Full text of DOI:10.1359/jbmr.2002.17.8.1372

JOURNAL OF BONE AND MINERAL RESEARCH
Volume 17, Number 8, 2002
©
2002 American Society for Bone and Mineral Research
Case Report
Synchrotron Radiation Microtomography Allows the
Analysis of Three-Dimensional Microarchitecture and
Degree of Mineralization of Human Iliac Crest Biopsy
Specimens: Effects of Etidronate Treatment
1
,2
3
1,4
5
3
S. NUZZO, M.H. LAFAGE-PROUST, E. MARTIN-BADOSA, G. BOIVIN, T. THOMAS,
3
1,2
C. ALEXANDRE, and F. PEYRIN
ABSTRACT
Quantitative microcomputed tomography using synchrotron radiation (SR ␮CT) was used to assess the effects
of a sequential etidronate therapy on both three-dimensional (3D) microarchitecture and degree of mineral-
ization of bone (DMB) in postmenopausal osteoporosis. Thirty-two iliac crest biopsy specimens were taken
from 14 patients with osteoporosis (aged 64 ؎ 1.8 years) before (baseline) and after 1 year of etidronate
treatment, and after 2 years of treatment for four of the patients. The samples were imaged at high spatial
resolution (voxel size ؍ 10 ␮m) using the microtomography system developed at the European Synchrotron
Radiation Facility (ESRF), Grenoble, France. Three-dimensional microarchitecture parameters were calcu-
lated and compared with those obtained from conventional histomorphometry. In addition, the DMB was
evaluated also in 3D. No significant statistical changes regarding bone mass and structural parameters were
observed in histomorphometry or 3D analyses. The distribution of the DMB in cortical and trabecular bone
showed a trend to a shift toward highest mineralization values after 1 year of etidronate treatment (3.88% and
1
.24% in cortical and trabecular bone, respectively). This trend was more evident after 2 years. The study also
showed that SR ␮CT is an accurate technique and the only one for quantifying both the mineralization and
the microarchitecture of bone samples at the same time in 3D. (J Bone Miner Res 2002;17:1372–1382)
Key words: bone architecture, degree of mineralization of bone, osteoporosis, quantitative microtomogra-
phy, etidronate
(
1–3)
INTRODUCTION
by DXA in patients with osteoporosis.
mechanisms of BP action remain not fully understood.
T IS well established that bisphosphonates (BPs) signifi- Indeed, although a significant increase in BMD of trabecular
However, the
(4)
I
cantly reduce fracture incidence by decreasing bone turn- sites was observed in the alendronate-treated patients, it
over and increasing bone mineral density (BMD) assessed was shown that no increase in bone mass (i.e., bone volume
to total volume [BV/TV] from quantitative histomorphom-
etry on iliac crest bone biopsy specimens) occurred in
a student for 3 months. All other authors have no conflict of patients after a 3-year treatment with alendronate as com-
Dr. Peyrin received a grant from Procter and Gamble to pay for
interest.
pared with the placebo group. Significant increases in BMD
1
ESRF, BP 220, Grenoble Cedex, France.
2
3
4
5
CREATIS, Bat Blaise Pascal, INSA, Villeurbanne Cedex, France.
INSERM E9901, St. Etienne Cedex 2, France.
Laboratori d’“ptica, Departament de F ´ı sica Aplicada i ”ptica, Universitat de Barcelona, Barcelona, Spain.
INSERM U403, Lyon Cedex 08, France.
1
372

3
D SYNCHROTRON ␮CT ANALYSIS OF ILIAC CREST BIOPSY SPECIMENS
1373
(
1,5–7)
also were observed in patients under other BPs,
cluding etidronate.
in- Hanover, NJ, USA) for 2.5 months. This 13-week cycle was
(8,2)
These data suggested that parameters repeated four times. Among the 14 patients, 4 patients were
other than true bone mass might explain changes in BMD or followed up 1 more year under a similar therapy and a third
reduced fracture incidence under treatment. Several factors biopsy was performed on the iliac crest that was operated on
account for bone quality including three-dimensional (3D) at baseline. These 14 patients were a subgroup of a popu-
microarchitecture, matrix composition, and degree of min- lation of 32 subjects with osteoporosis, submitted to etidr-
eralization of bone (DMB). In fact, by reducing osteoclastic onate, whom histomorphometric results at baseline and after
(
22)
resorption, BPs might lessen the probability of trabeculae 1 year of treatment were reported previously.
perforation in spongy bone,
After bi-
and in cortical bone it was opsy, bone samples were embedded in resin and processed
shown that they decrease bone porosity in ovariectomized for histomorphometry (see the following section). The 14
(9,10)
(
11)
(12)
primates.
In addition, Boivin et al.
recently showed biopsy specimens, which were selected from the original
that a 2- and 3-year treatment with alendronate was able to group for this study, were the ones with a substantial
improve the mean DMB (MDMB) as compared with pla- amount of bone available (at least more than one-half of the
cebo, thus leading to an increase in BMD in the absence of bone sample still remaining in the plastic block). These
actual changes in bone mass. As suggested by the authors, blocks (ϳ7 ϫ 7 ϫ 10 mm) were used for SR ␮CT mea-
this phenomenon certainly is related to a more complete surements.
secondary mineralization of the basic structural units al-
(13)
lowed by the reduced level of bone turnover. In this latter
work, the DMB was assessed on microradiography of bone
slices obtained from ground-down biopsy specimens, which
Bone histomorphometry
Iliac crest bone biopsy specimens were embedded in a
is the current reference technique. However, no data were resin made of purified glycol and methylmethacrylate at
available regarding concurrent changes in both 3D micro- 4°C. Nonserial sagittal 7-␮m-thick sections were carried out
architecture and DMB of human samples from patients on a Reichert microtome (Cambridge Instruments GmbH,
under BP therapy.
Nublock, Germany). Sections were stained with modified
Among the different techniques available, microcom- Goldner for subsequent measurements of conventional bone
(
23)
puted tomography (␮CT) is used increasingly to assess bone histomorphometry.
Measurements of BV (BV/TV, per-
(14)
microarchitecture. Synchrotron radiation ␮CT (SR ␮CT) cent of cancellous bone area) and calculation of structural
may provide 3D images of bone samples with a spatial indices (trabecular thickness [Tb.Th] and trabecular number
(15–17)
resolution as high as 1 ␮m.
Thus, specific image [Tb.N]) were carried out on an automatic image analyzer
analysis procedure may be applied to these 3D images to get (Biocom, Lyon, France). Five sections per individual were
morphological and topological quantitative parame- measured. The bone formation rate (BFR/bone surface
(
18,19)
ters.
SR ␮CT was used to study the relationships [BS]) was measured on five unstained sections under UV
between 3D architecture and the elastic properties of tra- light. It was calculated as the product of the mineral appo-
(20)
(21)
becular bone in humans
and rats.
So far, the possibil- sition rate (MAR) and of the mineralizing surface (MS/BS,
ity of obtaining quantitative information regarding the min- percentage of double-labeled bone perimeter Ϯ one-half
eral content of different areas of bone using ␮CT has not percentage of single-labeled bone perimeter).
been exploited. The technique relies on the physical prop-
All biopsy specimens were observed thoroughly to detect
erties of SR, which allows reconstructing quantitative maps any mineralization impairment or alteration in bone quality,
of the 3D distribution of linear absorption coefficient within that is, woven bone or microcallus.
volumetric samples. Because absorption depends on the
amount of mineral in bone, a suitable calibration is able to
relate the reconstructed gray levels in SR ␮CT images to the
Microtomographic image acquisition
local DMB.
Measurements were carried out on the high-resolution
In this context, our aim was to use SR ␮CT to analyze diffraction and topography beam-line ID19. The detailed
two elements of bone quality, that is, 3D microarchitecture description of the experimental device for tomographic
(24)
and DMB in patients with osteoporosis before and after a scans was reported elsewhere. The setup is based on a 3D
1
-year etidronate treatment.
parallel tomographic acquisition. A monochromatic X-ray
beam was selected using a double crystal silicon monochro-
mator, operating in the symmetrical Bragg reflection geom-
etry. The 3D distribution of the internal structure of the
sample was reconstructed from a set of 2D projections
under different angles of view. For each bone sample, 900
MATERIALS AND METHODS
Patients and treatment
Fourteen postmenopausal patients with osteoporosis radiographic images were acquired over an angular range of
(aged 64 Ϯ 1.8 years), with a mean of 2.1 Ϯ 1.3 vertebral 180°, (angular step, 0.2°). The transmitted X-ray beam after
crush fractures, underwent a bone iliac crest biopsy after the sample was recorded using a scintillator coupled to a 2D
double tetracycline labeling at baseline and after 1 year of CCD-based camera, the FRELON camera developed by the
(25)
sequential treatment including etidronate (400 mg/day for European SR facility (ESRF) Detector Group. It includes
4 days; Procter and Gamble, Cincinnati, OH, USA) fol- 1024 ء 1024 elements with a pixel size of 19 ␮m and a
1
lowed by a 1-g/day supplement of elemental calcium (Cal- dynamic range of 14 bits. For these experiments, the X-ray
cium Sandoz Forte; Sandoz Pharmaceuticals, Inc., East energy was set to 20 KeV and the optical system was fixed

1
374
NUZZO ET AL.
FIG. 1. Three-dimensional SR
␮CT reconstructed image from an il-
iac crest biopsy. (A–B) Three-
dimensional displays at two different
angles of view. (C–E) Two-
dimensional slices through the 3D
image. (C) Slice x ϭ 142; (D) slice
y ϭ 255; (E) slice z ϭ 26.
to get a pixel size of 10.13 ␮m. This pixel size yields a 10 back-projection algorithm. Volumes of ϳ600 ء 500 ء 900
mm ϫ 10 mm field of view for each radiograph. The cubic voxels (size, 10.13 ␮m) were reconstructed. As an
exposure time for one image was about 0.7 s so that the total illustration, Fig. 1 shows, respectively, a 3D display of bone
scan of each sample lasted for less than 15 minutes. During surfaces and three orthogonal slices through a reconstructed
the experiment, the operation mode of the machine was bone biopsy volume. Because of the quality of images in
multibunch mode with a current of ϳ200 mA. Dark current terms of spatial resolution and signal-to-noise ratio, bone
and reference images without the sample were taken regu- structure could be segmented easily from background by
larly to perform a flat field correction. The lacks of spatial simple thresholding. From a qualitative point of view, it is
homogeneity of the beam and of the individual pixel detec- noteworthy that differences in gray levels could be observed
tor responses for the most part were eliminated. The cor- on the slices both in cortical and in trabecular regions. This
rected projection images then were processed with an exact property allowed a clear visualization of the typical shapes
tomographic reconstruction algorithm based on 3D filtered of bone packets.

3
D SYNCHROTRON ␮CT ANALYSIS OF ILIAC CREST BIOPSY SPECIMENS
1375
the first slice of each volume reconstructed for comparison
to histomorphometry.
The correlation coefficients between the 3D and 2D ho-
mologous parameters were calculated to compare the dif-
ferent techniques used (histology, MIL, and direct method)
for the evaluation of 2D and 3D architectural parameters.
Analysis of the DMB
Calibration technique: At the difference to the CT num-
ber of clinical X-ray scanners, the gray levels in the 3D
reconstructed SR ␮CT images correspond to the linear
attenuation coefficient for a given energy. Because absorp-
tion depends on the mineral composition of bone, a suitable
calibration method allows estimating of the 3D distribution
of mineral content within a bone sample. To calibrate the
gray level (i.e., the absorption attenuation coefficient) into
the mineral part of bone tissue, we used phantoms of ho-
mogeneous water solutions with different dipotassium hy-
FIG. 2. Typical histogram of the gray levels of a 3D SR ␮CT image.
Analysis of 3D architecture
3
D SR ␮CT images ideally are suited to perform accurate drogen phosphate (K HPO ) concentrations. The samples
2 4
architectural analysis. For each patient, 3D architectural were imaged in the same experimental conditions used for
parameters were computed in selected regions of interest the bone samples (energy, 20 keV; pixel size, 10.13 ␮m).
(
ROIs) of trabecular bone. These ROIs were chosen within The method was compared with microradiography based on
(28,29)
the internal part of trabecular bone to avoid cracks and an aluminum step-wedge calibration.
For this purpose,
fragments due to biopsy taking and histomorphometry cut- the same 100-␮m-thick section of bone was radiographed
ting off. Because of these constraints, the size of each ROI both at the ESRF at 8 keV, and at the U403 INSERM Unit
was not the same for all biopsy samples. The number of (Lyon, France) using a nickel-filtered copper K74 radiation
voxels in the x, y, and z directions ranged from 300 to 500, with a wavelength of 1.54 Å (E ϳ 8 keV). Figure 3A
2
00 to 300, and 300 to 600, respectively. Figure 2 shows a illustrates the same cortical bone slice imaged with the two
typical histogram in a trabecular ROI. Because of the good techniques, displaying similar contrasts. The gray levels
contrast and the high spatial resolution of images, bimodal were converted into their corresponding degree of mineral-
histograms were obtained. The first peak (low gray levels) ization values by using both calibration methods. The his-
corresponds to resin and background absorption, and the tograms representing the number of pixels as a function of
second one corresponds to bone absorption. Thus, bone mineral content per cubic centimeter of bone, obtained from
structure was binarized by simple thresholding, with the the two techniques, were not directly at the same scale
same threshold for each sample of the series.
because the steps in DMB were different. To compare the
Different methods are available for architecture analy- results, we estimated the probability density function (pdf),
(
26)
(18)
sis.
The 3D mean intercept length (MIL) method
was respectively, from both histograms. Figure 3B shows the
used first and delivered morphometry and anisotropy pa- pdf’s of the DMB using the two methods in both ROIs
rameters. The 3D image was scanned by a 3D test grid for displayed in Fig. 3A. It may be seen that the two curves are
random orientations and the number of secants of parallel well overlaid and exhibit the same range of variation of the
3
test lines with bone were evaluated. The following param- DMB of 0.6–1.3 g/cm in cortical bone. The same result
eters were extracted assuming a parallel plate model: was found in both trabecular and cortical regions, as well as
Ϫ1
Ϫ1
BV/TV (%), BS/BV (mm ), Tb.Th (␮m), Tb.N (mm ), in total bone (cortical ϩ trabecular). The MDMB for the
trabecular separation (Tb.Sp; ␮m). This analysis was ap- different bone regions are reported in Table 1. The relative
plied to cubic subvolumes in trabecular ROIs.
errors were calculated as the absolute differences in DMB
Because 3D images provide complete information on the mean value between SR ␮CT and microradiography. The
BV, it is not necessary to make model assumptions, and we mean difference is ϳ4.21%, indicating a good agreement
also used direct 3D parameters. The Tb.Th* was computed between the two techniques.
(19)
according to the definition proposed by Hildebrand.
The
3D analysis of the DMB: The DMB was quantified sep-
algorithm does not require any special constraints regarding arately in the trabecular ROIs already used for architectural
both the size and the geometry of BVs. The same method analysis and in additional cortical ROIs carefully selected
was applied to the background instead of bone structure to from the 3D biopsy volumes. The number of voxels in the
evaluate a direct Tb.Sp*. The direct BS/BV* was calculated cortical ROIs along the x, y, and z directions ranged from
from a triangular mesh of the BSs. A direct Tb.N* also was 300 to 500, 200 to 300, and 60 to 120, respectively. The
calculated as BV/TV/Tb.Th*. Moreover, the connectivity of distribution of the gray levels within the different ROIs was
trabecular bone samples was estimated from the computa- obtained from the histogram of the bone images normalized
(27)
tion of the Euler number as previously described.
The by the TV. The gray levels corresponding to bone absorp-
connectivity was normalized to the TV and expressed (in tion were converted into concentrations of mineral content
Ϫ3
3
mm ). The architecture parameters also were computed on (in g/cm of bone) according to the calibration method. The

1
376
NUZZO ET AL.
FIG. 3.
Images of the same 100-␮m-thick
slice of cortical bone (A) SR image, (B) micro-
radiograph, and (C) corresponding pdf estimated
by the two calibration methods.
TABLE 1. MDMB AND BETWEEN BRACKETS THE SD,
RESULTS
CALCULATED FOR THE TRABECULAR AND CORTICAL REGION
OF THE SAME BONE FRAGMENT MEASURED BY USING SR
Conventional histomorphometric results
⌴
CT AND MICRORADIOGRAPHY
Data are summarized in Table 2. Bone mass and struc-
tural parameters did not show any significant change after
MDMB
SR ␮CT
␮Radiography
⌬MDMB
1
-year treatment with etidronate. In contrast, histodynamic
Trabecular
Cortical
Total
1.00 (0.05)
1.08 (0.05)
1.18 (0.06)
1.12 (0.07)
1.05 (0.06)
1.13 (0.08)
4.14%
4.11%
4.39%
measurements showed a significant decrease in BFR/BS
after 1-year treatment, confirming the expected etidronate-
induced reduction in bone turnover. This decrease in BFR
occurred through both a decrease in MAR (0.70 Ϯ 0.07
The relative error ⌬MDMB comparing results from the two
␮
m/day at baseline vs. 0.49 Ϯ 0.10 ␮m/day; p Ͻ 0.01) and
in MS/BS (8.1 Ϯ 4.9% at baseline vs. 3.5 Ϯ 2.6%; p Ͻ
.001). Thorough observation of the sections did not reveal
either mineralization impairment or signs of bone quality
histograms were averaged for each group. The changes damage such as woven bone, microcracks, or microcallus.
⌬DMB) after treatment were calculated as the relative Moreover, measurement of mineralization lag time, osteoid
percentage of the baseline value for each patient. Afterward, thickness, and osteoid volume ruled out focal and atypical
techniques also are indicated for each bone region.
0
(
the mean ⌬DMB value was determined for each group.
osteomalacia in these samples (data not shown).
Statistical analysis
3D architecture results
Descriptive statistics including, the mean, SD, minimum,
3D parameters calculated using MIL and direct methods
and maximum values were evaluated for each group of are reported in Tables 3 and 4, respectively. The statistical
patients for both microarchitecture parameters and DMB. analysis showed no significant changes in bone mass and
The three groups of patients, before treatment, after 1 year structural parameters after 1-year treatment as compared
and 2 years of etidronate (ETD) are, respectively, denoted with baseline. The correlation coefficients of the different
baseline, ETD1, and ETD2. To evaluate the effects of parameters are summarized in Table 5. Bone mass and
etidronate, the values of all parameters (architectural, DMB, structural parameters, measured with conventional histo-
and histodynamic parameters) obtained at baseline were morphometry, fairly correlated with those measured with
compared with those after a 1-year treatment (n ϭ 14) using SR ␮CT (BV/TV, rЈ ϭ 0.79 and p Ͻ 0.0001; Tb.N, rЈ ϭ
a T-Wilcoxon test. The Spearman correlation coefficient 0.78 and p Ͻ 0.0001; Tb.Sp, rЈ ϭ 0.82 and p Ͻ 0.0001)
was calculated to investigate potential correlations between except for Tb.Th (rЈ ϭ 0.5, not significant). High positive
the different techniques (i.e., conventional histomorphom- correlations were observed when comparing the 3D param-
etry, SR ␮CT direct, and MIL methods). Finally, a linear eters extracted by the direct method and the MIL approach.
regression analysis was carried out between BFR/BS and The weakest correlation was found for Tb.Th measure-
the DMB.
ments.

3
D SYNCHROTRON ␮CT ANALYSIS OF ILIAC CREST BIOPSY SPECIMENS
1377
TABLE 2. BONE MASS AND ARCHITECTURAL AND BONE FORMATION HISTOMORPHOMETRIC PARAMETERS IN PATIENTS AT
BASELINE AND AFTER 1 YEAR (ETD1) OR 2 YEARS (ETD2)
Baseline
n ϭ 14 patients
5.5 14.8
ETD1
Baseline
ETD1
ETD2
n ϭ 4 patients
BV/TV (%)
15.2
Ϯ
Ϯ
4.5
16.2
Ϯ
4.1
17.8
Ϯ
5.8
15.9 Ϯ 4.5
112.6 Ϯ 29.4
1.38 Ϯ 0.12
575.1 Ϯ 54.2
0.014 Ϯ 0.015
Tb.Th (␮m)
126.5 Ϯ 28.2
1.20 Ϯ 0.33
774.1 Ϯ 296.8
0.045 Ϯ 0.020
116.6 Ϯ 39.1
1.3 0.31
691.9 Ϯ 194.7
0.027 Ϯ 0.018*
130.6 Ϯ 14.9
1.28 Ϯ 0.20
668.4 Ϯ 158.2
0.070 Ϯ 0.015
114.0 Ϯ 41.8
1.41 Ϯ 0.26
593.3 Ϯ 113.7
0.019 Ϯ 0.013
Ϫ1
Tb.N (mm
)
Ϯ
Tb.Sp (␮m)
BFR/BS
3
2
(
␮m /␮m per day)
*
Versus baseline, p Ͻ 0.05.
TABLE 3. BONE MASS AND ARCHITECTURAL AND 3D PARAMETERS OBTAINED BY USING THE MEAN MIL METHOD, IN
PATIENTS AT BASELINE AND AFTER 1 YEAR (ETD1) OR 2 YEARS (ETD2)
Baseline
ETD 1 year
Baseline
ETD1
ETD2
3
D MIL
n ϭ 14 patients
n ϭ 4 patients
BV/TV (%)
14.5 Ϯ 5.9
(5.6–23.9)
23.5 Ϯ 5.9
(14.6–32.4)
91 Ϯ 23
(61–137)
1.5 Ϯ 0.4
(0.9–2.3)
13.9 Ϯ 4.6
(6.2–22.0)
24.5 Ϯ 6.3
(15.9–37.6)
85 Ϯ 22
(53–121)
1.6 Ϯ 0.4
(0.4–1.1)
18.5 Ϯ 1.6
(16.7–20.6)
19.5 Ϯ 1.44
(18.4–21.9)
92 Ϯ 18
(63–108)
1.8 Ϯ 0.3
(1.5–2.3)
17.1 Ϯ 4.1
(10.9–22.0)
23.4 Ϯ 8.4
(15.9–37.6)
93 Ϯ 23
(53–109)
1.8 Ϯ 0.1
(1.7–2.0)
21.2 Ϯ 4.7
(15.2–28.1)
20.8 Ϯ 3.2
(16.1–25.2)
99 Ϯ 16
(79–125)
2.1 Ϯ 0.2
(1.9–2.4)
(
min–max)
Ϫ1
BS/BV (mm
min–max)
Tb.Th (␮m)
min–max)
)
(
(
Ϫ1
Tb.N (mm
min–max)
Tb.Sp (␮m)
)
(
611 Ϯ 200
(351–1039)
582 Ϯ 188
(392–1062)
462 Ϯ 71
(351–538)
452 Ϯ 15
(436–477)
374 Ϯ 52
(317–442)
(
min–max)
3
D results on DMB
The gray level distribution allowing the study of bone
DISCUSSION
A few years ago, the World Health Organization added to
the definition of osteoporosis, characterized as a systemic
skeletal disease inducing a decrease in bone mass leading to
fracture, the notion of a deterioration of bone architec-
mineral content was evaluated from the bone peak of the
histogram, first normalized to the BV of each biopsy spec-
imen. The concentration in mineral substance after calibra-
3
tion was found in the range of 0.6–1.3 g/cm in both cortical
(30)
ture, thus emphasizing the role of trabecular connectivity
and trabecular bone. The averaged distributions of mineral
content for each group (baseline, ETD1, and ETD2) are
illustrated in Fig. 4. It is remarkable to note that changes in
the DMB under treatment differed between bone envelopes.
In cortical bone, values shifted toward higher degrees of
mineralization along with a decrease of the lowest degrees.
In trabecular bone, only the lowest values decreased.
MDMB are reported in Table 6. The differences between
the individual DMB values for the two groups were found in
the CI of 10%, for both trabecular and cortical envelopes.
After 2 years, the MDMB increased by ϳ12% and 8% in
cortical and trabecular bone, respectively. The peak corre-
in bone resistance to fracture. Indeed, some but not all
biomechanical properties correlate with bone density.
Ex vivo tests showed that 70–90% of the variance of bone
elastic modulus and strength is explained by linear and
(31,32)
power functions of bone density depending on the bone
(33,34)
site.
However, clinical studies showed a substantial
overlap between BMD values of individuals with bone
(35)
fractures and those without.
connectivity and orientation of the trabeculae) as well as
bone tissue intrinsic quality might explain the remaining
0–30% of the bone mechanical properties variance. In-
deed, studies evaluating concurrently bone mechanical
Trabecular 3D organization
(
2
sponding to the bone absorption region clearly showed a properties, bone mass, and microarchitecture showed that
shift toward the right side or highest mineralized values. bone strength and elastic modulus were better predicted
Interestingly, a significant negative relationship between when including microarchitecture parameters than when
(36–38)
BFR/BS and the DMB values was found in all etidronate- using bone mass alone.
treated patients (n ϭ 18) while no such relationship was veloped to visualize and measure microarchitecture includ-
observed at baseline (Fig. 5). ing histomorphometry, serial histological sections followed
Various techniques were de-

1
378
NUZZO ET AL.
TABLE 4. DIRECT 3D ARCHITECTURAL PARAMETERS IN PATIENTS AT BASELINE AND AFTER 1 YEAR (ETD1)
OR 2 YEARS (ETD2)
Baseline
ETD 1 year
Baseline
EDT1
ETD2
3
D Direct
n ϭ 14 patients
n ϭ 4 patients
BV/TV* (%)
15.2 Ϯ 6.4
(5.8–24.8)
20.3 Ϯ 4.8
(13.2–27.2)
180 Ϯ 40
(121–246)
1.5 Ϯ 0.4
(0.9–2.3)
701 Ϯ 62
(516–814)
4.5 Ϯ 2.2
(1.3–8.7)
14.5 Ϯ 5.1
(5.6–24.0)
24.5 Ϯ 6.5
(15.9–37.6)
172 Ϯ 37
(103–227)
1.6 Ϯ 0.4
(0.9–2.1)
682 Ϯ 64
(572–789)
5.4 Ϯ 3.6
(1.4–15.6)
19.4 Ϯ 2.0
(17.6–21.8)
17.5 Ϯ 1.7
(18.4–21.9)
196 Ϯ 15
(177–212)
1.8 Ϯ 0.3
(1.5–2.3)
684 Ϯ 83
(561–747)
4.9 Ϯ 2.0
(3.4–7.8)
18.0 Ϯ 5.6
(11.2–3.5)
20.2 Ϯ 7.6
(14.9–30.8)
178 Ϯ 54
(102–227)
1.9 Ϯ 0.2
(1.7–2.0)
660 Ϯ 40
(603–698)
5.0 Ϯ 1.8
(2.8–6.7)
22.2 Ϯ 6.0
(16.0–29.2)
18.1 Ϯ 2.7
(14.8–20.7)
187 Ϯ 27
(164–223)
2.1 Ϯ 0.2
(1.9–2.4)
573 Ϯ 56
(515–650)
10.2 Ϯ 6.4
(3.9–18.0)
(
min–max)
BS/BV* (mm
min–max)
Tb.Th* (␮m)
min–max)
Ϫ1
)
(
(
Ϫ1
Tb.N* (mm
min–max)
Tb.Sp* (␮m)
min—max)
Connectivity (mm
)
(
(
Ϫ3
)
(
min–max)
TABLE 5. SPEARMAN CORRELATION COEFFICIENTS BETWEEN MICROARCHITECTURAL PARAMETERS COMPUTED FROM
HISTOMORPHOMETRY (2D), 3D MIL, AND 3D DIRECT METHODS
BV/TV
BS/BV
Tb.Th
Tb.N
Tb.Sp
3
2
2
D MIL/3D Direct
D/3D MIL
D/3D MIL first slice
0.98*
0.79*
0.78*
0.98*
—
—
0.96*
0.50
0.84*
0.97*
0.78*
0.82*
0.87*
0.82*
0.84*
*
The statistical significance ( p Ͻ 0.0001) is indicated by *. “3D MIL first slice” corresponds to the parameters extracted from the
first slice of each volume reconstructed.
(
39)
(40)
by 3D reconstruction,
resonance imaging.
quantitative ␮CT,
or magnetic permit validation of SR ␮CT as compared with histology,
(
41,42)
(43,44)
According to Parfitt et al.,
the which is used widely and still considered as the standard
mathematic parallel plate model allows extrapolation of 3D technique for the study of bone microarchitecture. However,
parameters such as Tb.Th or Tb.N from bidimensional the values obtained in 2D measurements can significantly
measurements of trabecular bone perimeter and trabecular
bone area. This latter technique was used extensively in
humans and animals for studying osteoporosis pathophysi-
ology or treatments. Compared with reconstruction based
on serial histological sections, which is time consuming and
differ depending on the position and on the direction of the
slice. Thus, 3D analysis allows overcoming the strong vari-
ability of 2D measurements because of the larger number of
slices used in the computation.
The methods used for the calculation of morphological
parameters yielded significant differences for some param-
eters. Although the BV/TV and BS/BV parameters were
very close for 3D direct and indirect methods, the other
parameters presented substantial absolute variations. For
instance, the direct Tb.Th* generally was higher than the
Tb.Th derived from the parallel plate model. The same
behavior was already observed in a previous work reporting
microstructural data obtained from standard ␮CT on differ-
(45)
(46)
rarely performed, ␮CT, pioneered by Feldkamp et al.,
appears to be a suitable method for imaging and quantifying
(14)
trabecular bone microstructure in 3D
humans.
in animals and in
(
47)
The use of SR ␮CT as compared with standard ␮CT
offers the opportunity of an accurate investigation of local
bone mineral content in addition to microarchitectural anal-
ysis. Present results report the first simultaneous analysis of
both 3D microarchitecture and DMB in biopsy specimens
previously processed for conventional histomorphometry,
with no further destruction of the sample.
(49)
ent bone sites.
The orders of magnitude between the
values reported on iliac crest biopsies in the later work were
the same as in our findings. The differences between direct
and indirect methods suggest that the assumption of a par-
allel plate model is not completely appropriate. However,
We found fair correlations between histomorphometric
(
2D) and 3D homologous parameters, which were compa-
(48)
rable with previously reported results.
Interestingly, the
histomorphometric Tb.Th exhibited the weakest correlation most correlations between 3D parameters were quite high.
with the 3D MIL technique (rЈ ϭ 0.5). This correlation was A smaller correlation may be observed between Tb.Sp from
much higher when the 3D calculation was limited to the first the parallel plate model and its 3D direct homologous, and
slice closest to that of histology (rЈ ϭ 0.84). These results the values are quite different.

3
D SYNCHROTRON ␮CT ANALYSIS OF ILIAC CREST BIOPSY SPECIMENS
1379
FIG. 4. Distributions of the
DMB (in mineral content per cu-
bic centimeter of bone) in pa-
tients at baseline and after etidr-
onate treatment at (A) 1 year (n ϭ
1
4 patients, EDT1) and (B) 2
years (n ϭ 4 patients, ETD2).
TABLE 6. EFFECTS OF ETIDRONATE TREATMENT ON BONE MINERALIZATION
Baseline
n ϭ 14 patients
ETD1
Baseline
ETD1
ETD2
Mean DMB
SEM; g/cm )
3
(
n ϭ 4 patients
Cortical bone
0.871 (0.011)
0.940 (0.014)
0.901 (0.012)
ϩ3.88
0.951 (0.014)
ϩ1.24
0.846 (0.010)
0.880 (0.013)
0.902 (0.012)
ϩ4.09
0.910 (0.013)
ϩ1.39
0.948 (0.013)
ϩ12.07
0.952 (0.015)
ϩ8.26
Mean ⌬DMB (%)
Trabecular bone
Mean ⌬DMB (%)
Mean values of DMB and the SEM calculated for each averaged histogram over the different groups. The second line indicates the
mean ⌬DMB percentage of the baseline, averaged over the different groups.
In this study, after 1 year of etidronate treatment, neither yet for the quantification of the DMB. The present method,
bone volume nor microarchitecture parameters, including based on a monochromatic X-ray beam, provided by SR
connectivity, changed while the dynamic indices of bone makes this quantification possible directly in 3D because of
remodeling such as BFR/BS had significantly decreased. the lack of beam hardening.
This lack of effect on bone mass was not caused by the
In our study, the etidronate-induced changes did not reach
weak potency of etidronate because similar results were but were close to the level of statistical significance (p Ͻ
(
50)
reported with other more potent BPs such as pamidronate
0.05) after the first year of treatment. The MDBM tended to
(
4)
or alendronate. Thus, etidronate did not improve the con- increase in both envelopes. After 2 years of treatment the
nectivity parameters. However, in the absence of a control shift of the curves toward the high values became more
group, we cannot rule out the possibility that etidronate apparent, although the number of patients available was
prevented further microarchitectural deterioration.
very limited. These changes in mineralization under BPs
Using a gray-level scale, DMB currently is assessed by were thought related to an increase in the duration of the
microradiography, which is the gold standard for this eval- secondary mineralization phase because of the reduction in
(
12)
uation. However, before microradiography assessment, it bone remodeling. Data showing a significant negative rela-
is necessary to obtain regular, thin (100 ␮m thick), and tionship between BFR/BS and MDBM only in etidronate-
evenly polished bone slices, which is a very time- treated patients support this hypothesis. Moreover, it was
consuming procedure. Conversely, there is no need for assumed that this increase in MDBM could account for the
special sample preparation to perform 3D SR ␮CT images. improved bone strength. This assumption was supported by
In addition, 3D measurements give information from a stack the fact that the mineralization of bone matrix concurs with
(
51)
of virtual slices at the same time; thus, results are statisti- its mechanical properties
and by the decrease of fracture
(4,12)
cally more representative and robust than when obtained rate reported in BP-treated patients.
Indeed, it was
from a single slice. Currently, standard ␮CT is used to shown, in alendronate-treated minipigs that there was a
assess 3D bone microarchitecture but has never been used negative relationship between the level of bone remodeling

1
380
NUZZO ET AL.
ACKNOWLEDGMENTS
The authors thank the ESRF for financial support, J.
Baruchel for his valuable input into this project, and P.
Cloetens for his useful help during microtomography ex-
periments. We gratefully acknowledge R. Chagnon and D.
Rolhion for mechanical support, D. Fern a´ ndez-Carreiras,
Beam Line Instrumentation Software Support (BLISS), for
computing assistance, and the ESRF Detector Group for
assistance with the detector.
REFERENCES
1
. Black DM, Cummings SR, Karpf DB, Cauley JA, Thompson
DE, Nevitt MC, Bauer DC, Genant HK, Haskell WL, Marcus
R, Ott SM, Torner JC, Quandt SA, Reiss TF, Ensrud KE 1996
Randomised trial of effect of alendronate on risk of fracture in
women with existing vertebral fractures. Fracture Intervention
Trial Research Group. Lancet 7:1535–1541.
2
3
. Watts NB, Harris ST, Genant HK, Wasnich RD, Miller PD,
Jackson RD, Licata AA, Ross P, Woodson GC, Yanover MJ,
Mysim J, Kohse L, Rao MB, Steiger P, Chesnut CH 1990
Intermittent cyclical etidronate treatment of postmenopausal
osteoporosis. N Engl J Med 12:73–79.
. Harris ST, Watts NB, Genant HK, McKeever CD, Hangartner
T, Keller M, Chesnut CH III, Brown J, Eriksen EF, Hoseyni
MS, Axelrod DW, Miller PD 1999 Effects of risedronate
treatment on vertebral and non vertebral fractures in women
with postmenopausal osteoporosis: A randomized controlled
trial. Vertebral Efficacy with Risedronate Therapy (VERT)
Study Group. JAMA 13:1344–1352.
4
5
6
. Chavassieux PM, Arlot ME, Reda C, Wei L, Yates AJ,
Meunier PJ 1997 Histomorphometric assessment of the long-
term effects of alendronate on bone quality and remodeling in
patients with osteoporosis. J Clin Invest 15:1475–1480.
. Schneider PF, Fischer M, Allolio B, Felsenberg D, Schroder
U, Semler J, Ittner JR 1999 Alendronate increases bone den-
sity and bone strength at the distal radius in postmenopausal
women. J Bone Miner Res 14:1387–1393.
FIG. 5. Linear regression relationship between BFR/BS and DMB
A) in patients at baseline (n ϭ 14) and (B) in all patients treated by
etidronate (n ϭ 18; p Ͻ 0.01).
(
. Hanley DA, Ioannidis G, Adachi JD 2000 Etidronate therapy
in the treatment and prevention of osteoporosis. J Clin Densi-
tom 3:79–95.
and both the cortical and the trabecular bone resistance to
(
52)
fracture.
However, by reducing bone remodeling levels,
BPs might increase collagen cross-links as well and thus
improve bone matrix mechanical properties as it was re-
7. Crilly RG, Sebaldt RJ, Hodsman AB, Adachi JD, Brown JP,
Goldsmith CH, Hanley DA, Olszynski WO, Ste-Marie LG,
Stephenson GF 2000 Predicting subsequent bone density re-
sponse to intermittent cyclical therapy with etidronate from
initial density response in patients with osteoporosis. Osteo-
poros Int 11:607–614.
(53)
ported previously.
rect level of cross-linking for mineralization also was re-
In addition, the importance of a cor-
(
54,55)
ported,
emphasizing the major role of the mineral/
matrix interface.
8. Storm T, Thamsborg G, Steiniche T, Genant HK, Soresen OH
1
990 Effect of intermittent cyclical etidronate therapy on the
In conclusion, our results clearly showed that SR ␮CT is
a useful and innovative tool for bone quality analysis. It
allows information on both mineralization and trabecular
microarchitecture of the same bone samples. Moreover, we
found that etidronate, a first generation BP, was able to
induce changes in bone mineralization as early as the first
year of treatment, whereas no change in bone microarchi-
tecture parameters was observed. After 2 years of treatment,
the changes in mineralization are similar to those reported
bone mass and fracture rate in women with postmenopausal
osteoporosis. N Engl J Med 322:1265–1271.
. Barou O, Lafage-Proust MH, Martel C, Thomas T, Tirode F,
Laroche N, Barbier A, Alexandre C, Vico L 1999 Bisphos-
phonate effects in rat unloaded hindlimb bone loss model:
Three-dimensional microcomputed tomographic, histomor-
phometric, and densitometric analyses. J Pharmacol Exp Ther
9
2
91:321–328.
1
1
1
0. Parfitt AM 1992 Implications of architecture for the pathogen-
esis and prevention of vertebral fracture. Bone 13(Suppl 2):
(12)
4
1–47.
1. Balena R, Lafage MH, Markatos A, Seedor JG, Rodan GA
992 The effect of two years alendronate treatment on cortical
with alendronate,
suggesting that these effects may be
shared by drugs reducing bone remodeling. Whether this
increase in bone mineralization accounts for the increase in
bone strength and related decrease in fracture rate remains
to be determined.
1
bone remodeling. J Bone Miner Res 7:S329 (abstract).
2. Boivin GY, Chavassieux PM, Santora AC, Yates J, Meunier
PJ 2000 Alendronate increases bone strength by increasing the

3
D SYNCHROTRON ␮CT ANALYSIS OF ILIAC CREST BIOPSY SPECIMENS
1381
mean degree of mineralization of bone tissue in osteoporotic 33. Goldstein SA 1987 The mechanical properties of trabecular
women. Bone 27:687–694.
bone formed at remodelling sites in normal human iliac tra-
becular bone: Variations with age and sex. J Biomech 20:
1055–1061.
1
1
1
3. Meunier PJ, Boivin GY 1997 Bone mineral density reflects
bone mass but also the degree of mineralization of bone:
Therapeutic implications. Bone 21:373–377.
4. Ruegsegger P, Koller B, M u¨ ller R 1996 A microtomographic
system for the nondestructive evaluation of bone architecture.
Calcif Tissue Int 58:24–29.
34. Thomsen JS, Ebbesen EN, Mosekilde L 1998 Relationships
between static histomorphometry and bone strength measure-
ments in human iliac crest bone biopsies. Bone 22:153–163.
35. Melton LJ, Kan SH, Frye MA, Wahner HW, O’Fallon WM,
Riggs BL 1989 Epidemiology of vertebral fractures in women.
Am J Epidemiol 129:1000–1012.
5. Engelke K, Graeff W, Meiss L, Hahn M, Delling G 1993 High
spatial resolution imaging of bone mineral using computed
microtomography Comparison with microradiography and un- 36. Goulet RW, Goldstein SA, Ciarelli MJ, Kuhn JL, Brown MB,
decalcified histologic sections. Invest Radiol 28:341–349.
6. Peyrin F, Salom e´ M, Nuzzo S, Cloetens P, Laval-Jeantet AM,
Baruchel J 2000 Perspectives in Three-dimensional analysis of
Feldkamp LA 1994 The relationship between the structural
and orthogonal compressive properties of trabecular bone. LA
J Biomech 27:375–389.
1
1
bone samples using synchrotron radiation ␮CT. Cell Mol Biol 37. Ulrich D, Van Rietbergen B, Laib A, R u¨ egsegger PA 1999 The
4
6:1089–1102.
ability of three-dimensional structural indices to reflect me-
chanical aspects of trabecular bone. Bone 25:55–60.
7. Bonse U, Busch F, Gunnewig O, Beckmann F, Pahl R, Delling
G, Hahn M, Graeff W 1994 3D computed X-ray tomography 38. Wachter NJ, Augat P, Mentzel M, Sarkar MR, Krischak GD,
(
Ϫ4)
of human cancellous bone at 8 microns spatial and 10
energy resolution. Bone Miner 25:25–38.
Kinzl L, Claes LE 2001 Predictive value of bone mineral
density and morphology determined by peripheral quantitative
computed tomography for cancellous bone strength of the
proximal femur. Bone 28:133–139.
1
1
2
2
8. Hipp JA, Simmons CA 1997 Method-based differences in the
automated analysis of the three-dimensional morphology of
trabecular bone. J Bone Miner Res 12:942–947.
9. Hildebrand T, R u¨ egsegger P 1997 A new method for the
model independent assessment of thickness in three-
dimensional images. J Microsc 187:67–75.
39. Beck JD, Canfield BL, Haddock SM, Chen TJH, Kothari M,
Keaveny TM 1997 Three-dimensional imaging of trabecular
bone using the computer numerically controlled milling tech-
nique. Bone 21:281–287.
0. Kinney JH, Ladd AJ 1998 The relationship between three- 40. M u¨ ller R, Hildebrand T, R u¨ egsegger P 1994 Non-invasive
dimensional connectivity and the elastic properties of trabec-
ular bone. J Bone Miner Res 13:839–845.
bone biopsy: a new method to analyze and display the three-
dimensional structure of trabecular bone. Phys Med Biol 39:
145–164.
1. Kinney JH, Haupt DL, Balooch M, Ladd AJ, Ryaby JT, Lane
NE 2000 Three-dimensional morphometry of the L6 vertebra 41. Lang T, Augat P, Majumdar S, Ouyang X, Genant HK 1998
in the ovariectomized rat model of osteoporosis: Biomechani-
cal implications. J Bone Miner Res 15:1981–1991.
2. Thomas T, Barou O, Vico L, Alexandre C, Lafage-Proust MH
Noninvasive assessment of bone density and structure using
computed tomography and magnetic resonance. Bone 22:
149S–153S.
2
2
2
2
1
999 Recurrence of vertebral fracture with cyclical etidronate 42. Majumdar S, Newitt DC, Jergas MD, Gies AA, Chiu E, Osman
therapy in osteoporosis: Histomorphometry and X-ray micro-
analysis evaluation. J Bone Miner Res 14:198–205.
3. Parfitt AM, Glorieux F, Kannis H, Malluche A, Meunier PJ,
Ott SM, Recker RR 1987 Bone histomorphometry nomencla-
ture symbols and units Report of the ASBMR Histomorphom- 43. Parfitt AM, Mathews CHE, Villanueva AR, Kleerekoper M,
etry Nomenclature. J Bone Miner Res 2:595–610.
4. Salom e´ M, Peyrin F, Cloetens P, Odet C, Laval-Jeantet AM,
Baruchel J, Spanne P 1999 A synchrotron radiation microto-
mography system for the analysis of trabecular bone samples. 44. Kleerekoper M, Villanueva AR, Stanciu J, Rao DS, Parfitt AM
Med Phys 26:2194–2204.
5. Labiche JC, Segura-Purchades J, Van Brussel D, Moy JP 1996
FRELON camera: Fast read low noise. ESRF NewsLetter
D, Keltner J, Keyak JH, Genant HK 1995 Evaluation of
technical factors affecting the quantification of trabecular bone
structure using magnetic resonance imaging. Bone 17:417–
430.
Frame B, Rao DS 1983 Relationships between surface, vol-
ume, and thickness of iliac trabecular bone in aging and in
osteoporosis. J Clin Invest 72:1396–1409.
1985 The role of three-dimensional trabecular microstructure
in pathogenesis of vertebral compression fractures. Calcif Tis-
sue Int 37:594–597.
2
5:41–43.
45. DeHoff RT, Aigeltinger EH, Craig KR 1972 Experimental
determination of the topological properties of three-
dimensional microstructures. J Microsc 95:69–91.
2
2
6. Odgaard A 1997 Three-dimensional methods for quantifica-
tion of cancellous bone architecture. Bone 20:315–328.
7. Odgaard A, Gundersen HJG 1993 Quantification of connec- 46. Feldkamp LA, Goldstein SA, Parfitt AM, Jesion G, Kleereko-
tivity in cancellous bone with special emphasis on 3D recon-
structions. Bone 14:173–182.
8. Sissons HA, Jowsey J, Stewart L 1960 Quantitative microra-
per M 1994 The direct examination of three-dimensional bone
architecture in vitro by computed tomography. J Bone Miner
Res 4:3–11.
2
diography of bone tissue. In: Engstr o¨ m A, Cosslett V, Pattee H 47. Ito M, Nakamura T, Matsumoto T, Tsurusaki K, Hayashi K
(
eds.) X-Ray Microscopy and X-Ray Microanalysis. Elsevier,
1998 Analysis of trabecular microarchitecture of human iliac
bone using microcomputed tomography in patients with hip
arthrosis with or without vertebral fracture. Bone 23:163–169.
Amsterdam, The Netherlands, pp. 199–205.
2
9. Boivin G, Baud CA 1984 Microradiographic methods for
calcified tissue. In: Dickson GR (ed.) Methods of Calcified 48. M u¨ ller R, Van Campenhout H, Van Damme B, Van Der Perre
Tissue Preparation, pp. 391–412.
G, Dequeker J, Hildebrand T, Ruegsegger P 1998 Morpho-
metric analysis of human bone biopsies: A quantitative struc-
tural comparison of histological sections and micro-computed
tomography. Bone 23:59–66.
3
0. Consensus Development Conference Diagnosis 1993 Prohy-
laxis and treatment of osteoporosis. Am J Med 94:646–650.
1. Mosekilde L, Mosekilde L, Danielsen CC 1987 Biomechanical
3
competence of vertebral trabecular bone in relation to ash 49. Hildebrand T, Laib A, Muller R, Dequeker J, Ruegsegger P
density and age in normal individuals. Bone 8:79–85.
2. Wang XD, Masilamani NS, Mabrey JD, Alder ME, Agrawal
CM 1998 Changes in the fracture toughness of bone may not
1999 Direct three-dimensional morphometric analysis of hu-
man cancellous bone: Microstructural data from spine, femur,
iliac crest, and calcaneus. J Bone Miner Res 14:1167–1174.
3
be reflected in its mineral density, porosity and tensile prop- 50. Bravenboer N, Papapoulos SE, Holzmann P, Hamdy NA,
erties. Bone 23:67–72. Netelenbos JC, Lips P 1999 Bone histomorphometric evalua-

1
382
NUZZO ET AL.
tion of pamidronate treatment in clinically manifest osteopo-
rosis. Osteoporos Int 9:489–493.
1. Glimcher MJ 1998 The nature of the mineral phase in bone. In:
Glimcher MJ (eds.) Metabolic Bone Disease. Academic Press,
San Diego, CA, USA, pp. 23–50.
2. Lafage MH, Balena R, Battle MA, Shea M, Seedor JG, Klein
H, Hayes WC, Rodan GA 1995 Comparison of alendronate
and sodium fluoride effects on cancellous and cortical bone in
minipigs. A one-year study. J Clin Invest 95:2127–2133.
3. Rath NC, Balog JM, Huff WE, Huff GR, Kulkarni GB, Tierce
JF 1999 Comparative differences in the composition and bio-
mechanical properties of tibiae of seven- and seventy-two-
week-old male and female broiler breeder chickens. Poult Sci
ation of mineral crystallinity and collagen cross-linking of
bones in osteopetrotic toothless (tl/tl) rats and their improve-
ment after treatment with colony stimulating factor-1. Bone
20:127–132.
5
5
5
5. Wassen MH, Lammens J, Tekoppele JM, Sakkers RJ, Liu Z,
Verbout AJ, Bank RA 2000 Collagen structure regulates fibril
mineralization in osteogenesis as revealed by cross-link pat-
terns in calcifying callus. J Bone Miner Res 15:1776–1785.
5
5
Address reprint requests to:
Fran c¸ oise Peyrin, Ph.D.
ESRF, BP 220
7
8:1232–1239.
4. Wojtowicz A, Dziedzic-Goclawska A, Kaminski A, Stachow-
icz W, Wojtowicz K, Marks SC Jr, Yamauchi M 1997 Alter-
38043 Grenoble Cedex, France