Efficacy and safety of balloon kyphoplasty compared with non-surgical care for vertebral compression fracture (free): a randomised controlled trial

cacy and safety of balloon kyphoplasty compared with
non-surgical care for vertebral compression fracture (FREE):
a randomised controlled trial
Douglas Wardlaw, Steven R Cummings, Jan Van Meirhaeghe, Leonard Bastian, John B Tillman, Jonas Ranstam, Richard Eastell, Peter Shabe, Karen Talmadge, Steven Boonen Lancet 2009; 373: 1016–24
Background Balloon kyphoplasty is a minimally invasive procedure for the treatment of painful vertebral fractures,
which is intended to reduce pain and improve quality of life. We assessed the effi
cacy and safety of the procedure.
February 25, 2009 Methods Adults with one to three acute vertebral fractures were eligible for enrolment in this randomised controlled
trial at 21 sites in eight countries. We randomly assigned 300 patients by a computer-generated sequence to receive
See Comment page 982
kyphoplasty treatment (n=149) or non-surgical care (n=151). The primary outcome was the diff erence in change from
Orthopaedic Department,
baseline to 1 month in the short-form (SF)-36 physical component summary (PCS) score (scale 0–100) between the
Woodend Hospital, NHS
Grampian, Aberdeen, UK
kyphoplasty and control groups. Quality of life and other effi
cacy measurements and safety were assessed up to
(Prof D Wardlaw FRCSEd);
12 months. Analysis was by intention to treat. This trial is registered with ClinicalTrials.gov, number NCT00211211.
San Francisco Coordinating
Center, California Pacifi c
Findings 138 participants in the kyphoplasty group and 128 controls completed follow-up at 1 month. By use of
Medical Center Research
Institute, and Department of
repeated measures mixed eff ects modelling, all 300 randomised participants were included in the analysis. Mean
Epidemiology and
SF-36 PCS score improved by 7·2 points (95% CI 5·7–8·8), from 26·0 at baseline to 33·4 at 1 month, in the
Biostatistics, University of
kyphoplasty group, and by 2·0 points (0·4–3·6), from 25·5 to 27·4, in the non-surgical group (diff erence between
California, San Francisco, CA,
groups 5·2 points, 2·9−7·4; p<0·0001). The frequency of adverse events did not diff er between groups. There were
USA (Prof S R Cummings MD);
Dienst Orthopedie en
two serious adverse events related to kyphoplasty (haematoma and urinary tract infection); other serious adverse
Traumatologie, Algemeen
events (such as myocardial infarction and pulmonary embolism) did not occur perioperatively and were not related to
Ziekenhuis St Jan, Brugge,
Belgium (J Van Meirhaeghe MD);
Klinikum Leverkusen,
Leverkusen, Germany
Interpretation Our fi ndings suggest that balloon kyphoplasty is an eff ective and safe procedure for patients with acute
(Prof L Bastian MD); Medtronic
vertebral fractures and will help to inform decisions regarding its use as an early treatment option.
Spine LLC, Sunnyvale, CA, USA
(J B Tillman PhD, Funding Medtronic Spine LLC.
K Talmadge PhD); Swedish
National Competence Centre
for Musculoskeletal Disorders
body. The procedure can be done under general at Lund University Hospital,
Every year an estimated 1·4 million vertebral compression anaesthesia or conscious sedation, either as a day case, or Lund, Sweden (J Ranstam PhD);
fractures that cause pain and disability and diminish with an overnight stay, dependent on medical need. Academic Unit of Bone
Metabolism, University of
quality of life1,2 come to clinical attention worldwide.3 Although investigators have reported reduced pain and eld, Sheffi
Despite non-surgical management, including analgesia, improved function after kyphoplasty treatment,9–11 there (Prof R Eastell MD); Advanced
bed rest, physiotherapy, and back bracing, pain sometimes are no data from randomised trials assessing its effi Research Associates Inc,
Mountain View, CA, USA
resolves slowly, and can persist.4 The resulting vertebral and safety. We compared the effi cacy and safety of (P Shabe MS); and Leuven
deformity can cause height loss, kyphosis, reduced kyphoplasty with non-surgical management for the University Division of Geriatric
pulmonary function, and mobility and balance treatment of acute vertebral compression fractures, to Medicine, Katholieke
impairment.4–6 Vertebral fracture is associated with an test the hypothesis that kyphoplasty would result in Universiteit Leuven, Leuven,
Belgium (Prof S Boonen MD)
increased risk of future fractures.7 increased improvement in quality of life.
Since conventional open surgery for vertebral fractures Correspondence to: Prof Douglas Wardlaw, NHS is associated with risks resulting from open reduction Methods
Grampian, Orthopaedic and internal fi xation, it is usually reserved for fractures Participants
Department, Woodend Hospital, that cause neurological impairment. Balloon kyphoplasty We undertook a randomised controlled trial (Fracture Aberdeen AB15 6ZQ, UK is a minimally invasive procedure that is intended to Reduction Evaluation [FREE] trial) at 21 sites in eight reduce pain, disability, and vertebral deformity by use of countries between February, 2003, and December, 2005. catheters with infl atable bone tamps placed inside the Patients were eligible for enrolment if they had one to aff ected vertebral body. Balloon infl ation compacts the three vertebral fractures from T5 through L5. At least cancellous bone and pushes the endplates apart, which one fracture needed to have oedema assessed by MRI might partly restore height and correct angular deformity.8 and at least one had to show a 15% loss of height or Once the balloons have been removed, the resulting void more; single fractures were to meet both these criteria. is fi lled with viscous bone cement to stabilise the vertebral Patients with fractures due to osteopenia arising from www.thelancet.com Vol 373 March 21, 2009
primary or secondary osteoporosis, multiple myeloma, aged over 65 years.13 Prespecifi ed secondary outcome or osteolytic metastatic tumours were included. Painful measures at 1, 3, 6, and 12 months after randomisation fractures were diagnosed by investigators; patients with were SF-36 subscales (scaled 0–100); the 0–1-point up to three contiguous or non-contiguous fractures at EuroQol-5D (EQ-5D) quality-of-life questionnaire;14 any level could be included in the study if these self-rated back pain (average pain in past week) on a 0 additional fractures also had MRI signal changes, (no pain) to 10 (worst pain imaginable) visual-analogue progressive height loss, or pseudoarthrosis. Participants scale;15 back function by the 0–24-point Roland–Morris also had to have a back pain score of 4 points or more scale;16 and restricted activity days and bed rest because on a 0–10 scale.
of back pain during the previous 14 days.17 Back pain and Patients were excluded if they were younger than analgesic use were also assessed 5–10 days after 21 years of age; had chronic fractures (estimated fracture randomisation (control group) or after surgery. All age more than 3 months), pedicle fracture, previous adverse events and serious adverse events, defi ned per vertebroplasty, neurological defi cit, radicular pain, ISO14155, were reported; investigators assessed whether spinal cord compression, or canal narrowing; were they were related to device or procedure. Investigators taking uninterruptible anticoagulation therapy; had informed both the study sponsor and local ethical allergies to kyphoplasty materials or contraindications committees/institutional review boards of any serious to MRI; had dementia or were unable to walk before adverse events or serious adverse device eff ects. fracture (walking aids were allowed); or if their vertebral Dan Jolivette (Medtronic Spine LLC, Sunnyvale, CA, fractures were from primary bone tumours, osteoblastic USA) reviewed (in conjunction with the late Olof Johnell) metastases, or high energy trauma.
all safety issues related to the trial; data were reviewed Participants gave written informed consent before on a 6-month basis throughout the trial. All adverse enrolment. The protocol and consent form were events were categorised by body system according to approved by local ethics committees.
Standing lateral spinal radiographs were taken at baseline, 3 months, and 12 months. Two radiologists We randomly assigned study participants to receive independently made semiquantitative19 and quantitative kyphoplasty treatment or non-surgical care. Computer- morphometric20 assessments at a central laboratory generated randomisation was stratifi ed by sex, aetiology, (Bio-Imaging Technologies Inc, Newtown, PA, USA). A current treatment with corticosteroids, and any new or worsening fracture was defi ned by consensus that bisphosphonate treatment within 12 months before deformity increased by 1 Genant grade or more.19 Genant enrolment. A permuted block randomisation (stratifi ed is a semiquantitative assessment that describes normal as indicated) was generated before the study start by vertebrae (grade 0) or mild (grade 1, 20–25%), moderate Advanced Research Associates, (Mountain View, CA, (grade 2, 25–40%), or severe (grade 3, more than 40%) USA), the statistical contract research organisation, by deformity in any vertebral vertical dimension. The readings use of SAS PROC PLAN.
from the two radiologists were highly concordant (κ=0·80). Kyphoplasty was done with introducer instruments, If semiquantitative readings diff ered, a decrease of 20% infl atable bone tamps, and polymethylmethacrylate bone and 4 mm in any vertical dimension2 or a third expert cement and delivery devices (manufactured by Medtronic resolved disagreements.21 Because bone cement is Spine LLC, Sunnyvale, CA, USA), by a percutaneous, radio-opaque, we could not blind treatment assignment. bilateral, transpedicular, or extrapedicular approach.12 Bone cement extravasation was assessed by investigators Most procedures were done under general anaesthesia. with intraoperative fl uoroscopy and postoperative Six patients had conscious or deep sedation with local radiographs and included vertebrae with any noticeable anaesthesia. cement outside the vertebral borders.
All participants received analgesics, bed rest, back braces, physiotherapy, rehabilitation programmes, and walking Statistical analysis
aids according to standard practices of participating From a pilot study of 42 patients randomised to
hospitals. Investigators referred participants for treatment kyphoplasty or non-surgical treatments, 75 patients per
with calcium and vitamin D supplements and antiresorptive
group provided 80% power with a two-sided α of 5% to or anabolic agents. Subsequent clinical fractures were to detect a 0·5 SD for the 1-month diff erence in change for be treated according to original assignment.
the SF-36 PCS score. To compensate for possible loss, The primary endpoint was the diff erence in change 300 patients were enrolled. At least 200 patients with from baseline to 1 month in the short-form (SF)-36 fractures caused by osteoporosis were needed to focus physical component summary (PCS) scale between the the study on osteoporosis-related fractures without kyphoplasty and control groups. The SF-36 PCS is a excluding other causes such as fractures related to validated global quality-of-life measure weighted on cancer.
physical abilities. Although scaled from 0 to 100, the Endpoints were analysed by intention to treat, including expected range is between 8 and 59 points for participants all data available from all (300) randomised patients. We www.thelancet.com Vol 373 March 21, 2009
used repeated measures analysis of variance with mixed models that assumed a compound symmetry covariance 1279 patients assessed for eligibility structure to undertake an analysis of the primary and secondary endpoints of the full analysis set, which contained unbalanced data (ie, some patients had missing 655 did not meet inclusion criteria data at some timepoints).22 Treatment, visit (ie, hospital 209 refused to pa rticipate 115 had other reasons appointment), and treatment by visit interaction were included in the model. Analyses of diff erence (between groups) in change from baseline scores included baseline 300 enrolled and randomised as a covariate. We used the t-test to compare the calculated means at every timepoint. Patient proportions (adverse events, drugs, baseline fractures) were compared by the stratifi ed Cochran-Mantel-Haenszel χ² balloon kyphoplasty non-surgical care test. All analyses included randomisation stratifi cation factors as covariates; no adjustments were made for 11 did not complete follow-up 23 did not complete follow-up multiple tests.23 1 protocol violation (dementia) 9 withdrew and underwent surgery Because not all vertebrae were readable by radiologists, 10 did not receive surgery 6 withdrew (2 had inadequate the frequency of new or worsening fractures, including source documents) 2 could not comply 3 could not comply worsening index fractures, was analysed in participants 1 died before receipt with data available for at least seven vertebrae (T5 to L5) 1 adverse event before receipt at baseline and 12 months; the proportion of patients 1 adverse event before receipt with subsequent fractures was tested with Fisher's Exact test. SAS version 8.0 was used for all statistical analyses. 138 completed follow-up 128 completed follow-up This trial is registered with ClinicalTrials.gov, number 4 did not complete follow-up 11 did not complete follow-up Role of the funding source
5 withdrew and underwent surgery 2 could not comply Medtronic Spine LLC (JBT, KT, Dan Jolivette) contributed 1 could not comply (inadequate to study design, data monitoring, and reporting of source documents) results, and paid for statistical analysis (Advanced 1 lost (inadequate source documents) Research Associates Inc, Mountain View, CA, USA). All authors had complete access to data and provided all 134 completed follow-up 117 completed follow-up analyses requested. An independent statistician (JR) received the entire data set and verifi ed the statistical analyses and the primary endpoint data by comparing a 3 did not complete follow-up 2 did not complete follow-up 10% random sample with case report forms. The publication committee (DW, SRC, JVM, LB, RE, JR, and 1 withdrew 1 lost (inadequate source SB), which did not include the sponsor, reviewed and approved the fi nal version and had fi nal responsibility for the decision to submit for publication.
131 completed follow-up 115 completed follow-up Results
300 patients were screened and randomly assigned to
7 did not complete follow-up 4 did not complete follow-up balloon kyphoplasty (n=149) or non-surgical care (n=151; fi gure 1). Table 1 shows the baseline characteristics of 1 lost (inadequate source the enrolled participants. Of the ten patients assigned to kyphoplasty treatment who did not have the procedure, one underwent vertebroplasty because the 124 completed follow-up 111 completed follow-up investigator judged that kyphoplasty was no longer feasible (fi gure 1). 14 patients assigned to the control group withdrew and underwent surgery; fi ve of these patients had 1-month assessments. There were no diff erences in baseline SF-36 PCS, EQ-5D, 149 included in ITT analysis 151 included in ITT analysis Roland–Morris, or back pain scores, or in the number of baseline fractures between those who completed 12-month follow-up and those who discontinued (data Figure 1: Trial profi le
Notes in parentheses are patients considered protocol violations.
not shown). There were also no diff erences in these www.thelancet.com Vol 373 March 21, 2009
measures between the 15 patients who underwent Kyphoplasty
surgery diff erent from that assigned and the participants who received assigned treatment (data not shown).
For the primary outcome measure, eight patients had missing baseline data, 39 had missing 1-month data (34 terminated and fi ve did not complete SF-36), and 75 had missing 12-month data (65 terminated and ten did Primary osteoporosis not complete SF-36). Two participants in each group had Secondary osteoporosis all data missing. Eleven patients included in the dataset violated the protocol (three kyphoplasty and six control Bisphosphonate use for patients with inadequate source documents, one Glucocorticoid use kyphoplasty patient had dementia, and one had an Baseline fractures unfractured vertebral body treated). All 300 randomised participants were included in the intention-to-treat analysis by use of mixed eff ects modelling that includes patients with incomplete follow-up data.
Fractures were a mean of 5·6 weeks (SD 4·4) old at Baseline fracture location† randomisation in the kyphoplasty group and 6·4 weeks Thoracic (T5–T9) (5·2) old in the non-surgical group. At baseline, of Thoracolumbar junction (T10–L2) 127 (59%) 338 index vertebrae that were available for Genant assessment, 236 (70%; kyphoplasty n=113, control n=123) Treated fractures per patient were grade 2 or more (more than 25% deformity) and 99 (29%; kyphoplasty n=49, control n=50) were grade 3 (more than 40% deformity). The mean time between randomisation and kyphoplasty procedure was 6·8 days (4·5) and 1-month assessments were made 28·8 days Data are n (%) or mean (SD). N/A=not applicable. Groups were similar at (8·9) after surgery. 1-month assessments were made a baseline with the exception of multiple fractures. *One patient had a fourth mean of 35·6 days (8·9) and 32·5 days (7·4) after index fracture identifi ed between screening and planned surgery. †Kyphoplasty randomisation for the kyphoplasty and control groups, N=214, control N=195 (ie, number of index fractures identifi ed at baseline). ‡Ten kyphoplasty patients did not receive surgery.
More patients in the control group received walking Table 1: Patient characteristics
aids, back braces, physical therapy, and analgesics during follow-up than did patients in the kyphoplasty group. (1·0–4·1; p=0·0012) more in the kyphoplasty group than Similar proportions of patients in each group received in the non-surgical group at 1 month and 12 months, treatments for osteoporosis (table 2).
respectively. Patients in the kyphoplasty group reported The improvement in mean SF-36 PCS score from 2·9 fewer days of restricted activity per 2 weeks (1·3–4·6; baseline to 1 month was 5·2 points (95% CI 2·9–7·4) p=0·0004) because of back pain at 1 month than did more in the kyphoplasty group than in the non-surgical controls, although the diff erence in improvement was no group (p<0·0001; fi gure 2). The mean diff erences in longer signifi cant at 12 months (1·6 days, –0·1 to 3·3; improvement between the groups were 4·0 points p=0·0678). A mean of 2·5 fewer days of restricted activity (1·6–6·3; p=0·0008), 3·2 points (0·9–5·6; p=0·0064), per 2 weeks (1·2–3·8; p<0·0001) was reported during the and 1·5 points (–0·8 to 3·9; p=0·208) at 3, 6, and year for patients in the kyphoplasty group than in 12 months, respectively. The SF-36 PCS score improved controls.
during the year by a mean of 3·5 points (1·6–5·4; Back pain score decreased by 2·2 points (1·6–2·8; p=0·0004 vs control) more in the kyphoplasty group than p<0·0001) more in the kyphoplasty group than in controls in controls. There was a signifi cant interaction between at 1 week and by 0·9 points (0·3–1·5; p=0·0034) after treatment and follow-up time (p=0·0104); this interaction 12 months (fi gure 2). The kyphoplasty group also had a suggests that the treatment eff ect over the year was not greater reduction than had controls in the percentage of uniform across follow-up because of an early patients needing narcotic analgesics between 1 month improvement in the kyphoplasty group.
and 6 months (fi gure 2).
Compared with controls, the kyphoplasty group had Several SF-36 subscale scores improved more in the greater improvements in quality of life as assessed by the kyphoplasty group than in the non-surgical group. EQ-5D questionnaire from baseline to 1 month (diff erence Averaged across 12 months, patients assigned to between groups 0·18 points, 0·08–0·28; p=0·0003) and kyphoplasty had greater improvements than controls for from baseline to 12 months (0·12 points, 0·01–0·22; body pain (diff erence between groups 9·2 points, 3·9–14·6; p=0·0252; fi gure 2). The Roland–Morris score improved p=0·0008), role physical (12·5, 4·8–20·2; p=0·0016), by 4·0 points (2·6–5·5; p<0·0001) and 2·6 points vitality (5·2, 0·2–10·1; p=0·0399), and social function www.thelancet.com Vol 373 March 21, 2009
Non-pharmacological medical therapies Miscellaneous aids* Bed rest (≥1 day per 14 days) Combination (non-opioid and opioid) Osteoporosis therapies ≥3-month bisphosphonate use ≥12-month bisphosphonate use Parathyroid hormone Data are n/N (%). Patients might have received more than one treatment. *For example, wheelchairs, hospital beds, transcutaneous electrical nerve stimulation, or crutches.
Table 2: Non-surgical treatments received in kyphoplasty and non-surgical care groups
(11·4, 4·0–18·9; p=0·0026). There was no signifi cant related or device-related myocardial infarc treatment–time interaction for these outcomes. For pulmonary embolisms, neurological injuries, or deaths. physical function, the interaction was signifi cant Two serious adverse events were attributed to (p=0·0381); the kyphoplasty treatment eff ect for physical kyphoplasty; a soft tissue haematoma at the surgical function was 9·3 points (2·4–16·2; p=0·0081) at 1 month site and a postoperative urinary tract infection that and 1·9 points (–5·2 to 9·0; p=0·603) at 12 months.
needed intervention. Cement extravasation occurred in During follow-up, 21 (14%) participants in the kypho- 51 (27%) of 188 vertebrae treated in 48 patients; all cases plasty group had new clinical vertebral fractures; nine were asymptomatic. Most were endplate or discal (6%) underwent additional kyphoplasty (six within leakages; there was one foraminal leakage and none to 3 months and three more within 6 months of initial the spinal canal and no cement embolisms.
treatment).
Plain radiographs were available for 143 (96%) and Discussion
140 (93%) participants at baseline, 130 (87%) and 107 (71%) This randomised controlled trial showed that in patients at 3 months, and 120 (81%) and 103 (68%) at 12 months in with acute, painful, vertebral fractures, balloon kyphoplasty the kyphoplasty and non-surgical groups, respectively. Of improved quality of life, function, mobility, and pain more these, 115 who underwent kyphoplasty and 95 controls had rapidly than did non-surgical management, with signifi cant data available for at least seven vertebrae at baseline and diff erences in improvement between the groups at 12 months. At 12 months, 38 of 115 (33%) patients in the 1 month. For most outcome measures, the diff erences kyphoplasty group and 24 of 95 (25%) in the non-surgical between kyphoplasty treatment and control were group had new or worsening radiographic vertebral diminished at 12 months because the non-surgical group fractures (7·7% diff erence, 95% CI –4·5 to 20·0; improved over time, probably as a result of fracture healing. There were two serious adverse events related to The overall frequencies of adverse events (table 3) did kyphoplasty treatment (haematoma and urinary tract not diff er between the kyphoplasty and control groups. infection); other serious adverse events (such as myocardial About a third of patients in each group had a serious infarction and pulmonary embolism) did not occur adverse event, none of which occurred periopera tively; perioperatively and were not related to procedure. fi ve kyphoplasty patients had myocardial infarc tion Kyphoplasty treatment did not result in a signifi cant and three had pulmonary embolism (earliest occurrence increase in new radiographic vertebral fractures at 1 year 46 days postoperative). Thus, there were no procedure- compared with controls.
www.thelancet.com Vol 373 March 21, 2009
Treatment p<0·0001 Treatment p=0·0009 Total EQ-5D score -36 physical component summary score Number of patients
with data available*
Kyphoplasty 143 136 131 127 119
144 136 131 127 120 Control 1 49 125 110 110 106 1 49 125 111 111 106 Treatment p<0·0001 Treatment p<0·0001 oland–Morris scoreR educed activity (days)R Number of patients
with data available*
Kyphoplasty 146 129 118 111 103
146 128 126 125 117 Control 1 49 124 107 109 101 1 42 118 107 109 105 Treatment p<0·0001 Back pain score 3 Number of patients
Follow-up (months) Follow-up (months) with data available*
Kyphoplasty 148 137 136 132 128 121
140 103 114 120 124 115 Control 1 50 137 128 114 113 105 1 46 131 115 106 112 101 Figure 2: Quality of life, disability, and back pain at baseline and after kyphoplasty treatment or non-surgical care
Group calculated means and 95% CIs are shown for balloon kyphoplasty (n=149) and non-surgical control (n=151) groups for (A) the short-form (SF)-36 physical
component summary scores (scale 0–100); (B) total EuroQol-5D (EQ-5D) scores (scale 0–1·0); (C) Roland–Morris scores (scale 0–24); (D) the number of days (within
past 2 weeks) patients reported spending greater than half the day in bed because of back pain; and (E) back pain (0–10 scale; 0=no pain). The proportions and
95% CIs of patients are shown for (F) patients taking opioid drugs to control back pain. In panels A and B, higher scores indicate improvement. For all other panels,
lower values indicate improvement. In panels A–E, the treatment p value refers to the average treatment eff ect diff erence during follow-up. The treatment by visit
p value relates to a time-related change of this diff erence. A signifi cant treatment by visit interaction indicates that the treatment eff ect diff erence is not constant
throughout the year. p values for each timepoint comparison are shown in panel F. *Numbers of patients with data available are shown to provide information about
the amount of missing data that exists for each measure.
www.thelancet.com Vol 373 March 21, 2009
diff erent methods used for estimation,25 estimated Kyphoplasty Control
minimally clinically important diff erences have ranged between 3·5 and 4·3 points for the SF-36 PCS,25 1·0 and Adverse events within 12 months 2·5 points for back pain,15,25 and between 2 and 3 points Withdrew because of adverse event for the Roland–Morris scale.16 The SF-36 physical function Serious adverse events* within 12 months and EQ-5D estimates are 15 and 0·08 points, respectively.26,27 At 1 month, the kyphoplasty treatment eff ect exceeded the smallest estimate for all measures Cardiovascular and vascular disorders apart from SF-36 physical function, and exceeded the Coronary heart disease largest estimate for EQ-5D, Roland–Morris, and SF-36 PCS. The eff ect of kyphoplasty treatment was higher Pulmonary embolism than the smallest estimate for EQ-5D and Roland–Morris scales throughout the 12 months of follow-up. For SF-36 PCS, the kyphoplasty treatment eff ect met the smallest estimate for 3 months and when averaged over 12 months. The eff ect of kyphoplasty on back pain exceeded the Clostridium infection smallest estimate up to 6 months. From reports of restricted activity days, we estimate that patients in the Urinary tract infection kyphoplasty group had about 60 fewer days of restricted activity during the year than had controls.
Nervous system disorders The results of this randomised trial are similar to those Psychiatric disorders of two small controlled but non-randomised studies that Respiratory disorders show that kyphoplasty treatment was associated with greater improvement in back pain and physical functioning than non-surgical management for at least Serious adverse events that resulted in death 6 months11 and 12 months.10 Vertebroplasty, an alternative treatment for vertebral fractures, consists of percutaneous needle placement into fractured vertebrae with infusion of bone cement. One small randomised trial showed that vertebroplasty reduced disability and improved quality of life at 2 weeks *An adverse event was serious if it resulted in death, life-threatening injury, or compared with non-surgical treatment,28 but crossovers permanent impairment, or if it required extended hospital stay or intervention to precluded long-term comparisons.
prevent impairment. †Patients might have had multiple serious adverse events (SAEs). MedDRA catergories listed are those where there was a procedure or Vertebral fractures change the biomechanics of the device-related SAE, any SAE occuring in 2% or more patients treated, or any SAE spine,29 which might increase the risk of additional where there was a diff erence between groups with a p value less than 0·2. ‡Event vertebral fractures.7 The eff ect of kyphoplasty treatment was judged to be related to kyphoplasty procedure.
on this risk has been unclear. Two non-randomised Table 3: Patients with adverse events in the kyphoplasty and
studies reported fewer subsequent fractures after non-surgical care groups
kyphoplasty than non-surgical care,10,11 whereas other uncontrolled studies have suggested that kyphoplasty Patients in this study had substantially reduced baseline might heighten the risk.30 In our study, the rate of quality of life compared with other patients with chronic subsequent fracture was numerically higher in the diseases.13 At 1 month, kyphoplasty treatment resulted in kyphoplasty group but was not signifi cantly diff erent mean SF-36 PCS scores (33 points) that were close to from that of controls, although it is worth noting that this Swedish age-matched normative values (scores of study was not powered to detect diff erences in fractures 39 points for women 70–74 years and 37 points for women between the two groups. Although most patients used more than 74 years); the score in the control group at bisphosphonates or other osteoporosis treatments, the 1 month was 27 points.24 Similar normative values are rate of new radiographic fracture during the following reported for the US population.13 year was high (about 30%), as has been seen in other Minimally clinically important diff erences are often patients with painful vertebral fractures.10,11 A probable used to assess the clinical signifi cance for outcome explanation for this high rate is that patients included in measures since they could be diff erent from statistical this study had symptomatic vertebral fractures, whereas diff erences. These diff erences have been estimated for in other reports, incident fractures were identifi ed several of the outcome measures used in this study. morphometrically at baseline.7 The high subsequent Although they were developed for degenerative spinal fracture rate underscores the importance of treatments conditions other than acute vertebral fractures and the specifi cally intended to reduce risk of future fractures in eff ects of kyphoplasty, and they might vary because of osteoporotic patients who qualify for kyphoplasty.
www.thelancet.com Vol 373 March 21, 2009
This study has several limitations. Because the and owns stock options in Medtronic Inc. JR is employed by the Swedish intervention was not blinded, we cannot rule out the National Competence Centre for musculoskeletal disorders at Lund possibility that knowledge of the treatment assignment University Hospital, Sweden, an organisation that has received compensation for work by Medtronic Spine LLC. RE has received might have determined patient responses to questions honoraria for consulting from Medtronic Spine LLC and has received or radiologist assessments of new vertebral fractures, research funding or consultation honoraria from Lilly, Pfi zer, Novartis, which could have contributed to the greater Amgen, Procter and Gamble, and Organon. PS is employed by and is an improvements seen in the kyphoplasty group. However, owner of Advanced Research Associates Inc, an organisation compensated for work by Medtronic Spine LLC. KT was employed by and other potential biases—eg, the high frequency of new owned stock and stock options in Kyphon Inc and is currently employed vertebral fractures (similar in both groups)—might have by Medtronic Spine LLC. SB received honoraria for consulting from decreased the apparent improvements in pain and Kyphon and Medtronic Spine LLC, received research funding or grant disability after kyphoplasty. Out of 300 patients support from Amgen, Eli Lilly, Kyphon, Medtronic Spine LLC, Merck, Novartis, Procter and Gamble, Sanofi -Aventis, Servier, and randomised, 235 (78%) completed the study with more Roche–GlaxoSmithKline.
patients in the control group discontinuing than in the Acknowledgments
kyphoplasty group; however, baseline pain, function, This study was sponsored and funded by Medtronic Spine LLC. We thank and quality of life were similar between those who the patients who consented to participate in the FREE trial and all completed 12-month follow-up and those who participating staff at the investigational centres. We wish to acknowledge and dedicate this manuscript to the late Olof Johnell, whose contributions Non-surgical treatment was not as the original principal investigator were invaluable. We thank M Nevitt standardised; for generalisability, every study centre was and P Lips for initial review of the data and manuscript and G Jiang for asked to provide non-surgical care consistent with local providing third readings of radiographic images. SB is senior clinical practices. We present 12-month data of a study with a investigator of the Fund for Scientifi c Research (FWO-Vlaanderen). DW and SRC are co-primary authors. JR, biostatistician at the Swedish planned minimum of 24 months follow-up.
National Competence Centre for musculoskeletal disorders at Lund Thus, compared with non-surgical management, University Hospital, Sweden, did the independent statistical review.
balloon kyphoplasty resulted in improvements in References
quality of life and disability measures and reduction of 1 Borgstrom F, Zethraeus N, Johnell O, et al. Costs and quality of life back pain in patients with acute painful vertebral associated with osteoporosis-related fractures in Sweden.
Osteoporos Int 2006; 17: 637–50.
fractures; however, diff erences in improvement between Silverman SL, Minshall ME, Shen W, Harper KD, Xie S. The kyphoplasty and non-surgical control groups diminished relationship of health-related quality of life to prevalent and incident by 1 year. These fi ndings will help to inform decisions vertebral fractures in postmenopausal women with osteoporosis: results from the Multiple Outcomes of Raloxifene Evaluation Study. about the use of balloon kyphoplasty as an early Arthritis Rheum 2001; 44: 2611–19.
treatment option for this patient population.
Johnell O, Kanis JA. An estimate of the worldwide prevalence and disability associated with osteoporotic fractures. Osteoporos Int 2006; DW, JVM, and LB contributed to data collection and interpretation. Lyritis GP, Mayasis B, Tsakalakos N, et al. The natural history of the SRC, JBT, JR, SB, RE, and PS contributed to data interpretation and osteoporotic vertebral fracture. Clin Rheumatol 1989; analysis. SRC and KT contributed to the study design. DW, SRC, JBT, 8 (suppl 2): 66–69.
JR, SB, and KT contributed to manuscript development. JVM, LB, RE, Schlaich C, Minne HW, Bruckner T, et al. Reduced pulmonary and PS contributed to manuscript revisions. KT contributed to the function in patients with spinal osteoporotic fractures. Osteoporos Int study interpretation. DW, SRC, JVM, LB, JR, SB, and RE made the 1998; 8: 261–67.
decision to submit the manuscript for publication. All authors Lyles KW, Gold DT, Shipp KM, Pieper CF, Martinez S, Mulhausen PL. approved the fi nal manuscript for submission.
Association of osteoporotic vertebral compression fractures with FREE study investigators
impaired functional status. Am J Med 1993; 94: 595–601.
Vienna, Austria—S Becker. Brugge, Belgium—J Van Meirhaeghe. Liège, Lindsay R, Silverman SL, Cooper C, et al. Risk of new vertebral Belgium—V Bex, L Collignon, F Sacré. Caen, France—H Huet, fracture in the year following a fracture. JAMA 2001; 285: 320–23.
C Marcelli. Lille, France—P Chastanet, B Cortet. Paris, Voggenreiter G. Balloon kyphoplasty is eff ective in deformity France—M Cohen-Solal, J Laredo. Toulouse, France—B Mazières, correction of osteoporotic vertebral compression fractures. Spine
2005; 30: 2806–12.
J Railhac. Berlin, Germany—D Felsenberg. Hannover, Germany—L Bastian, C Krettek. Heidelberg, Germany—C Kasperk, 9 Garfi n SR, Buckley RA, Ledlie J. Balloon kyphoplasty for symptomatic vertebral body compression fractures results in rapid, signifi cant, and P Meeder. Rosenheim, Germany—G Regel. Catania, Italy—T Russo. sustained improvements in back pain, function, and quality of life for Verona, Italy—P Bartolozzi. Tilburg, Netherlands—P N M Lohle. Falun, elderly patients. Spine 2006; 31: 2213–20.
Sweden—P Fritzell. Malmö, Sweden—R Hasserius, O Johnell, 10 Grafe IA, Da Fonseca K, Hillmeier J, et al. Reduction of pain and M Karlsson, A Ohlin. Stockholm, Sweden—G Ordeberg. Uppsala, fracture incidence after kyphoplasty: 1-year outcomes of a prospective Sweden—M Cornefjord, P Försth. Aberdeen, UK—F Smith, D Wardlaw. controlled trial of patients with primary osteoporosis. Osteoporos Int London, UK—K Lam, T Sabharwal. Oxford, UK—J Wass, D Wilson. 2005; 16: 2005–12.
Confl ict of interest statement
11 Komp M, Ruetten S, Godolias G. Minimally invasive therapy for DW has received honoraria for consulting from Medtronic Spine LLC and functionally unstable osteoporotic vertebral fracture by means of kyphoplasty: prospective comparative study of 19 surgically and Cryolife, and has received research funding from Medtronic Spine LLC, 17 conservatively treated patients. J Miner Stoff wechs 2004; Zimmer, Apatec, and Cryolife. SRC has received consulting and honoraria 11 (suppl 1): 13–15 (in German).
from Medtronic Spine LLC, and research support, consultation, or 12 Garfi n SR, Yuan HA, Reiley MA. New technologies in spine: honoraria from Lilly, Pfi zer, Novartis, Amgen, Merck, Procter and kyphoplasty and vertebroplasty for the treatment of painful Gamble, Zelos, Roche, GlaxoSmithKline, and Organon. JVM and LB have osteoporotic compression fractures. Spine 2001; 26: 1511–15.
received honoraria for consulting from Medtronic Spine LLC. JBT was 13 Ware J, Kosinski M, Keller S. SF-36 physical and mental health employed by and owned stock and stock options in Kyphon Inc (now summary scales: a user's manual. Boston, MA: Health Assessment Medtronic Spine LLC) and is currently employed by Medtronic Spine LLC Lab, New England Medical Center, 1994.
www.thelancet.com Vol 373 March 21, 2009
14 EuroQol Group. EQ-5D user guide. Rotterdam, Netherlands: 23 Rothman KJ. Statistics in nonrandomized studies. Epidemiology EuroQol Group, 1996.
1990; 1: 417–18.
15 Farrar JT, Young JP, LaMoreaux L, Werth JL, Poole RM. Clinical 24 Sullivan M, Karlsson J, Taft C. SF-36 Halsoenkat. Svensk manual importance of changes in chronic pain intensity measured on an och tolkningsguide (Swedish manual and interpretation guide). 11-point numerical pain rating scale. Pain 2001; 94: 149–58.
Goteborg: Sahlgrenska University Hospital, 2002.
16 Roland M, Fairbank J. The Roland–Morris disability questionnaire 25 Copay AG, Glassman SD, Subach BR, Berven S, Schuler TC, and the Oswestry disability questionnaire. Spine 2000; 25: 3115–24.
Carreon LY. Minimum clinically important diff erence in lumbar 17 Nevitt MC, Ettinger B, Black DM, et al. The association of spine surgery patients: a choice of methods using the Oswestry radiographically detected vertebral fractures with back pain and disability index, medical outcomes study questionnaire short form function: a prospective study. Ann Intern Med 1998; 128: 793–800.
36, and pain scales. Spine J 2008; 8: 968–74.
18 International Conference on Harmonisation of Technical 26 Walters SJ, Brazier JE. Comparison of the minimally important Requirements for Registration of Pharmaceuticals for Human Use. diff erence for two health state utility measures: EQ-5D and SF-6D. MedDRA data retrieval and presentation: points to consider. Qual Life Res 2005; 14: 1523–32.
ICH-endorsed guide for MedDRA users on data output. http:// 27 Davidson M, Keating JL, Eyres S. A low back-specifi c version of the www.ich.org/LOB/media/MEDIA2841.pdf (accessed Dec 17, 2008).
SF-36 physical functioning scale. Spine 2004; 29: 586–94.
19 Genant HK, Wu CY, van Kuijk C, Nevitt MC. Vertebral fracture 28 Voormolen MH, Mali WP, Lohle PN, et al. Percutaneous assessment using a semiquantitative technique. J Bone Miner Res vertebroplasty compared with optimal pain medication treatment: 1993; 8: 1137–48.
short-term clinical outcome of patients with subacute or chronic 20 Eastell R, Cedel SL, Wahner HW, Riggs BL, Melton LJ. painful osteoporotic vertebral compression fractures. The VERTOS Classifi cation of vertebral fractures. J Bone Miner Res 1991; study. AJNR Am J Neuroradiol 2007; 28: 555–60.
6: 207–15.
29 Briggs AM, Wrigley TV, van Dieen JH, et al. The eff ect of 21 Jiang G, Eastell R, Barrington NA, Ferrar L. Comparison of osteoporotic vertebral fracture on predicted spinal loads in vivo. methods for the visual identifi cation of prevalent vertebral fracture Eur Spine J 2006; 15: 1785–95.
in osteoporosis. Osteoporos Int 2004; 15: 887–96.
30 Fribourg D, Tang C, Sra P, Delamarter R, Bae H. Incidence of 22 Mallinckrodt CH, Clark WS, David SR. Accounting for dropout bias subsequent vertebral fracture after kyphoplasty. Spine 2004; using mixed-eff ects models. J Biopharm Stat 2001; 11: 9–21.
www.thelancet.com Vol 373 March 21, 2009

Source: https://www.unfallchirurgie.ch/sites/unfallchirurgie.ch/files/FREE_-_Lancet_1.pdf