J Neurosurg (Spine 3) 96:309–312, 2002

Increased pedicle screw pullout strength with vertebroplasty

augmentation in osteoporotic spines

JOHN S. SARZIER, M.D., AVERY J. EVANS, M.D., AND DAVID W. CAHILL, M.D.
Department of Neurosurgery, Division of Interventional Neuroradiology, University of South Florida,
Tampa, Florida

Object. The authors conducted a biomechanical study to evaluate pedicle screw pullout strength in osteoporo-
tic cadaveric spines. Nonaugmented hemivertebrae were compared with pressurized polymethylmethacrylate (PMMA)–
augmented hemivertebrae.
Methods. Six formalin-fixed cadaveric thoracolumbar spines at least two standard deviations below the mean bone
mineral density (BMD) for age were obtained. Radiographic and BMD studies were correlated to grades I, II, and III
osteoporosis according to the Jekei scale. Each of the 21 vertebrae underwent fluoroscopic placement of 6-mm
transpedicular screws with each hemivertebra serving as the control for the contralateral PMMA-augmented hemiver-
tebra. Pedicle screws were then evaluated for biomechanical axial pullout resistance.
Augmented hemivertebrae axial pullout forces were increased (p = 0.0005). The mean increase in pullout force was
181% for Grade I, 206% for Grade II, and 213% for Grade III osteoporotic spines. Augmented Grade I osteoporotic
spines demonstrated axial pullout forces near those levels reported in the literature for nonosteoporotic specimens.
Augmented Grade II osteoporotic specimens demonstrated increases to levels found in nonaugmented vertebrae with
low-normal BMD. Augmented Grade III osteoporotic specimens had increases to levels equal to those found in
nonaugmented Grade I vertebrae.
Conclusions. Augmentation of osteoporotic vertebrae in PMMA-assisted vertebroplasty can significantly increase
pedicle screw pullout forces to levels exceeding the strength of cortical bone. The maximum attainable force appears
to be twice the pullout force of the nonaugmented pedicle screw for each osteoporotic grade.

KEY WORDS • pedicle screw • polymethylmethacrylate • vertebroplasty

INCE the introduction of intrapedicular fixation by
S Roy-Camille, et al.,15 in 1963, the use of pedicle
screw instrumentation has become increasingly pop-
ular and effective in the management of vertebral frac-
tures, scoliotic deformities, metastatic disease, and degen-
erative disorders.6,23 The advantage of this construct is the
achievement of a shorter, more rigid spinal fixation that
leads to a faster incorporation of bone graft and subse-
quent segmental fusion.11 The key to the fixation, howev-
er, lies in the strength of purchase obtained by the screws
in the pedicle and in the trabecular bone of the VB.12 There
are several variables that can contribute to the loss of pedi-
cle fixation.4,9–11,15,17,24 Of these variables, osteoporosis is
frequently found in patients undergoing spinal fusion and
has been implicated as a cause of hardware failure at an
unknown rate. Loss of purchase and screw loosening in
older patients with degenerative spondylosis have been
reported to occur intraoperatively at a rate of 1.7% and
postoperatively at a rate of 3.8%.23 Because of the
increased use of pedicle fixation systems in elderly pa-
Abbreviations used in this paper: BMD = bone mineral density;
CT = computerized tomography; PMMA = polymethylmethacry-
late; VB = vertebral body.
tients with degenerative disease, the problem has become
more prevalent.
Several investigators have sought to augment the in-
tegrity of the screw–bone interface by the insertion of
milled bone, matchstick bone, biodegradable composite,
calcium phosphate, or PMMA into the fractured, stripped,
or native vertebrae, each with varying degrees of suc-
cess.1,2,5,10,14,16,19,21,25 In several clinical and laboratory stud-
ies the use of PMMA has been reported to produce statis-
tically significant increases in axial pullout resistance and
transverse bending stiffness. Its use, however, has been
limited to repair after intraoperative screw stripping or
fracture of the pedicle, and/or with only small amounts of
nonpressurized PMMA being applied. Its limited applica-
tion has been based on concern for infection, giant cell
reaction causing bone resorption, irremovable hardware,
or extracorporeal PMMA extrusion causing pulmonary
embolus or mechanical or thermal irritation.5,20 In addi-
tion, Soshi, et al.,16 have reported insignificant increases in
pullout force in severely osteoporotic vertebrae and sug-
gest that patients with Jekei Grade III osteoporosis not be
considered candidates for transpedicular fixation, even
with augmentation.
Recent advances in percutaneous PMMA-assisted ver-
tebroplasty of osteoporotic compression fracture have in-

J. Neurosurg: Spine / Volume 96 / April, 2002 309

 J. S. Sarzier, A. J. Evans, and D. W. Cahill

creased the degree of VB filling, the safety of the proce-

dure, and the structural stability and density of the VB. It
is postulated that combining a formal vertebroplasty—that
is, maximum filling of the trabecular space with PMMA—
with pedicle screw placement in osteoporotic vertebrae
could result in resistance to pullout forces significantly
higher than previously reported and in all grades of osteo-
porosis.
We classified the severity of osteoporosis on lateral radio-
graphs according to ther Jekei scale:7 Grade 0, a normal
pattern of transverse and vertical trabecula; Grade I, de-
creased transverse trabecula with prominent appearance of
vertical trabecula and endplates; Grade II, similar decrease
in vertical and transverse trabecula; and Grade III, near dis-
appearance of the transverse trabecular pattern as well as
unclear vertical trabecula appearing as ground glass.

Materials and Methods

Six formalin-fixed cadaveric spines were obtained from the Uni-
versity of South Florida Department of Anatomy. The specimens
were examined for a history of disease. The spines were evaluated
radiographically for metastatic, metabolic, or structural disease that
would otherwise compromise bone integrity. The specimens were
then assessed for osteopenia by using quantitative CT scanning. The
four spines chosen for testing were at least two standard deviations
below the mean BMD for age. Plain radiographic data and BMD
values were correlated with the Jekei scale for osteoporosis. Speci-
men 1 had Grade I, Specimen 2 had Grade II, and Specimens 3 and
4 had Grade III osteoporosis.
The spines were freed of all soft tissue, and the pedicles were
exposed bilaterally. A unilateral fluoroscopic vertebroplasty was
performed via a transpedicular approach in which a 13-gauge nee-
dle and cranioplastic PMMA were used (Codman and Shurluff,
Randolph, MA). The augmented pedicle was chosen randomly. The
augmentation was performed as described by Jensen, et al.,8 at each
level from T-12 to L-5, leaving the contralateral pedicle to act as a
control for each vertebral segment. A mean volume of 4 ml (range
3.25–6 ml) of PMMA was injected into each level based on the
maximum filling of the hemivertebrae as seen on fluoroscopy as
well as volumes reported by Jensen, et al., for bi- and unipedicular
approaches.18 An equal-sized needle was passed to a similar depth
and angle in the control pedicle under fluoroscopic guidance. Be-
fore the PMMA could harden, 6  40–mm self-tapping transpedic-
ular pedicle screws (Synthes Spine, Paoli, PA) were placed into the
augmented and control pedicles in the standard fashion. All screws
were inserted to a symmetrical depth and angle confirmed on later-
al and cephalocaudal fluoroscopy and CT scanning (Fig. 1). Three
vertebrae in which screws were incorrectly placed, three vertebrae
with CT evidence of VB fracture, and four vertebrae with evidence
of PMMA extension across the midline to involve the control screw
were discarded. The VBs were disarticulated from each other be-
cause of restrictions imposed by the materials testing machine (In-
stron, Canton, MA).
Individual VBs were secured in a PMMA cast and an aluminum
fixation device and mounted on a materials testing machine. The
screw heads were attached to a stainless-steel ring by using the col-
lar and nut used for operative fixation of the rod to the screw. A lin-
ear displacement force was applied along the long axis of the screw
at a rate of 1 mm/minute. Load values were constantly recorded by
uniaxial load cell and arranged graphically on a computer until the
force of pullout resistance peaked and declined and the screw had
been displaced, as measured by uniaxial cell. The maximum pullout
forces of the augmented and nonaugmented/control specimens were
analyzed using the Student paired t-test.
Results
During axial pullout testing of the control screws, all
FIG. 1. Axial CT scan of a VB in which hemivertebroplasty and
bilateral pedicle screw placement were performed.
cases of failure occurred as the threads of the pedicle
screws stripped through the trabecular bone of the pedicle
and VB. There was no evidence of cortical fracture or
compromise in the control specimens. Of the PMMA-aug-
mented vertebrae, nine of 21 screws failed at areas other
than the pedicle. In these cases the VB demonstrated a
cortical fracture and subsequent shearing of the VB be-
tween the augmented and nonaugmented sections of the
trabecular vertebrae. The pedicle screw, PMMA, and sur-
rounding trabecular and cortical bone were distracted en
bloc. For statistical analysis, the force measured at the
time of cortical failure was recorded as the maximum pull-
out force because the recorded axial forces declined after
cortical failure and there was no evidence of stripping or
movement of the augmented pedicle screw. The remain-
ing 12 PMMA-augmented screws demonstrated peak ax-
ial forces at the time of cortical failure at the base of the
pedicle, with the subsequent removal of the osseous pedi-
cle, PMMA, and trabeculae en bloc. None of the aug-
mented pedicle screws stripped out of the pedicle or VB,
as has been previously described in the literature.
Comparative results of control and PMMA-augmented
vertebrae are summarized in Table 1. The mean axial pull-
out force in the Grade I osteoporotic control side was
764.7 N (range 464–896 N). The PMMA-augmented side
had a mean maximum axial force of 1363 N (range 1072–
1620 N). The mean increase in pullout strength was 181%
(range 161–231%). In the Grade II osteoporotic spines,
the mean maximum pullout force on the control side
was 412 N (range 288–512 N). The augmented side had
a mean maximum axial force of 856 N (512–1216 N).
The mean increase in pullout strength for Grade II was
206% (range 159–292%). In the Grade III osteoporotic
spines, the mean maximum pullout force for the control
side was 289 N (range 144–570 N). The augmented side

310 J. Neurosurg: Spine / Volume 96 / April, 2002

Pedicle screw augmentation with PMMA

demonstrated a mean maximum force of 576 N (range TABLE 1

352–832 N). The mean increase in pullout strength for Bone mineral density, axial pullout forces, and mode of failure in
Grade III osteoporosis was 213% (range 124–333%). The control and PMMA-augmented transpedicular screws
paired Student t-test demonstrated a significant difference
between the augmented and control sides (p = 0.0005). Pedicle Screws (N) % Increase
Specimen BMD in Pullout Mode of
The number of screws tested did not allow for statistical No. (mg/ml) Control Augmented Force Failure*
analysis for each osteoporotic grade. There was no signif-
icant difference in the degree of osteoporosis and the in- 1
crease in axial pullout force. There was no evidence of T-12 82 764 1232 161 pedicle
L-1 97 896 1472 164 pedicle
correlation between the location of cortical failure, the L-2 70 768 1280 163 pedicle
maximum PMMA-augmented pullout forces, and the de- L-3 104 880 1620 184 pedicle
gree of osteoporosis. The BMD was noted to have a linear L-4 57 464 1072 231 VB
correlation with the native axial pullout force. L-5 92 816 1504 184 VB
2
T-12 66 492 780 159 pedicle
Discussion L-1 62 352 608 173 pedicle
The effectiveness of pedicle screw–fixated instrumenta- L-2 76 416 1216 292 VB
L-3 75 512 1168 228 VB
tion is critically dependent on the bone–screw interface.10 L-4 60 288 512 178 VB
Rather than metal fatigue, this interface is the variable 3
most likely to lead to instrumentation failure and thus T-12 66 320 560 269 VB
pseudarthrosis or loss of spinal alignment, necessitating L-1 59 224 480 175 VB
possible reoperation in the patient with osteoporosis.14 L-2 42 144 480 214 VB
Since the advent of intrapedicular fixation, there have L-3 65 386 832 333 VB
L-4 53 192 416 216 VB
been several cadaveric studies in which investigators have L-5 73 226 608 217 VB
found a correlation between pedicle screw failure and the 4
decreased BMD of the vertebrae seen in osteoporosis.3,13, L-1 48 288 680 236 VB
16,22,25
In 1990, Coe, et al.,3 reported the linear relationship L-2 62 570 832 146 pedicle
between BMD measured by dual photon absorptiometry L-3 55 285 352 124 pedicle
and the maximum pullout strength expressed in the fol- L-4 50 256 528 206 VB
lowing formula: pullout force = 43.6
499 (BMD). They * pedicle = maximum axial pullout force obtained at fracture of cortex at
reported the strongest correlation in specimens instru- base of pedicle; VB = maximum axial pullout force obtained at fracture of
mented with spinous process wiring, Cotrel-Dubousset cortex in VB.
screws, and Steffee transpedicular screws, whereas in
those specimens instrumented with laminar hooks there
was little correlation between BMD and maximum fixa- et al.,16 Jekei Grade III osteoporotic cadaveric spines were
tion force. This difference was attributed to the high corti- noted to have poor fixation irrespective of the size of pedi-
cal bone content of the lamina, which was noted to be less cle screw used or augmentation with pressurized PMMA as
affected by osteoporosis than cancellous bone.3 Other in- described by Zindrick, et al.25 It has been suggested that
vestigators have reinforced the linear association between such patients are not candidates for transpedicular instru-
BMD, as measured by quantitative CT scanning, micro- mentation even when PMMA augmentation was perform-
density measurements, or plain radiographic changes ed.14 The implication of this statement appeared contrary to
(Jekei scale) and pedicle fixation.13,16,22 Nonetheless, a the noted correlation between BMD and pullout force.
clinically applicable critical threshold for BMD and sub- Analysis of the data in this study demonstrates that the
sequent probability of pedicle screw failure have yet to be axial pullout strength of pedicle screws in cadaveric osteo-
elucidated.3,13,16,22 porotic spines can be increased by a mean of 200% when
Several investigators have sought, with varying degrees maximum amounts of PMMA are used to augment the
of success, to restore or augment the integrity of the screws via a transpedicular approach. Contrary to the find-
screw–bone interface in fractured or stripped pedicles in ings published by Soshi, et al.,16 even the most severely
vivo and in vitro.1,9,14,16,19,21,25 In 1986, Zindrick, et al.,25 osteoporotic spines in our study demonstrated statistical-
reported that stripped pedicle screws could be restored to ly significant increases in pullout force. The PMMA-in-
within 5% of their baseline pullout strength in cadaveric duced increase in the trabecular bone density consistently
specimens when the pedicle screw track was filled with increased the pullout resistance to levels that exceeded the
PMMA. They found that pressurized injection of 2.5 ml of strength of the cortical bone resulting in fracture of the VB
PMMA into the cancellous bone of the posterior VB just or fracture at the pedicle–VB junction. The failure mode
anterior to the pedicle increased the mean pullout force by was therefore shifted from screw purchase to cortical bone
96% compared with similar nonaugmented osteoporotic fracture. Consistently, the cortical failure occurred at a
vertebrae.24 Subsequently, several other investigators have force twice that required for screw pullout in nonaug-
reported results that support this finding.9,14,16,17,19,21 mented vertebrae in each grade.
When the patient suffers from significant osteopenia, The clinical significance of these results remains un-
however, restoration to baseline strength or even doubling clear, because there is no clinically applicable model with
of pullout force may be inadequate to anchor the pedicle which to predict the forces that each pedicle screw must
screw and secure a rigid construct, which is necessary for support; therefore, a threshold BMD value that indicates
subsequent osseous fusion. In the study reported by Soshi, the need for augmentation cannot be elucidated at present.

J. Neurosurg: Spine / Volume 96 / April, 2002 311


In addition, the mode of failure investigated in this study
(maximum pullout force) may not accurately represent the
failure mode seen in constructs used clinically. Many in-
vestigators believe that cyclic transverse bending more
accurately mimics the clinical situation. This will be eval-
uated in a subsequent study. Bone density in the trabecu-
la, pedicle, and cortex, however, has undeniably signifi-
cant importance in the biomechanics of screw behavior in
both modes of testing.
The PMMA augmentation as used in this study has had
limited clinical acceptance because of the risks of extrava-
sation and subsequent thermal neural injury, giant cell
reaction and bone resorption, and the permanence of the
material. At present, however, there is no substance pos-
sessing equal compressive and tensile strength that can
positively affect bone remodeling in osteoporotic bone.
Conclusions
1) Vertebroplasty, or maximal filling of the trabecula
with PMMA, can increase the pullout strength of pedicle
screws to twofold over that of nonaugmented, similar
grade osteoporotic vertebrae. 2) The mode of failure in
axial pullout is then shifted to cortical bone fracture in
spines of all osteoporotic grades. 3) The maximally aug-
mented axial pullout forces in severely osteoporotic verte-
brae approach values reported in the literature for nonaug-
mented, minimally osteoporotic spines.
Acknowledgments
The authors thank Richard Parker of Synthes Spine for supplying
pedicle screws and instruments as well as Shriners Hospital, Tampa,
Florida, for allowing access to materials testing machines.
References
1. Ashman RB, Galpin RD, Corin JD, et al: Biomechanical analy-
sis of pedicle screw instrumentation systems in a corpectomy
model. Spine 14:1398–1405, 1989
2. Brantley AG, Mayfield JK, Koeneman JB, et al: The effects of
pedicle screw fit. An in vitro study. Spine 19:1752–1758, 1994
3. Coe JD, Warden KE, Herzig MA, et al: Influence of bone min-
eral density on the fixation of thoracolumbar implants. A com-
parative study of transpedicular screws, laminar hooks, and spi-
nous process wires. Spine 15:902–907, 1990
4. Dick W: The “fixateur interne” as a versatile implant for spine
surgery. Spine 12:882–900, 1987
5. Halvorson TL, Kelley LE, Thomas KA, et al: Effects of bone
mineral density on pedicle screw fixation. Spine 19:2415–2420,
1994
6. Hirabayashi S, Kumano K, Kuroki T: Cotrel-Dubousset pedicle
screw system for various spinal disorders. Merits and problems.
Spine 16:1298–1304, 1991
7. Itami Y, Ohata Y: The epidemiology and physical condition in
osteoporosis. J Jap Orthop Assoc 38487–489, 1964
8. Jensen ME, Evans AJ, Mathis JM, et al: Percutaneous poly-
methylmethacrylate vertebroplasty in the treatment of osteopo-
rotic vertebral body compression fractures: technical aspects.
AJNR Am J Neuroradiol 18:1897–1904, 1997
312
J. S. Sarzier, A. J. Evans, and D. W. Cahill
9. Kinnard P, Ghibely A, Gordon D, et al: Roy-Camille plates
in unstable spinal conditions. A preliminary report. Spine 11:
131–135, 1986
10. Kostuik JP, Errico TJ, Gleason TF: Techniques of internal fix-
ation for degenerative conditions of the lumbar spine. Clin Or-
thop 203:219–231, 1986
11. Louis R: Fusion of the lumbar and sacral spine by internal fixa-
tion with screw plates. Clin Orthop 203:18–33, 1986
12. Moore D, Maitra R, Farjo L: Restoration of pedicle screw fixa-
tion with an in situ setting calcium phosphate cement. Spine
15:1696–1705, 1997
13. Okuyama K, Sato K, Abe E, et al: Stability of transpedicular
screwing for the osteoporotic spine. An in vitro study of the
mechanical stability. Spine 8:2240–2245, 1993
14. Pfeifer BA, Krag MH, Johnson C: Repair of failed transpedicu-
lar screw fixation: A biomechanical study comparing poly-
methylmethacrylate, milled bone, and matchstick bone recon-
struction. Spine 19:350–353, 1994
15. Roy-Camille R, Saillant G, Mazel C: Internal fixation of the
lumbar spine with pedicle screw plating. Clin Orthop 203:
7–17, 1986
16. Soshi S, Shiba S, Kondo H, et al: An experimental study on
transpedicular screw fixation in relation to osteoporosis of the
lumbar spine. Spine 16:1335–1341, 1991
17. Steffee AD, Biscup RS, Sitkowski DJ: Segmental spine plates
with pedicle screw fixation. A new internal fixation device for
disorders of the lumbar and thoracolumbar spine. Clin Orthop
203:45–53, 1986
18. Tohmeh AG, Mathis JM, Fenton DC, et al: Biomechanical effi-
cacy of unipedicular versus bipedicular vertebroplasty for the
management of osteoporotic compression fractures. Spine 24:
1772–1776, 1999
19. Weinstein JN, Rydevik BL, Rauschning W: Anatomic and tech-
nical considerations of pedicle screw fixation. Clin Orthop
284:34–46, 1992
20. Wilkes RA, Mackinnon JG, Thomas WG: Neurological deteri-
oration after cement injection into a vertebral body. J Bone
Joint Surg (Br) 76:155, 1994
21. Wittenberg RH, Lee K-S, Shea M, et al: Effect of screw diam-
eter, insertion technique, and bone cement augmentation of
pedicle screw fixation strength. Clin Orthop 296:278–287,
1993
22. Yamagata M, Kitahara H, Minami S, et al: Mechanical stability
of the pedicle screw fixation systems for the lumbar spine.
Spine 7:S51–S54, 1992
23. Yuan HA, Garfin SR, Dickman CA, et al: A historical cohort
study of pedicle screw fixation in thoracic, lumbar, and sacral
spinal fusions. Spine 19:S2279–S2296, 1994
24. Zdeblick TA, Kunz DN, Cooke ME, et al: Pedicle screw pull-
out strength. Correlation with insertional torque. Spine 18:
1673–1676, 1993
25. Zindrick MR, Wiltse LL, Widell EH, et al: A biomechanical
study of intrapeduncular screw fixation in the lumbosacral
spine. Clin Orthop 203:99–112, 1986
Manuscript received August 22, 2000.
Accepted in final form November 15, 2001.
Address reprint requests to: David W. Cahill, M.D., Department
of Neurosurgery, University of South Florida, 4 Columbia Drive
Suite 730, Tampa, Florida 33606. email: dcahill@com1.med.usf.edu.
J. Neurosurg: Spine / Volume 96 / April, 2002