Note: This copy is for your personal non-commercial use only.

To order presentation-ready
copies for distribution to your colleagues or clients, contact us at www.rsna.org/rsnarights.

ORIGINAL RESEARCH
Osteoporotic Compression
Fractures: Outcomes after Single-

䡲 MUSCULOSKELETAL IMAGING
versus Multiple-Level Percutaneous
Vertebroplasty1
Anand K. Singh, MD
Purpose: To compare single- and multiple-level percutaneous verte-
Thomas K. Pilgram, PhD
broplasty (PV) in terms of pain relief, activity level, and
Louis A. Gilula, MD
analgesic use in patients with osteoporotic vertebral com-
pression fractures (VCFs).

Materials and Institutional review board approval and informed consent

Methods: were obtained, and the study was HIPAA compliant. One
hundred seventy-three patients (mean age at treatment,
73.8 years ⫾ 11.9 [standard deviation]) with 422 symp-
tomatic osteoporotic VCFs underwent 204 treatment ses-
sions for over 4 years. Pain immediately before and after
PV was measured by using a visual analogue scale (VAS).
Pain degree, activity level, and analgesic use were assessed
at 2 weeks and 1, 3, 6, 12, and 24 months after PV by using
telephone interview questionnaires. Data were analyzed
by using a combination of paired t tests, analysis of vari-
ance, contingency tables, and ␹2 tests.

Results: Findings of 172 PV treatment sessions for 149 patients

(mean age at treatment, 73.4 years ⫾ 12), 110 (74%) of
whom were women, were assessed; 32 treatment cases
were lost to follow-up or lost owing to death. A single
fracture level was treated at 65 sessions; two fracture
levels, at 52 sessions; and three or more fracture levels, at
55 sessions. The mean VAS pain score decreased signifi-
cantly (P ⬍ .001), from 76 ⫾ 21 before to 19 ⫾ 27 imme-
diately after PV. Of the outcomes reported at 24 months,
82% (64 of 78 treatment sessions) were marked to com-
plete resolution of the initial pain, 51% were complete
cessation of analgesic use, and 51% were increased activ-
ity levels. These results did not differ greatly over time or
when stratified into groups according to the number of
fracture levels treated.

Conclusion: PV performed at a single fracture level and that performed

at multiple fracture levels were equally effective in facilitat-
ing long-term pain relief, increased activity level, and de-
creased analgesic use in patients with osteoporotic VCFs.

娀 RSNA, 2006

1
From Barnes-Jewish Hospital (A.K.S.) and Washington
University School of Medicine (T.K.P., L.A.G.), Mallinckrodt
Institute of Radiology, 510 S Kingshighway Blvd, St Louis,
MO 63110. Received December 8, 2004; revision re-
quested January 25, 2005; revision received April 8; ac-
cepted May 2; final version accepted June 24.

姝 RSNA, 2006

Radiology: Volume 238: Number 1—January 2006 211

MUSCULOSKELETAL IMAGING: Vertebroplasty for Osteoporotic Compression Fractures
O
steoporosis is the leading cause of
vertebral compression fractures
(VCFs) in the United States and
results in loss of independence and de-
creased quality of life among elderly in-
dividuals (1). An estimated 700 000
vertebral osteoporotic fractures occur
every year, and this number is expected
to increase fourfold during the next 50
years with the aging of individuals in the
“Baby Boomer” generation (1). Immo-
bility due to fracture pain is a common
cause of morbidity in this population,
making these individuals at risk for
pneumonia, deep vein thrombosis, and
pulmonary embolism (2). In addition,
patients experience loss of indepen-
dence and decreased quality of life (1–
6). Conservative therapy for VCF often
consists of bed rest, analgesic use, and
physical therapy (3–7), with the failure
of such therapies necessitating the use
of other treatments.
Percutaneous vertebroplasty (PV)
with polymethylmethacrylate (PMMA)
is a radiologically guided technique that
can be used in patients in whom conser-
vative therapy has been nonsuccessful
(2–9). PV with PMMA has been used to
treat a variety of lesions, including neo-
plastic and osteoporotic VCFs. In the
United States, PV is used primarily for
treatment of osteoporotic VCFs (9).
Study (2,3,5–10) results have shown
substantial pain relief and increased
mobility during the first 72 hours after
vertebroplasty in up to 90% of patients.
The practice of treating multiple
fracture levels in a single patient by us-
ing PV appears to be commonly re-
ported in the medical literature. This is
because the likelihood of having addi-
tional fractures increases once a patient
has a fracture compared with this likeli-
hood in age-matched control subjects
who have had previous fractures but
have not undergone PV (11). The risk of
having a second fracture is fivefold; in
individuals with two fractures, the risk
increases to 12-fold (11). Given the
prevalence of osteoporosis, the number
of patients with multiple osteoporotic
VCFs is expected to be high. Investiga-
tors in many studies of PV have based
their conclusions regarding the effec-
tiveness of PV on data from multiple-
212
level vertebroplasty that were grouped
with data from single-level vertebro-
plasty (5–7,10,12–18).
Some investigators (10) have sug-
gested that better outcomes are achieved
in patients treated at a single fracture
level than in those treated at multiple
levels. The purpose of this study was to
compare single- and multiple-level PV in
terms of pain relief, activity level, and
analgesic use in patients with osteopo-
rotic VCFs.
Materials and Methods
Patient Population
For more than 4 years—from June 9,
1998, to July 1, 2002—173 patients
(mean age at time of treatment, 73.8
years ⫾ 11.9 [standard deviation]) un-
derwent a total of 204 PV treatment
sessions for osteoporotic compression
fractures at Mallinckrodt Institute of
Radiology. Ninety-six of the 204 treat-
ment sessions were performed before
we decided to collect data pertinent to
our study. The remaining 108 treatment
sessions were performed with the inten-
tion of following up the treated patients
prospectively to collect similar data. In-
stitutional review board approval and
informed consent to collect data on both
patient groups for our Health Insurance
Portability and Accountability Act–
compliant study were obtained. Tele-
phone follow-up was initiated Septem-
ber 15, 2000, and ended March 29,
2004.
Patients with back pain believed to
be secondary to osteoporotic compres-
sion fractures had been fully examined
by their primary clinician. The patients
had undergone conservative treatment
for the fracture(s) that consisted of bed
rest, analgesic use, and physical therapy
before vertebroplasty was considered.
Those in whom conservative therapy
had failed underwent a formal examina-
tion for possible PV. Their medical his-
tories, physical examination findings,
and coagulation parameters were re-
viewed. Examination of all available ra-
diographs, bone scintigrams, and mag-
netic resonance (MR) images was
performed. In each case, the senior au-
Singh et al
thor (L.A.G.), who had more than 15
years of experience interpreting such
images, compared fracture levels with
pain symptoms. The degree and charac-
ter of the vertebral compression, the
visibility of the pedicles, the presence of
cortical destruction, the edema on MR
images, and the increased uptake on
bone scintigrams were assessed. Indica-
tions for PV included focal, severe, in-
tractable pain at the level of a known
compression fracture without definite
radicular signs or symptoms of bone im-
pinging on the spinal cord or nerves.
Contraindications to PV included infec-
tion, pain not related to a fracture, true
radicular symptoms of bone compress-
ing the spinal cord or nerves, and/or
unstable fractures involving the poste-
rior columns (19).
PV Technique
All PVs were performed either by or
under the supervision of the senior au-
thor (L.A.G.), who has extensive expe-
rience with the technique, having began
treating patients with PV in 1998. A
similar PV technique was used in all pa-
tients and at all treated levels. Patients
were placed in the prone position on the
C-arm fluoroscopic table. The patients’
vital signs were constantly monitored
while conscious sedation was induced
by a nurse trained in conscious seda-
Published online
10.1148/radiol.2381042078
Radiology 2006; 238:211–220
Abbreviations:
PMMA ⫽ polymethylmethacrylate
PV ⫽ percutaneous vertebroplasty
VAS ⫽ visual analogue scale
VCF ⫽ vertebral compression fracture
Author contributions:
Guarantors of integrity of entire study, A.K.S., L.A.G.;
study concepts/study design or data acquisition or data
analysis/interpretation, all authors; manuscript drafting or
manuscript revision for important intellectual content, all
authors; approval of final version of submitted
manuscript, all authors; literature research, A.K.S.; clinical
studies, A.K.S., L.A.G.; statistical analysis, A.K.S., T.K.P.;
and manuscript editing, all authors
Address correspondence to L.A.G.
(e-mail: gilulal@mir.wustl.edu ).
Authors stated no financial relationship to disclose.
Radiology: Volume 238: Number 1—January 2006
MUSCULOSKELETAL IMAGING: Vertebroplasty for Osteoporotic Compression Fractures
tion. Sedation was induced with neuro-
leptic analgesics such as fentanyl citrate
(Sublimaze; Abbott Laboratories, North
Chicago, Ill) and midazolam (Versed;
Hoffmann-LaRoche Pharmaceuticals,
Manati, Puerto Rico). Patients were
kept alert enough to state whether or
not they were experiencing pain during
the procedure.
With fluoroscopic guidance, the ver-
tebra(e) in question was localized for
visualization of the pedicle(s) at the lev-
el(s) to be treated. The skin overlying
the area was prepared and draped in
sterile fashion. The skin and periosteum
overlying the pedicle were anesthetized
with a 1:1 solution of lidocaine 1% (As-
tra-Zeneca, Wilmington, Del) and bu-
pivacaine hydrochloride 0.25% (Abbott
Laboratories). After the skin incision,
an 11- or 13-gauge Jamshidi-type bone
biopsy trocar was advanced with fluoro-
scopic control until its tip reached the
lamina posterior to the pedicle. Once
this position was confirmed, the trocar
was advanced centrally into the verte-
bral body with lateral and frontal fluoro-
scopic guidance.
Intraosseous “blush” venography
was then performed by using 0.5–1.0
mL of iohexol (Omnipaque 180; Ny-
comed, Princeton, NJ), which was in-
jected through the trocar to prevent the
trocar from being placed in the main
venous structures and to plan the subse-
quent injection of PMMA (20). PMMA
powder—Osteobond Copolymer Bone
Cement (Zimmer, Warsaw, Ind) or, in
earlier cases, Cranioplastic (Plastics
One, Roanoke, Va)—was mixed with
7 g of barium sulfate powder, which had
been sterilized with dry heat to increase
opacity (21). Cranioplastic was used in
the 16 early treatment sessions, and Os-
teobond Copolymer Bone Cement was
used in the remaining sessions. The bar-
ium sulfate powder was broken into fine
particles and then combined with the
PMMA powder. In the earlier cases
only, tobramycin (Nebcin; Eli Lilly, Indi-
anapolis, Ind) (1.2 g) was then added to
the mixture. The entire combination of
materials was then mixed to form a
toothpaste-like substance.
The PMMA mixture was then placed
through the back of a 20-mL syringe and
Radiology: Volume 238: Number 1—January 2006
Singh et al
backfilled into a 10-mL syringe, with were documented. Fracture age and
care taken to expel air from the mix- fracture levels were measured in a way
ture. The 10-mL syringe was then similar to the way in which they were
mounted onto a screw flange–type injec- measured in the study of Evans et al
tor with a hub adaptor modeled after (12), who defined acute (⬍2 weeks),
the LeVeen screw-type syringe (Boston subacute (2 weeks to 2 months), early
Scientific, Boston, Mass) (22), and the chronic (⬎2 months to 1 year), and late
remainder of the PMMA mixture was chronic (⬎1 year) fractures.
placed into a cold water bath for later
use. The PMMA mixture was injected
with lateral fluoroscopic guidance through
the trocar (Figure). The injection was
terminated when PMMA entered the
posterior quarter of the vertebral body
or when there was leakage outside the
vertebral body into the paravertebral
space or intervertebral disks. If leakage
was noted, either the injection was
halted for 1–2 minutes to allow the
PMMA to harden and plug the leak or
the needle was repositioned. Minor ex-
tension of PMMA into the intervertebral
disk space, paravertebral soft tissues,
or paravertebral veins was not consid-
ered to represent clinically important
sequelae (8,23).
If the PMMA injected through a sin-
gle pedicle did not cross the midline,
then the contralateral pedicle was ac-
cessed in a similar manner, as just de-
scribed, and the same steps were re-
peated to fill the vertebral body. After
the PV procedure, the patient remained
immobile for 1–2 hours and was dis-
charged from an outpatient recovery
area when he or she was judged to be
stable or able to ambulate. The physi-
cian who performed the PV established
stability by means of continuous moni-
toring of vital signs and follow-up clini-
cal examination. If the patient was an
inpatient, he or she remained in the
hospital until being discharged by the
referring physician. No clinically impor-
tant complications of PV were observed
early or late after the procedure. No Placement of transpedicular trocars for multiple-
patient required surgery because of the fracture-level PV with PMMA. (a) Lateral fluoro-
small leakages that occurred. scopic view shows single trocars have been placed
in each fractured vertebral body by using a trans-
Data Collection pedicular approach. The stylet for the lowermost
The patient and procedural characteris- trocar has been removed in preparation for PV.
tics of each PV were recorded by using a (b) Lateral fluoroscopic view shows filling of the
standard form that was completed at two lowermost vertebral bodies with PMMA
the time of the procedure and on which through the trocars. The uppermost vertebral body
has not been filled yet, and the trocar stylet
the patient’s age and sex and the num-
remains.
ber and location of fractures treated
213
MUSCULOSKELETAL IMAGING: Vertebroplasty for Osteoporotic Compression Fractures Singh et al

Before undergoing PV, the patients ized spreadsheets (Microsoft Excel; Mi- included reported pain degree, activity
rated their pain level by using a standard crosoft, Seattle, Wash). Treatment ses- level, and analgesic use after PV. Pre-
visual analogue scale (VAS), on which a sions were divided into three groups for treatment and immediate posttreat-
score of 0 indicated no pain and a score of analysis: one fracture level, two fracture ment pain was analyzed by using VAS
100 indicated maximal pain (5,12,13,15– levels, and three or more fracture levels pain scores. Mean scores of pre- and
17). They repeated this rating approxi- treated. Measures of treatment success posttreatment pain and of change in
mately 1 hour after the procedure. Then,
a research assistant trained to conduct
Table 1
telephone interviews contacted the pa-
tients by telephone at fixed intervals after Clinical Parameters of Patient Population
their treatment sessions—at 2 weeks and
Parameter Value*
1, 3, 6, 12, and 24 months—and using an
institutional review board–approved Total no. of treatment sessions 204
questionnaire designed specifically for Death terminated follow-up† 28
this study recorded their responses to Lost to all follow-up† 4
treatment in terms of a variety of factors, Total no. of treatment sessions included in analysis 172
including pain degree, activity level, and No. of treatment sessions with female patients‡ 126 (73)
analgesic use. Patient age at treatment session (y)†
⬍60 27 (13)
As stated earlier, some treatment
61–70 38 (19)
sessions were performed before we de-
71–80 79 (39)
cided to follow up patients prospectively
ⱖ81 60 (29)
for this study. The imperfect overlap of
Mean§ 73.4 ⫾ 12
starting the study with the procedure
No. of fracture levels treated‡
dates meant that there were missing fol- 1 65 (38)
low-up data on the earliest treatment 2 52 (30)
sessions for some of the early time peri- ⱖ3 55 (32)
ods and on the latest sessions for some Fracture locations㛳
of the later time periods. However, fol- T2 vertebra 1 (0.2)
low-ups of more than 50% of the ses- T3 vertebra 1 (0.2)
sions were conducted at four or more T4 vertebra 1 (0.2)
time periods. T5 vertebra 8 (2)
During each telephone interview, T6 vertebra 19 (5)
the patient rated his or her pain as gone T7 vertebra 20 (5)
or as better than, the same as, or worse T8 vertebra 26 (6)
than it was before the PV. In the past, T9 vertebra 20 (5)
many of our elderly patients had had T10 vertebra 26 (6)
difficulty reporting a reliable VAS pain T11 vertebra 27 (6)
score over the telephone. Because of T12 vertebra 62 (15)
this, a telephone questionnaire that ad- L1 vertebra 73 (17)
L2 vertebra 50 (12)
dressed the described pain relief cate-
L3 vertebra 40 (9)
gories was developed. To allow ade-
L4 vertebra 36 (9)
quate time for the treatment effect to
L5 vertebra 12 (3)
occur, at each time interval except 2
Thoracic spine (T1–T10) 122 (29)
weeks, the patients rated their activity
Thoracolumbar junction (T11–L2) 212 (50)
level as follows: more active than, the Lower lumbar spine (L3–L5) 88 (21)
same as, or less active than before the Thoracic spine, total 211 (50)
procedure. For assessment of analgesic Lumbar spine, total 211 (50)
use, the patients were asked whether or Total no. of fractures treated 422
not they were currently taking any pain Mean no. of fractures treated per patient 2.1
medicine, because the entire cohort had Fracture age at treatment time#
been taking some form of analgesic Acute (⬍2 wk) 1 (0.5)
medication before undergoing PV. Subacute (2 wk to 2 mo) 36 (18)
Early chronic (⬎2 mo to 1 y) 109 (54)
Statistical Analyses Late chronic (⬎1 y) 56 (28)
All data were collected and stored in a Unknown 0
common database by using computer- (Table 1 continues)

214 Radiology: Volume 238: Number 1—January 2006

MUSCULOSKELETAL IMAGING: Vertebroplasty for Osteoporotic Compression Fractures Singh et al

Table 1 (continued)
power, and the ␹2 tests had 80% power
Clinical Parameters of Patient Population for the detection of a 25% difference
(eg, 65% vs 90% improved) at an ␣ of
Parameter Value*
.05. For all statistical testing, results
Fracture age by treatment subgroup were reported to be significant at P ⬍
One fracture level treated (n ⫽ 80) .05. All statistical analyses were per-
Acute 0 formed by using JMP 4.0 software (SAS
Subacute 19 (24) Institute, Cary, NC).
Early chronic 44 (55)
Late chronic 17 (21)
Two fracture levels treated (n ⫽ 54) Results
Acute 1 (2)
Subacute 8 (15) Final Study Cohort
Early chronic 29 (54)
A total of 204 treatment sessions for
Late chronic 16 (30)
422 symptomatic osteoporotic VCFs
Three or more fracture levels treated (n ⫽ 68)
were performed in 173 patients. Four
Acute 0
Subacute 9 (13)
cases were lost to follow-up, and there
Early chronic 36 (53) were 28 deaths from causes unrelated
Late chronic 23 (34) to PV during the follow-up period. An
Unknown equal proportion of deaths were seen in
Acute 2 the one-fracture-level and three-or-
Subacute ... more-fracture-level treatment groups,
Early chronic ... with fewer deaths noted in the two-frac-
Late chronic ... ture-level group. A total of 172 treat-
No. of follow-up interviews‡ ment sessions performed in 149 pa-
1 18 (10) tients were included in our analysis. The
2 36 (21) number of follow-up interviews per
3 26 (15) treatment session ranged from one to
4 18 (10) six, with five being the most common
5 47 (27) number of interviews performed (for 47
6 27 (16) [27%] treatment sessions) (Table 1). In
Time interval‡ the majority of cases (for 92 [53%]
2 weeks 110 (64)
treatment sessions), four or more fol-
1 month 108 (63)
low-up interviews were conducted; the
3 months 106 (62)
mean number of follow-ups conducted
6 months 113 (66)
was 3.7. Nearly two-thirds of all the pa-
12 months 122 (71)
tients were successfully contacted (range,
24 months 78 (45)
62%–71% of patients contacted) for a
* Numbers in parentheses are percentages. telephone interview at each time interval
†
Values are numbers of treatment sessions (of total of 204). except 24 months, at which 45% of pa-
‡
Values are numbers of treatment sessions (of total of 172). tients were contacted (Table 1).
§
Mean patient age ⫾ standard deviation.
㛳
Values are numbers of fractures (of total of 422). General Features
#
Values are numbers of treatment sessions (of total of 202).
The mean age of the studied patient co-
hort was 73.4 years ⫾ 12 (standard de-
viation); 110 (74%) of the patients were
pain level were calculated. The change group at each follow-up time period by women. Similar numbers of treatment
in VAS pain scores was tested for statis- using contingency tables. sessions were performed in each treat-
tical significance within each group by For statistical testing of long-term ment level group: 65 sessions were per-
using paired t tests. Group differences change, treatment sessions were cate- formed in the one-fracture-level group;
in the change in VAS pain score accord- gorized into two groups: single fracture 52 sessions, in the two-fracture-level
ing to number of fracture levels treated level and multiple fracture levels group; and 55 sessions, in the three-or-
were tested by using analysis of vari- treated. Outcomes also were catego- more-fracture-level group. Comparable
ance. Longer-term changes in pain de- rized into two groups: improved and un- numbers of thoracic and lumbar frac-
gree, activity level, and analgesic use improved. The combining of categories tures were treated. Half the fractures
were examined according to treatment was designed to increase statistical occurred in the thoracolumbar junction:

Radiology: Volume 238: Number 1—January 2006 215

MUSCULOSKELETAL IMAGING: Vertebroplasty for Osteoporotic Compression Fractures
21% occurred in the lower lumbar
spine, and 29% occurred in the thoracic
spine. The mean number of fractures
treated per patient was 2.1. The major-
ity of fractures were in the early chronic
(⬎2 months to 1 year) phase, regard-
less of the number of levels treated. For
two treatment sessions, the fracture age
was unknown.
Pain Relief
The mean VAS pain score for all treat-
ments was 76 ⫾ 21 before and 19 ⫾ 27
after treatment (P ⬍ .001, paired t test)
(Table 2). The decrease in pain follow-
ing treatment was similar among all
treatment groups (P ⫽ .50, analysis of
variance). Longer-term pain relief was
recorded by using categorical evalua-
tions. Pain was reported as gone or bet-
ter by 77%– 85% of all the respondents
throughout the follow-up period (Table
3). At most time periods, a greater pro-
portion of the respondents in whom
multiple fracture levels were treated re-
ported improvement compared with the
proportion of respondents in whom one
fracture level was treated; however, the
difference never exceeded 6% and was
not significant (P ⬎ .40 for all cases, ␹2
test). A minority (2%–5%) of the re-
spondents reported having worse pain
after the PV. The only exception to this
finding was the proportion of respon-
dents who reported having worse pain
among those patients who had two frac-
ture levels treated and were followed up
at 24 months.
Table 2
Pain Relief after Vertebroplasty
Mean VAS Pain Score
Treatment Group Before Vertebroplasty*
Overall 76 ⫾ 21
One fracture level treated 73 ⫾ 24
Two fracture levels treated 77 ⫾ 20
Three or more fracture levels
treated 77 ⫾ 19
* Data are mean scores ⫾ standard deviations.
†
P values for significance of change in VAS pain score, calculated by using t test. P ⫽ .50 for significance of differences in
changes according to treatment group (one fracture level vs two fracture levels vs three or more fracture levels treated), as
calculated at analysis of variance.
216
Activity Level
The majority (54%–70%) of the respon-
dents reported having increased activity
levels after the procedure at every time
interval (Table 4). At all follow-up
times, a larger proportion of respon-
dents with a single fracture level treated
reported having increased activity levels
compared with the proportion of re-
spondents in whom multiple levels were
treated. The difference ranged from
3%–17% and approached but did not
achieve significance at 12 months (P ⫽
.06, ␹2 test). A minority of respondents
(5%–9%) had decreased activity levels
throughout the follow-up period, with
exceptions observed in only a few sub-
groups: in patients who had one and
two fracture levels treated and were fol-
lowed up at 1 month, in patients who
had two fracture levels treated and
were followed up at 3 months, in pa-
tients who had three fracture levels
treated and were followed up at 6
months, and in patients who had two
fracture levels treated and were fol-
lowed up at 24 months.
Analgesic Use
The percentages of respondents who re-
ported ceasing to use analgesics after
PV increased until the 3-month fol-
low-up period, after which they stabi-
lized (Table 5). The overall percentage
of respondents who ceased using anal-
gesics was 32% (subgroup range, 23%–
37%) at 2 weeks and increased to ap-
proximately 55% at both the 3-month
After Vertebroplasty* Mean Change
19 ⫾ 27 ⫺57
14 ⫾ 24 ⫺59
20 ⫾ 27 ⫺57
24 ⫾ 29 ⫺53
Singh et al
(subgroup range, 50%–59%) and 6-month
(subgroup range, 49%– 62%) intervals.
A larger proportion of respondents in
the multiple-treatment-level group than
in the single-treatment-level group re-
ported analgesic use at all time periods
except 12 months, but the differences
were small (1%– 8%) and not significant
(P ⱖ .39, ␹2 test).
Discussion
To our knowledge, in only one study
(12) have investigators outlined the dif-
ferences between patients treated for
single- versus multiple-level osteopo-
rotic VCFs. We could not find any re-
view in which the prospective results for
one or two treated levels were com-
pared with the results for more than
two levels treated during the same ses-
sion. We believe that at the time we
performed the study, the data in our
investigation represented data collected
for some of the longest continuous fol-
low-up periods reported in the litera-
ture.
In this study, the pain relief re-
ported by the total population of pa-
tients according to VAS scores was sig-
nificant and similar across all treatment
subgroups. This magnitude of relief has
been replicated in several other studies
(5,12,13,15–17) in which the same pain
scale was used. Barr et al (10) achieved
initial pain relief in 95% of patients
treated for osteoporotic VCF and stable
pain relief for 18 months in 94% of pa-
tients. Grados et al (16) found pain re-
lief to be unchanged during a long-term
follow-up period of 48 months in 25 pa-
tients. Perez-Higueras et al (15) ob-
served large decreases in VAS pain
scores, which were detected early
(third day after PV) and preceded con-
P Value† sistent pain relief for 5 years in a small
patient cohort. Given these results, we
⬍ .001
believed that it was acceptable to mea-
⬍ .001
sure the initial pain relief by using VAS
⬍ .001
scores and then to track patients on the
basis of pain measurements by using
⬍ .001
simpler criteria that patients could
readily understand and describe during
the 24-month follow-up period. The
suitability of this protocol was further
validated by the 77%– 85% of respon-
Radiology: Volume 238: Number 1—January 2006
MUSCULOSKELETAL IMAGING: Vertebroplasty for Osteoporotic Compression Fractures Singh et al

Table 3

Pain Relief up to 2 Years after Vertebroplasty

Time after Vertebroplasty
Pain Status 2 Weeks 1 Month 3 Months 6 Months 12 Months 24 Months

Overall
Gone 17/110 (16) 26/108 (24) 29/106 (27) 37/113 (33) 48/122 (39) 28/78 (36)
Better 69/110 (63) 57/108 (53) 61/106 (58) 58/113 (51) 52/122 (43) 36/78 (46)
Same 20/110 (18) 23/108 (21) 13/106 (12) 15/113 (13) 17/122 (14) 10/78 (13)
Worse 4/110 (3) 2/108 (2) 3/106 (3) 3/113 (3) 5/122 (4) 4/78 (5)
One fracture level treated
Gone 8/44 (18) 10/40 (25) 9/35 (26) 12/41 (29) 18/43 (41) 13/34 (38)
Better 25/44 (57) 20/40 (50) 20/35 (57) 21/41 (51) 17/43 (40) 16/34 (47)
Same 9/44 (20) 10/40 (25) 4/35 (11) 8/41 (20) 6/43 (14) 4/34 (12)
Worse 2/44 (5) 0 2/35 (6) 0 2/43 (5) 1/34 (3)
Two fracture levels treated
Gone 3/35 (9) 9/37 (24) 11/36 (31) 13/37 (35) 13/39 (34) 6/21 (29)
Better 26/35 (74) 19/37 (51) 20/36 (55) 19/37 (51) 18/39 (46) 10/21 (48)
Same 6/35 (17) 8/37 (22) 4/36 (11) 3/37 (8) 6/39 (15) 2/21 (9)
Worse 0 1/37 (3) 1/36 (3) 2/37 (6) 2/39 (5) 3/21 (14)
Three or more fracture levels treated
Gone 6/31 (20) 7/31 (23) 9/35 (26) 12/35 (34) 17/40 (43) 9/23 (39)
Better 18/31 (58) 18/31 (58) 21/35 (60) 18/35 (51) 17/40 (43) 10/23 (44)
Same 5/31 (16) 5/31 (16) 5/35 (14) 4/35 (12) 5/40 (12) 4/23 (17)
Worse 2/31 (6) 1/31 (3) 0 1/35 (3) 1/40 (2) 0
Single vs multiple treatment levels treated
One fracture level
Improved (gone or better) 33/44 (75) 30/40 (75) 29/35 (83) 33/41 (80) 35/43 (81) 29/34 (85)
Unimproved (same or worse) 11/44 (25) 10/40 (25) 6/35 (17) 8/41 (20) 8/43 (19) 5/34 (15)
Multiple fracture levels
Improved (gone or better) 53/66 (80) 53/68 (78) 61/71 (86) 62/72 (86) 65/79 (82) 35/44 (80)
Unimproved (same or worse) 13/66 (20) 15/68 (22) 10/71 (14) 10/72 (14) 14/79 (18) 9/44 (20)
P value* .51 .73 .68 .43 .90 .51

Note.—Data are numbers of follow-ups within each subgroup with given responses. Numbers in parentheses are percentages of follow-ups in each subgroup with given response.
* P values for differences in improvement between group with one fracture level treated and group with multiple fracture levels treated, as calculated at ␹2 analysis.

dents who reported having partial to
complete pain relief throughout the
study, similar to the results observed in
multiple other studies (2,7,10,12,13,17).
Vertebroplasty has been shown to
provide early pain relief and increased
ambulation (7,12,13,17). Ninety-one
percent to 95% of the patients in our
study had the same or increased activity
levels after PV, and the majority (range,
54%–70%) of respondents had im-
proved activity at all time intervals. This
range of improved activity is consistent
with results described in prior reports
(7). McGraw et al (13) observed im-
proved ambulation in 93% of patients
after treatment. Evans et al (12) ob-
served a similar substantial effect on
ambulation and the ability to perform
routine daily activities in a retrospective
review of data for 245 treated patients
with osteoporosis. Some study investi-
gators (12,17) analyzed posttreatment
activity by using formal outcome mea-
sures and found improved activity lev-
els. In addition, each subgroup in our
study maintained the benefits of PV in
terms of activity level for 24 months,
and these benefits did not differ accord-
ing to the number of fracture levels
treated. These findings are consistent
with those observed by previous au-
thors and show that vertebroplasty per-
formed for treatment of a prospective
cohort of osteoporotic VCFs can be ex-
pected to facilitate pain relief and the
associated benefit of improved activity.
The percentages of patients with
complete to partial posttreatment pain
relief in our study population, as re-
flected by decreased analgesic use, were
similar to findings described in prior re-
ports. In most prior series, pain relief
was substantial but not complete (2,7,
10,12,13,17). This correlates with the
number of patients who were able to
completely cease using analgesics after
PV in our study. Amar et al (7) observed
large posttreatment decreases in anal-
gesic use in 63% of their patient cohort.
Evans et al (12) also observed a sub-
stantial decrease in pain medication
use: Nearly 46% of patients claimed to
have stopped using pain medication af-
ter the procedure; this percentage is
similar to the percentages of patients in
our study who made this claim at almost
all follow-up time intervals. In our
study, the percentage of patients who
completely ceased using analgesics

Radiology: Volume 238: Number 1—January 2006 217

MUSCULOSKELETAL IMAGING: Vertebroplasty for Osteoporotic Compression Fractures Singh et al

gradually increased for 3– 6 months and
then stabilized to a level in the range of
values reported in the other described
studies. The gradual increase in analge-
sic use cessation may have been due to
progressive fracture healing and associ-
ated pain relief.
Some practitioners may expect pa-
tients who are treated at multiple frac-
ture levels to have worse outcomes than
patients who are treated at a single frac-
ture level owing to factors such as in-
creased thoracic kyphosis in patients
with multiple fractures. Our treatment
protocol involved the examination of
each patient for focal pain at each frac-
tured level and the correlation of these
findings with MR imaging and/or scinti-
graphic findings. In this way, we were
able to tailor each patient’s treatment to
his or her clinically relevant fractured
levels. Very few patients returned for
additional treatment of levels noted to
be abnormal at radiologic and clinical
evaluation. This is the likely explanation
for the consistent results regarding pain
relief, analgesic use, and activity level—
regardless of the number of fracture lev-
els treated—reported in our study.
A potential weakness of this study is
that it was not completely longitudinal.
However, it is not uncommon to have
missing data when large cohorts of el-
derly patients are studied (7). Grados et
al (16) collected data on only 60% of
their original patients during follow-up,
whereas Evans et al (12), in their large
retrospective review, reported on only
half of their original patient cohort. In
our study, the patient recruitment pe-
riod of almost 4 years, the follow-up
interview period of more than 31⁄2
years, and the loss of follow-up in some
patients yielded complete follow-up
data (six follow-up reports) on just 27
(16%) treatment sessions. However,
four or more follow-up reports were
completed on the majority (n ⫽ 92
[53%]) of patients; the mean number of
follow-ups conducted was 3.7. Neither
the patient recruitment protocol nor the
clinical examination procedure changed
during the study period, however, so it
is likely that the data collected at all
follow-up times were comparable. The
fact that the results as a whole were
remarkably consistent also supports the
credibility of the data.
The results obtained in this study
represent data on only those patients in
each treatment subgroup who were
contacted by telephone at each time in-
terval. Therefore, the patients within
each subgroup who were contacted at
one time point were not necessarily the
exact same patients who were con-
tacted at the next time point. It is possi-
ble that because the entire population
was not sampled at every time point,

Table 4

Activity Level after Vertebroplasty

Time after Vertebroplasty
Activity Level 1 Month 3 Months 6 Months 12 Months 24 Months

Overall
More active 58/108 (54) 73/105 (70) 69/113 (61) 77/122 (63) 40/79 (51)
Same 40/108 (37) 24/105 (23) 38/113 (34) 37/122 (30) 34/79 (43)
Less active 9/108 (9) 8/105 (7) 6/113 (5) 8/122 (7) 5/79 (6)
One fracture level treated
More active 24/39 (61) 27/35 (77) 26/41 (63) 32/43 (74) 19/36 (53)
Same 11/39 (28) 6/35 (17) 14/41 (34) 8/43 (19) 15/36 (42)
Less active 4/39 (11) 2/35 (6) 1/41 (3) 3/43 (7) 2/36 (5)
Two fracture levels treated
More active 16/37 (43) 24/36 (67) 24/37 (65) 22/39 (57) 10/20 (50)
Same 17/37 (46) 8/36 (22) 12/37 (32) 15/39 (38) 8/20 (40)
Less active 4/37 (11) 4/36 (11) 1/37 (3) 2/39 (5) 2/20 (10)
Three or more fracture levels treated
More active 18/31 (58) 22/34 (65) 19/35 (54) 23/40 (58) 11/23 (48)
Same 12/31 (39) 10/34 (29) 12/35 (34) 14/40 (35) 11/23 (48)
Less active 1/31 (3) 2/34 (6) 4/35 (12) 3/40 (7) 1/23 (4)
Single vs multiple fracture levels treated
One fracture level
Improved (more active) 24/39 (62) 27/35 (77) 26/41 (63) 32/43 (74) 19/36 (53)
Unimproved (same or less active) 15/39 (38) 8/35 (23) 15/41 (37) 11/43 (26) 17/36 (47)
Multiple fracture levels
Improved (more active) 34/68 (50) 46/70 (66) 43/72 (60) 45/79 (57) 21/43 (49)
Unimproved (same or less active) 34/68 (50) 24/70 (34) 29/72 (40) 34/79 (43) 22/43 (51)
P value* .25 .23 .70 .06 .73

Note.—Data are numbers of follow-ups within each subgroup with given response. Numbers in parentheses are percentages of follow-ups in each subgroup with given response.
* P values for differences in activity level improvement between group with one fracture level treated and group with multiple fracture levels treated, as calculated at ␹2 analysis.

218 Radiology: Volume 238: Number 1—January 2006

MUSCULOSKELETAL IMAGING: Vertebroplasty for Osteoporotic Compression Fractures Singh et al

Table 5

Analgesic Use after Vertebroplasty

Time after Vertebroplasty
Analgesic Use 2 Weeks 1 Month 3 Months 6 Months 12 Months 24 Months

Overall
No use 35/108 (32) 47/107 (44) 58/105 (55) 62/113 (55) 65/122 (53) 39/76 (51)
Use 73/108 (68) 60/107 (56) 47/105 (45) 51/113 (45) 57/122 (47) 37/76 (49)
One fracture level treated
No use 16/43 (37) 19/39 (49) 20/35 (57) 20/41 (49) 24/43 (56) 18/34 (53)
Use 27/43 (63) 20/39 (51) 15/35 (43) 21/41 (51) 19/43 (44) 16/34 (47)
Two fracture levels treated
No use 12/35 (34) 16/37 (43) 18/36 (50) 23/37 (62) 17/39 (44) 10/21 (48)
Use 23/35 (66) 21/37 (57) 18/36 (50) 14/37 (38) 22/39 (56) 11/21 (52)
Three or more fracture levels treated
No use 7/30 (23) 12/31 (39) 20/34 (59) 19/35 (54) 24/40 (60) 11/21 (52)
Use 23/30 (77) 19/31 (61) 14/34 (41) 16/35 (46) 16/40 (40) 10/21 (48)
Single vs multiple fracture levels treated
One fracture level
No use 16/43 (37) 19/39 (49) 20/35 (57) 20/41 (49) 24/43 (56) 18/34 (53)
Use 27/43 (63) 20/39 (51) 15/35 (43) 21/41 (51) 19/43 (44) 16/34 (47)
Multiple fracture levels
No use 19/65 (29) 28/58 (48) 38/70 (54) 42/72 (58) 41/79 (52) 21/42 (50)
Use 46/65 (71) 30/58 (52) 32/70 (46) 30/72 (42) 38/79 (48) 21/42 (50)
P value* .39 .97 .78 .33 .41 .80

Note.—Data are numbers of follow-ups within each subgroup with given response. Numbers in parentheses are percentages of follow-ups in each subgroup with given response.
* P values for differences in analgesic use reduction between group with one fracture level treated and group with multiple fracture levels treated, as calculated at ␹2 analysis.

measurement error in which a few pa-
tients with poor outcomes were not
sampled could have occurred. How-
ever, the positive outcomes docu-
mented throughout the study were rela-
tively stable across the treatment
subgroups, despite the different combi-
nations of patients whose data were
sampled at each time interval; thus, this
type of error was unlikely. Also, many
of the outcome results generated, in-
cluding percentages of patients with
pain relief and percentages of patients
who ceased using analgesics, were simi-
lar to those obtained in other studies.
Selection bias could have contrib-
uted to the treatment and examination
of patients within certain subgroups
who had less severe or fewer comor-
bidities and thus may have been ex-
pected to have better long-term results.
However, the similar clinical makeups
of the patients in our study population,
regardless of the number of fractures,
and the resultant positive outcomes
seen in all groups independent of the
number of levels treated made this bias
less likely.
There was worsening pain and re-
duced activity level in some of the pa-
tient subgroups at various time inter-
vals. Follow-up of our patient subgroups
revealed the worsening conditions to be
transient. In addition, results of studies
in the literature suggest that patients
treated with PV may have an increased
relative risk of fracture at adjacent un-
treated levels (6). This is expected to
result in worse pain in more patients in
whom multiple fracture levels were
treated; however, the reported worsen-
ing of these symptoms was not concen-
trated in any one subgroup in our study.
The absolute numbers of patients with
such results were small, and the per-
centages of these patients at some time
points may have simply reflected a small
sample size at that time.
It should be recognized that those
patients who stated that they had worse
symptoms usually had other comorbidi-
ties that were the source of their pain
rather than failed vertebroplasty. In ad-
dition, the senior author of this article
has encountered many patients who
stated that their symptoms worsened
after PV, only to find that they had de-
veloped a new fracture. The failure to
follow up such patients after PV may be
erroneously attributed to treatment fail-
ure, when in fact new fractures that
may necessitate additional treatment
may be present. Symptomatic worsen-
ing may be caused by a combination of
comorbidities and new fractures. We
believe that further studies to address
this issue of persistent or worsening
pain may be warranted for such pa-
tients.
Our data collection protocol, with
the exception of the initial use of VAS
pain scores, was based on elementary
outcome measures. A research assis-
tant trained in telephone follow-up in-
terviewing found this format to be rela-
tively easy to apply in this patient
population, and the study questionnaire
itself was created to minimize recall er-
ror. However, validated assessments of
the global status of patients would have
been preferable. Only formalized pain
assessments have been performed reli-
ably in many studies in which the VAS
was used (5,12,13,15–17). Although

Radiology: Volume 238: Number 1—January 2006 219

MUSCULOSKELETAL IMAGING: Vertebroplasty for Osteoporotic Compression Fractures
this tool is more accurate when it is
used with patients who are seen in per-
son, many of our study patients lived
several hours from our medical center,
so a simpler, more reliable assessment
tool that could be used over the tele-
phone was needed. In addition, formal
outcome measures of other parameters,
such as mobility and analgesic use,
which are used with small numbers of
patients in some studies, may be diffi-
cult to assess in a large prospective pa-
tient cohort with long-term follow-up.
Perhaps the use of an instrument such
as the Roland scale proposed by
Kallmes and Jensen (24) may be an op-
tion for future studies.
PV is a highly effective treatment in
the hands of skilled personnel despite
being invasive and the possible associ-
ated complications. However, a true
measurement of its effectiveness cannot
be performed until a randomized con-
trol trial in which some patients un-
dergo PV while others participate in a
sham procedure, with an emphasis on
validated outcome measures and possi-
ble placebo effects, is undertaken.
Some authors (24,25) have attempted
to simulate this type of study on a small
scale and achieved inconclusive results.
Only until such a study is conducted will
we be able to determine what, if any,
true benefit patients attain after under-
going successful vertebroplasty.
We believe our study data are the
first to demonstrate the equal effective-
ness of PV performed at a single fracture
10. Barr JD, Barr MS, Lemley TJ, McCann RM.
level and PV performed at multiple frac-
ture levels—in terms of pain relief, activ-
ity level, and analgesic cessation—in a
large patient cohort. The results of the
11. Lindsay R, Silverman S, Cooper C, et al. Risk
current study lend validity to the practice
of treating multiple levels of osteoporotic
VCF, with the expectation that many pa-
12. Evans AJ, Jensen ME, Kip KE, et al. Verte-
tients will have favorable treatment out-
comes, improved qualities of life, and de-
creased morbidity and mortality.
References
13. McGraw JK, Lippert J, Minkus K, Rami P,
1. Riggs BL, Melton LJ 3rd. The worldwide
problem of osteoporosis: insights afforded
220
by epidemiology. Bone 1995;17(5 suppl):
505S–511S.
2. Jensen ME, Evans AJ, Mathis JM, Kallmes
DF, Cloft HJ, Dion JE. Percutaneous poly-
methylmethacrylate vertebroplasty in the
treatment of osteoporotic vertebral com-
pression fractures: technical aspects. AJNR
Am J Neuroradiol 1997;18:1897–1904.
3. Mathis JM, Barr JD, Belkoff SM, Barr MS,
Jensen ME, Deramond H. Percutaneous
vertebroplasty: a developing standard of
care for vertebral compression fractures.
AJNR Am J Neuroradiol 2001;22:373–381.
4. Dion J. Percutaneous vertebroplasty. Medi-
camundi 2001;45(2):21–28.
5. Cortet B, Cotten A, Boutry N, et al. Percuta-
neous vertebroplasty in the treatment of os-
teoporotic vertebral compression fractures:
an open prospective study. J Rheumatol
1999;26(10):2222–2228.
6. Uppin A, Hirsch J, Centenera L, Pfiefer B,
Pazianos A, Choi I. Occurrence of new verte-
bral body fracture after percutaneous verte-
broplasty in patients with osteoporosis. Ra-
diology 2003;226:119 –124.
7. Amar AP, Larsen DW, Esnaashari N, Albu-
querque FC, Lavine SD, Teitelbaum GP. Per-
cutaneous transpedicular polymethylmethac-
rylate vertebroplasty for the treatment of
spinal compression fractures. Neurosurgery
2001;49(5):1105–1114.
8. Peh WC, Gilula LA, Peck D. Percutaneous
vertebroplasty for severe osteoporotic verte-
bral body compression fractures. Radiology
2002;223:121–126.
9. Murphy KJ, Deramond H. Percutaneous ver-
tebroplasty in benign and malignant disease.
Neuroimaging Clin N Am 2000;10(3):535–
545.
Percutaneous vertebroplasty for pain relief
and spinal stabilization. Spine 2000;25(8):
923–928.
of new vertebral fracture in the year follow-
ing fracture. JAMA 2001;285:320 –323.
bral compression fractures: pain reduction
and improvement in functional mobility after
percutaneous polymethylmethacrylate verte-
broplasty—retrospective report of 245
cases. Radiology 2003;226:366 –372.
Davis T, Budzik R. Prospective evaluation of
pain relief in 100 patients undergoing percu-
Singh et al
taneous vertebroplasty. J Vasc Interv Radiol
2002;13:883– 886.
14. Martin JB, Jean B, Sugiu K, et al.
Vertebroplasty: clinical experience and fol-
low up results. Bone 1999;25(2 suppl):11S–
15S.
15. Perez-Higueras A, Alvarez L, Rossi RE,
Quinones D, Al-Assir I. Percutaneous
vertebroplasty: long-term clinical and ra-
diological outcome. Neuroradiology 2002;
44(11):950 –954.
16. Grados F, Depriester C, Cayrolle G, Hardy
N, Deramond H, Fardellone P. Long-term
observations of vertebral osteoporotic frac-
tures by percutaneous vertebroplasty. Rheu-
matology (Oxford) 2000;39:1410 –1414.
17. Zoarski GH, Snow P, Olan WJ, et al. Percu-
taneous vertebroplasty for osteoporotic
compression fractures: quantitative pro-
spective evaluation of long-term outcomes. J
Vasc Interv Radiol 2002;13:139 –148.
18. Kaufmann TJ, Jensen ME, Schweickert PA,
Marx WF, Kallmes DF. Age of fracture and
clinical outcomes of percutaneous vertebro-
plasty. AJNR Am J Neuroradiol 2001;22:
1860 –1863.
19. Appel NB, Gilula LA. Percutaneous verte-
broplasty in patients with spinal canal com-
promise. AJR Am J Roentgenol 2004;182:
947–951.
20. Peh WC, Gilula LA. Additional value of a
modified method of intraosseous venography
during percutaneous vertebroplasty. AJR
Am J Roentgenol 2003;180:87–91.
21. Leibold RA, Gilula LA. Sterilization of bar-
ium for vertebroplasty: an effective, reliable,
and inexpensive method to sterilize powders
for surgical procedures. AJR Am J Roentge-
nol 2002;179:198 –200.
22. Schallen EH, Gilula LA. Vertebroplasty: re-
usable flange converter with hub lock for
injection of polymethylmethacrylate with
screw-plunger syringe. Radiology 2002;
222(3):851– 855.
23. Hodler J, Peck D, Gilula LA. Midterm out-
come after vertebroplasty: predictive value
of technical and patient-related factors. Ra-
diology 2003;227(3):662– 668.
24. Kallmes DF, Jensen ME. Percutaneous ver-
tebroplasty. Radiology 2003;229:27–36.
25. Diamond TH, Champion B, Clark WA. Man-
agement of acute osteoporotic vertebral
fractures: a nonrandomized trial comparing
percutaneous vertebroplasty with conserva-
tive therapy. Am J Med 2003;114:257–265.
Radiology: Volume 238: Number 1—January 2006