Biomaterials 26 (2005) 36993712

Static and fatigue mechanical behavior of bone cement with

elevated barium sulfate content for treatment of vertebral
compression fractures
S.M. Kurtz
a,b,
, M.L. Villarraga
a,b
, K. Zhao
c
, A.A. Edidin
b,d
a
Exponent, Inc., 3401 Market Street, Suite 300, Philadelphia, PA 19104, USA
b
School of Biomedical Engineering, Science, and Health Systems, Drexel University, 3141 Chestnut Street, Philadelphia, PA 19104, USA
c
Exponent, Inc., 149 Commonwealth Drive, Menlo Park, CA 94025, USA
d
Kyphon Inc., 1221 Crossman Avenue, Sunnyvale, CA 94089, USA
Received 7 July 2004; accepted 24 September 2004
Available online 21 November 2004
Abstract
The use of bone cement to treat vertebral compression fractures in a percutaneous manner requires placement of the cement under
uoroscopic image guidance. To enhance visualization of the ow during injection and to monitor and prevent leakage beyond the
connes of the vertebral body, the orthopedic community has described increasing the amount of radiopacier in the bone cement.
In this study, static tensile and compressive testing, as well as fully reversed fatigue testing, was performed on three PMMA-based
bone cements. Cements tested were Simplex
s
P with 10% barium sulfate (Stryker Orthopedics, Mahwah, NJ) which served as a
control; Simplex
s
P with 36% barium sulfate prepared according to the clinical recommendation of Theodorou et al.; and KyphX
HV-R with 30% barium sulfate (Kyphon Inc., Sunnyvale, CA). Static tensile and compressive testing was performed in accordance
with ASTM F451-99a. Fatigue testing was conducted in accordance with ASTM F2118-01a under fully reversed, 710-, 715-, and
720-MPa stress ranges. Survival analysis was performed using three-parameter Weibull modeling techniques. KyphX HV-R was
found to have comparable static mechanical properties and signicantly greater fatigue life than either of the two control materials
evaluated in the present study. The static tensile and compressive strengths for all three PMMA-based bone cements were found to
be an order of magnitude greater than the expected stress levels within a treated vertebral body. The static and fatigue testing data
collected in this study indicate that bone cement can be designed with barium sulfate levels sufciently high to permit uoroscopic
visualization while retaining the overall mechanical prole of a conventional bone cement under typical in vivo loading conditions.
r 2004 Elsevier Ltd. All rights reserved.
Keywords: Polymethylmethacrylate(PMMA); Bone cement; Fatigue; Tension; Compression; Spine; Kyphoplasty; Vertebroplasty; Mechanical
behavior; Elastic modulus; Ultimate strength; Vertebrae
1. Introduction
Acrylic bone cement, based on polymethyl methacry-
late (PMMA), has been used in joint reconstruction
since the 1950s [1]. Charnley popularized the use of
PMMA bone cement for xation of the femoral stem in
total hip replacements during the 1960s [2]. Today, over
60 types of bone cements are commercially used around
the world, primarily for the treatment of joint diseases
and long bone fractures [3]. Recently, however, PMMA-
based bone cements have also been introduced into
spine reconstructive surgery as part of the treatment for
vertebral compression fractures [46].
When employed as part of the treatment of compres-
sion fractures in the spine, bone cements are currently
injected into a vertebral body under real-time uoro-
scopic image guidance [6]. Consequently, orthopedic
ARTICLE IN PRESS
www.elsevier.com/locate/biomaterials
0142-9612/$ - see front matter r 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.biomaterials.2004.09.055

Corresponding author. Exponent, Inc., 3401 Market Street, Suite

300, Philadelphia, PA 19104, USA. Tel.: +215 446 3325; fax:
+215 446 3380.
E-mail address: skurtz@exponent.com (S.M. Kurtz).
researchers have advocated increasing the radiopacier
content of the cements to enhance visualization of the
ow during injection and to monitor and prevent
leakage beyond the connes of the vertebral body [5].
The injection of bone cement into vertebrae raises new
questions regarding ow visualization, as well as
regarding the mechanical requirements of bone cements
for applications in the spine.
Barium sulfate (BaSO
4
) and zirconium dioxide (ZrO
2
)
are the two most common radiopaciers used in
contemporary bone cements, and usually comprise
about 1015% of the formulation on a weight basis
[3]. However, for improved visualization under uoro-
scopy, clinicians have increased the radiopacier content
of the bone cements to over 30% by hand mixing
additional vials of barium sulfate, zirconium dioxide,
and tantalum powder with commercially available bone
cements prior to injection. Although the size, shape, and
coatings of radiopacier particles have been associated
with the fatigue and fracture behavior of PMMA-based
cements [7], comparatively little research has been
conducted on the effect of increased radiopacier
content [8].
The purpose of this study was to investigate the static
and fatigue mechanical properties of PMMA-based
bone cements with increased BaSO
4
content, which
may be suitable for the treatment of vertebral compres-
sion fractures. The mechanical properties of bone
cements have been studied extensively as a function of
formulation and testing condition, as reviewed by Lewis
[9,10]. In this study, we employed a recently published
ASTM standard testing technique (F2118-01a) to
evaluate the fatigue behavior of bone cements modied
by the addition of BaSO
4
. We also evaluated the tensile
and compressive behavior of the bone cements to
determine whether increased barium sulfate content
compromised the static mechanical properties. The
primary goal of this research was to determine the static
and fatigue mechanical properties of bone cement
containing levels of barium sulfate tuned for uoro-
scopic visualization.
2. Materials and methods
2.1. Bone cement formulations and powder morphology
Two commercial formulations of PMMA-based bone
cements that are relevant to the treatment of vertebral
compression fractures were studied. Simplex
s
P (Stryker
Orthopedics, Mahwah, NJ) served as a control. Sim-
plex
s
P has a BaSO
4
content of 10%. The second bone
cement evaluated in this study was a PMMA-based
formulation with a BaSO
4
content of 30% for enhanced
visualization during uoroscopic procedures (KyphX
HV-R, Kyphon Inc.). KyphX HV-R is reported by the
manufacturer to have an extended dough time (a factor
of two times greater than Simplex) to facilitate place-
ment of the cement into a fractured vertebral body.
In addition to these two neat cements, we also mixed
BaSO
4
with the Simplex
s
P to prepare a formulation
that was representative of previous clinical practice for
the treatment of vertebral compression fractures [5].
This clinical formulation was prepared according to the
clinical protocol published by Theodorou et al. [5] by
measuring 14.7 g of the neat bone cement powder,
adding a 6 g aliquot of sterile BaSO
4
(Biotraces: Bryan
Corporation, Woburn, MA), and mixing with 10 ml of
Simplex
s
P monomer. The radiopacier content for the
clinical formulation was 36%w/w BaSO
4
for Sim-
plex
s
P.
Powder samples of each bone cement were examined
using scanning electron microscopy (SEM) to evaluate
the dispersion of barium sulfate. Carbon lm tape was
placed on three individual SEM stubs and a sample of
powder (less than 0.5 mg) of each type of bone cement
formulation was poured individually on a designated
stub. Each of the stubs was knocked against the end of a
table to remove excess powder, so as to achieve a single
monolayer of powder on the reective tape. The powder
was sputter-coated with carbon for visualization with
SEM. A JSM-6300 FV scanning electron microscope
(SEM) (JEOL USA, Peabody, MA), operating at 10 kV,
was used to view the powder samples starting at 100
and up to 8000 . The highest magnication (8000 )
was used to perform additional analyses with energy
dispersive spectroscopy (EDS).
2.2. Mechanical test specimen preparation
Specimen preparation is well known to affect the
fatigue properties of bone cement [9,10]. Consequently
all of the specimens were prepared using identical hand
mixing techniques and molded into their nal shape. We
chose hand mixing for the present study over vacuum
mixing because clinical treatment of vertebral compres-
sion fractures involves the hand mixing of additional
barium sulfate by the surgeon prior to injection into the
spine [5]. The mixing and molding processes occurred in
an environmentally controlled laboratory in a fume
hood. Temperature was maintained at 22 1C and 30%
relative humidity. Custom silicone molds (RTV664A,
GE Silicones) were prepared for the tensile and fatigue
specimens in accordance with ASTM F2118-01a. In
contrast, stainless steel molds were used to produce the
compression specimens in accordance with ASTM
F451-99a.
The liquid and powder components were mixed using
a sterile, inert spatula and bowl, with a consistent
mixing time of less than 1 min (conrmed by stopwatch).
Before the dough-like state, the cements were
deposited over the molds and forced into the mold with
ARTICLE IN PRESS
S.M. Kurtz et al. / Biomaterials 26 (2005) 36993712 3700
a at paddle until the mold cavities were completely
lled. The specimens were extracted using a cylindrical,
at-end punch after a 60-min polymerization period.
The same engineer mixed and molded all the specimens
consecutively.
Tensile and fatigue specimens were fabricated
using the same molds, as described in ASTM F2118-
01a. The 65-mm-long tensile and fatigue specimens
had a nominal diameter of 8.5 mm and a gage section
that was 5 mm in diameter and 10 mm long. Cylindrical
compression specimens, 6 mm in diameter and 12 mm
in height, were molded in accordance with ASTM
F451-99a.
2.3. Porosity evaluation
All specimens were qualied for testing based on their
porosity, as assessed using calibrated orthogonal radio-
graphs, using the procedure outlined in ASTM F2118-
01a. The resolution of the radiographic measurements
was 70.01 mm. We excluded tensile, compressive, and
fatigue specimens with radiographic evidence of internal
voids of greater than 1 mm, dened as macropores in
a review by Lewis [9]. Lewis also dened defects of less
than 1 mm in diameter as micropores. We also
excluded compression specimens having visible surface
defects or discontinuities of greater than 0.5 mm, as per
ASTM F451-99a. Tensile or fatigue specimens with
surface defects of greater than 0.25 mm were excluded as
per ASTM F2118-01a.
We further characterized the size and distribution of
micropores (less than 1 mm in diameter) within
qualied compression specimens. Seven compression
specimens for each bone cement, radiographically
prescreened such that they contained no macropores,
were further analyzed to determine the greatest dimen-
sion of the largest visible defect (micropores) on
radiographic views.
The porosity of the pre-screened compression speci-
mens was then further analyzed using microCT
(microCT 40: Scanco, Switzerland) at 12 mm resolution.
Each specimen was scanned in two axial positions,
located at 4 mm from each end of the specimen. At each
axial location, 20 slices were obtained, thus scanning a
total region of 480 mm at each position. Each stack of 20
two-dimensional slices was then reconstructed to create
a three-dimensional solid, and volumetric porosity
values were determined for each reconstruction.
Chi-square analysis was used to evaluate whether
bone cement formulation signicantly inuenced the
incidence of macropores during hand mixing. Differ-
ences in defect sizes and porosity values were statistically
evaluated as a function of cement using analysis of
variance (ANOVA). The statistical analyses were con-
ducted using Statview software (Version 5.0.1: SAS
Institute, Cary, NC).
2.4. Compression testing
Testing was performed using a calibrated MTS Test
Star IIM servo-hydraulic test system (MTS, Eden
Prairie, MN) with a 15-N capacity load cell. The steel
compression platens were made parallel to each other
using a feeler gage. Load frame compliance was
accounted for by using a calibrated strain gage. The
gage length was 10 mm and its range was 71.0 mm
(Model No. 3542, Epsilon Corp, Jackson, WY); it was
mounted across the test platens to measure compressive
strain.
Following the guidelines specied in ASTM F451-
99a, a test speed of 20 mm/min was used. Test
conditions were 22 1C and 20% relative humidity. Eight
(n 8) compression specimens of each formulation were
randomly selected from the pool of available pre-
screened specimens, as described in the previous section.
The testing order of the selected specimens was then
randomized.
Elastic modulus, yield stress, and compressive
strength were calculated as per ASTM F451-99a. These
metrics for the three materials were analyzed using
univariate analysis of variance (ANOVA), coupled with
post hoc t-tests to compare pairs of materials. The
statistical analyses were conducted using JMP software
(Version 5.0: SAS Institute, Cary, NC). In all of the
analyses, a p-value of 0.05 was used as the threshold of
signicance.
2.5. Tension testing
Tensile tests were performed using the same Test Star
load frame as was used for compression testing. Self-
aligned, hydraulic grips were used, and a contacting
extensometer (Model No. 3542, Epsilon Corp, Jackson,
WY) was attached to the gage region of the tensile
specimens to measure local strain. Five specimens of
each bone cement (n 5) were tested at a constant
displacement rate of 0.75 mm/min, which was a speed
that caused fracture within the
1
2
to 5-min timeframe that
the ASTM D638-02 standard suggests. The actual time
to failure for these experiments was approximately
12 min.
Tensile testing was performed both at room tempera-
ture with 20% relative humidity and at body tempera-
ture saturated with phosphate buffered saline (PBS).
The room temperature tension tests were intended to
replicate the environment during the compression tests,
whereas the body temperature saline tests were intended
to duplicate the conditions of the fatigue tests.
Tensile tests were performed approximately 30 days
after molding. Room temperature test specimens were
stored in a stable laboratory environment of 22 1C and
20% humidity, whereas the specimens for body tem-
perature tests were soaked for at least 25 days in 37 1C
ARTICLE IN PRESS
S.M. Kurtz et al. / Biomaterials 26 (2005) 36993712 3701
PBS that was balanced at a pH of 7.4. This protocol was
specied in ASTM F2118-01a which suggests soaking
for at least 7 days.
An environmental chamber, maintained at 3771 1C,
was allowed to stabilize overnight so that the xtures
reached the desired target temperature. An incubator,
also set at 37 1C, was placed a few feet away from the
test system and contained the tensile specimens soaking
in a 37 1C PBS bath. A spray bottle of PBS for
saturating specimens during testing was also stored in
the incubator. Body temperature testing was conducted
in air inside the environmental chamber, with the
presoaked specimen sprayed to ensure a uniform coating
of PBS on the entire specimen throughout the testing.
A SEM, described in the powder morphology section,
was used to image the carbon-coated fracture surface of
all of the room temperature-tested tensile specimens.
Images were taken at 20 , 100 , 300 and 2000 .
The tensile strength (stress at failure), % elongation
(strain at failure), and elastic modulus were calculated
for each specimen. The tensile test metrics were analyzed
using univariate analysis of variance (ANOVA), coupled
with post hoc t-tests to compare pairs of materials. The
statistical analyses were conducted using JMP software
(Version 5.0: SAS Institute, Cary, NC) running on a
personal computer. In all of the analyses, a p-value of
0.05 was used to judge signicance.
2.6. Fatigue testing
Fully reversed, tension-compression fatigue testing
was conducted on the three bone cements at 2 Hz in
accordance with ASTM F2118-01a. Stress levels of
710, 15, and 20 MPa were selected for this study,
consistent with previous testing plans reported for bone
cement in the literature [9]. Although ASTM F2118-01a
recommends that a minimum of eight specimens be
tested at each stress level for each cement (i.e., 24
specimens per cement), our study was designed with a
target of 15 specimens per stress level, corresponding to
a total sample size of 135 for the study (15 specimens/
stress level 3 stress levels 3 cements). We also
prepared extra specimens as potential replacements in
the event of unforeseen circumstances during testing.
Subsequent to specimen molding, all qualied speci-
mens were soaked in a 37 1C phosphate buffered saline
(PBS) solution that was pH balanced to 7.4. Specimens
soaked for a minimum of 7 days, following the
recommendation in ASTM F2118-01a. The soaking
time was tracked as a potential nuisance variable for
our subsequent statistical analyses.
Two servohydraulic load frames were used in the
testing. Each load frame was equipped with aligned,
hydraulic test grips and identical specimen chambers
designed to maintain the specimens in PBS at 3771 1C
for the duration of the fatigue experiment. The
specimens were randomized to avoid potential con-
founding due to specimen manufacture, test order, or
test machine.
An optical microscope (Leica S8AP0: Leica Micro-
systems Inc., Bannockburn, IL) was used to examine
each fatigue specimen after testing. A representative
sample of each type of bone cement and loading
magnitude was chosen for further SEM analysis (nine
specimens, total), using the instrument described in the
powder morphology section. SEM images were taken at
100 , 300 and 2000 .
Statistical analysis of the fatigue data, based on the
recommendations of ASTM F2118-01a, was conducted
upon completion of the experimental phase of this
study. Statistical analyses were performed using JMP
Discovery software (SAS Institute, Cary, NC), as well as
using the LIFEREG procedure of SAS software (SAS
Institute, Cary, NC). For all analyses, a p-value of 0.05
was considered signicant.
The fatigue life (mean7SD) was calculated using the
logarithmic transformation of the data as recommended
in section 10 in ASTM F2118-01a. Comparisons
between mean fatigue life of the three specimens were
performed using t-tests at each stress level, assuming a
log-normal distribution of the fatigue life and neglecting
the impact of censoring on the data.
Univariate and parametric survival analyses were
conducted to evaluate the effect of bone cement
formulation and cyclic stress level on the fatigue life.
The survival analyses took into account the censoring
effect of the single Simplex
s
P specimen test, which was
suspended prior to failure. KaplanMeier survival
curves, as well as Weibull curves, were constructed for
the three bone cements at each stress level. We also
calculated the fatigue strength distribution for the three
bone cements at each stress level, based on the three-
parameter Weibull model tted to the data. A nonpara-
metric survival analysis was performed, in which the
fatigue data for all three stress levels were pooled to
provide an overall assessment of signicance for each
bone cement. Finally, nonparametric survival analyses,
employing Wilcoxon tests, were conducted to evaluate
the effect of nuisance variables (e.g., soaking time, mold
used to prepare the specimens, and the testing machine)
on the fatigue life of the three bone cements.
3. Results
3.1. Bone cement formulations and powder morphology
All three bone cement formulations exhibited similar
powder morphology under SEM. Polymer beads
and white agglomerates were evident in all samples
(Figs. 1AC). The agglomerates observed in the samples
were conrmed to be barium sulfate by EDS. The three
ARTICLE IN PRESS
S.M. Kurtz et al. / Biomaterials 26 (2005) 36993712 3702
bone cements contained polymer beads that were under
100 mm diameter, and the barium sulfate presented as an
agglomerate of particles with typical length scales on the
order of 1 mm. The SEM images of polymer beads and
barium sulfate in this study were comparable to images
presented by Ku hn in his recent monograph [3]. Kuhns
monograph also indicates that the PMMA beads
typically range in size from 1 to 125 mm in diameter,
which is consistent with our observations.
3.2. Porosity evaluation
In both radiographs and microCT, there were clear
differences in the radioopacity of the Simplex
s
P bone
cement (containing 10% radiopacier) as compared to
the KyphX HV-R and the clinical formulations, which
contained 30% and 36% radiopacier, respectively (Fig.
2). From chi-square analysis, we observed signicant
differences in the prevalence of macropores (41 mm)
within the as molded cylindrical and dogbone speci-
mens of three bone cements (Tables 1A and B). The
overall incidence of macropores was 72% (152/212) in
cylindrical specimens, and 44% (159/359) in dog bone
specimens. For both cylindrical and dogbone specimens,
KyphX HV-R consistently had the lowest incidence of
macropores (Tables 1A and B). When molding cylind-
rical specimens, the clinical formulation had the highest
incidence of macropores, whereas Simplex
s
P had the
highest incidence of macropores when molding dogbone
specimens. However, we found no correlation between
the barium sulfate content and the incidence of
macroporosity, suggesting that other factors, such as
chemical formulation, also inuence the generation of
pores greater than 1 mm.
Among the pre-screened compression specimens,
there was no signicant difference in the maximum pore
size, based on analysis of all the plane radiographs.
The average maximum pore size determined radio-
graphically was 0.76 mm (range: 0.680.81 mm) for all
specimens that passed the initial pre-screening for
macroporosity.
MicroCT imaging was found to be an effective
technique for nondestructively evaluating the porosity
of bone cement. Using microCT, we were able to
visualize not only the size, geometry, and distribution of
micropores, but were also able to identify agglomera-
tions of barium sulfate in the two-dimensional images
(white regions in Fig. 2) of the cylindrical specimens.
However, we detected no signicant difference in
porosity between the three bone cement formulations
using microCT (Table 2). The average porosity deter-
mined from the 3-D microCT reconstructions was 5.7%
(range: 5.32 and 6.23%, Table 2).
3.3. Compression testing
The compressive true stressstrain curves of the three
bone cements exhibited similar qualitative features,
including an initial linear elastic regime, followed by
yielding and a local maximum (Fig. 3). The compressive
ARTICLE IN PRESS
Fig. 1. Scanning electron micrographs of (A) Simplex
s
P, (B) KyphX HV-R, and (C) clinical formulation at 300 magnication.
S.M. Kurtz et al. / Biomaterials 26 (2005) 36993712 3703
strength of all three bone cements exceeded the 70 MPa
minimum specication from ASTM F451-99a (Table 3).
The average elastic modulus, yield stress, and max-
imum compressive strengths of the three bone cements
were all signicantly different from one another
(po0:05; Table 3). The KyphX HV-R bone cement
exhibited signicantly higher elastic modulus, yield
stress, and maximum compressive strength than either
ARTICLE IN PRESS
Table 1
Incidence of macropores in molded cylindrical and dogbone specimens fabricated from three bone cements
Bone cement Specimen type Number molded Number containing micropores only Incidence of macropores (%)
Simplex
s
P (Control) Cylindrical 89 22 75.3
Dogbone 143 69 51.7
KyphX HV-R Cylindrical 68 30 55.9
Dogbone 109 71 34.8
Clinical formulation Cylindrical 55 8 85.5
Dogbone 107 60 43.9
Table 2
Summary porosity results for three bone cement formulations measured from compression specimens using microCT imaging (all were
radiographically prescreened to exclude the presence of macropores)
Cement Number of readings Mean of percent porosity StDev (%) Minimum (%) Maximum (%)
Simplex
s
P (Control) 14 6.23 1.44 4.27 8.25
KyphX HV-R 14 5.32 2.64 2.70 13.25
Clinical formulation 14 5.53 1.10 3.87 7.67
Fig. 2. MicroCT scan slices of (A) Simplex
s
P, (B) KyphX HV-R, and (C) clinical formulation at 12 mm resolution. The specimens are 6 mm in
diameter.
S.M. Kurtz et al. / Biomaterials 26 (2005) 36993712 3704
the Simplex
s
P or the clinical formulation (Table 3).
Increasing the barium sulfate content of Simplex
s
P
resulted in a signicant, but not substantial, increase in
compressive properties (Table 3). For instance, increas-
ing the barium sulfate content of Simplex
s
P from 10%
(as-manufactured) to 36% had the effect of increasing
the average elastic modulus and maximum compressive
strength by only 13% and 5%, respectively (Table 3).
3.4. Tension testing
The tensile stressstrain curves of the three bone
cements under dry, room temperature and saline, body
temperature conditions exhibited similar qualitative
features, including a linear elastic regime, followed by
brittle fracture (Figs. 4A and B). However, the dry
specimens clearly exhibit higher modulus and higher
tensile strength at a shorter elongation (Table 4).
The elastic modulus was signicantly affected by the
test environment (po0:05), but not by the bone cement
formulation (Table 4). Overall, the average modulus
for all the bone cement samples tested at room tem-
perature was determined to be 2690 MPa (range:
24402950 MPa); the wet specimens tested at 37 1C
had, on average, an elastic modulus of 1470 MPa (range:
13601520 MPa), corresponding to a 45% reduction as
compared with dry room temperature testing.
Testing environment, bone cement formulation, as
well as the interaction between test condition and
formulation were all signicant variables in our ANO-
VA modeling for the tensile strength (po0:05; Table 4).
The ultimate strengths of both KyphX HV-R and
Simplex
s
P were signicantly higher than the clinical
formulation (po0:05). Overall, the ultimate tensile
ARTICLE IN PRESS
Fig. 3. Compressive stressstrain curves for Simplex, KyphX HV-R,
and clinical formulations. Testing was conducted in room tempera-
ture air.
Table 3
Summary of uniaxial compressive properties as mean7SD
Bone cement
formulation
Compressive
modulus
(MPa)
2% Offset
yield strength
(MPa)
Compressive
strength (MPa)
Simplex 9076 9774 97 (4)
KyphX HV-R 10573 11174 111 (4)
Clinical
formulation
9675 10277 102 (7)
(All differences between materials summarized here are signicant,
po0.05).
Testing was conducted at room temperature in air.
Fig. 4. Tensile stressstrain curves for Simplex, KyphX HV-R, and
clinical formulations. Testing was conducted in room temperature air
(A) and body temperature saline (B).
S.M. Kurtz et al. / Biomaterials 26 (2005) 36993712 3705
strengths of KyphX HV-R and Simplex
s
P were
not signicantly different (p40:05; Table 5). The
tensile strength of the clinical formulation was between
22% and 25% lower than the neat Simplex
s
P bone
cement.
Testing environment and bone cement formulation
were found to be signicant parameters in our ANOVA
modeling of elongation to failure (po0:05; Table 4). The
elongation of Simplex was signicantly higher than the
clinical formulation (po0:05; Table 4). We did not
observe a consistent ranking of the average elongation
to failure for the three bone cements at both test
temperatures (Table 4).
The fracture surfaces of all three formulations were
comparable and characteristic of brittle fracture (Fig. 5).
Aggregates of barium sulfate were visible as lighter
colored patches or white spots on the fracture surfaces
even when viewed at 20 (Fig. 5C). This was consistent
with what was observed in the two-dimensional images
obtained via microCT scanning (Fig. 2). EDS analysis of
white spots or light-colored patches on the fracture
surfaces conrmed that they were agglomerations of
barium sulfate. At 300 and 2000 magnications we
observed barium surface particulates similar to those
observed in the SEM evaluation of bone cement
powders (Figs. 1AC). We also observed complete
PMMA beads inside the cross-sections of pores and in
cross-sections on the fracture surfaces of all three
formulations, indicating that the fracture propagated
around the polymer beads and through them.
3.5. Fatigue testing
During the execution of the fatigue testing program,
six specimens failed due to experimental difculties:
three failed prior to testing during placement in the
hydraulic grips, and the other three specimens failed
prematurely (within 13106 cycles) on a single day due
to malfunctioning of one of the hydraulic systems. After
repair of the hydraulic system, testing proceeded with-
out further irregularities. Excluding the six specimens,
which failed for reasons unrelated to cement perfor-
mance, valid fatigue data were collected for a total of
131 bone cement specimens, as summarized in Table 5.
Of the 131 specimens, 130 were fatigue tested to
failure. For one (Simplex
s
P) specimen, testing was
suspended after 1,556,921 cycles (considered a run-out).
One of the clinical formulation specimens and three of
the KyphX HV-R specimens were noted to fracture in
the grip region of the specimen. The remaining 126
specimens failed in the gage region.
KyphX HV-R had a signicantly greater mean fatigue
life (po0:05) as compared with both the clinical
formulation and the Simplex
s
P bone cements at
10 MPa (Fig. 6), when calculated using a log-normal
transformation as recommended in ASTM F451-99a.
However, these comparisons were slightly biased by the
run-out of one of the Simplex specimens. At the higher
stress levels, the fatigue life of KyphX HV-R was
signicantly greater than that of the clinical formulation
(po0:05), but not higher than the Simplex
s
P (p40:05).
Analysis using a three-parameter Weibull model,
taking into account censoring, consistently ranked the
characteristic fatigue life of the three cements as follows
for all three load levels (Fig. 7, Table 6): KyphX HV-
R4Simplex
s
P4clinical formulation. Evaluation of the
Weibull slopes at all three cyclic stress levels suggested
the following ranking (Table 7): KyphX HV-RESim-
plex
s
P4clinical formulation. In all cases, the predicted
fatigue life of KyphX HV-R was greater than that for
the clinical formulation; the predicted fatigue life of
KyphX HV-R was greater than that for Simplex for the
710 and 20 MPa stress conditions.
ARTICLE IN PRESS
Table 4
Summary of uniaxial tensile properties as mean7SD
Bone cement type Test environment Tensile modulus (MPa) Tensile strength (MPa) Elongation at maximum tension (%)
Simplex
s
P 22 1C, 20% RH 24407190 3271 1.2570.08
37 1C, PBS 1360770 3272 2.2970.6
KyphX HV-R 22 1C, 20% RH 29507360 3771 1.3670.23
37 1C, PBS 15207230 3171 1.8070.11
Clinical formulation 22 1C, 20% RH 26707380 2473 0.8370.11
37 1C, PBS 15107370 2573 1.5770.19
(All differences between materials summarized here are signicant, po0.05).
Testing was conducted at room temperature in air as well as at body temperature in saline (PBS).
Table 5
Fatigue test data collected for each bone cement type at each stress
level
No. of specimens
Bone cement 10 MPa 15 MPa 20 MPa
KyphX HV-R 15 16 16
Simplex
s
P 15 14 15
Clinical formulation 13 15 12
S.M. Kurtz et al. / Biomaterials 26 (2005) 36993712 3706
Parametric survival analyses indicated that the fatigue
life of the KyphX HV-R bone cement was signicantly
different from the other two bone cements, but not at all
stress levels. The p-values for the individual parametric
comparisons are summarized in Table 8. The statistics
from the three stress levels, when combined using
parametric test methods and using the Weibull failure
distribution, show an overall signicant difference
between KyphX HV-R and Simplex
s
P, as well as
between KyphX HV-R and the clinical formulation.
Nuisance experimental factors, such as soaking time, the
type of load frame, and the mold used to prepare the
samples, had no signicant effect on the survival times
of the bone cements (p40:2).
In general, the fatigue surfaces of all three formula-
tions were planar and consistent with brittle failure (Fig.
8). Fracture surface striations were typically visible
under SEM for all the fatigue specimens. We identied
possible fracture initiation sites in the Simplex
s
P and
the clinical formulation specimens, but we were not able
to determine the crack initiation source on the KyphX
HV-R samples. The clinical formulation samples loaded
at 15 and 20 MPa sometimes showed crack initiation in
large barium sulfate deposits, but micropores were the
most likely crack initiation sites in the Simplex
s
P
specimens. In general, KyphX HV-R specimens frac-
tured perpendicular to the long axis of the samples, with
minor step discontinuities at lower loading levels. The
Simplex
s
P and clinical formulation samples fractured
unevenly with larger step discontinuities, especially
towards the end of the fracture; these fractures often
changed direction, bisecting larger pores.
Stellate cracks radiated from the pores on the fracture
surface of Simplex
s
P (neat) specimens; in each of these
cases, at least one crack extended to the edge. These
pores were probable sources of crack initiation. The
ARTICLE IN PRESS
Fig. 5. Scanning electron micrographs of the tensile specimen fracture surfaces: (A) Simplex
s
P, (B) KyphX HV-R, and (C) clinical formulation at
20X magnication. Light-colored patches correspond to barium sulfate, whereas the darker features are pores. The specimens are 6 mm in diameter.
Fig. 6. Mean (7SD) Fatigue Life calculated as per ASTM F 2118-01a
using a log-normal transformation. *Note: mean is slightly biased due
to run-out of one specimen.
S.M. Kurtz et al. / Biomaterials 26 (2005) 36993712 3707
stellate cracks occurred in one of the samples of clinical
formulation (10 MPa), but did not occur in any of the
KyphX HV-R specimens. Cracks connected small pores
on the fracture surfaces of Simplex
s
P (neat); this was
also true for the lower loading levels of the clinical
formulation.
The addition of barium sulfate was apparent in both
the KyphX HV-R and the clinical formulation samples.
Fracture surfaces of the KyphX HV-R and the clinical
formulation samples were white, rough, and chalky due
to barium sulfate aggregates; in contrast, the Simplex
s
P
specimens had glassy or wet fracture surface. Barium
sulfate aggregates were visible at 20x on the fracture
surface of all KyphX HV-R specimens and larger
deposits were visible to the naked eye on all clinical
formulation specimens.
4. Discussion
The results of this study indicate that the static and
fatigue properties of PMMA-based bone cement are not
necessarily compromised by the addition of barium
ARTICLE IN PRESS
Fig. 7. Weibull Curves for (A) 710 MPa, (B) 715 MPa, and (C) 720 MPa.
Table 6
Weibull characteristic fatigue life for clinical formulation, Simplex
s
P,
and KyphX HV-R
Cyclic stress
level (MPa)
Weibull fatigue life (cycles)
Clinical
formulation
Simplex
s
P KyphX HV-R
710 131,745 221,606 360,553
715 18,739 26,549 59,340
720 795 2,087 3,821
Table 7
Weibull slope for clinical formulation, Simplex
s
P, and KyphX HV-R
Cyclic stress
level (MPa)
Weibull slope
Clinical
formulation
Simplex
s
P KyphX HV-R
710 0.659 0.823 1.04
715 0.707 1.37 0.963
720 0.473 1.27 1.40
S.M. Kurtz et al. / Biomaterials 26 (2005) 36993712 3708
sulfate to improve visualization during injection into the
spine. We found that adding radiopacier to Simplex
had the result of lowering the tensile strength and
fatigue life of the bone cement. Large agglomerations of
barium sulfate, which acted as local stress concentra-
tions or fatigue initiation sites (Fig. 8c), are hypothe-
sized to be responsible for the reduced mechanical
performance of the clinical formulation as compared
with Simplex
s
P. On the other hand, the tensile,
compressive, and fatigue properties of KyphX HV-R
were comparable or superior to Simplex
s
P. Conse-
quently, our data indicate that bone cement formula-
tion, when tailored for a specic spinal application, can
successfully neutralize the effects of elevated barium
sulfate on the tensile and fatigue properties.
Careful radiographic prescreening, veried by mi-
croCT analysis, ruled out porosity as a potential
confounding factor in our experiments. We found no
signicant difference in the average porosity of the three
bone cements, which were, on average, between 5.32%
and 6.23%. Lewis [11] reviewed 13 previous studies
reporting the porosity of Simplex
s
P. For manual
mixing techniques, the average porosity for Simplex
s
P
was reported between 9.4% and 12.5%; with centrifuga-
tion or mechanical mixing techniques, the porosity
ranged between 0.5% and 6.5% [11]. Thus, our porosity
measurements were lower than those previously re-
ported for manual mixing and fell within the range
associated with centrifugation or mechanical mixing.
The lower porosity of our samples, relative to values
published in the literature, is likely related to our
exclusion of samples containing macropores of 1 mm or
greater, as required by ASTM F2118-01a. It should be
emphasized that macropores of greater than 1 mm may
be observed in PMMA-based bone cements in vivo.
Thus, the specimen selection criteria for the present
study may not encompass the full range of aw sizes that
could be deployed clinically. Nevertheless, to control for
the potentially confounding effect of pore size, speci-
mens that were selected for this study were randomly
selected from the population of pre-screened candidates.
Thus, the results of our microCT study should be
interpreted as providing porosity content of hand-mixed
bone cements containing only micropores. We conclude,
based on the results of our study, that the addition of
barium sulfate does not signicantly inuence the
ARTICLE IN PRESS
Table 8
Summary of Weibull survival analysis results (p-values) comparing
KyphX HV-R with clinical formulation and Simplex
Comparison Cyclic stress levels (MPa)
710 715 720
KyphX HV-R vs. Simplex
s
P cement NS 0.01 0.03
KyphX HV-R vs. clinical formulation
cement
NS 0.04 NS
(NS=not signicant).
Fig. 8. Scanning electron micrographs of representative fatigue specimen fracture surfaces (715 MPa): (A) Simplex
s
P, (B) KyphX HV-R, and (C)
clinical formulation at 20X magnication. Light-colored patches correspond to barium sulfate, whereas the darker features are pores. The specimens
are 6 mm in diameter.
S.M. Kurtz et al. / Biomaterials 26 (2005) 36993712 3709
average microporosity of PMMA-based bone cement.
On the other hand, based on the incidence of macro-
pores in three bone cements (Table 1), there is some
evidence that the addition of barium sulfate to
Simplex
s
P can inuence the generation of pores greater
than 1 mm. The clinical signicance of macroporosity
for the treatment of vertebral compression fractures
remains unclear. Unlike in joint arthroplasty, there are
insufcient data at the present time to make a statement
as to the clinical signicance of macropores in the
treatment of vertebral compression fractures.
Because the properties of bone cements developed
specically for spinal applications have not previously
been reported, we chose Simplex
s
P bone cement as the
control for our study because of its long successful
clinical history in joint replacements and for the
extensive body of literature documenting its properties
under a wide range of testing conditions. For example,
the compressive modulus and compressive strength
of Simplex
s
P measured in our series of experiments
compared favorably with previously published
data. According to Lewis review [11], the compressive
modulus for Simplex
s
P has been reported to range
between 2540 and 3010 MPa depending upon the
specimen preparation and curing conditions;
similarly, the compressive strength for Simplex
s
P is
reported to range between 84.8 and 114.3 MPa [11].
Therefore, the compressive properties we measured
for Simplex
s
P were in general agreement with the
published literature.
Similarly, the tensile modulus, tensile strength, and
percent elongation of Simplex
s
P measured in our series
of experiments compared favorably with previously
published data. According to a recent review article,
the tensile modulus for Simplex
s
P has been reported to
range between 1583 and 3080 MPa depending upon the
specimen preparation and curing conditions; this range
compared favorably with the value of 2440 MPa we
obtained at room temperature. While the modulus
obtained in our tests at body temperature (1364 MPa)
is lower than the moduli summarized by Lewis, it is
important to note that of the summarized modulus tests
reported being done at body temperature none of them
indicated that the specimens soaked for the same
duration. We measured the tensile strength of Sim-
plex
s
P to be 32 MPa at room and body temperature,
which falls within Lewis reported range of 27 and
46 MPa. In particular, one study utilized similar condi-
tions to our study: the cement was mixed manually and
aged in saline at 37 1C for 35 days [12]. The average
strength for these samples was 30.2 MPa, which falls
within 6% of the average value we obtained under
comparable conditions. Finally, the values for elonga-
tion at maximum tension for both of our test conditions
(1.25% at RT, and 2.29% at BT) were within the
reported range (0.862.49%) in Lewis literature review.
Overall, the tensile properties of Simplex
s
P (Table 4)
were in general agreement with the extensive published
literature for this biomaterial.
We employed the same specimen geometry for our
tensile tests as was used in our fatigue experiments,
based on ASTM F2118-01a. Utilizing the same speci-
men size and shape for fatigue and tension (as per
ASTM F2118-01a) has the benet of enabling a single
mixing and molding protocol, thus improving
specimen uniformity and decreasing variability.
Furthermore, the geometry of the fatigue specimens in
ASTM F2118-01a is smaller than the sizes recom-
mended in ASTM D638-02, thereby requiring less bone
cement per specimen. Based on the favorable results
achieved for Simplex in this study in comparison with
the literature, we advocate using ASTM F2118-01a
fatigue specimen geometry in future uniaxial tension
studies of bone cement.
Our fatigue experiments were conducted using a new
standard (ASTM F2118-01a), complicating the compar-
ison of our data with studies in the literature that were
formed using a wide range of different test conditions.
The stress levels chosen for testing (10, 15 and 20 MPa)
in our study were executed under fully reversed cyclic
loading at 2 Hz following recommendations in the
standard; these choices of stress amplitude and fre-
quency were also well documented in the literature [10].
Several studies conducted fully reversed cyclic loading at
15 MPa for Simplex
s
P, reporting a range of 23324,218
cycles; our 19,027 mean cycles to failure at 715 MPa
fall within this previously reported range [1319]. More
specically, a study by Jacobs et al. has reported data
for Simplex
s
P that are directly comparable to our
study, since it was performed under very similar
conditions, including hand-mixing (HM), directly mold-
ing into nal dimensions (DIM), cylindrical cross-
section (CYL), tested in reversed tension/compression
at 15 MPa, and at a frequency of 2 Hz [18]. The mean
number of stress cycles to fracture for their study is
18,000, which is within 6% of our mean reported value.
Lewis did not review reports of mean cycles to fracture
at fully reversed cyclic loading with amplitude 10 and
20 MPa to compare with our results, but our reported
results (133,990 cycles at 10 MPa and 1770 cycles at
20 MPa) are generally consistent with the data reviewed
by Lewis [10]. It is important to note that the studies
reviewed by Lewis differ markedly in protocol (e.g.,
mixing, specimen fabrication method, sample cong-
uration, stress magnitude, and test frequency). Overall,
our fatigue results for Simplex
s
P are generally compar-
able with previous studies.
This study shares many of the limitations commonly
attributed to studies of bone cement properties, namely
the sensitivity of the results to specimen preparation,
testing rate, and testing environment. All of these
factors have been documented to inuence the mechan-
ARTICLE IN PRESS
S.M. Kurtz et al. / Biomaterials 26 (2005) 36993712 3710
ical properties of bone cement [9,10], but to address
these potential sources of variability, we have relied on
standardized test methods and used a control material
(Simplex
s
P) with a well-documented set of mechanical
properties. Nevertheless, the mechanical properties
reported this study must be interpreted within a tightly
controlled set of experimental conditions (e.g., hand
mixing, molding of specimens, radiographic pre-screen-
ing), necessary to facilitate comparisons between
PMMA bone cements historically used for joint
replacements and new biomaterials intended for spine
applications.
The KyphX HV-R bone cement formulation eval-
uated in this study is intended specically for the
treatment of vertebral compression fractures. The
stresses placed upon bone cement in the spine are
generally compressive and much lower in magnitude
than in the cement mantle of an arthroplasty stem. For
example, at the distal end of a hip stem, the local cement
stresses can exceed 10 MPa. In addition, these stresses
may be fully reversed as the femur bends around the
stem [20,21]. In the spine, the stresses in the bone cement
within a vertebral body are typically axially oriented and
compressive [22,23]. Furthermore, the magnitude of
bone cement stresses in the spine is predicted to be
approximately 1 MPa or less [22,23], which is an order
of magnitude less than the tensile strength of the
material. Note that the fully reversed, 710-, 715-,
and 720-MPa stress ranges in this study, which are
based upon the recommendations of ASTM F2118-01a,
are also about an order of magnitude more severe in
magnitude than the loading conditions to which bone
cement can be subjected in the spine. Despite these
severe fatigue loading conditions, the fatigue life of
KyphX HV-R was found to be comparable to or greater
than Simplex
s
P. Taken together, the static and fatigue
testing data herein indicate that KyphX HV-R is not at
risk of static or fatigue fracture during typical in vivo
loading conditions in the spine.
5. Conclusions
The addition of barium sulfate radiopacier to
improve ow visualization does not necessarily com-
promise the static and fatigue properties of PMMA-
based bone cements used for the treatment of vertebral
compression fractures. KyphX HV-R exhibited compar-
able or superior tensile, compressive, and fatigue
properties when compared with neat Simplex
s
P. Add-
ing radiopacier directly to Simplex
s
P in the clinical
formulation lowered the tensile strength and fatigue life
of the resulting bone cement. We hypothesize that
agglomerations of barium sulfate particles in the clinical
formulation acted as local stress concentrations or
fatigue crack initiation sites, and were likely responsible
for the lower tensile and fatigue properties, relative to
neat Simplex
s
P.
Acknowledgment
Supported by a Research Grant from Kyphon, Inc.
References
[1] Haboush EJ. A new operation for arthroplasty of the hip based
on biomechanics, photoelasticity, fast-setting dental acrylic, and
other considerations. Bull Hosp Jt Dis 1953;14(1):24277.
[2] Charnley J. Total hip replacement by low-friction arthroplasty.
Clin Orthop 1970;72:721.
[3] Ku hn K- D. Bone cements: up-to-date comparison of physical
and chemical properties of commercial materials. Berlin: Springer;
2000.
[4] Watts NB, Harris ST, Genant HK. Treatment of painful
osteoporotic vertebral fractures with percutaneous vertebroplasty
or kyphoplasty. Osteoporos Int 2001;12(6):42937.
[5] Theodorou DJ, Theodorou SJ, Duncan TD, Garn SR, Wong
WH. Percutaneous balloon kyphoplasty for the correction of
spinal deformity in painful vertebral body compression fractures.
Clin Imaging 2002;26(1):15.
[6] Garn SR, Yuan HA, Reiley MA. New technologies in spine:
kyphoplasty and vertebroplasty for the treatment of painful
osteoporotic compression fractures. Spine 2001;26(14):15115.
[7] Kim HY, Yasuda HK. Improvement of fatigue properties of
poly(methyl methacrylate) bone cement by means of plasma
surface treatment of llers. J Biomed Mater Res 1999;48(2):
13542.
[8] May B, Subrata, S. The effect of increased barium sulfate on the
mechanical properties of bone cement. In: 28th annual meeting of
the Society for Biomaterials; 2002 April 2427. Tampa, FL, USA,
2002.
[9] Lewis G. Properties of acrylic bone cement: state of the art review.
J Biomed Mater Res 1997;38(2):15582.
[10] Lewis G. Fatigue testing and performance of acrylic bone-cement
materials: state-of-the-art review. J Biomed Mater Res
2003;66B(1):45786.
[11] Lewis G. Properties of acrylic bone cement: state of the art review.
Appl Biomater 1997;38(2):15582.
[12] Krause W, Hoffman A. Antibiotic impregnated acrylic bone
cements: a comparative study of the mechanical properties.
J Bioactive Compat Polym 1989;4:34561.
[13] Davies JP, Burke DW, OConnor DO, Harris WH. Comparison
of the fatigue characteristics of centrifuged and uncentrifuged
Simplex P bone cement. J Orthop Res 1987;5(3):36671.
[14] Davies JP, OConnor DO, Burke DW, Jasty M, Harris WH. The
effect of centrifugation on the fatigue life of bone cement in the
presence of surface irregularities. Clin Orthop 1988;229:15661.
[15] Davies JP, OConnor D, Burke DW, Harris WH. Comparison of
centrifuged and vacuum mixed Simplex P. In: Transactions of the
34th annual meeting of the Orthopaedic Research Society; 1988,
Atlanta, GA, 1988. p. 221.
[16] Davies JP, OConnor DO, Burke DW, Greer JA, Harris WH.
Comparison and optimization of three centrifugation systems for
reducing porosity of Simplex P bone cement. J Arthroplasty
1989;4(1):1520.
[17] Davies JP, Jasty M, OConnor DO, Burke DW, Harrigan TP,
Harris WH. The effect of centrifuging bone cement. J Bone Jt
Surg Br 1989;71(1):3942.
ARTICLE IN PRESS
S.M. Kurtz et al. / Biomaterials 26 (2005) 36993712 3711
[18] Jacobs CR, Davis BR, Naidu S, Dauckardt RH, Pellegrini VD.
Small amounts of PBMA increase the fatigue lifeetime of acrylic
bone cement. In: Transactions of the 44th annual meeting of the
Orthopaedic Research Society; 1998. New Orleans, LA, 1998.
[19] Lewis G, Janna S, Carroll M. Effect of test frequency on the in
vitro fatigue life of acrylic bone cement. Biomaterials 2003;24(6):
11117.
[20] Harrigan TP, Kareh JA, OConnor DO, Burke DW, Harris WH.
A nite element study of the initiation of failure of xation in
cemented femoral total hip components. J Orthop Res 1992;10(1):
13444.
[21] Mann KA, Werner FW, Ayers DC. Modeling the tensile behavior
of the cementbone interface using nonlinear fracture mechanics.
J Biomech Eng 1997;119(2):1758.
[22] Villarraga M, Cripton P, Kurtz S, Edidin A. The biomechanical
effects of kyphoplasty on single thoracolumbar vertebrae: a
parametric nite element analysis. In: 49th annual meeting of the
Orthopaedic Research Society; 2003. New Orleans, 2003. p. 29.
[23] Villarraga M, Bellezza AJ, Harrigan, TP, Cripton PA, Kurtz SM,
Edidin, A. The biomechanical effects of kyphoplasty on treated
and adjacent non-treated vertebral bodies. J Spinal Disord Tech
2005, in press.
ARTICLE IN PRESS
S.M. Kurtz et al. / Biomaterials 26 (2005) 36993712 3712