip fractures and in controls. J Bone Miner Res 2000;15:32–40. [26] Mori S, Harruff R, Ambrosius W, Burr DB. Trabecular bone volume and microdamage accumulation in the femoral heads of women with and without femoral neck fractures. Bone 1997;21:521–6.
1061
[27] Zhang ZM, Li ZC, Jiang LS, Jiang SD, Dai LY. Micro-CT and mechanical evaluation of subchondral trabecular bone structure between postmenopausal women with osteoarthritis and osteoporosis. Osteoporos Int 2010;21:1383–90. [28] Fazzalari NL, Parkinson IH. Femoral trabecular bone of osteoarthritic and normal subjects in an age and sex matched group. Osteoarthritis Cartilage 1998;6:377–82. [29] Tsangari H, Findlay DM, Fazzalari NL. Structural and remodeling indices in the cancellous bone of the proximal femur across adulthood. Bone 2007;40:211–7. [30] Cohen A, Dempster DW, Muller R, Guo XE, Nickolas TL, Liu XS, et al. Assessment of trabecular and cortical architecture and mechanical competence of bone by highresolution peripheral computed tomography: comparison with transiliac bone biopsy. Osteoporos Int 2010;21:263–73. [31] Laib A, Ruegsegger P. Calibration of trabecular bone structure measurements of in vivo three-dimensional peripheral quantitative computed tomography with 28microm-resolution microcomputed tomography. Bone 1999;24:35–9. [32] MacNeil JA, Boyd SK. Accuracy of high-resolution peripheral quantitative computed tomography for measurement of bone quality. Med Eng Phys 2007;29:1096–105. [33] Thomsen JS, Laib A, Koller B, Prohaska S, Mosekilde L, Gowin W. Stereological measures of trabecular bone structure: comparison of 3D micro computed tomography with 2D histological sections in human proximal tibial bone biopsies. J Microsc 2005;218:171–9. [34] Sode M, Burghardt AJ, Nissenson RA, Majumdar S. Resolution dependence of the non-metric trabecular structure indices. Bone 2008;42:728–36. [35] Chappard C, Marchadier A, Benhamou CL. Side-to-side and within-side variability of 3D bone microarchitecture by conventional micro-computed tomography of paired iliac crest biopsies. Bone 2008;43:203–8. [36] Chavassieux PM, Arlot ME, Meunier PJ. Intersample variation in bone histomorphometry: comparison between parameter values measured on two contiguous transiliac bone biopsies. Calcif Tissue Int 1985;37:345–50. [37] de Vernejoul MC, Kuntz D, Miravet L, Goutallier D, Ryckewaert A. Bone histomorphometric reproducibility in normal patients. Calcif Tissue Int 1981;33: 369–74. [38] Visser MJ, Roelofs JM, Peters JJ, Lentferink MF, S.A. D.. Sampling variation in bone histomorphometry. Metab Bone Dis Relat Res 1980(2 S):429–34. [39] Burghardt Andrew J, Buie Helen R, Laib Andres, Majumdar Sharmila, Boyd Steven K. Reproducibility of direct quantitative measures of cortical bone microarchitecture of the distal radius and tibia by HR-pQCT. Bone 2010;47:519–28. [40] Issever AS, Link TM, Kentenich M, Rogalla P, Burghardt AJ, Kazakia GJ, et al. Assessment of trabecular bone structure using MDCT: comparison of 64- and 320slice CT using HR-pQCT as the reference standard. Eur Radiol 2010;20:458–68. [41] Issever AS, Link TM, Kentenich M, Rogalla P, Schwieger K, Huber MB, et al. Trabecular bone structure analysis in the osteoporotic spine using a clinical in vivo setup for 64-slice MDCT imaging: comparison to microCT imaging and microFE modeling. J Bone Miner Res 2009;24:1628–37.
Contents lists available at SciVerse ScienceDirect
Bone j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / b o n e
Original Full Length Article
Comparison of 2D and 3D bone microarchitecture evaluation at the femoral neck, among postmenopausal women with hip fracture or hip osteoarthritis☆ Stéphanie Boutroy a,⁎, Nicolas Vilayphiou a, Jean-Paul Roux a, Pierre D. Delmas a, Hubert Blain b, Roland D. Chapurlat a, Pascale Chavassieux a a b
INSERM, UMR 1033, Université de Lyon, Lyon, France Department of Internal Medicine and Geriatrics, University Hospital of Montpellier, University Montpellier 1, Montpellier, France
a r t i c l e
i n f o
Article history: Received 20 January 2011 Revised 30 June 2011 Accepted 25 July 2011 Available online 2 August 2011 Edited by: Rene Rizzoli Keywords: Femoral neck Microarchitecture Microcomputed tomography Hip osteoarthritis Hip fracture
a b s t r a c t Objectives: High resolution peripheral quantitative tomography (HR-pQCT) is used more widely to assess microarchitecture, but we are lacking comparisons between HR-pQCT and histomorphometry, which is considered the gold standard. They have only been assessed on different anatomical regions. The purpose of our study was to assess the microarchitecture and the relative contribution of cortical and trabecular bone in hip fracture with this 3D imaging technique, compared with the 2D histomorphometry. Material and methods: We compared the distribution of cortical and trabecular bone in the ultradistal femoral neck samples (~ 3 mm thick) obtained after total hip replacement in 21 hip osteoarthritis (HOA, 66 ± 8 yrs) and 20 hip fracture (HF, 79 ± 8 yrs) menopausal women by a direct 3D evaluation method (HR-pQCT: XtremeCT, Scanco Medical AG) and by histomorphometry, performed and averaged on three 10 μm-thick sections 800 μm apart. Results: Significant correlations were found between both techniques for trabecular bone volume, number, thickness, separation and cortical thickness (0.51 b r′ b 0.81, p b 0.01). The connectivity was also significantly correlated (r′ = 0.58, p b 0.001) between both techniques, as well as the trabecular bone pattern factor measured in 2D with the structural model index (SMI) measured in 3D (r′ = 0.62, p b 0.001). However HRpQCT overestimated the absolute value of most parameters, with higher values being even more overestimated. The agreement between the two techniques was weak for cortical porosity. With the 3D measurements we found that trabecular bone volume was 43% lower in HF than HOA (p b 0.01), associated with loss of trabecular connectivity (− 50%, p b 0.01) and a more rod-like structure (SMI, 22%, p b 0.01), mainly at the inferior (34%, p b 0.01) and posterior (22%, p b 0.05) quadrants. Cortical thickness was found to be lower in the posterior quadrants (− 22%, p b 0.05) and tended to be lower in HF than in HOA at the inferior quadrant (− 14%, p = 0.08), but it was still the highest at the inferior quadrant in both groups. In conclusion, 3D methods confirmed the alteration of trabecular and cortical bone found by histomorphometry in HF compared with HOA and the frequency of the rod-like structure in HF. © 2011 Elsevier Inc. All rights reserved.
Introduction The evaluation of bone microarchitecture has gained increased attention in the past years, with the recognition that the usual areal bone mineral density (aBMD) measurements failed to identify most fragility fractures [1–3]. Indeed, several limitations are associated with aBMD measurements: aBMD values are influenced by bone size as DXA does ☆ This work was supported in part by unrestricted grants from Eli Lilly and by the French Ministry of Health (Projet Hospitalier de Recherche Clinique Régional, Languedoc-Roussillon, 2003, UF-7755). ⁎ Corresponding author at: INSERM UMR 1033, Hôpital Edouard Herriot, Pavillon F, 69437 Lyon cedex 03, France. Fax: + 33 4 72 11 74 32. E-mail addresses: [email protected] (S. Boutroy), [email protected] (N. Vilayphiou), [email protected] (J.P. Roux), [email protected] (H. Blain), [email protected] (R.D. Chapurlat), [email protected] (P. Chavassieux). 8756-3282/$ – see front matter © 2011 Elsevier Inc. All rights reserved. doi:10.1016/j.bone.2011.07.037
not measure true volumetric BMD; DXA cannot distinguish cortical and trabecular bone compartments and moreover DXA cannot directly measure cortical and trabecular architecture. However structural changes are important determinants of bone strength and have been associated with hip fracture independently of aBMD [4]. Bone histomorphometry allows for assessment of bone microarchitecture and remains the gold standard for this purpose. It also permits a dynamic evaluation by estimating the level of bone remodeling. However, histomorphometry has inherent limitations relative to its 2D design. Measurements are based on the assumption of a plate-like structure of bone and they are limited to a few slices thus representing a very small fraction of the biopsy. Additionally, histomorphometry precludes the assessment of non-metric parameters able to provide insight into bone quality, such as the structure model index. By conventional 2D histomorphometric analysis of femoral neck biopsies, we have previously shown that in addition to cortical
1056
S. Boutroy et al. / Bone 49 (2011) 1055–1061
thinning, loss of trabecular bone mass and connectivity may distinguish women with hip fracture (HF) from women with hip osteoarthritis (HOA) [5]. To overcome the 2D histomorphometric limitations, those femoral neck samples have been scanned with a 3D device. Three-dimensional analysis of the femoral neck structure by high-resolution peripheral quantitative computed tomography will assess the microarchitecture and the relative contribution of cortical and trabecular bone in women with hip fracture when compared to women with hip HOA. Material and methods Subjects The subjects and bone biopsies procedures were described in detail in our earlier study, in which cortical and trabecular bone distribution at the femoral neck in fractured and hip osteoarthritis women was assessed by histomorphometry [5]. Briefly, femoral neck biopsies were obtained during arthroplasty for non-traumatic hip fracture (HF), i.e. fall from standing height or less, (HF; n = 20; aged 79.3 ± 8.0 years old) or hip osteoarthritis (HOA; n = 21; aged 65.9 ± 7.9 years old) in postmenopausal women. For both HOA and HF patients, femoral neck biopsies were taken at the ultradistal part of the femoral neck, beginning at the cut face prepared for the insertion of the prosthesis at the base of the neck, thus leading to a bone slice of 10 mm thick approximately. Among the 20 HF, there were 19 cervical fractures and 1 trochanteric fracture. Undecalcified femoral neck samples were fixed in 70% ethanol and embedded in methylmethacrylate [6]. The protocol was approved by an independent Ethics Committee, and all patients gave written informed consent before participation. Measurement of 3D bone microarchitecture Three-dimensional bone microarchitecture of embedded femoral neck biopsies was measured using a high-resolution peripheral quantitative computed tomography system (XtremeCT, Scanco Medical AG®, Brüttisellen, Switzerland), with a nominal isotropic voxel size of 82 μm, using the standard in vivo scanning protocol (60 kVp, 900 μA) [7]. According to the bone microarchitecture regional differences in femoral neck [5,8], each section was divided into four quadrants (superior, anterior, inferior and posterior). CT slices were visually matched with histomorphometric analysis to have consistent regions of interest (Fig. 1), leading to mean ROI of 33 ± 6 CT slices (2.7 ± 0.5 mm). Trabecular and cortical parameters were first assessed in their respective whole compartment and then on the 4 quadrants separately. Image processing included a Gaussian filter (support = 1, σ = 0.8). Cortical bone was segmented manually from trabecular bone on a slice-by-slice basis and both volumes were thresholded separately to delineate bone from non-bone voxels (Fig. 1) [9]. As shown in Fig. 1, the global threshold used for trabecular bone segmentation (337 mg.HA/cm 3) tended to slightly overestimate the trabecular structure in order to keep an intact connectivity with the relatively low resolution in regard of the structure measured, whereas the
global threshold for cortical bone segmentation (511 mg.HA/cm 3) was chosen to differentiate the cortical porosity [10]. To avoid potential errors in trabecular spacing measurement due to the relatively small ROI, mirror images of the sample were added at the 2 extremities of the sample, otherwise trabecular spacing may have been underestimated [11]. The trabecular bone volume ratio (BV/TV) was determined by dividing the number of voxels representing trabecular bone by the total number of voxels in the trabecular compartment. Methods used to process the trabecular microarchitectural data were based on direct measurements of structure. Trabecular number (Tb.N*, 1/mm), thickness (Tb.Th*, μm) and separation (Tb.Sp*, μm) measurements were based on the distance transformation method, where maximal spheres are filled into the segmented object [12], thus not relying on an assumed model (rod/plate) type. Non-metric indices for trabecular bone structure as SMI [13] and Conn.D [14] were also calculated from the segmented images. Mean cortical thickness (Ct.Th, μm) was calculated using an annular method [15] where mean cortical volume is divided by the periosteal surface area, thus being irrespective of any pores. Cortical porosity was calculated as the complement of the cortical bone volume ratio (1 −(BVcortical/TVcortical)) in the compact-appearing cortex (Ct.Po, %). Two HF biopsies were excluded because the material was insufficient to be measured in 3D, with a total of 21 HOA and 18 HF as a result. Moreover, in most HF biopsies, the cortex from the superior quadrant was impaired or missing. Thus, cortical thickness and porosity were measured in only 3 samples at this site. Histomorphometry Serial 10 μm-thick cross-sectional slices were cut at three different levels of the bone sample, 800 μm apart, and were stained with Goldner trichrome [16]. Measurements were done separately on the four quadrants, the mean value of each sample being the average of the 4 quadrants among the 3 analyzed slices. The trabecular ROI corresponded to a rectangular area of maximum 31 mm², centered in each quadrant and cortical measurements were evaluated in the area corresponding to the trabecular ROI, as shown in Fig. 1. Parameters of bone structure and micro-architecture were performed using Bone and MorphoExpert softwares (Explora Nova, La Rochelle, France) and standard abbreviations for bone histomorphometry were used [17]. Outcome parameters for trabecular bone were volume (BV/TV,%), thickness (Tb.Th; μm), separation (Tb.Sp; μm), number (Tb.N; 1/mm) based on Parfitt's formulae, i.e. derived from volume and surfaces measurements [18]. After skeletonization, we evaluated the number of nodes (N.Nd/TV, 1/mm²) and the trabecular bone pattern factor (TBPf; 1/mm), both reflecting the connectivity/topology of the network. Outcome parameters for cortical bone were thickness (Ct.Th; μm) and porosity (Ct.Po, %) representing the cortical area occupied by Haversian canals. Statistical analysis Descriptive statistics were summarized by means and standard deviations. Regarding the distribution of the variables, non-parametric
Fig. 1. Cross-section of a femoral neck. A — Histomorphometric slice showing the four quadrants and the regions of interest for trabecular and cortical bone (rectangle and its projection on the cortex respectively) B — HR-pQCT gray-level slice of the same sample C–E — 3D reconstruction of the whole trabecular (C) and cortical (porosity shown in gray) (D) bone and of each quadrants (E).
S. Boutroy et al. / Bone 49 (2011) 1055–1061
tests were performed. The relationships between 2D and 3D parameters were studied using Spearman correlation (r′) in the entire data set and within-group, and 2D and 3D values were compared by paired Wilcoxon Signed Rank test. Bland–Altman plots were provided for qualitative analysis of accuracy (Fig. 2). The differences among women with HF and HOA were assessed by Wilcoxon Signed Rank test. In each group, the regional comparisons between the 4 quadrants (3 quadrants for cortex because of impaired materiel) were performed by Friedman test, followed when significant by paired Wilcoxon Signed Rank test for pairwise comparisons. The differences among women with HF and HOA were also assessed in a subset of the population ranging from 63 to 81 years old to have groups comparable in age. Statistical analyses were performed using SPSS software (version 16.0). Results Comparison of 2D and 3D parameters Despite mean values differences between 2D and 3D measurements (Tables 1a and 1b), significant correlations (r′) were observed between both techniques, ranging from 0.51 (p = 0.001) for trabecular thickness to 0.81 (p b 0.001) for trabecular bone volume (Fig. 2).
1057
Moreover, 2D TBPf was correlated with the 3D SMI (r′ = 0.62, p b 0.001), and the number of nodes was associated with the connectivity density (r′ = 0.58, p b 0.001). 3D mean cortical thickness of the whole cortical compartment was almost similar to 2D cortical thickness (+2.4% in 3D vs 2D) with a coefficient of correlation of 0.72 (p = 0.004). Cortical porosity measured in 3D and 2D was not correlated for the full sample, yet measurements were only available in a subset of the population because of impaired cortical bone in the superior quadrant in HF. When analyzed by quadrants, cortical porosity measured in 2D and 3D was correlated for the inferior and anterior quadrants (r′ = 0.37–0.54, p b 0.05). These correlations were almost the same when studied only in the HF or hip HOA group (data not shown). Bland–Altman plots were provided to visualize the agreement between 2D and 3D measurements. Dotted lines represented the lower and higher limit of agreement, i.e. average difference ± 1.96 standard deviation of the difference (Fig. 2). For most parameters, differences between the 2 techniques were smaller for lower values and increased with higher values of the measured parameters (Figs. 2B–D and F). Tb.Th measured in 3D was systematically higher than measured in 2D. There was no systematic disagreement for cortical porosity (Fig. 2E). A few values in the upper range were out of the limit of agreement between the two techniques.
Fig. 2. Correlation and Bland–Altman plots between 2D and 3D methods. (A) — Correlation of BV/TV measured by 2D vs. 3D methods in the whole sample. (B–D) — Bland–Altman plot of BV/TV (B), Tb.N (C) and Tb.Th (D) assessed in the whole sample. (E–F) — Bland–Altman plot of Ct.Po (E) and Ct.Th (F) assessed in the inferior quadrant.
1058
S. Boutroy et al. / Bone 49 (2011) 1055–1061
Table 1a Mean values and correlations between 2D and 3D parameters in the entire sample.
BV/TV (%) Trabecular number (1/mm) Trabecular thickness (μm) Trabecular separation (μm) TBPf (2D)/SMI (3D) N.Nd/TV (2D)/Conn.D (3D) Cortical thickness (μm) a Cortical porosity (%) a
2D values
3D values
% Difference
Correlation
Mean ± SD
Mean ± SD
3D vs. 2D
r′
11.1 ± 7.1 0.51 ± 0.20 312 ± 44 2358 ± 1096 2.8 ± 0.6 0.33 ± 0.25 1902 ± 355 9.9 ± 4.7
43.2⁎⁎
0.81⁎⁎ 0.69⁎⁎ 0.51⁎⁎ 0.67⁎⁎ 0.62⁎⁎ 0.58⁎⁎ 0.72⁎
7.8 ± 4.2 0.78 ± 0.31 94 ± 19 2067 ± 1989 3.6 ± 4.0 0.80 ± 0.63 1857 ± 513 12.6 ± 4.2
− 35.0⁎⁎ 231.1⁎⁎ 14.1⁎ NA NA 2.4 − 21.7
0.06
NA: non-applicable. a Measurements were only available in a subset of the population because of impaired cortical bone in the superior quadrant in HF. ⁎ p b 0.01. ⁎⁎ p ≤ 0.001.
Within-group regional differences in 3D parameters When comparing the 4 quadrants in the HOA group, we observed significant regional differences in trabecular BV/TV (p = 0.025), thickness (p = 0.01) and SMI (p = 0.002) (Fig. 3). The same pattern was observed in the HF group but it reached significance only for trabecular thickness (p = 0.001). BV/TV and trabecular thickness were the highest in the inferior quadrant and were significantly associated with a more plate-like structure in the HOA group as shown by a lower value of SMI. In both HOA and HF groups, we observed strong regional differences in cortical thickness (p b 0.001 and p = 0.009 respectively) and porosity (p b 0.001 and p = 0.006, respectively) (Fig. 4). In both groups, cortical thickness was the highest in the inferior quadrant, where cortical porosity was the lowest. Differences in 3D parameters between HF and hip HOA samples Trabecular bone On the entire femoral neck (Table 2), HF samples had 43% lower trabecular bone volume than HOA samples (p = 0.005), due to loss of trabeculae, which were significantly thinner (−11%, p b 0.01). Similar trends were observed for each quadrant (Fig. 3). These results were also associated with a loss of the trabecular connectivity (−50%, p = 0.009), mainly in the superior and inferior quadrants (− 60% and −43% respectively, p b 0.05). This was characterized by a more rodlike structure as shown by the significant higher SMI (+ 22%, p = 0.002) in the HF group than in the HOA group, mainly in the inferior and posterior quadrants (34% and 22% respectively, p b 0.05) (Fig. 3).
Table 1b Mean values and correlations between 2D and 3D cortical parameters by quadrants.
Ct.Th — Post (μm) Ct.Th — Inf (μm) Ct.Th — Ant (μm) Ct.Po — Post (%) Ct.Po — Inf (%) Ct.Po — Ant (%)
2D values
3D values
% Difference
Correlation
Mean ± SD
Mean ± SD
3D vs. 2D
r′
1064 ± 413
1372 ± 434
3234 ± 1244
2956 ± 695
1206 ± 514
1479 ± 369
22.7⁎
0.44⁎
11.2 ± 5.5 9.8 ± 5.6 14.8 ± 9.2
11.0 ± 8.4 5.1 ± 4.1 15.4 ± 8.5
− 1.8 − 48.2⁎⁎ 4.1
0.02 0.37⁎ 0.54⁎⁎
29.0⁎⁎ − 8.6
0.71⁎⁎ 0.84⁎⁎
Ct.Th = cortical thickness, Ct.Po = cortical porosity, Post = posterior quadrant, Inf = inferior quadrant, Ant = anterior quadrant. ⁎ p b 0.05. ⁎⁎ p ≤ 0.001.
Cortical bone When compared to HOA, cortical thickness was significantly lower in the posterior quadrant (−22%, p = 0.025) and tended to be lower in the inferior quadrant (−14%, p = 0.08) in HF. Mean cortical porosity assessed in 3D was not significantly different between HF and HOA in any quadrants (posterior, inferior and anterior) but was lower in HF than in HOA (−39 to −42%, p b 0.05) when assessed by histomorphometry. Subset of HF and HOA with comparable age As HF were in average 13 years older than HOA, we determined a subset of biopsies ranging from 63 to 81 years old, including 10 HOA and 12 HF (respectively 71 ± 6 and 75 ± 5 years old, p = 0.13). With the 2D method, most of the trabecular parameters (−39%, −20%, +103% for BV/TV, Tb.Th and Tb.Sp respectively, p b 0.05) remained significantly impaired in HF compared to HOA, as well as cortical parameters (−30%, −25%, −44% for Ct.Th posterior, Ct.Th anterior and Ct.Po inferior, p b 0.05). With the 3D method, most of the trabecular and cortical parameters were also impaired in the HF group but the results were not significant (− 23%, −5%, −13%, −41%, for BV/TV, Tb.Th, Ct.Th posterior and Ct.Po inferior respectively), unless for SMI (2.7 ± 0.4 vs 3.0 ± 0.4, respectively HOA and HF, p = 0.03). Discussion In this study, the 3D method confirmed the cortical thinning and the loss of trabecular connectivity found by histomorphometry in postmenopausal women with a femoral neck fracture when compared to women with hip osteoarthritis [5]. In addition 3D analysis outlined the prevalence of a disconnected rod-like structure at the femoral neck in HF women compared to HOA women. Bone fragility relates to both cortical and trabecular bone, but their contributions to bone strength vary throughout the skeleton. At the proximal femur, a study of the distribution of maximal principal and effective strain by finite element analysis has found that the contributions of the trabecular and cortical bone to bone strength are quite similar [19]. However, the load distribution between cortical and trabecular bone varies strongly within the proximal femur, cortical bone carrying as low as 30% at the subcapital region, to approximately 50% at the mid-neck and up to 96% at the base of the neck [20]. These computational results have recently been supported by Holzer et al. [21]. Therefore, while focusing on femoral neck, a thick cortical shell is preferred over a dense trabecular core [4,19,22]. However, in an earlier study of younger individuals [23], the role of the trabecular bone was greater than in the analysis of Holzer et al. This may reflect trabecular bone loss with age. The axis of greatest strain is the inferoanterior to superoposterior direction in both stance and fall configuration, which results in a
S. Boutroy et al. / Bone 49 (2011) 1055–1061
1059
Fig. 3. 3D trabecular microarchitecture of femoral neck biopsies by quadrants from women with HF or HOA. Comparison HF vs. HOA (* p b 0.05), comparison between quadrants ( p b 0.05), mean ± SEM. Sup = superior quadrant, Post = posterior quadrant, Inf = inferior quadrant, Ant = anterior quadrant.
similar stress/strain distribution [20]. Several studies suggested that hip fractures may be initiated at the superior neck, by buckling of the thin superior cortex [4]. However, femoral neck is a complex structure and trabecular bone may have consequences for cortical bone resistance. Indeed, trabecular bone may be seen as buttressing the cortex, thus stiffening it and reducing the risk of buckling. Jordan et al. found that the lesser femoral neck fracture risk observed in HOA may be conferred principally through structural effects on trabecular bone, as they observed an increased trabecular bone mass and connectivity in HOA at this site [24]. In our study, we found not only a significant decrease in cortical thickness in patients with HF compared to HOA but also an impairment of trabecular bone. Other studies have focused on trabecular microarchitecture at the femoral head [25–27], or at the intertrochanteric region [28,29] and consistently with our study, Zhang et al. [27] reported, at the femoral head, higher BV/TV, trabecular thickness and a more plate-like structure in HOA compared to HF, but no difference in Conn.D. We observed correlations between histomorphometry and 3D direct measurements of the femoral neck biopsies. Most previous studies have shown a link between μCT measurements and bone histomorphometry measurements. However, comparisons between HR-pQCT and histomorphometry have only been assessed on different anatomical regions [30], thus reporting either non-significant or weak correlations [12]. Laib et al. [31], and more recently MacNeil et al. [32],
reported higher correlation between μCT measurements (with a resolution of 28 and 19 μm) and 3D-pQCT (165 μm) or HR-pQCT (82 μm) measurements on ex vivo femoral head and radii, respectively. In the latter study, they showed that application of direct measurement techniques from HR-pQCT scans gave good correlations for BV/TV (r = 0.84) and Tb.N* (r = 0.97) [32]. However HR-pQCT overestimated the absolute value of most parameters, as we also observed in our study, with higher value being even more overestimated. Similar to our study, for mean BV/TV values ranging from 5 to 20%, they observed differences between the two techniques ranging from 0 to almost 15% for higher values of BV/TV. This was particularly true for Tb.Th*, with trabeculae found to be 2 to 3 times thicker than measured by histomorphometry, but still correlated with trabecular width measured in 2D. This overestimation may be due to partial volume effect as the resolution of the device is 82 μm whereas the thickness of trabeculae is close to 100 μm. We also found a lower number of trabeculae with the 3D than the 2D methods. This can also be related to partial volume effect, with relatively small trabeculae not correctly depicted. On the other hand, the trabecular number may be overestimated on thin slices, in 2D: a single trabeculae can be cut twice and thus counted as two trabeculae by histomorphometry. The number of nodes and the trabecular bone pattern factor measured by histomorphometry reflect the connectivity/topology of the network and can correspond to density connectivity (Conn.D) and
Fig. 4. 3D cortical thickness and porosity of femoral neck biopsies by quadrants from women with HF or HOA. Comparison HF vs. HOA (* p b 0.05), comparison between quadrants ( p b 0.05), mean ± SEM. a measurements were only available in a subset (n = 3) of the HF because of impaired cortical bone in the superior quadrant. Sup = superior quadrant, Post = posterior quadrant, Inf = inferior quadrant, Ant = anterior quadrant.
1060
S. Boutroy et al. / Bone 49 (2011) 1055–1061
Table 2 3D trabecular microarchitecture of femoral neck biopsies from women with HF or HOA (mean values and percentage differences).
BV/TV (%) Trabecular number (1/mm) Trabecular thickness (μm) Trabecular separation (μm) SMI Conn.D
HF
HOA
% Difference
Mean ± SD
Mean ± SD
HF vs HOA
7.9 ± 4.3 0.48 ± 0.16 293 ± 31 2339 ± 785 3.1 ± 0.4 0.22 ± 0.16
13.9 ± 7.9 0.59 ± 0.17 329 ± 47 1878 ± 572 2.5 ± 0.7 0.43 ± 0.22
− 42.8⁎⁎ − 17.7 − 10.9⁎ 24.5 22.0⁎⁎ − 49.7⁎
⁎ p b 0.01. ⁎⁎ p ≤ 0.005.
Structure Model Index (SMI) assessed in 3D. Indeed, we found that SMI and Conn.D performed quite well with their homologous 2D parameters: trabecular bone pattern factor and number of nodes. Even better relationship (r = 0.95) was ascertained by Thomsen et al. between TBPf and SMI measured with an isotropic 20 μm resolution on proximal tibial metaphysis necropsies [33]. Sode et al. found that SMI and Conn.D measured by HR-pQCT at 82 μm resolution on ex vivo distal tibia strongly correlated with that measured at 16 μm by μCT [34]. MacNeil et al. reported similar relationship for Conn.D but not for SMI measured by HR-pQCT and μCT at 19 μm on cadaver radii [32]. One of the limitations of our study is the small sample size. The age difference between women with HF and HOA and the lack of healthy controls have already been addressed in details [5]. Indeed, most parameters obtained in 3D were altered in HF compared to HOA but were no more significant, unless for SMI (p = 0.03) when studied in a subset of HF and HOA of comparable age. Other limitations can explain some moderate relationships found in our study between HR-pQCT and histomorphometry. First, the measurements of histomorphometric parameters were based on plate model assumptions whereas 3D measurements were directly measured without model assumption. Second, histomorphometric parameters may not be representative of the whole sample and results may be biased by regional variation. At the iliac crest, several studies have shown an intra-individual difference of 15–30% in trabecular bone volume between contiguous biopsies [35–38]. In order to obtain the best representativeness, histomorphometry measurements were performed on 3 separated slices covering the whole tissue area measured in 3D (in the first, middle and last slices of the 3D ROI). Yet, differences observed in cortical thickness, especially in the inferior quadrant, can partly be attributed to the difference in ROI between 2D and 3D. Indeed, in 2D, Ct.Th was measured in a rectangle centered on the quadrant, i.e. where the cortex is thicker, whereas in 3D, Ct.Th was measured in the entire quadrant, thus including thinner regions near the posterior and anterior quadrants. Third, discrepancies might be explained by the resolution of the 3D device (82 μm) which was probably not sufficient to depict accurately some parameters such as small cortical pores and non-metric parameters. Sode et al. showed that non-metric parameters were dependent on the resolution and may be limited by partial volume effect; therefore they should be interpreted with caution [34]. The impact of partial volume effect on the measurements is related to the amount of tissue i.e., higher value of BV/TV results in an increased overestimation of itself, as shown in Fig. 2. The lack of agreement between histomorphometry and HRpQCT for cortical porosity (Fig. 2E) is likely related to the size of pores detectable by each technique. Indeed in 3D, only pores bigger than 82 μm × 82 μm × 82 μm can be measured whereas smaller pores, which can be observed at different densities in both low and high porosity samples, were depicted by histomorphometry. This is consistent with the higher porosity observed in HOA than in HF patients when measured in 2D but not in 3D. Moreover, cortical thinning concomitant to cortical porosity in HF, so-called trabecularized cortex, may partly contribute to their similar or lower cortical porosity
compared to HOA. Reproducibility of cortical porosity was not computed in this study but was reported in vivo by Burghardt et al. to be 12% at the radius and 4% at the tibia in women aged 61± 10 yrs [39]. We acknowledged that reproducibility may have been affected by the age of the population and the measurement site, yet as opposed to in vivo measurements, we were not influenced by movement artifacts, which play a major role in the reproducibility. The clinical device used in this study with an isotropic resolution of 82 μm is already used to measured bone microarchitecture in vivo, at peripheral sites. Recent studies have shown that microarchitecture measurements acquired using high resolution multi-detector CT (MDCT) imaging available in vivo correlate strongly with those assessed using either μCT or HR-pQCT [40,41]. Our results apply only to femoral neck fractures, but not to trochanteric fracture. There are some evidences showing that trochanteric fractures are more dependent on trabecular bone impairment than femoral neck fractures, which are more associated with cortical bone degradation. Altogether, these HR-pQCT results confirm those obtained by histomorphometry, showing that hip fragility results not only from cortical but also from trabecular bone deterioration. In addition, the HR-pQCT results show a more rod-like and disconnected structure of the femoral neck in postmenopausal women with hip fracture compared to women with hip osteoarthritis. According to recent progress provided by MDCT with a resolution close to HR-pQCT, the present results suggest that cortical and trabecular impairments involved in femoral neck fragility would be detected by non-invasive assessment of the 3D microarchitecture deterioration. Conflict of interest The authors have no conflict of interest. Acknowledgments The authors thank F. Bonnel, F. Canovas, M. Chammas, P. Maury, C. Belloc, R. Louahem and N. Portero-Muzy for their help in sample collection and preparation. This work was supported in part by unrestricted grants from Eli Lilly and by the French Ministry of Health (Projet Hospitalier de Recherche Clinique Régional, LanguedocRoussillon, 2003, UF-7755). References [1] Schuit SC, van der Klift M, Weel AE, de Laet CE, Burger H, Seeman E, et al. Fracture incidence and association with bone mineral density in elderly men and women: the Rotterdam Study. Bone 2004;34:195–202. [2] Siris ES, Miller PD, Barrett-Connor E, Faulkner KG, Wehren LE, Abbott TA, et al. Identification and fracture outcomes of undiagnosed low bone mineral density in postmenopausal women: results from the National Osteoporosis Risk Assessment. Jama 2001;286:2815–22. [3] Stone KL, Seeley DG, Lui LY, Cauley JA, Ensrud K, Browner WS, et al. BMD at multiple sites and risk of fracture of multiple types: long-term results from the Study of Osteoporotic Fractures. J Bone Miner Res 2003;18:1947–54. [4] Mayhew PM, Thomas CD, Clement JG, Loveridge N, Beck TJ, Bonfield W, et al. Relation between age, femoral neck cortical stability, and hip fracture risk. Lancet 2005;366:129–35. [5] Blain H, Chavassieux P, Portero-Muzy N, Bonnel F, Canovas F, Chammas M, et al. Cortical and trabecular bone distribution in the femoral neck in osteoporosis and osteoarthritis. Bone 2008;43:862–8. [6] Zanelli JM, Pearson J, Moyes ST, Green J, Reeve J, Garrahan NJ, et al. Methods for the histological study of femoral neck bone remodelling in patients with fractured neck of femur. Bone 1993;14:249–55. [7] Boutroy S, Bouxsein ML, Munoz F, Delmas PD. In vivo assessment of trabecular bone microarchitecture by high-resolution peripheral quantitative computed tomography. J Clin Endocrinol Metab 2005;90:6508–15. [8] Bell KL, Loveridge N, Power J, Garrahan N, Meggitt BF, Reeve J. Regional differences in cortical porosity in the fractured femoral neck. Bone 1999;24:57–64. [9] Tommasini SM, Hu B, Nadeau JH, Jepsen KJ. Phenotypic integration among trabecular and cortical bone traits establishes mechanical functionality of inbred mouse vertebrae. J Bone Miner Res 2009;24:606–20. [10] Bouxsein ML, Boyd SK, Christiansen BA, Guldberg RE, Jepsen KJ, Muller R. Guidelines for assessment of bone microstructure in rodents using microcomputed tomography. J Bone Miner Res 2010;25:1468–86.
S. Boutroy et al. / Bone 49 (2011) 1055–1061 [11] Harrigan TP, Jasty M, Mann RW, Harris WH. Limitations of the continuum assumption in cancellous bone. J Biomech 1988;21:269–75. [12] Hildebrand T, Laib A, Muller R, Dequeker J, Ruegsegger P. Direct three-dimensional morphometric analysis of human cancellous bone: microstructural data from spine, femur, iliac crest, and calcaneus. J Bone Miner Res 1999;14:1167–74. [13] Hildebrand T, Ruegsegger P. Quantification of bone microarchitecture with the structure model index. Comput Methods Biomech Biomed Engin 1997;1:15–23. [14] Odgaard A, Gundersen HJ. Quantification of connectivity in cancellous bone, with special emphasis on 3-D reconstructions. Bone 1993;14:173–82. [15] Laib A, Hauselmann HJ, Ruegsegger P. In vivo high resolution 3D-QCT of the human forearm. Technol Health Care 1998;6:329–37. [16] Chavassieux P, Arlot ME, Meunier PJ. Osteoporosis. 2nd ed. San Diego, CA: Academic Press; 2001. [17] Parfitt AM, Drezner MK, Glorieux FH, Kanis JA, Malluche H, Meunier PJ, et al. Bone histomorphometry: standardization of nomenclature, symbols, and units. Report of the ASBMR Histomorphometry Nomenclature Committee. J Bone Miner Res 1987;2:595–610. [18] Parfitt AM, Mathews CH, Villanueva AR, Kleerekoper M, Frame B, Rao DS. Relationships between surface, volume, and thickness of iliac trabecular bone in aging and in osteoporosis. Implications for the microanatomic and cellular mechanisms of bone loss. J Clin Invest 1983;72:1396–409. [19] Verhulp E, van Rietbergen B, Huiskes R. Load distribution in the healthy and osteoporotic human proximal femur during a fall to the side. Bone 2008;42:30–5. [20] Lotz JC, Cheal EJ, Hayes WC. Stress distributions within the proximal femur during gait and falls: implications for osteoporotic fracture. Osteoporos Int 1995;5: 252–61. [21] Holzer G, von Skrbensky G, Holzer LA, Pichl W. Hip fractures and the contribution of cortical versus trabecular bone to femoral neck strength. J Bone Miner Res 2009;24:468–74. [22] Van Rietbergen B, Huiskes R, Eckstein F, Ruegsegger P. Trabecular bone tissue strains in the healthy and osteoporotic human femur. J Bone Miner Res 2003;18: 1781–8. [23] Martens M, Van Audekercke R, Delport P, De Meester P, Mulier JC. The mechanical characteristics of cancellous bone at the upper femoral region. J Biomech 1983;16: 971–83. [24] Jordan GR, Loveridge N, Bell KL, Power J, Dickson GR, Vedi S, et al. Increased femoral neck cancellous bone and connectivity in coxarthrosis (hip osteoarthritis). Bone 2003;32:86–95. [25] Ciarelli TE, Fyhrie DP, Schaffler MB, Goldstein SA. Variations in three-dimensional cancellous bone architecture of the proximal femur in female hip fractures and in controls. J Bone Miner Res 2000;15:32–40. [26] Mori S, Harruff R, Ambrosius W, Burr DB. Trabecular bone volume and microdamage accumulation in the femoral heads of women with and without femoral neck fractures. Bone 1997;21:521–6.
1061
[27] Zhang ZM, Li ZC, Jiang LS, Jiang SD, Dai LY. Micro-CT and mechanical evaluation of subchondral trabecular bone structure between postmenopausal women with osteoarthritis and osteoporosis. Osteoporos Int 2010;21:1383–90. [28] Fazzalari NL, Parkinson IH. Femoral trabecular bone of osteoarthritic and normal subjects in an age and sex matched group. Osteoarthritis Cartilage 1998;6:377–82. [29] Tsangari H, Findlay DM, Fazzalari NL. Structural and remodeling indices in the cancellous bone of the proximal femur across adulthood. Bone 2007;40:211–7. [30] Cohen A, Dempster DW, Muller R, Guo XE, Nickolas TL, Liu XS, et al. Assessment of trabecular and cortical architecture and mechanical competence of bone by highresolution peripheral computed tomography: comparison with transiliac bone biopsy. Osteoporos Int 2010;21:263–73. [31] Laib A, Ruegsegger P. Calibration of trabecular bone structure measurements of in vivo three-dimensional peripheral quantitative computed tomography with 28microm-resolution microcomputed tomography. Bone 1999;24:35–9. [32] MacNeil JA, Boyd SK. Accuracy of high-resolution peripheral quantitative computed tomography for measurement of bone quality. Med Eng Phys 2007;29:1096–105. [33] Thomsen JS, Laib A, Koller B, Prohaska S, Mosekilde L, Gowin W. Stereological measures of trabecular bone structure: comparison of 3D micro computed tomography with 2D histological sections in human proximal tibial bone biopsies. J Microsc 2005;218:171–9. [34] Sode M, Burghardt AJ, Nissenson RA, Majumdar S. Resolution dependence of the non-metric trabecular structure indices. Bone 2008;42:728–36. [35] Chappard C, Marchadier A, Benhamou CL. Side-to-side and within-side variability of 3D bone microarchitecture by conventional micro-computed tomography of paired iliac crest biopsies. Bone 2008;43:203–8. [36] Chavassieux PM, Arlot ME, Meunier PJ. Intersample variation in bone histomorphometry: comparison between parameter values measured on two contiguous transiliac bone biopsies. Calcif Tissue Int 1985;37:345–50. [37] de Vernejoul MC, Kuntz D, Miravet L, Goutallier D, Ryckewaert A. Bone histomorphometric reproducibility in normal patients. Calcif Tissue Int 1981;33: 369–74. [38] Visser MJ, Roelofs JM, Peters JJ, Lentferink MF, S.A. D.. Sampling variation in bone histomorphometry. Metab Bone Dis Relat Res 1980(2 S):429–34. [39] Burghardt Andrew J, Buie Helen R, Laib Andres, Majumdar Sharmila, Boyd Steven K. Reproducibility of direct quantitative measures of cortical bone microarchitecture of the distal radius and tibia by HR-pQCT. Bone 2010;47:519–28. [40] Issever AS, Link TM, Kentenich M, Rogalla P, Burghardt AJ, Kazakia GJ, et al. Assessment of trabecular bone structure using MDCT: comparison of 64- and 320slice CT using HR-pQCT as the reference standard. Eur Radiol 2010;20:458–68. [41] Issever AS, Link TM, Kentenich M, Rogalla P, Schwieger K, Huber MB, et al. Trabecular bone structure analysis in the osteoporotic spine using a clinical in vivo setup for 64-slice MDCT imaging: comparison to microCT imaging and microFE modeling. J Bone Miner Res 2009;24:1628–37.