Two-Year Results of a Randomized Placebo-Controlled Trial of Vertebroplasty for Acute Osteoporotic Vertebral Fractures

Two‐Year Results of a Randomized Placebo‐ControlledTrial of Vertebroplasty for Acute OsteoporoticVertebral FracturesFéline Kroon,1,2 Margaret Staples,1,3 Peter R Ebeling,4 John D Wark,5 Richard H Osborne,6 Peter J Mitchell,7

Chris HR Wriedt,8 and Rachelle Buchbinder1,3

1Monash Department of Clinical Epidemiology, Cabrini Institute, Victoria, Australia2Department of Rheumatology, Leiden University Medical Centre, Leiden, the Netherlands3Department of Epidemiology and Preventive Medicine, School of Public Health and Preventive Medicine, Monash University, Victoria, Australia4NorthWest Academic Centre, University of Melbourne, Western Health, Victoria, Australia5University of Melbourne Department of Medicine, and Bone and Mineral Medicine, Royal Melbourne Hospital, Victoria, Australia6Public Health Innovation, Population Health Strategic Research Centre, Deakin University, Victoria, Australia7University of Melbourne, Department of Radiology, Royal Melbourne Hospital, Victoria, Australia8I‐Med Network, Victoria, Australia

ABSTRACTWe previously reported the results of a randomized controlled trial that found no benefit of vertebroplasty over a sham procedure foracute osteoporotic vertebral fractures up to 6 months. We report here the 12‐month and 24‐month clinical outcomes of this trial.Eligible participants (n¼ 78) were randomly assigned to receive either vertebroplasty (n¼ 38) or a sham procedure (n¼ 40).Randomization was stratified by treatment center, sex, and symptom duration (<6 weeks or �6 weeks). Participants, investigators(except the treating radiologists), and outcome assessors were blinded to group assignments. Enrolment occurred betweenApril 2004 andOctober 2008with follow‐up completedOctober 2010. The primary outcomewas overall painmeasured on a scale of 0(no pain) to 10 (maximal imaginable pain). Secondary outcomes included pain at rest and at night, disability, quality of life, perceivedrecovery, and adverse events, including incident clinically apparent vertebral fractures. At 12 and 24 months, complete data wereavailable for 67 (86%) and 57 (73%) participants, respectively. At 12 months participants in the active group improved by 2.4� 2.7(mean� SD) units in overall pain comparedwith 1.9� 2.8 units in the shamgroup, adjusted between‐groupmean difference (MD) 0.3(95% confidence interval [CI], –0.9 to 1.5), whereas at 24 months participants in the active group had improved by 3.0� 3.1 unitscompared with 1.9� 3.0 units in the sham group, MD 1.1 (95% CI, –0.3 to 2.4). No significant between‐group differences wereobserved for any of the secondary efficacy outcomes at 12 or 24months. Therewere no between‐group differences in incident clinicalvertebral fractures up to 24 months (active: n¼ 14, sham: n¼ 13), although the study had inadequate power for this outcome. Theseresults provide further evidence that the use of this treatment in routine care is unsupported. © 2014 American Society for Bone andMineral Research.

KEY WORDS: CLINICAL TRIALS; OSTEOPOROSIS; INJURY/FRACTURE HEALING; THERAPEUTICS

Introduction

Vertebroplasty, the percutaneous injection of polymethylme-thacrylate (PMMA) into the affected vertebral body, was an

accepted treatment for painful osteoporotic vertebral fracturesfor a decade before the first randomized controlled trials wereperformed to determine its efficacy and safety.(1–4) Open trialsthat have compared vertebroplasty with usual care have showedmixed results.(1,2,5–7) However, the results of two randomizedplacebo‐controlled trials found no benefit of vertebroplasty over

a sham procedure up to 1 month in the INVEST trial(4) and up to6 months in an Australian‐based trial.(3) Furthermore, anindividual patient data meta‐analysis using data from thesetwo trials also failed to show an advantage of vertebroplasty overplacebo for participants with recent onset fracture (<6 weeks) orfor those in severe pain (�8 out of 10).(8)

In this work we report the preplanned 12‐month and 24‐month clinical outcomes from the Australian‐based randomizedplacebo‐controlled trial. We also report pain and functionoutcomes related to cement volume and cement leakage.

Received in original form October 14, 2013; revised form November 29, 2013; accepted December 5, 2013. Accepted manuscript online December 11, 2013.Address correspondence to: Rachelle Buchbinder, MBBS (Hons), MSc, PhD, FRACP, Monash Department of Clinical Epidemiology, Suite 41 Cabrini Medical Centre,183 Wattletree Rd, Malvern, Victoria, Australia 3144. E‐mail: [email protected]

CLINICAL TRIALS JJBMR

Journal of Bone and Mineral Research, Vol. 29, No. 6, June 2014, pp 1346–1355DOI: 10.1002/jbmr.2157© 2014 American Society for Bone and Mineral Research

1346

Patients and Methods

Study design

We performed a double‐blind, randomized, parallel‐group,placebo‐controlled trial of vertebroplasty for painful osteoporot-ic vertebral fractures to determine its efficacy in terms ofalleviating pain and improving physical functioning, and itssafety, particularly the excess risk of further fracture over 2 years.Follow‐up for 2 years was planned. The protocol,(9) and results upto 6months,(3) have been published. In brief, enrolment occurredbetween April 2004 and the end of October 2008, with follow‐upcompleted at the end of October 2010.

Participants

The recruitment methods have been described.(3,9) Participantswere recruited from hospital inpatient and emergency depart-ments as well as general practitioners and specialists. Inclusioncriteria were the presence of acute onset back pain of no morethan 12 months duration with one or two recent vertebralfractures, defined as vertebral collapse greater than or equal tograde 1 according to the grading system of Genant andcolleagues,(10) and edema, a fracture line, or both within thevertebral body on magnetic resonance imaging (MRI).(9)

Eligible participants were randomly assigned to undergoeither vertebroplasty or a sham procedure with the randomiza-tion stratified by treatment center, sex, and duration ofsymptoms (<6 weeks or �6 weeks). The participants, inves-tigators (except the radiologists performing the procedures), andoutcome assessors were blinded to the group assignments. Thehuman research ethics committee at each participating centerapproved the study, and all participants provided writteninformed consent.

Interventions

There were four participating sites, and experienced interven-tional radiologists performed all procedures according to astandardized protocol. All treating radiologists had undertakenformal training in vertebroplasty, had appropriate certification,and were actively performing the procedure. For percutaneousvertebroplasty, the left pedicle of the fracture site was identifiedwith the use of a metallic marker. A 25G needle was used toinfiltrate the skin overlying the pedicle, and a 23G needle wasused to infiltrate the periosteum of the posterior lamina. Anincision was made in the skin, and a 13G needle was placedposterolaterally relative to the eye of the pedicle. Gentle tappingguided the needle through the pedicle into the anterior twothirds of the fractured vertebral body. Anterior‐posterior andlateral images were recorded with the needle in the correctposition. Prepared PMMA (approximately 3mL) was slowlyinjected into the vertebral body, and satisfactory infiltration ofthe vertebral body was confirmed radiographically. A bipedicularapproach was used only if there was inadequate instillation ofcement with the unipedicular approach. Injection was stoppedwhen substantial resistance was met or when the cementreached the posterior quarter of the vertebral body; injectionwasalso stopped if cement leaked into the extraosseous structures orveins. The volume of cement injected and cement leakageduring the procedure were recorded. All participants in thevertebroplasty group received cephalothin, administered intra-venously immediately after PMMA injection.

Participants assigned to the sham intervention received thesame procedures as those in the vertebroplasty group up to theinsertion of the 13G needle to rest on the lamina. The centralsharp stylet was then replaced with a blunt stylet. To simulatevertebroplasty, the vertebral bodywas gently tapped, and PMMAwas prepared so that its smell permeated the room.

After the intervention, all participants received usual care.Treatment decisions were made at the discretion of the treatingphysician, who received up‐to‐date guidelines on the manage-ment of osteoporosis. Analgesia was given according to standardpractice, and its use was recorded.

Outcome assessment

Baseline data collected by a blinded assessor included sex, birthdate, height, weight, risk factors for osteoporosis, smokingstatus, alcohol use, medication use, history of fractures, bonemineral density (current or within the previous year), and thenature of the vertebral fractures. Participants were mailedquestionnaires at 1 week and 1, 3, 6, 12, and 24 months after theprocedure and all efforts were made to recover non‐returnedquestionnaires.

The primary outcome was the score for overall pain (over thecourse of the previous week) measured on a scale of 0 to 10(with 0 indicating no pain, 10 indicating the maximumimaginable pain, and 1.5 as the minimal clinically importantdifference).(11,12) Secondary outcomes included pain at rest andpain in bed at night, themodified 23‐item version of the Roland‐Morris Disability Questionnaire (RDQ, in which scores rangefrom 0 to 23, with higher numbers indicating worse physicalfunctioning, and 2 to 3 points representing the minimalclinically important difference).(13) Quality of life was measuredwith the Quality of Life Questionnaire of the EuropeanFoundation for Osteoporosis (QUALEFFO), a 41‐item verte-bral‐fracture‐specific and osteoporosis‐specific questionnaire(in which scores range from 0 to 100, with lower scoresindicating a better quality of life);(14) the Assessment of Qualityof Life (AQoL) questionnaire, a validated instrument that issensitive to changes in the frail elderly (scores range from 0 to 1,with 1 indicating perfect health and 0.06 representing theminimal clinically important difference);(15) and the EuropeanQuality of Life 5 Dimensions (EQ‐5D) scale (scores range from 0to 1, with 1 indicating perfect health and 0.074 representing theminimal clinically important difference).(16) At baseline, 12months, and 24 months, participants also did a Timed Up & Gotest, which measures the time required to rise from a standardarm chair, walk 3 meters, turn around, return to the chair, and sitdown again.(17)

Perceived recovery with respect to pain, fatigue, and overallhealth wasmeasured on seven‐point ordinal scales ranging from “agreat deal worse” to “a great deal better.” Responses of “moderatelybetter” or “a great deal better” were classified as successfuloutcomes. Adverse events, including incident clinical fractures, wereassessed at each time point with the use of open‐ended questions.Todetermine the incidenceof new radiologically apparent vertebralfractures, all participants underwent plain film examination of thethoracic and lumbosacral spine at 12 and 24 months. The results ofthe radiographic outcomes have been reported separately (StaplesM, Howe B, Ringler M, et al., unpublished data). Success ofblinding was assessed at the end of the study by asking theparticipants which treatment they thought they had. The responseoptions were “real treatment,” “placebo treatment,” and “unsure.”

Journal of Bone and Mineral Research RCT OF VERTEBROPLASTY FOR ACUTE OSTEOPOROTIC VERTEBRAL FRACTURES 1347

Sample size and statistical analysis

The primary end point was the score for overall pain at 3 months.A sample of 24 participants per group was required for 80%power to show at least a 2.5‐unit advantage of vertebroplastyover placebo with respect to pain (SD 3.0, two‐sided significancelevel 5%).

All analyses were performed according to the intention‐to‐treat principle. Baseline differences between the groups wereassessed with the use of parametric or nonparametric tests, asappropriate. We report here the changes from baseline to 12months and 24 months in pain and QUALEFFO, AQoL, EQ‐5D,RDQ scores, and the Timed Up & Go Test. Between‐groupchanges were compared using multiple linear regressionanalyses. Estimates of between‐groupmean differences adjustedfor baseline values and the stratification variables are presented,together with 95% confidence intervals (CIs). All results arepresented as improvements from baseline. Analyses wererepeated with missing values at 12 and 24 months imputedusing the multivariate normal multiple imputation methodsimplemented in Stata (Stata Corporation, Inc., Collage Station, TX,USA).

We compared measures of a perceived successful outcome(reports of feeling moderately better or a great deal betterversus no change or feeling worse) after vertebroplasty andafter the sham procedure by calculating the relative risks ateach time point, using log binomial regression.(18) We alsoperformed similar analyses among participants who reportedan improvement in pain or function that was greater than 2.5units or >30%. The latter was a post hoc analysis requested byreviewers of our previous results article.(3) All reported p valuesare two‐sided and have not been adjusted for multiple testing.Analyses were performed with the use of Stata software version11 (StataCorp).

Blinding success was assessed at 24 months using the JamesBlinding index, which takes a value of 1 for complete blindingand 0 for complete lack of blinding.(19) We also performed posthoc analyses to determine whether there was any relationshipbetween volume of cement injected and outcome or betweencement leakage and outcome. Volume injectedwas split into twogroups based upon the median volume injected (2.5mL) into ahigh‐volume and low‐volume group. Comparisons were madefor mean change in pain from baseline to each time point to 24months for high‐volume versus low‐volume and cement leakageversus no leakage.

Results

Of 468 participants assessed, 78 met inclusion criteria and wererandomly assigned to either vertebroplasty (n¼ 38) or theplacebo arm (n¼ 40). Figure 1 shows the flow of participantsthrough the study from assessment for eligibility through to the24‐month follow‐up. During follow‐up 7 participants in theplacebo arm and 5 in the vertebroplasty arm died; all wereassessed as being unrelated to the trial. In addition, 1 participantfrom each armwithdrewbefore the 24‐month assessment. At the12‐month assessment, complete data were available for 67 (86%)participants (33 [87%] from the vertebroplasty group and 34[85%] from the sham procedure group). At the 24‐monthassessment, complete data were available for 57 (73%)participants (29 [76%] from the vertebroplasty group and 28[70%] from the sham procedure group).

The baseline characteristics of the two treatment groups weresimilar.(3) Table 1 compares the baseline characteristics of thosewho completed the 24‐month follow‐up by treatment groupcompared with the 21 participants who failed to complete it.There were no important differences between completers andnoncompleters with respect to demographic characteristics;however, noncompleters tended to have greater pain anddisability and poorer quality of life. There were no between‐group differences in self‐reported bisphosphonate use at 12 or24 months (25/34 and 21/28 participants in the placebo groupand 23/33 and 21/29 participants in the vertebroplasty groupreported taking bisphosphonates at 12 and 24 months,respectively).

Similar to the results at 6 months, there were no significantbetween‐group differences for any primary or secondary efficacyoutcome at 12 and 24months (Table 2). Results shown are for theintention‐to‐treat analysis for participants with available data. Forthe primary outcome of overall pain, at 12months participants inthe active group improved by 2.4� 2.7 units (on a 0 to 10 scale)compared with 1.9� 2.8 units in the sham group, adjustedbetween‐group mean difference 0.3 (95% CI, –0.9 to 1.5),whereas at 24 months participants in the active group hadimproved by 3.0� 3.1 units compared with 1.9� 3.0 units in thesham group, adjusted between‐group mean difference 1.1 (95%CI, –0.3 to 2.4). Figure 2 displays the distribution of individualresults for overall change in pain from baseline to 12 and 24months, respectively. The analysis using imputed data gavesimilar results (data not shown). The analyses for participantswho reported an improvement in pain or function greater than2.5 units or >30% also showed no between‐group differences(Table 3).

At the end of 24 months follow‐up there had been 14 incidentclinical vertebral fractures in the vertebroplasty group and 13 inthe sham procedure group (Table 4). Fifty‐five participantscompleted the assessment of blinding question; 23 (42%)correctly guessed their group allocation; 22 (40%) were unsure;and 10 (18%) gave the wrong group (Table 5). The Blinding Indexwas 0.59 (95% CI, 0.47–0.71), indicating adequate blinding at theend of the study.

Median cement volumewas higher for those who experiencedleakage during the procedure (3.3mL versus 2.2mL; Wilcoxonrank sum test p¼ 0.001). There was an increased odds of cementleakage with higher volume (odds ratio [OR] 2.8; 95% CI, 1.3–6.1).There were no differences in any outcome measures forvertebroplasty participants receiving a low total volume ofcement or for those who experienced cement leakage during theprocedure (Table 6).

Discussion

We found no beneficial effects of vertebroplasty over a shamprocedure at 12 or 24 months among patients with painfulosteoporotic vertebral fractures. These results were consistentacross all primary and secondary efficacy outcomes, and are inkeeping with our short‐term trial results.(3) In addition there wereno between‐group differences in clinically apparent newvertebral fractures up to 24 months. We found no relationshipbetween treatment response and volume of cement injected,although there was an increased risk of cement leakage withhigher volumes of injected cement.

Whereas earlier controlled, before‐after studies,(20,21) andsome,(5,7) but not all open randomized controlled trials,(1,2)

1348 KROON ET AL. Journal of Bone and Mineral Research

Screened for eligibility (n = 468)

Excluded (n = 390) Did not meet inclusion criteria (n = 248)

No vertebral fracture (n = 114); fracture> 12 months or failed MRI criteria (n = 67); no significant pain (n = 24); MRI contraindication to vertebroplasty (n = 10); > 2 new fractures (n = 10); significant other health problems (n = 8); active malignancy (n = 6); dementia (n = 4); not correctable coagulation disorder (n =1); neurological complications (n = 1); significant trauma (n = 2); previous vertebroplasty (n = 1)

Refused to participate (n = 141) Died (n = 1)

1 month assessment (n=35)

3 did not return questionnaire



Vertebroplasty intervention (n=38: 26, 7, 2 and 3 at Centres 1-4

respectively)

Placebo intervention (n=40: 27, 8, 3 and 2 at Centres 1-4

respectively)

Randomisation (stratified by centre, gender and duration of symptoms <6 or ≥6 weeks)

Baseline assessment (n = 78)

1 week assessment (n=37)


1 week assessment (n=37)

3 did not return questionnaires

1 died (acute MI)





1 died (chest infection)





1 died (oesophageal cancer) 1 died (stroke); 1 withdrew (too ill);





12month assessment (n=33)




2 died (CCF, gallbladder infection)

4 died (2 natural causes,1 cancer; 1 unstated cause)

1 died (unstated cause) 1 withdrew (dementia), 1 died (natural causes)

Fig. 1. Flow of participants in the trial.


Table 1. Baseline Characteristics of the Trial Completers (by Group) Versus the Noncompleters

Vertebroplasty(n¼ 29)

Sham procedure(n¼ 28)

Noncompleters(n¼ 21) pa

Age in years, mean� SD 76.7� 9.4) 77.7� 9.2) 77.2� 9.2) 0.50Female, n (%) 23 (79) 24 (86) 15 (71) 0.29Duration of back pain (weeks), median (IQR) 9 (4–13) 9.5 (2.6–25) 9 (3–14) 0.45Duration of symptoms <6 weeks, n (%) 9 (31) 9 (32) 7 (33) 0.88Any medication for osteoporosis, n (%) 27 (93) 27 (96) 18 (86) 0.50Calcium supplements 19 (66) 19 (68) 7 (33) 0.001Vitamin D 13 (44) 14 (48) 19 (76) 0.06Bisphosphonates 25 (86) 25 (89) 13 (65) 0.02

Taking opioids for pain, n (%) 25 (86) 21 (75) 18 (86) 0.61Past or current glucocorticoids, n (%) 9 (31) 12 (43) 8 (38) 0.92Duration of glucocorticoid use (years), median (IQR) 2.5 (0.2–8.5) 1 (0.2–10) 9.5 (3.3–23.3) 0.04Previous fractures, n (%)One or more vertebral fractures 15 (52) 15 (54) 9 (43) 0.44

Hip 3 (10) 3 (11) 10 (47) 0.89Wrist 7 (14) 3 (18) 3 (14) 0.50Ribs 6 (21) 6 (21) 2 (10) 0.24Otherb 8 (n¼ 5) 10 (n¼ 7) 3 (n¼ 3) 0.50

T‐score for BMD <2.5 SD, n (%) (n¼ 28) (n¼ 21)Lumbar 18 (64) 6 (76) 8 (62) 0.59Femoral neck 11 (39) 11 (52) 6 (46) 0.94

Number of treated vertebral bodies, n (%)1 26 (90) 24 (86) 14 (67)2 3 (10) 4 (14) 7 (33)

Severity of fracture, n (%) (n¼ 32) (n¼ 32) (n¼ 26) 0.64Mild 8 (25) 8 (25) 9 (35)Moderate 16 (50) 16 (50) 11 (42)Severe 8 (25) 8 (25) 6 (23)

Fracture appearance, n (%) (n¼ 32) (n¼ 32) (n¼ 28) 0.94Biconcave 4 (13) 2 (6) 2 (7)Crush 3 (9) 8 (25) 5 (18)Wedge 25 (78) 22 (69) 21 (75)

Smoking status, n (%) 0.34Never 15 (52) 11 (39) 7 (33)Former 11 (38) 14 (50) 13 (62)Current 3 (10) 3 (11) 1 (5)

Alcohol use, n (%) 0.24Never 4 (14) 8 (29) 8 (38)Sometimes 13 (45) 13 (46) 9 (43)Daily 12 (41) 7 (25) 4 (19)

Pain score (0–10, 10¼worst pain), mean� SDOverall 7.0� 2.2 6.9� 2.3 8.0� 2.0 0.06In bed at night 4.3� 2.8 3.5� 3.2 5.3� 3.0 0.07At rest 4.3� 2.3 4.8� 2.8 5.4� 2.9 0.20

QUALEFFO (0–100, 100¼worst quality of life), mean� SDTotal score 53.1� 11.9 55.9� 16.9 66.5� 15.7 0.002

Pain 68.1� 16.6 71.3� 17.9 79.0� 13.9 0.03Physical function 47.2� 14.7 51.3� 22.5 60.7� 22.7 0.03Leisure, social activities 63.2� 22.0 68.5� 22.7 82.9� 19.0 0.003General health perception 74.1� 16.7 73.8� 16.9 90.9� 10.2 <0.001Mental function 41.9� 14.7 41.5� 15.9 51.5� 19.0 0.02

AQoL (–0.04 to 1.0, 1¼perfect health), mean� SD 0.38� 0.24 0.27� 0.27 0.21� 0.23 0.08RDQ (0–23, 23¼worst function), mean� SD 17.4� 2.8 (n¼ 22) 17.3� 3.9 (n¼ 20) 18.5� 2.1 0.04EQ‐5D (0–1, 1¼perfect health), mean� SD 0.42� 0.28 (n¼ 22) 0.31� 0.33 (n¼ 20) 0.10� 0.29 0.003Timed Up & Go test in seconds, mean� SD 20.6� 8.8 (n¼ 29) 21.4� 13.0 (n¼ 25) 25.8� 13.4 0.12

IQR¼ interquartile range; BMD¼bone mineral density; QUALEFFO¼Quality of Life Questionnaire of the European Foundation for Osteoporosis;AQoL¼Assessment of Quality of Life; RDQ¼ Roland‐Morris Disability Questionnaire; EQ‐5D¼ European Quality of Life 5 Dimensions scale.

aValue of p for difference between completers and noncompleters.bIncludes upper limb (n¼ 8), pelvis (n¼ 6), lower limb (n¼ 6), and jaw (n¼ 1).


Table

2.Outcomes

at12

Mon

thsan

d24

Mon

ths,According

toInterven

tionGroup

Outcomemeasure

12Mon

ths

24Mon

ths

Cha

ngein

verteb

roplasty

grou

p(n¼33

)a

Cha

ngein

sham

proced

ure

grou

p(n¼34

)a

Adjusted

betw

een‐grou

pmeandifferen

ce(95%

CI)b

Cha

ngein

verteb

roplasty

grou

p(n¼29

)a

Cha

ngein

sham

proced

ure

grou

p(n¼28

)a

Adjusted

betw

een‐grou

pmeandifferen

ce(95%

CI)b

Pain

scorec

Overall

2.4�2.7

1.9�2.8

0.3(–0.9to

1.5)

3.0�3.1

1.9�3.0

1.1(–0.3to

2.4)

Atrest

1.6�2.5

1.2�3.4

0.4(–0.8to

1.6)

1.7�2.6

1.6�3.1

0.5(–0.8to

1.7)

Inbe

dat

nigh

t2.1�3.1

0.7�3.1

0.7(–0.5to

1.9)

2.2�2.8

0.6�2.9

1.1(–0.1to

2.3)

QUALEFFO

totalscore

d6.7�12

.28.8�13

.3–1

.3(–7.0to

4.3)

5.9�10

.74.6�15

.02.1(–4.4to

8.5)

AQoL

scoree

0.1�0.3

0.2�0.3

0.1(0.0

to0.2)

0.1�0.3

0.1�0.3

–0.1

(–0.2to

0.1)

RDQ

scoref

2.0�5.7

2.6�6.9

–0.5

(–4.3to

3.2)

2.6�7.0

2.7�5.6

0.3(–3.5to

4.1)

EQ‐5Dscoreg

0.2�0.4

0.2�0.4

0.0(–0.2to

0.2)

0.2�0.4

0.2�0.4

0.0(–0.2to

0.2)

Timed

Up&Gotest

(secon

ds)

–2.6�12

.24.3�13

.4–0

.6(–6.3to

5.2)

3.5�17

.14.7�9.7

1.2(–4.3to

6.8)

Outcomemeasure

12Mon

ths

24Mon

ths

Verteb

roplasty

n(%

)Sh

amproced

ure

n(%

)Re

lativ

erisk

(95%

CI)i

Verteb

roplasty

n(%

)

Sham

proced

ure

n(%

)Re

lativ

erisk

(95%

CI)i

Perceivedpa

inh

Better

15(45)

15(44)

1.0(0.7

to1.9)

12(41)

10(36)

1.2(0.6

to2.2)

Nochan

ge15

(45)

17(50)

14(48)

12(43)

Worse

3(9)

2(6)

3(10)

6(21)

Values

aremean�SD

unless

othe

rwiseindicated.

CI¼

confi

denceinterval;Q

UALEFFO¼Qua

lityof

Life

Que

stionn

aire

oftheEu

rope

anFo

unda

tionforOsteo

porosis;AQoL

¼Assessm

entof

Qua

lityof

Life;R

DQ¼Ro

land

‐Morris

Disab

ility

Que

stionn

aire;

EQ‐5D¼Eu

rope

anQua

lityof

Life

5Dim

ension

sscale.

a Positive

values

forthechan

gescores

represen

tan

improv

emen

tfrom

baselin

e.bTh

ead

justed

betw

een‐grou

pdifferen

ceswerecalculated

usingmultip

lelin

earregression

adjusted

forstratifi

catio

nvaria

bles

andba

selin

evalues.P

ositive

values

favo

rtheverteb

roplasty

grou

p.c Painwas

assessed

onascaleof

0to

10,w

ithhigh

ernu

mbe

rsindicatin

gmorepa

inan

dwith

1.5as

theminim

alclinicallyim

portan

tdifferen

ce.

dScores

ontheQUALEFFOrang

efrom

0to

100,

with

high

erscores

indicatin

gworse

quality

oflife.

eScores

ontheAQoL

questio

nnaire

rang

efrom

–0.04to

1.0,

with

1indicatin

gpe

rfecthe

alth

and0.06

represen

tingtheminim

alclinicallyim

portan

tdifferen

ce.

f Scoreson

theRD

Qrang

efrom

0to

23,w

ithhigh

erscores

indicatin

gworse

physicalfunctio

ning

and2to

3po

intsrepresen

tingtheminim

alclinicallyim

portan

tdifferen

ce.The

values

werecalculated

ontheba

siso

f26

/24pa

rticipan

tsin

theverteb

roplasty

grou

pan

d22

/19in

thesham

proced

uregrou

pat

12/24mon

ths.

gScores

ontheEQ

‐5Dqu

estio

nnaire

rang

efrom

0to

1,with

1indicatin

gpe

rfecthe

alth

and0.07

4represen

tingtheminim

alclinicallyim

portan

tdifferen

ce.The

values

werecalculated

ontheba

sisof

26/24

participan

tsin

theverteb

roplasty

grou

pan

d21

/20in

thesham

proced

uregrou

pat

12/24mon

ths.

hPa

inwas

classified

as“better”ifthepa

rticipan

tind

icated

thatthepa

inwas

mod

eratelyor

agreatd

ealbettertha

nbe

fore

theinterven

tionan

das

“worse”ifthepa

inwas

repo

rted

tobe

mod

eratelyor

agreatd

eal

worse

than

before

theinterven

tion.

i The

relativ

eriskisforthecompa

rison

of“better”with

“nochan

ge”or

“worse”(w

ith“better”de

fine

dapriorias

beingasuccessful

outcom

e).


that have compared vertebroplasty to usual care have suggestedthat the treatment may be effective, they are likely to have beenliable to performance and detection bias. Lack of blinding in trialsthat primarily include subjective patient‐reported outcomessuch as pain and function are known to overestimate treatmentbenefit by about 25%,(22,23) an effect likely to be magnified withan invasive procedure.(24) Taking these factors into consider-ation, our results are entirely in keeping with the results of allprevious controlled trials and before‐after studies.

The major strength of our trial was the blinding of participantsand outcome assessors, which we confirmed by formalassessment of blinding success at the conclusion of the trial.In retrospect we perhaps could have assessed blinding successearlier, as participants undergoing subsequent spinal X‐rays mayhave become aware of their treatment at the time of that X‐ray.However, the results of the Blinding Index indicates that blindingwas adequate.

Publication of our initial results together with the results ofanother placebo‐controlled trial that reported similar findingshave generated considerable controversy,(25–31) although mostif not all of the proposed criticisms have been refuted.(8,32–38) Anunfounded criticism of our trial has been that it suffered fromselection bias because 30% of potentially eligible participantsdeclined to participate in the study. This type of bias, also calledsample selection bias, may undermine the external validity of astudy if those who choose to participate in a study are different

in a way that influences outcome, from those that do not.Although we cannot entirely exclude this type of bias, it isunlikely that the negative results of our study were due toinclusion of participants less likely to benefit from vertebro-plasty for several reasons. At baseline, included participants hadsignificant pain and disability, and, consistent with previouscontrolled studies, all participants were required to have boneedema and/or fracture line in the affected vertebrae onMRI, or apositive bone scan. As well, the generalizability of our findings isunlikely to be different from other controlled trials ofvertebroplasty or kyphoplasty, because they have experiencedsimilar rates of refusal,(1,39) and included participants had similarbaseline characteristics.(1,2,7,20,21,39) It is also unlikely that oursham intervention had any sustained treatment benefit becauseonly short‐acting local anesthetic was used, and the trajectory ofimprovement in the sham treatment group in our trial wassimilar to what has been observed in the control arms of openrandomized trials that have compared vertebroplasty to usualcare.(1,2,7)

It has also been suggested that the negative results of our trialare explicable on the basis of inadequate cement volume.(25)

However, both our trial and previous studies(40,41) have failed toconfirm an association between volume of cement andtreatment efficacy. In addition we found that cement leakagewas almost three times more likely with higher cementvolumes.

Although we did not find a between‐group difference inclinically apparent incident vertebral fractures, we had inade-quate power for this outcome. Randomized controlled trials ofvertebroplasty compared with usual care that have measuredthis outcome have reported conflicting results.(1,2,7,42) Voormo-len and colleagues(1) observed two adjacent fractures in thevertebroplasty group and none in the usual care group at 2weeks’ follow‐up. Rousing and colleagues(2) observed three newvertebral fractures (two of them adjacent) in the vertebroplastygroup and one nonadjacent new fracture in the usual care groupafter 3 months (RR for a new fracture, 2.9; 95% CI, 0.3–25.7),whereas Klazen and colleagues(42) reported no difference in newfractures in patients receiving either vertebroplasty or usual careafter a mean of 11.4 months. On the other hand, Blasco andcolleagues(7) reported a much higher risk of clinically apparentincident vertebral fractures over 1 year in those who receivedvertebroplasty compared to usual care (OR, 25.67; 95% CI, 3.0–216.8), with most occurring within 180 days of the procedure.They also reported an increased risk of new radiological vertebralfractures (OR, 2.78; 95%CI, 1.0–7.6). Ameta‐analysis of safety datafrom all available randomized trials may provide more preciseinformation, although in the absence of safety data from largeplacebo‐controlled trials with adequate follow up, the true risk offurther vertebral fractures may be inestimable.

In conclusion, this trial found no beneficial effect ofvertebroplasty over a sham procedure up to 24 months amongpatients with painful osteoporotic vertebral fractures. There wasno discernible increased risk of further clinically apparentvertebral fractures although our trial was underpowered forthis outcome. Longer‐term billing claims data that have beenanalyzed to take into account selection bias, have also failed todemonstrate any long‐term benefits in terms of reducedmortality or major medical complications while observing anincreased use of health care utilization.(43) Taken together withthe known potential risks of vertebroplasty, these findingsprovide further evidence that the use of this treatment in routinecare is unsupported.

Fig. 2. Distribution of individual results by treatment group for change inoverall pain from baseline to 12 and 24 months.


Table 4. New Clinical Fractures Over the 2‐Year Duration of the Trial by Treatment Group

Vertebroplasty Sham procedure

1 w 1 m 3 m 6 m 12 m 24 m Total 1 w 1 m 3 m 6 m 12 m 24 m Total

Vertebra 1 2 2 4 3 2 14 1 4 2 1 3 2 13Hip 1 1 0Ribs 1 1 2 4 1 2 3Pelvis 1 1 2 2Sternum 1 1 1 1Shoulder 1 1 1 1Waist 1 1 0Elbow 0 1 1

w¼week; m¼month(s).

Table 3. ParticipantsWho Improved by<2.5 Units or�2.5 Units and<30% or�30%With Respect to Overall Pain, Pain at Rest, Night Painand RDQ by Treatment Group

12 Months 24 Months

Vertebroplastyn (%)

Shamprocedure

n (%)Relative risk(95% CI)a

Vertebroplastyn (%)

Shamprocedure

n (%)Relative risk(95% CI)a

Overall pain<2.5 units 18 (55) 21 (62) 1.19 (0.67–2.10) 10 (34) 14 (50) 1.31 (0.83–2.07)�2.5 units 15 (45) 13 (38) 19 (66) 14 (50)

Pain at rest<2.5 units 21 (64) 23 (68) 1.12 (0.58–2.18) 16 (55) 16 (57) 1.05 (0.58–1.88)�2.5 units 12 (36) 11 (32) 13 (45) 12 (43)

Night pain<2.5 units 19 (58) 27 (791) 2.06 (0.95–4.45) 16 (55) 20 (71) 1.57 (0.77–3.20)�2.5 units 14 (42) 7 (21) 13 (46) 8 (20)

RDQ<2.5 units 9 (36) 12 (50) 1.28 (0.78–2.10) 10 (45) 8 (42) 0.94 (0.55–1.62)�2.5 units 16 (64) 12 (50) 12 (55) 11 (58)

Overall pain<30% 17 (52) 18 (53) 1.03 (0.62–1.70) 9 (31) 11 (39) 1.14 (0.77–1.67)�30% 16 (48) 16 (47) 20 (69) 17 (61)

Pain at rest<30% 13 (39) 16 (47) 1.14 (0.75–1.74) 8 (28) 11 (39) 1.19 (0.82–1.73)�30% 20 (61) 18 (53) 21 (74) 17 (61)

Night pain<30% 13 (39) 16 (47) 1.14 (0.75–1.74) 9 (31) 13 (46) 1.29 (0.84–1.96)�30% 20 (61) 18 (53) 20 (67) 15 (54)

RDQ<30% 18 (55) 15 (43) 0.80 (0.50–1.27) 15 (50) 11 (37) 0.79 (0.50–1.24)�30% 15 (45) 20 (57) 15 (50) 19 (63)

RDQ¼Roland‐Morris Disability Questionnaire; CI¼ confidence interval.aThe relative risk is for the comparison of “better” compared with “no change” or “worse” (with “better” defined a priori as being a successful outcome).

Table 5. Perceived Treatment by Treatment Group (Assessed at 24 Months)

Vertebroplasty Sham procedure Total

Believed they received vertebroplasty 16 8 24Believed they received placebo 2 7 9Unsure which treatment 11 11 22Total 29 26a 55

aTwo participants in the sham procedure group did not respond to this question.


Disclosures

RHO is the recipient of a National Health and Medical ResearchCouncil Population Health Career Development Award; RB is therecipient of a National Health and Medical Research CouncilPractitioner Fellowship. All other authors state that they have noconflicts of interest.

Acknowledgments

This work was supported by grants from the National Health andMedical Research Council of Australia (284354), Arthritis Australia,the Cabrini Education and Research Institute, and Cook Australia.None of the funders had any role in the design of the trial,the collection or analysis of the data, the preparation of themanuscript, or the decision to submit the manuscript forpublication. We thank Dr. David Connell, Ms. Lainie Wengier,Dr. James Burnes, Dr.Michelle Leech, Dr.WinstonChong,Dr. StuartLyon, and members of the radiology departments of CabriniHospital, Royal Melbourne Hospital, Monash Medical Centre, andthe Alfred Hospital for their participation in the trial.

Authors’ roles: Study design: RB, PRE, JDW, RHO, PJM, andCHRW. Study conduct: RB, PRE, JDW, PJM, and CHRW. Datacollection: RB. Data analysis: MS. Data interpretation: RB, MS, FK,PRE, JDW, RHO, PJM, and CHRW. Draftingmanuscript: FK, MS, andRB. Revisingmanuscript content: PRE, JDW, RHO, PJM, and CHRW.

Approving final version ofmanuscript: RB, MS, FK, PRE, JDW, RHO,PJM, and CHRW. RB takes responsibility for the integrity of thedata analysis.

References

1. Voormolen M, Mali W, Lohle P, et al. Percutaneous vertebroplastycompared with optimal pain medication treatment: short‐termclinical outcome of patients with subacute or chronic painfulosteoporotic vertebral compression fractures. The VERTOS Study.AJNR Am J Neuroradiol. 2007;28:555–60.

2. Rousing R, Andersen M, Jespersen S, Thomsen K, Lauritsen J.Percutaneous vertebroplasty compared to conservative treatment inpatients with painful acute or subacute osteoporotic vertebralfractures: three‐months follow‐up in a clinical randomized study.Spine. 2009;34:1349–54.

3. Buchbinder R, Osborne R, Ebeling P, et al. A randomized trial ofvertebroplasty for painful osteoporotic vertebral fractures. N Engl JMed. 2009;361:557–68.

4. Kallmes D, Comstock B, Heagerty P, et al. A randomized trial ofvertebroplasty for osteoporotic spinal fractures. N Engl J Med.2009;361:569–79.

5. Klazen C, Lohle P, de Vries J, et al. Vertebroplasty versus conservativetreatment in acute osteoporotic vertebral compression fractures(Vertos II): an open‐label randomised trial. Lancet. 2010 Sep25;376(9746):1085–92.

6. Farrokhi MR, Alibai E, Maghami Z. Randomized controlled trial ofpercutaneous vertebroplasty versus optimal medical management

Table 6. Mean� SD Change Scores From Baseline and Adjusted Between Group Differences for Main Outcomes Within theVertebroplasty Group for High/Low Cement Volume and No Leakage/Leakage Groups

High cement volume (>2.5mL) Low cement volume (�2.5mL)Between‐group meandifference (95% CI)bn Mean� SDa n Mean� SDa

12 monthsOverall pain 17 2.0� 2.4 16 2.8� 3.0 –0.8 (–2.7 to 1.2)Pain at rest 17 2.8� 3.1 16 1.4� 3.0 1.3 (–0.8 to 3.5)Night pain 17 1.6� 2.6 16 1.6� 2.5 0.0 (–1.8 to 1.8)Roland Morris 11 4.6� 4.2 14 3.1� 5.9 1.6 (–2.7 to 5.9)

24 monthsOverall pain 15 2.7� 2.8 14 3.3� 3.6 –0.6 (–3.0 to 1.9)Pain at rest 15 2.9� 2.3 14 1.4� 3.2 1.6 (–0.5 to 3.7)Night pain 15 2.1� 2.1 14 1.4� 3.0 0.7 (–1.3 to 2.7)Roland Morris 9 4.3� 5.8 13 4.2� 6.8 0.1 (–5.6 to 5.9)

Cement leakage No cement leakageBetween‐group meandifference (95% CI)bn Mean� SDa n Mean� SDaa

12 monthsOverall pain 13 2.0� 2.8 20 2.6� 2.6 –0.6 (–2.6 to 1.4)Pain at rest 13 2.7� 3.1 20 1.8� 3.0 0.9 (–1.3 to 3.2)Night pain 13 2.2� 2.8 20 1.2� 2.3 1.1 (–0.7 to 2.9)Roland Morrisc 8 4.8� 5.0 17 3.3� 5.3 1.5 (–3.1 to 6.1)

24 monthsOverall pain 12 2.8� 3.1 17 3.1� 3.3 –0.3 (–2.7 to 2.2)Pain at rest 12 2.9� 2.1 17 1.6� 3.2 1.3 (–0.9 to 3.4)Night pain 12 2.5� 2.0 17 1.2� 2.8 1.3 (–0.6 to 3.3)Roland Morrisc 7 4.6� 6.5 15 4.1� 6.3 0.4 (–5.6 to 6.5)

CI¼ confidence interval.aPositive values for mean change indicate improvements from baseline.bPositive values for between‐group differences favor the high‐volume/cement leakage groups, negative values favor the low‐volume/no cement

leakage groups.cModified, 23‐question Roland‐Morris Disability Questionnaire.


for the relief of pain and disability in acute osteoporotic vertebralcompression fractures. J Neurosurg Spine. 2011;14:561–9.

7. Blasco J, Martinez‐Ferrer A, Macho J, et al. Effect of vertebroplasty onpain relief, quality of life, and the incidence of new vertebral fractures:a 12‐month randomized follow‐up, controlled trial. J Bone Miner Res.2012;27:1159–66.

8. Staples M, Kallmes D, Comstock B, et al. Effectiveness ofvertebroplasty using individual patient data from two randomisedplacebo controlled trials: meta‐analysis. BMJ. 2011;342:d3952.

9. Buchbinder R, Osborne R, Ebeling P, et al. Efficacy and safety ofvertebroplasty for treatment of painful osteoporotic spinal fractures:a randomised controlled trial. BMC Musculoskel. 2008;9:156.

10. Genant HK, Wu CY, van Kuijk C, Nevitt MC. Vertebral fractureassessment using a semiquantitative technique. J Bone Miner Res.1993;8:1137–48.

11. Grotle M, Brox JI, Vøllestad NK. Concurrent comparison ofresponsiveness in pain and functional status measurements usedfor patients with low back pain. Spine. 2004;29:E492–501.

12. Huskisson E. Measurement of pain. J Rheumatol. 1982;9:768–9.

13. Trout AT, Kallmes DF, Gray LA, et al. Evaluation of vertebroplasty witha validated outcome measure: the Roland‐Morris Disability Ques-tionnaire. AJNR Am J Neuroradiol. 2005;26:2652–7.

14. Lips P, Cooper C, Agnusdei D, et al. Quality of life in patients withvertebral fractures: validation of the Quality of Life Questionnaire ofthe European Foundation for Osteoporosis (QUALEFFO). WorkingParty for Quality of Life of the European Foundation for Osteoporosis.Osteopor Int. 1999;10:150–60.

15. Hawthorne G, Osborne R. Population norms and meaningfuldifferences for the Assessment of Quality of Life (AQoL) measure.Aust N Z J Public Health. 2005;29:136–42.

16. Walters S, Brazier J. Comparison of the minimally importantdifference for two health state utility measures: EQ‐5D and SF‐6D.Qual Life Res. 2005;14:1523–32.

17. Podsiadlo D, Richardson S. The timed “Up & Go”: a test of basicfunctional mobility for frail elderly persons. J Am Geriatr Soc.1991;39:142–8.

18. McNutt L, Wu C, Xue X, Hafner J. Estimating the relative risk in cohortstudies and clinical trials of common outcomes. Am J Epidemiol.2003;157:940–3.

19. James D, Bloch D, Lee K, Kraemer H, Fuller R. An index for assessingblindness in a multi‐centre clinical trial: disulfiram for alcoholcessation: a VA cooperative study. Stat Med. 1996;15:1421–34.

20. Álvarez L, Alcarez M, Pérez‐Higueras A, et al. Percutaneousvertebroplasty: functional improvement in patients with osteoporot-ic compression fractures. Spine. 2006;31:1113–8.

21. Diamond TH, Champion B, Clark WA. Management of acuteosteoporotic vertebral fractures: a nonrandomized trial comparingpercutaneous vertebroplasty with conservative therapy. Am J Med.2003;114:257–65.

22. Psaty B, Prentice R. Minimizing bias in randomized trials. JAMA.2010;304:793–4.

23. Wood L, Egger M, Gluud L, et al. Empirical evidence of bias intreatment effect estimates in controlled trials with differentinterventions and outcomes: meta‐epidemiological study. BMJ.2008;36:601–5.

24. Moerman DE, Jonas WB. Deconstructing the placebo effect andfinding the meaning response. Ann Intern Med. 2002;136:471–6.

25. Baerlocher M, Munk P, Liu D, et al. Clinical utility of vertebroplasty:need for better evidence. Radiology. 2010;255:669–74.

26. Baerlocher M, Munk P, Liu D. Trials of vertebroplasty for vertebralfractures. N Engl J Med. 2009;361:2098.

27. Bono C, Heggeness M, Mick C, Resnick D, Watters W. Newly releasedvertebroplasty RCTs: a tale of two trials. Spine J. 2010;10:238–2340.

28. Clark W, Lyon S, Burnes J. Trials of vertebroplasty for vertebralfractures. N Engl J Med. 2009;361:2097–8.

29. Clark W, Diamond T, McNeil H, Gonski P, Schlaphoff G, Rouse J.Vertebroplasty for painful acute osteoporotic vertebral fractures:recent Medical Journal of Australia editorial is not relevant to thepatient group that we treat with vertebroplasty. Med J Aust.2010;192:334–7.

30. Gangi A, Clark W. Have recent vertebroplasty trials changed theindications for vertebroplasty? Cardiovasc Intervent Radiol.2010;33:677–80.

31. Carragee E. The vertebroplasty affair: the mysterious case of thedisappearing effect size. Spine J. 2010;10:191–2.

32. Buchbinder R, Kallmes D. Vertebroplasty: when randomized placebo‐controlled trial results clash with common belief. Spine J. 2010;10:241–3.

33. Buchbinder R, Osborne R, Kallmes D. Invited editorial presents anaccurate summary of the results of two randomised placebo‐controlled trials of vertebroplasty. Med J Aust. 2010;192:338–41.

34. Buchbinder R, Osborne R, Staples M, Response to Letters to the Editor[in reference to the following article: Buchbinder et al. A randomizedtrial of vertebroplasty for painful osteoporotic vertebral fractures. NEngl J Med. 2009;361:557–68. DOI: 10.1056/NEJMoa0900429]. NewEngl J Med. 2009; 361: 2099. Available from: http://www.nejm.org/doi/full/10.1056/NEJMoa0900429#t¼letters.

35. Kallmes D, Buchbinder R, Jarvik J, Heagerty P, Turner J, Osborne R.Response to “Randomized Vertebroplasty Trials: Bad News or ShamNews?”. AJNR Am J Neuroradiol. 2009;30:1809–10.

36. Kallmes D, Jarvik J, Osborne R, et al. Clinical utility of vertebroplasty:elevating the evidence. Radiology. 2010;255:675–80.

37. Kallmes D, Buchbinder R, Miller F. Viewpoint: Randomisedcontrolled trials using invasive control interventions should beincluded in Cochrane Reviews. Cochrane Database Syst Rev. 2011;8:ED000030.

38. Miller F, Kallmes D, Buchbinder R. Vertebroplasty and the placeboresponse. Radiology. 2011 Jun;259(3):621–5.

39. Wardlaw D, Cummings S, Van Meirhaeghe J, et al. Efficacy and safetyof balloon kyphoplasty comparedwith non‐surgical care for vertebralcompression fracture (FREE): a randomised controlled trial. Lancet.2009;373:1016–24.

40. Kaufmann TJ, Trout AT, Kallmes DF. The effects of cement volume onclinical outcomes of percutaneous vertebroplasty. AJNR Am JNeuroradiol. 2006;27:1933–7.

41. Al‐Ali F, Barrow T, Luke K. Vertebroplasty: what is important and whatis not. AJNR Am J Neuroradiol. 2009;30:1835–9.

42. Klazen C, Venmans A, de Vries J, et al. Percutaneous vertebroplasty isnot a risk factor for new osteoporotic compression fractures: resultsfrom VERTOS II. AJNR Am J Neuroradiol. 2010;31:1447–50.

43. McCullough B, Comstock B, Deyo R, KreuterW, Jarvik J. Majormedicaloutcomes with spinal augmentation vs conservative therapy. JAMAIntern Med. 2013;173:1514–21.


http://www.nejm.org/doi/full/10.1056/NEJMoa0900429#t&x003D;letters



Documents

Two-Year Results of a Randomized Placebo-Controlled Trial of Vertebroplasty for Acute Osteoporotic Vertebral Fractures