Percutaneous Vertebroplasty1

Radiology

David F. Kallmes, MD2

Mary E. Jensen, MD

This review, aimed at current practitioners of vertebroplasty, highlights recent

Index terms: changes in patient work-up and procedural techniques that have streamlined the
Radiology and radiologists, How I authors’ clinical practice. Preprocedural work-up, including history, physical exam-
Do It ination, and adjunctive imaging techniques, are discussed. Technical details are
Spine, fractures, 30.4111, 30.4112
reviewed, including types of equipment, techniques of needle placement, and utility
Spine, interventional procedures,
30.1269 of venography. Postprocedural issues are noted, including risk of subsequent frac-
Spine, vertebroplasty, 30.1269 ture after vertebroplasty, long-term outcome of cement in the vertebral body, and
utility of prophylactic vertebroplasty. Finally, the current state of evidence in support
Published online of the efficacy of vertebroplasty are discussed, with particular attention to the need
10.1148/radiol.2291020222
Radiology 2003; 229:27–36
for ongoing clinical trials.
© RSNA, 2003
Abbreviation:
PMMA ⫽ polymethylmethacrylate

1
From the Department of Radiology, The first percutaneous vertebroplasty of which we are aware was performed in Europe in
University of Virginia Health Sciences
1984 and reported in the literature in 1987 (1), and the first vertebroplasty in North
Center, PO Box 800170, Charlottes-
ville, VA 22908. Received March 26, America was performed in 1993 and reported in 1997 (2). There are currently over 253
2002; revision requested June 5; revi- published reports focused on vertebroplasty. Most of the literature regarding percutaneous
sion received September 26; accepted vertebroplasty is based on results in early technical reports (2,3) and case series (4), which
November 6. Address correspon- include methods for patient selection, procedural details, and postprocedural care. Since
dence to M.E.J. (e-mail: mej4u@virginia
.edu). the time these previous studies were published, numerous modifications in patient eval-
M.E.J. is a paid consultant (co-chair of
uation and procedural technique have been made to better define the appropriate patient
the Scientific Advisory Board) for Par- population, to decrease surgery time, and to optimize overall patient care. Some of these
allax Medical, which makes vertebro- modifications have been published in the literature, but many have not.
plasty products, and a shareholder in Our target audience for this review is current practitioners of vertebroplasty who already
the company that owns Parallax Med-
have some knowledge of the basic indications, techniques, and vertebroplasty literature.
ical.
Our goal is to identify specific areas where the approach to vertebroplasty has changed
Current address:
2
Department of Radiology, Mayo substantially over the past several years, with an emphasis on technical changes and
Clinic, Rochester, Minn. clinical research. In addition, the current state of evidence on the efficacy of vertebroplasty
© RSNA, 2003 will be discussed, with specific focus on the need for future clinical trials.
While this review, by necessity, contains our biases and opinions on the ideal way to run
a vertebroplasty service, we do not want to imply that our techniques are the only valid
ones. However, as it will become clear in the following review, we have made substantial
efforts recently to justify our methods by careful study of our clinical database, which
comprises approximately 530 treatment levels in 320 unique patients, and to report
important lessons from these studies.

PATIENT EVALUATION AND PREPROCEDURAL WORK-UP

A listing of appropriate clinical features for patients being considered for vertebroplasty
can be found in the American College of Radiology Standards 2000 –2001 (5). In our practice,
by and large, we adhere to clinical indications in the American College of Radiology
standards. However, our approach to patient evaluation has changed substantially since
1993. During the early development period of percutaneous vertebroplasty, the typical
patient presented with subacute or chronic back pain that was unresponsive to medical
therapy, with a new fracture documented on a conventional radiograph (2). Selection
criteria included focal discomfort at palpation over the spinous process of the involved
vertebra and absence of radicular symptoms or neurologic deficits. Computed tomography
(CT) or magnetic resonance (MR) imaging was often performed to evaluate for nerve root
compression or retropulsed fragments. While most of these features are still considered
relevant in the patient work-up, substantial changes in our approach to patient selection
have been made on the basis of our clinical experience. These revisions involve imaging
evaluation, physical examination, and determination of the duration of pain prior to the
performance of vertebroplasty.

27
Radiology

Figure 1. Images in an elderly woman with low back pain. (a) Lateral radiograph shows multiple lumbar compression fractures of indeterminate
age. Clinical examination demonstrated nonfocal tenderness over the lower back. (b) Posteroanterior and oblique bone scan images show marked
uptake at L1 (arrows). After treatment of this single vertebra, the patient’s pain was relieved.

Adjunctive Imaging for multiple fractures of uncertain age (Fig 1). fragments. Canal compromise from ret-
Identification of Symptomatic In some cases, especially with multiple ropulsed bone is not considered an abso-
Fractures severe compression fractures, exact label- lute contraindication, provided there is
ing of vertebrae on bone scans is difficult, no cord or nerve root compression or
Although much has been written about
although use of a radioactive and ra- neurologic symptoms. To our knowl-
imaging procedures required prior to ver-
diopaque marker helps make identifica- edge, there are no written reports of im-
tebroplasty, the role of such imaging re-
tion possible. mediate or delayed movement of the
mains largely speculative (empiric) at this
To our knowledge, there currently ex- fracture fragment; however, if substantial
point. Patients referred to us have often
ist no data regarding the use of MR im- retropulsion is present, we will proceed
undergone a wide array of radiologic stud-
aging in the evaluation of patients for with vertebroplasty but are careful to
ies, including conventional radiography,
consideration of vertebroplasty, although keep the acrylic within the ventral aspect
bone scintigraphy, CT, and/or MR imag-
some investigators (7) have suggested that of the vertebral body and away from the
ing. We have tried to determine which
edema seen on MR images is predictive of fracture lines.
studies are the most helpful in identifying
a favorable response to vertebroplasty. CT provides information about frac-
who is most likely to respond to treatment.
Whereas MR imaging is sensitive for the ture involvement of the pedicles and pos-
Patients with a documented new or sub-
detection of acute compression fractures, terior elements, which may help deter-
acute fracture on a conventional radio-
we have noted a number of cases where mine the appropriate needle path. CT
graph and who meet the clinical criteria
MR imaging has demonstrated normal allows measurement of the pedicular di-
regarding pain pattern usually proceed to
(fatty) marrow signal intensity on T1- and ameter, which may influence the size of
vertebroplasty without undergoing other
T2-weighted images, while the bone scan the needle chosen for puncture, particu-
imaging. Adjunctive imaging is indicated
was abnormal. After treatment of the ver- larly in the more gracile thoracic vertebral
in patients with single or multiple fractures
tebrae that demonstrated increased activ- pedicles. However, we recently reviewed
of uncertain age or when serial conven-
ity, a good clinical response was noted in our series of thoracic vertebroplasties and
tional radiographs are unavailable. Results
these patients (Fig 2). However, we have found that the size of the needle used did
from physical examination alone may be
not evaluated the use of imaging tech- not result in a difference in complication
misleading in this setting. Either bone scan
niques such as short-tau inversion-recov- rate (8). Forty-seven vertebral bodies were
imaging or MR imaging is potentially use-
ery or contrast material– enhanced fat sat- treated in 34 patients. Eleven-gauge nee-
ful. There is only one published report of
urated MR imaging. What remains unclear
which we are aware regarding the use of dles were used in 40 (85%) of 47 treat-
is the appropriateness of these imaging
scintigraphy in preprocedural evaluation ments, while 13-gauge needles were used
modalities with regard to age of fracture.
of patients being considered for vertebro- in seven (15%) treatments. Postvertebro-
plasty (6). In that small retrospective se- plasty fracture involving the pedicle used
Role of CT in Vertebroplasty
ries, a high percentage of patients (94%) for needle access was noted in one (2%) of
achieved nearly complete pain relief after The primary indication for CT prior to 47 treatments; this pedicle had been tra-
vertebroplasty of those vertebral levels that vertebroplasty is to evaluate the integrity versed by using a 13-gauge needle. We
showed increased uptake of tracer, even in of the posterior wall of the vertebral body favor the larger-gauge needle over the
a challenging patient population with and to assess posterior displacement of smaller needle because, in our experi-

28 䡠 Radiology 䡠 October 2003 Kallmes and Jensen

Radiology

Figure 2. Images in a 50-year-old woman being treated with high-dose

steroids who presented with worsening back pain. (a) Lateral radiograph
shows multiple thoracic and lumbar compression fractures of indetermi-
nate age. Sagittal (b) T1-weighted (750/12) and (c) turbo spin-echo T2-
weighted (4,500/112) MR images show no evidence of edema to indicate
a new fracture; however, (d) posteroanterior bone scan demonstrates
intense radionuclide uptake at L3 (arrow). (e) Fluoroscopic spot image.
After injection of polymethylmethacrylate (PMMA) (arrows), the patient
noted marked relief of her pain.

tients who were hospitalized for pain

control requiring parenteral narcotics
were excluded from this requirement and
were treated acutely. This conservative
approach was instituted because of con-
cern about the risk-benefit ratio of verte-
broplasty; even though complications
are rare, vertebroplasty results in a per-
manent medical implant. The natural
history of osteoporotic compression frac-
ence, it is easier to direct precisely during amounts of extraosseous methacrylate ture is, in a substantial percentage of pa-
placement. might be performed in asymptomatic tients, spontaneous resolution of pain
Postvertebroplasty CT has been recom- patients. within 4 – 6 weeks (11,12).
mended by some authors for postproce- In recent years we have noted an in-
dural documentation, although there is creasing number of patients to whom
Duration of Pain Prior vertebroplasty is offered early after frac-
no evidence in the literature to suggest
to Vertebroplasty ture. Typically, these patients are referred
that such a policy affects clinical practice.
We reserve CT for those patients who From its inception, vertebroplasty has from physicians who have clinical expe-
remain symptomatic after vertebroplasty, been reserved for treatment of patients rience with vertebroplasty, have been ex-
especially in cases of possible nerve root who have failed a course of conservative tremely pleased with the outcomes, and
irritation from methacrylate. CT is very medical treatment (2–5,9,10). Although would like to avoid the use of potent
sensitive to the presence of small amounts no defined waiting period was rigorously analgesics or immobilization in their el-
of methacrylate, however, and unnec- observed, most patients were treated derly patients. Further, we frequently are
essary interventions because of small 6 –12 weeks after the onset of pain. Pa- asked to perform early vertebroplasty by

Volume 229 䡠 Number 1 Percutaneous Vertebroplasty 䡠 29

 patients who have previously been
treated successfully and have sustained a
subsequent fracture. In most cases, we
will proceed with early vertebroplasty in
these patients. We recently have ana-
lyzed our patient outcomes as a function
Radiology

of fracture age (13). Even though subjec-

tive pain relief was reported as excellent
regardless of fracture age, patients with
more chronic fractures failed to improve
with regard to use of analgesics. This lack
of decrease in medication requirement
noted in patients with more chronic frac-
tures may have resulted from chemical
dependency, which suggests some disad-
vantage to delaying vertebroplasty. We
no longer require a failure of medical
therapy prior to our offer to perform ver-
tebroplasty; however, adopting such a
course may result in nonpayment by
Medicare. In addition, adoption of early
vertebroplasty might result in substantial
increases in the number of such proce- Figure 3. Lateral radiographs in an elderly woman with acute low back pain. (a) Severe anterior
dures performed, with resultant increases wedge deformity of L2 during weight bearing is demonstrated, but restoration of height and pain
in societal costs for treatment of painful relief occurred in the recumbent position. T2-weighted MR image (not shown) depicted a large
intravertebral cavity, consistent with Kummell osteonecrosis. (b) After vertebroplasty, height
compression fractures. restoration and vertebral stability are shown, with resolution of symptoms.

Physical Examination and

Vertebroplasty
that alleviation of the bone pain unmasks gle-plane unit, biplane monitoring of flu-
In the past, focal pain elicited by pal- pain associated with facet disease, since pa- oroscopic images decreases procedural
pation over the spinous process of the tients usually gain relief after facet injec- time and enables orthogonal visualiza-
fractured vertebra has been used as a pa- tion. tion of the injection. The availability of
tient-selection criterion. Patients have digital subtraction angiography allows
been excluded from treatment when TECHNICAL CONSIDERATIONS documentation of needle placement and
their point tenderness has been located evaluation of the trabecular space and
remote to the affected vertebra. The pres- epidural veins. In cases of osteolytic me-
Vertebroplasty methods described in the
ence of radicular pain involving the tastases or treatment of cervical or upper
literature have evolved on the basis of the
lower extremities or low back pain that thoracic vertebrae, needle placement
predominant European (3,4,15) and
radiates to the hip may disqualify a pa- may be facilitated by using CT guidance
American (2,10,16) experiences. Techni-
tient or lead to a different intervention, (15,16) or CT fluoroscopy. Regardless of
cal differences are mostly minor and re-
such as facet injections. the modality used for needle placement,
lated to the availability of products and
In our experience, however, a physical the injection of PMMA should always be
equipment used, as well as the operators’
examination is not always entirely sensi- performed with direct fluoroscopic con-
training and personal style. No one
tive or specific for determination of pa- trol. We have attempted injection by us-
method is “right” or “wrong,” provided a
tients who will have a good outcome af- ing CT fluoroscopy but did not feel con-
good embolization technique is used and
ter vertebroplasty. We have evaluated fident that the PMMA distribution was
certain guidelines are respected. It is
our clinical data in retrospective fashion, adequately visualized. Cement may flow
highly recommended that the reader re-
identifying 10 patients where no local pain in a cephalic and/or caudal orientation,
view the American College of Radiology
was present over the fracture site. We com- which would be difficult to identify with
“Standard for the Performance of Verte-
pared this group with 90 patients who real-time transverse CT. Some operators
broplasty” (5), because essential informa-
demonstrated focal tenderness (37). We have performed serial injections of small
tion is contained in this important doc-
failed to detect a significant difference in aliquots of acrylic by using intermittent
ument.
outcome between these two groups. In ad- CT scanning, with relative success (16).
dition, we frequently noted nonlocalizing
Radiographic Visualization
pain patterns in patients with Kummell os-
Needles
teonecrosis, in which radicular pain, hip Complications are more likely to occur
pain, or pain several levels from the frac- when visualization of needle placement Needle selection is operator depen-
ture is present (Fig 3). We surmise that the or cement injection is poor. Therefore, dent. To our knowledge, there are no
relief of radicular pain is due to stabiliza- operators should use the highest quality studies on comparison of performance
tion of an unstable fracture. We have also fluoroscopy available to them and avoid among needle types that might guide se-
noticed that after successful vertebroplasty, poor-quality imaging systems such as lection. Multiple needles are available
patients often develop paravertebral pain older bedside units. Although vertebro- that are excellent for vertebroplasty. Im-
that radiates to the hip (14). We suspect plasty can be performed by using a sin- portant attributes include the shape of

30 䡠 Radiology 䡠 October 2003 Kallmes and Jensen

 time allows the powder to dissolve into
the liquid, preventing their separation
during injection. Such separation may
lead to the formation of a powder plug
within the needle.
Radiology

Opacification
Perhaps the most critical attribute that
facilitates safe vertebroplasty is excellent
opacification of cement. Authors of early
reports (1– 4) suggested use of either bar-
ium sulfate and powdered tungsten or
tantalum. We have observed that ideal
Figure 4. Needles suitable for vertebroplasty are supplied with a visualization of cement is achieved by
variety of stylets: A, single bevel; B, multibevel point; C, diamond using relatively large particles of barium,
point; D, threaded stylet. (Image courtesy of Parallax Medical.)
on the order of 1 mm in diameter, which
can be tracked easily during slow injec-
tion of cement. Smaller particles or finely
sifted opacifiers provide a gray back-
the tip of the stylet (Fig 4) and the can- an “off-label” manner. In most instances, ground to the cement, but this gray back-
nula, as well as the type of handle. We approved devices can be used in routine ground is difficult to discern against the
prefer to use a system that includes two clinical practice without the need for an overlying tissues. As such, we have aban-
types of stylets. The first stylet has a sharp investigational device exemption (IDE) doned the use of tungsten powder. Trac-
multibeveled or “diamond-shaped” tip from the FDA. If in doubt, however, it is ers (Parallax Medical) is composed of var-
and facilitates entry into the pedicle. In suggested that practitioners check with ious sizes of barium sulfate particles and
our experience, single-beveled stylets local institutional review boards prior to has been approved by the U.S. Food and
tend to slide off the pedicle. Once we starting a vertebroplasty practice. In ad- Drug Administration for cement opacifi-
have traversed the pedicle, we typically dition, it is suggested that practitioners cation. Another barium product is of-
remove the multibeveled stylet and place discuss the regulatory status of PMMA fered by Bryan (Woburn, Mass). Barium is
a single-beveled stylet. Although there with patients as part of the consent pro- already present in the Stryker-Howmedica
are no data to support this, we believe cess. Finally, if vertebroplasty is to be per- PMMA product, but it is not a sufficient
that the single bevel allows one to steer formed as part of a clinical trial, then an amount for opacification, and use of addi-
the needle tip slightly (Fig 5). IDE is required. tional barium is recommended.
Also available is a coaxial system with a There are at least four PMMA products
curved nitinol cannula (Cook, Blooming- currently available, including Secour Antibiotics
ton, Ind) for facilitating cross-midline ac- (Parallax Medical), Codman Cranioplas-
cess or specific placement (Fig 6). With tic (Johnson and Johnson, Bracknell, We routinely add tobramycin (Nebcin;
this device, care should be used to avoid England), Osteobond (Zimmer, Warsaw, Eli Lilly, Indianapolis, Ind) to the cement
puncture of the lateral wall of the verte- Ind) and Surgical Simplex P (Stryker- mixture, on the basis of information in
bral body. While most needle cannulas Howmedica, Limerick, Ireland). Impor- the surgical literature supporting this
have a square distal shape, one available tant differences are seen among these practice (17). Other practitioners advo-
cannula has a beveled distal end (Cook) products with regard to polymerization cate use of intravenously administered
that may allow one to direct cement in a time. The Stryker-Howmedica and Zim- antibiotics (7), but we reserve use of these
given direction. Multiple handle shapes for patients who are substantially immu-
mer products have relatively rapid poly-
are available, including standard grip de- nocompromised. We have encountered
merization, wherein the cement becomes
signs (Cook; Manan Medical Products, one case of iatrogenic infection, with
too viscous to inject within 5–7 minutes.
Wheeling, Ill), whereas other manufac- Staphylococcus epidermidis, among 250
This polymerization time can be pro-
turers offer novel designs such as an awl consecutive patients treated with verte-
longed by refrigerating the powdered
handle (Parallax Medical, Scotts Valley, broplasty (unpublished data, 2002). This
polymer pack and liquid monomer vial
Calif). Last, we use 10-cm-long needles in single patient was taking multiple immu-
prior to use or by placing syringes filled
most patients but favor 15-cm-long nee- nosuppressive medications and thus was
with the acrylic in an ice bath. The rapid
dles when treating lower lumbar verte- at high risk. In similar situations in the
polymerization of the cement may limit
brae in larger patients. future, we will use intravenous antibiot-
the ability to inject multiple levels with a
ics in addition to antibiotics placed into
single mix. If using the Codman product,
Vertebroplasty with PMMA the cement.
we recommend the slow-polymerization
Currently, the only biomaterial used type, which allows 17–20 minutes of
working time (unpublished data, 1999).
Injection Devices
for vertebroplasty in the United States is,
to our knowledge, PMMA. PMMA is ap- Because it takes time for the powdered Although various methods have been
proved for surgical implantation in mul- PMMA component to dissolve in the liq- proposed for cement injection, the ma-
tiple bone locations. However, there is no uid monomer, the manufacturer of Sec- jority of our experience has been gained
commercially available cement approved our recommends addition of a “solvation by using 1-mL syringes. The 1-mL sy-
for percutaneous vertebroplasty. Use of time” of approximately 2–3 minutes after ringes are inexpensive, require minimal
PMMA in vertebroplasty is performed in mixing and before injection. This added storage space, and allow exquisite tactile

Volume 229 䡠 Number 1 Percutaneous Vertebroplasty 䡠 31

Radiology

Figure 5. (a) Lateral radiograph shows initial trajectory (arrow) of the needle, which places the tip anteriorly at the midportion of the vertebral
body. However, use of a beveled stylet, with bevel face pointing upward, deflects the tip downward. (b) Lateral radiograph shows that final position
of the cannula approximates the anterior inferior corner of the vertebral body (arrow).

feedback during injection, which we con- cently performed a retrospective review

sider to be important to prevent large in which we compared vertebroplasty
amounts of cement extravasation. There performed with and that performed
are several commercially available injec- without venography (18). We demon-
tion devices (from Parallax Medical, strated no significant differences in fre-
Cook, and Stryker-Howmedica) for the quency or amount of venous extravasa-
delivery of cement. These injection de- tion and no difference in clinical
vices increase the distance between the outcome between patients in whom
operator and the x-ray tube; facilitate an- venography was performed versus that in
teroposterior fluoroscopy during injec- patients in whom no venography was
tion, because the operator’s hands are performed. Furthermore, in cases of a
out of the field; and allow a single con- preexisting cavity or endplate fracture,
nection of the injector to the needle. Use contrast medium injected during venog-
of 1-mL syringes rather than an injection raphy may be impossible to wash out
device is largely determined by operator prior to cement injection, rendering it
preference. difficult to visualize the barium-opacified
cement.
VERTEBROPLASTY TECHNIQUE Although we no longer consider it nec-
essary to perform venography prior to Figure 6. Lateral radiograph shows curved
Vertebral Venography cement infusion, some operators may stylet, which allows the operator to deposit
PMMA precisely at superior (arrowheads) and
Some practitioners (2,18) of percutane- find the venogram to be comforting, as it inferior (arrows) endplates. The referring phy-
ous vertebroplasty described the use of defines the exact point where the basiver- sician specifically requested that reinforce-
intraosseous venography prior to cement tebral plexus exits the vertebral body and ment of these locations be performed prior to a
infusion, to map the venous outlets from outlines the paraspinal venous system. surgical procedure.
the vertebral body (Fig 7). On the basis of Previously, authors (19) have reported
the venographic findings, the operator the use of venography to help detect di-
would gain confidence in his or her ability rect venous communications and predict opacified cement from the initial injec-
to prevent extraosseous cement extravasa- PMMA flow characteristics and potential tion obscures visualization.
tion, since the outlets would be known sites of egress. To speed procedure time and eliminate
already. Alternatively, the needle could be the need for separate injections, many
repositioned if injection showed a large di- practitioners have adopted a unipedicu-
Needle Placement
rect venous connection. Some practitio- lar technique for vertebroplasty. This
ners even suggested protective venous em- Unipedicular versus bipedicular vertebro- technique involves placement of the nee-
bolization with gelatin foam sponges or plasty.—Authors of early reports (1– 4) of dle tip in the midline of the ventral as-
other embolic agents prior to cement injec- vertebroplasty described bipedicular ver- pect of the vertebral body by using a
tion. tebroplasty with separate cement infu- transpedicular approach, with the expec-
Although we routinely used venogra- sions into both hemivertebra with the tation that the central portion of the ver-
phy for several years during the develop- use of two needles. Bipedicular injections tebra can be filled (20). The technique is
ment of vertebroplasty (2), we have aban- were performed in an effort to maximize slightly different when comparing lum-
doned its use over the past 2 years. volume of cement placed into the verte- bar to thoracic vertebra. In the lumbar
Because of our extensive experience with bra. However, two needle placements region, the appropriate oblique approach
the technique, we gained a reliable un- and two injections result in relatively can be found by angling the anteropos-
derstanding of the most likely routes for long procedures. Further, monitoring of terior tube laterally until the “scotty-
venous extravasation, including epidural the second injection is often problem- dog” profile is seen over the pedicle, with
and paravertebral routes. We have re- atic, given that the indwelling barium- approximately 20° of lateral angulation.

32 䡠 Radiology 䡠 October 2003 Kallmes and Jensen

 vertebrae at a single session. We know of
no published study in which the safety of
multilevel vertebroplasty has been investi-
gated. We have studied the relative efficacy
of single- versus multilevel vertebroplasty
(unpublished data, 2002). Three groups
Radiology

were studied, including patients treated at

a single level at one session, patients
treated at multiple levels at a single session,
and patients treated at multiple levels at
multiple sessions. When performing mul-
tilevel vertebroplasty, we routinely place
multiple needles at once before preparing
the cement. We then inject multiple
pedicles sequentially, using the same batch
of cement. In this retrospective review,
we demonstrated equivalent pain relief
Figure 7. Prone intraosseous venograms obtained during (a) anteroposterior and (b) lateral among the three groups. To our surprise,
injections of contrast material prior to vertebroplasty shows rapid filling of intraosseous venous however, we noted that mobility was sig-
complex followed by egress into paravertebral veins (PVV) and basivertebral plexus (BVP). Filling nificantly more impaired for the single-
of inferior vena cava (IVC) demonstrates how cement migration into the venous system may level group than for the other groups. The
result in pulmonary embolization.
reason for this difference between groups is
unknown.

The pedicle is punctured in its midpor- ence when performing “high-volume”

tion, with the needle tip placed in the vertebroplasty, defined as injection into a OUTCOMES
upper one-third of the pedicle, just me- vertebral body of more than 3 mL, versus
Risk of Subsequent Fracture
dial to the lateral pedicular border. Sub- “low-volume” vertebroplasty, defined as
stantial angulation is more difficult in injection of less than 3 mL (unpublished The effect of cement deposition into
thoracic vertebrae, because the pedicles data, 2002). Although we have not yet an osteoporotic fracture on the risk for
are oriented in the straight anteroposte- correlated such volumes to vertebral level subsequent fractures at other levels re-
rior direction. The pedicular outline may or percentage of collapse, we did not de- mains unclear. There is theoretical con-
become difficult to visualize with mini- tect clinically important differences in cern that diminishment of the compli-
mal angulation. In our experience, the outcome between the low- and high-vol- ance of one vertebra by means of cement
proper obliquity is achieved by angling ume groups. These data may offer com- injection may place the remainder of the
the anteroposterior tube laterally until fort to inexperienced practitioners who axial skeleton at greater risk for collapse.
the pedicle projects over the medial one- want to minimize risk for extraosseous Previously, authors (28) have noted a
fifth of the vertebral body (19). cement extravasation by “underfilling” small increased risk of new-onset frac-
We have performed a retrospective re- the vertebral body. tures in the vicinity of the treated level
view of cases performed with a unipe- Potential cardiovascular changes with but did not identify specific levels in re-
dicular versus a bipedicular technique PMMA administration.—Authors of re- lation to treated vertebrae. We have ret-
(20). We were unable to demonstrate any ports in the orthopedic literature (23–25) rospectively reviewed our data to identify
difference in clinical outcome between have shown cardiovascular compromise 58 patients who returned with new frac-
the two groups, even though there was a due to instillation of large amounts of tures following vertebroplasty. One-half
small statistically significant difference in PMMA during hip arthroplasty. We have of these new fractures were adjacent to
percentage of vertebral body filling when studied a large cohort, including 142 ver- the initial vertebroplasty level (Fig 9). We
comparing unipedicular and bipedicular tebroplasties in 78 patients, where de- reviewed a separate series of patients pre-
procedure results. On the basis of these tailed cardiovascular data were docu- senting with multiple osteoporotic frac-
considerations, we use the unipedicular mented before, during, and after PMMA tures prior to vertebroplasty and found
approach in essentially all cases. injection (26). We noted no change in that 68% of fractures were at contiguous
Volume of cement.—The amount of ce- mean arterial blood pressure or heart rate levels (Fig 8) (29). These latter data sug-
ment required for good clinical outcome at any point in time. There was a statis- gest a strong trend toward “clustering” of
has never been systematically studied, to tically significant decrease in percentage fractures as part of the natural history of
our knowledge. Testing of cadaveric of oxygen saturation 10 minutes after osteoporosis. As such, a finding that 50%
spines suggests that up to 8 mL of cement PMMA injection; however, this differ- of new postvertebroplasty fractures occur
is required to achieve biomechanical in- ence was very small, with mean prepro- at adjacent levels may simply represent
tegrity (21). However, the risk of ex- cedure saturation of 98.0% and mean the normal distribution of new-onset
traosseous extravasation of cement in- postprocedural saturation of 97.4%, and fractures. We conclude that, at this time,
creases with increasing volumes of was considered clinically irrelevant. there remains no compelling evidence to
cement injected (22). To minimize the Multilevel vertebroplasty.—Currently, many suggest that vertebroplasty results in
risk for such extravasation, we tend to practitioners routinely treat multiple frac- higher risk of subsequent fracture as com-
place relatively small amounts of cement tures at a single session. Not uncommonly, pared with the risk in untreated patients
into a given vertebral body. we will treat two to three levels at one and that proof of such a claim will re-
We have recently reviewed our experi- session (Fig 8) and have treated up to five quire large-scale, prospective studies.

Volume 229 䡠 Number 1 Percutaneous Vertebroplasty 䡠 33

Radiology

Figure 8. Multilevel vertebroplasty in a 60-year-old woman with steroid-dependent chronic obstructive pulmonary disease,
who presented with severe midthoracic back pain unresponsive to narcotic analgesia. (a) Sagittal T1-weighted (750/12) MR
image shows three adjacent thoracic compression fractures at T7, T8, and T9. (b) Lateral radiograph shows vertebroplasty of
thoracic vertebrae. Patient’s pain resolved after vertebroplasty of all three levels. She returned 1 month later with new lumbar
pain, and a new L4 fracture was treated. (c) Patient returned 1 month later with new fractures of L1 and L2, identified on
sagittal T1-weighted (750/12) MR image, which were injected. Ultimately, she went on to fracture three more vertebral bodies.

Long-term Follow-up of Cement gle case in which there was apparent com- say et al (31) reported on a large series of
pression of the injected cement. The patients who were followed up after an
PMMA represents a permanent medical
cement morphology was unchanged in the index fracture; within 1 year after the
implant. Although there is a long history
remainder of the levels. initial fracture, approximately 25% of pa-
of surgical implantation of PMMA, scant
These data indicate that in the major- tients experienced a new fracture. Exact
literature exists regarding the long-term be-
ity of patients treated with percutaneous locations of new fractures were not re-
havior of cement as used in vertebroplasty.
vertebroplasty, there is stability of degree ported. We and others have noted that
We consider it of paramount importance
of vertebral compression, kyphosis, and approximately 20%–25% of patients re-
to understand the long-term sequelae of
cement morphology. In a minority of pa- turn with new, painful fractures after be-
cement deposition, in order to appropri-
tients, there may be progressive kyphosis ing treated with vertebroplasty (29). As
ately counsel patients and referring physi-
and compression at the treated level, noted earlier, approximately one-half of
cians prior to vertebroplasty. Grados et al
which indicates the need for prospective these new fractures are at sites not adja-
(28) reported radiographic follow-up in 34
study with complete clinical and radio- cent to the previously treated vertebra.
vertebrae treated with vertebroplasty in 25
graphic follow-up evaluation after verte- Thus, since only a minority of patients
patients at a mean follow-up of 48 months.
broplasty. return with new painful fractures, and,
They noted no progression of vertebral de-
since it is impossible to predict which
formity in any of the injected vertebrae.
levels will undergo subsequent fracture,
We performed a similar study in which we
Prophylactic Vertebroplasty we do not consider prophylactic verte-
identified 20 levels treated in 10 patients
broplasty to be justifiable. Furthermore,
who had undergone conventional radio- The notable pain relief achieved with
Medicare will not reimburse for prophy-
graphic follow-up at a mean of 1.3 years vertebroplasty raises the question of
lactic vertebroplasty at this time.
(30). Sixteen (80%) of 20 vertebral body whether we should routinely perform
compression fractures were stable with re- prophylactic vertebroplasty at nonfrac-
spect to the degree of kyphosis and com- tured levels in patients who present with
Outcomes Measures
pression. Two (10%) levels showed moder- painful fractures. Prophylactic vertebro-
ately increased central endplate collapse plasty might be justified if one could ac- The majority of studies on vertebro-
without a change in the degree of kypho- curately identify patients at extremely plasty have relied on rudimentary out-
sis. Two (10%) levels showed progressive high risk for new-onset fractures and comes measurements, typically includ-
collapse associated with increased kypho- could predict which nonfractured levels ing pain relief, change in mobility, and
sis. One of these levels represented the sin- will undergo spontaneous fracture. Lind- change in medication requirements. For

34 䡠 Radiology 䡠 October 2003 Kallmes and Jensen

 taneous vertebroplasty (unpublished data,
2002).
Institutional review board approval
was obtained for a prospective, random-
ized, blinded trial to compare percutane-
ous vertebroplasty and sham vertebro-
Radiology

plasty. Patients were randomly assigned

to either the vertebroplasty or the sham
condition. The sham condition included
fluoroscopically guided placement of a
25-gauge needle and infiltration of the
pedicle with 10 mL of 0.25% bupivicaine
(Abbott Laboratories, North Chicago, Ill),
without placement of either the vertebro-
plasty needle or cement. Methacrylate
monomer (Secour; Parallax Medical) was
opened in the angiography suite to sim-
ulate cement preparation. To simulate ce-
ment deposition, localized pressure was
placed on the patient’s back and opera-
tors gave verbal clues typical of those
given during cement injection. After 14
days, patients who failed to respond to
the initial treatment (vertebroplasty or
sham) were crossed over to the other pro-
cedure but remained blinded to treat-
Figure 9. Lateral radiographs in 35-year-old, obese, steroid-depen-
ment type. Subjects were asked to guess
dent woman with asthma who presented with severe lumbar pain.
(a) Acute compression fractures of the superior endplate of L1 (upper which procedure they underwent ini-
arrows) and inferior endplate of L4 (lower arrows) are shown. After tially, to determine whether blinding was
vertebroplasty, the patient returned to full activity, with relief of her attained.
pain; 2 weeks later, however, she was admitted for recurrent back Nine patients were screened. Five pa-
pain. (b) Repeat radiograph shows new superior endplate fractures of tients agreed to enroll in the trial. All
L2 and L3 (arrowheads).
patients had documented fractures; a
bone scan was used in cases where the
age of the fracture was unknown. Three
most patients, accurate recollection of questionnaire was readily accepted by both patients were initially randomly assigned
pain severity is difficult. Ideal outcome patients and interviewers, but also that sig- to the sham procedure. One of these ex-
measures would be sensitive and specific nificant short-term improvements in func- perienced a new fracture 48 hours after
for subtle changes in health status. Mea- tion were achieved after vertebroplasty the sham procedure and was excluded
sures of global health status, such as the (unpublished data, 2002). Future investiga- from the trial. Two patients in the sham
SF-36 Health Survey, are difficult to ad- tors should examine whether any of these group gained minimal pain relief from
minister in patients with severe back instruments can be administered reliably the sham procedure and thus crossed
pain. There exist no vertebroplasty data in patients treated with vertebroplasty. over to vertebroplasty. Pain relief after
obtained with the SF-36, but authors of a vertebroplasty in these two patients was
recent article on kyphoplasty (32) re- minimal. Two other patients were ini-
ported use of the SF-36 and found signif- Sham Trial tially assigned to undergo vertebroplasty.
icant changes in physical functioning Pain relief after vertebroplasty was mini-
and pain domains. Several respected authorities have noted mal and both of these patients crossed
Even without use of relatively cumber- the lack of controlled prospective trials for over to the sham procedure. One of these
some questionnaires such as the SF-36, vertebroplasty (36). In our local environ- two patients gained complete pain relief
the vertebroplasty literature could be en- ment, one third-party payer has recently following the sham procedure. All five
hanced by using validated disease-spe- concluded that insufficient evidence is patients guessed that they had under-
cific outcomes instruments. Other spe- available to support vertebroplasty, and gone the sham condition as the initial
cialties use a variety of back pain–specific payment has been denied. The apparent treatment.
disability instruments, including the Os- clinical benefits of vertebroplasty may re- We conclude that enrollment of pa-
twestry Back Pain Index (33), the Roland flect either the natural history of painful tients in a sham-controlled trial of percu-
Scale (34), and the Osteoporosis Quality compression fractures, where spontaneous taneous vertebroplasty is feasible. We
of Life Questionnaire (35). We have re- resolution is common, or the placebo ef- have some concern that selection bias,
cently gained experience with the Roland fect. We and others (34) have proposed a where patients with lesser degrees of pain
Scale, which is a validated low back pain– sham-controlled trial to demonstrate that would be more likely to enroll than pa-
related measurement tool that is easy to the benefits of vertebroplasty are real. Re- tients with severe pain, may dilute the
administer in person and on the tele- cently, we have carried out a pilot study to apparent effect of vertebroplasty. We
phone (34). We noted, in a series of 16 demonstrate the feasibility of enrolling pa- conclude that patients are unable to ac-
consecutive patients, not only that the tients into a sham-controlled trial of percu- curately discern whether sham or true

Volume 229 䡠 Number 1 Percutaneous Vertebroplasty 䡠 35

 vertebroplasty is being performed. Com- 8. Kallmes DF, Schweickert PA, Marx WF, et et al. Blood clearance and acute pulmo-
plete pain relief can be achieved with the al. Vertebroplasty in the mid- and upper nary toxicity of methylmethacrylate in
thoracic spine. AJNR Am J Neuroradiol dogs after simulated arthroplasty and in-
sham procedure, even after failure of ver- 2002; 23:1568 –1576. travenous injection. J Bone Joint Surg Am
tebroplasty. The placebo effect may play 9. Cortet B, Cotten A, Boutry N, Flipo RM, 1973; 55:1621–1628.
an important role in determining the Duquesnoy B, Chastanet P, Delcambre B. 25. Phillips HF, Cole PV, Lettin AW. Cardio-
Radiology

outcome of vertebroplasty. Percutaneous vertebroplasty in the treat- vascular effects of implanted acrylic bone
ment of osteoporotic vertebral compres- cement. BMJ 1971; 3:460 – 461.
sion fractures: an open prospective study. 26. Kaufmann TJ, Jensen ME, Ford G, Gill LL,
CONCLUSION J Rheumatol 1999; 26:2222–2228. Marx WF, Kallmes DF. Cardiovascular ef-
10. Jensen ME, Dion JE. Percutaneous verte- fects of polymethylmethacrylate use in
broplasty in the treatment of osteopo- percutaneous vertebroplasty. AJNR Am J
Percutaneous vertebroplasty has been rotic compression fractures. Neuroimag- Neuroradiol 2002; 23:601– 604.
embraced by the North American radiol- ing Clin N Am 2000; 10:547–568. 27. Kaufmann TJ, Jensen ME, Ford G, Gill LL,
ogy community within the past decade. 11. Silverman SL. The clinical consequences Marx WF, Kallmes DF. Cardiovascular ef-
Although the basic principles behind ver- of vertebral compression fracture. Bone fects of polymethylmethacrylate use in
1992; 13(suppl 2):S27–S31. percutaneous vertebroplasty. AJNR Am J
tebroplasty remain unchanged, the tech- 12. Patel UF, Skingle SF, Campbell GAF, Crisp Neuroradiol 2002; 23:601– 604.
nical aspects have been dramatically af- AJF, Boyle IT. Clinical profile of acute ver- 28. Grados F, Depriester C, Cayrolle G, Hardy
fected by operator experience, product tebral compression fractures in osteopo- N, Deramond H, Fardellone P. Long-term
development, and critical evaluation of rosis. Br J Rheumatol 1991; 30:418 – 421. observations of vertebral osteoporotic
13. Kaufmann TJ, Jensen ME, Schweickert fractures treated by percutaneous verte-
large series of patients. Although ques- PA, Marx WF, Kallmes DF. Age of fracture broplasty. Rheumatology (Oxford) 2000;
tions concerning the safety of vertebro- and clinical outcomes of percutaneous 39:1410 –1414.
plasty have been answered, its efficacy vertebroplasty. AJNR Am J Neuroradiol 29. Jensen ME, Kallmes DF, Short JG, et al.
and durability remain clouded owing to 2001; 22:1860 –1863. Percutaneous vertebroplasty does not in-
14. Do HM, Kallmes DF, Marx WF, Jensen crease the risk of adjacent level fracture: a
the lack of randomized controlled trials ME. Percutaneous vertebroplasty in the
and uncertainty over the role of the pla- retrospective study (abstr). In: ASNR An-
treatment of patients with vertebral os- nual Meeting Program. Oak Brook, Ill:
cebo effect. Radiologists have spear- teonecrosis (Kummell’s disease). Neuro- American Society of Neuroradiology,
headed the effort behind the validation surgical Focus/Journal of Neurosurgery 2000; 4.
1999; 7(1):article 2.
of vertebroplasty. It remains incumbent 30. Marx WF, Short JG, Kallmes DF, et al.
15. Gangi A, Kastler BA, Dietemann JL. Per- Long term plain film follow-up of pa-
on us to silence any doubts about the role cutaneous vertebroplasty guided by a tients treated with percutaneous verte-
vertebroplasty should play in patient care combination of CT and fluoroscopy.
broplasty: evaluation for changes in de-
through our continued thoughtful ques- AJNR Am J Neuroradiol 1994; 15:83– 86.
gree of vertebral compression, vertebral
tioning evaluation of this procedure. 16. Barr JD, Barr MS, Lemley TJ. Combined
kyphosis, and cement morphology. Pre-
CT and fluoroscopic guidance for percu-
sented at the ASNR 39th Annual Meeting,
taneous vertebroplasty. American Society
References Boston, Mass, April 23–27, 2001.
of Neuroradiology Annual Meeting 1996.
31. Lindsay R, Silverman SL, Cooper C, et al.
1. Galibert P, Deramond H, Rosat P, Le Gars 17. Shapiro SA. Cranioplasty, vertebral body
Risk of new vertebral fracture in the year
D. Preliminary note on the treatment of replacement, and spinal fusion with to-
following a fracture. JAMA 2001; 285:
vertebral angioma by percutaneous bramycin-impregnated methylmethacry-
acrylic vertebroplasty. Neurochirurgie late. Neurosurgery 1991; 28:789 –791. 320 –323.
1987; 33:166 –168. [French] 18. Gaughen JR Jr, Jensen ME, Schweickert 32. Lieberman IH, Dudeney S, Reinhardt MK,
2. Jensen ME, Evans AJ, Mathis JM, Kallmes PA, Kaufmann TJ, Marx W, Kallmes DF. Bell G. Initial outcome and efficacy of
DF, Cloft HJ, Dion JE. Percutaneous poly- Relevance of antecedent venography in “kyphoplasty” in the treatment of pain-
methylmethacrylate vertebroplasty in the percutaneous vertebroplasty for the treat- ful osteoporotic vertebral compression
treatment of osteoporotic vertebral body ment of osteoporotic compression frac- fractures. Spine 2001; 26:1631–1638.
compression fractures: technical aspects. tures. AJNR Am J Neuroradiol 2002; 23: 33. Fairbank JCF, Couper JF, Davies JBF,
AJNR Am J Neuroradiol 1997; 18:1897– 594 – 600. O’Brien JP. The Oswestry low back pain
1904. 19. Heatwole EV, McGraw JK, Patzik SB, et al. disability questionnaire. Physiotherapy
3. Cotten A, Boutry N, Cortet B, et al. Per- Predictive value of intraosseous venogra- 1980; 66:271–273.
cutaneous vertebroplasty: state of the art. phy prior to percutaneous vertebroplasty 34. Roland MF, Morris R. A study of the nat-
RadioGraphics 1998; 18:311–320. (abstr). J Vasc Interv Radiol 2001; 12:S11. ural history of back pain. I. Development
4. Deramond H, Depriester C, Toussaint P, 20. Kim AK, Jensen ME, Dion JE, Schweickert PA, of a reliable and sensitive measure of dis-
Galibert P. Percutaneous vertebroplasty. Kaufmann TJ, Kallmes DF. Unilateral trans- ability in low-back pain. Spine 1983;
Semin Musculoskelet Radiol 1997; 1:285– pedicular percutaneous vertebroplasty: ini- 8:141–144.
296. tial experience. Radiology 2002; 222:737– 35. Cook DJ, Guyatt GH, Adachi JD, et al.
5. Barr JD, Mathis JM, Barr MS, et al. Stan- 741. Development and validation of the mini-
dard for the performance of percutaneous 21. Belkoff SM, Mathis JM, Jasper LE, De- osteoporosis quality of life questionnaire
vertebroplasty. In: American College of ramond H. The biomechanics of verte- (OQLQ) in osteoporotic women with
Radiology standards 2000 –2001. Reston, broplasty: the effect of cement volume on back pain due to vertebral fractures: Os-
Va: American College of Radiology, 2000; mechanical behavior. Spine 2001; 26: teoporosis Quality of Life Study Group.
441– 448. 1537–1541. Osteoporos Int 1999; 10:207–213.
6. Maynard AS, Jensen ME, Schweickert PA, 22. Murphy KJF, Deramond H. Percutaneous 36. Jarvik JG, Deyo RA. Cementing the evi-
Marx WF, Short JG, Kallmes DF. Value of vertebroplasty in benign and malignant dence: time for a randomized trial of ver-
bone scan imaging in predicting pain re- disease. Neuroimaging Clin N Am 2000; tebroplasty. AJNR Am J Neuroradiol 2000;
lief from percutaneous vertebroplasty in 10:535–545. 21:1373–1374.
osteoporotic vertebral fractures. AJNR 23. Convery FR, Gunn DR, Hughes JD, Mar- 37. Gaughen JR Jr, Jensen ME, Schweickert
Am J Neuroradiol 2000; 21:1807–1812. tin WE. The relative safety of polymeth- PA, Kaufmann TJ, Marx WF, Kallmes DF.
7. Mathis JM, Barr JD, Belkoff SM, Barr MS, ylmethacrylate: a controlled clinical Lack of preoperative spinous process ten-
Jensen ME, Deramond H. Percutaneous study of randomly selected patients derness does not affect clinical success of
vertebroplasty: a developing standard of treated with Charnley and ring total hip percutaneous vertebroplasty. J Vasc In-
care for vertebral compression fractures. replacements. J Bone Joint Surg Am 1975; terv Radiol 2002; 13:1135–1138.
AJNR Am J Neuroradiol 2001; 22: 1:57– 64.
373–381. 24. McLaughlin RE, DiFazio CA, Hakala MF,

36 䡠 Radiology 䡠 October 2003 Kallmes and Jensen