SEARCH

SEARCH BY CITATION

Keywords:

  • CEMENT AUGMENTATION;
  • VERTEBRAL COMPRESSION FRACTURE;
  • META-ANALYSIS;
  • VERTEBROPLASTY;
  • KYPHOPLASTY

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

Cement augmentation is a controversial treatment for painful vertebral compression fractures (VCF). Our research questions for the meta-analysis were: Is there a clinical and statistical difference in pain relief, functional improvement, and quality of life between conservative care and cement augmentation for VCF and, if so, are they maintained at longer time points? We conducted a search of MEDLINE from January 1980 to July 2011 using PubMed, Cochrane Database of Systematic Reviews and Controlled Trials, CINAHL, and EMBASE. Searches were performed from Medical Subject Headings. Terms “vertebroplasty” and “compression fracture” were used. The outcome variables of pain, functional measures, health-related quality of life (HRQOL), and new fracture risk were analyzed. A random effects model was chosen. Continuous variables were calculated using the standardized mean difference comparing improvement from baseline of the experimental group with the control group. New vertebral fracture risk was calculated using log odds ratio. Six studies met the criteria. The pain visual analog scale (VAS) mean difference was 0.73 (confidence interval [CI] 0.35, 1.10) for early (<12 weeks) and 0.58 (CI 0.19, 0.97) for late time points (6 to 12 months), favoring vertebroplasty (p < 0.001). The functional outcomes at early and late time points were statistically significant with 1.08 (CI 0.33, 1.82) and 1.16 (CI 0.14, 2.18), respectively. The HRQOL showed superior results of vertebroplasty compared with conservative care at early and late time points of 0.39 (CI 0.16, 0.62) and 0.33 (CI 0.16, 0.51), respectively. Secondary fractures were not statistically different between the groups, 0.065 (CI −0.57, 0.70). This meta-analysis showed greater pain relief, functional recovery, and health-related quality of life with cement augmentation compared with controls. Cement augmentation results were significant in the early (<12 weeks) and the late time points (6 to 12 months). This meta-analysis provides strong evidence in favor of cement augmentation in the treatment of symptomatic VCF fractures. © 2013 American Society for Bone and Mineral Research


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

Vertebroplasty is the percutaneous injection of acrylic cement into the vertebral body for the treatment of painful compression fractures. A similar procedure is kyphoplasty, where, before cement augmentation, vertebral body reduction is obtained using a balloon or other mechanical device. Vertebroplasty is now widely used for management of painful osteoporotic fractures and in fractures secondary to neoplasm. Publication of two randomized controlled trials (RCT) in 2009 demonstrating limited effectiveness of vertebroplasty over sham treatment led to restrictions of its use in North America.1, 2 Recent guidelines from the American Academy of Orthopedic Surgery based on an evidence-based approach recommended “against vertebroplasty for patients who present with an osteoporotic spinal compression fracture …” However, since the publication of those RCTs and guidelines, several other trials have been published with contrary conclusions.2–5

Although the authors of the current study rarely perform vertebroplasty or kyphoplasty, we have noted that in select severely disabled patients, excellent pain relief and rapid return to function result after vertebroplasty. Juxtaposing these observations to the results of RCTs published in 2009 prompted our interest in performing an unbiased systematic review and meta-analysis. The research questions were: Is there a clinical difference in pain relief, functional improvement, and overall quality of life between conservative care and cement augmentation for painful osteoporotic compression fractures? In addition, are the results maintained over long time points?

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

Selection criteria

Articles of only randomized control trials comparing either vertebroplasty or kyphoplasty to conservative or sham treatment for osteoporotic compression fractures were identified and reviewed. Further inclusion criteria included use of a validated outcome measure. Studies that investigated treatment of compression fractures as a result of neoplasm were excluded. Two reviewers independently reviewed the abstracts and full text of articles to determine eligibility based on these criteria. If a consensus could not be reached, a third reviewer was used to resolve the disagreement.

Search strategy

Two independent reviewers conducted a computerized search of MEDLINE using PubMed, Cochrane Database of Systematic Reviews, Cochrane Central Register of Controlled Trials, CINAHL, and EMBASE (Fig. 1). The electronic databases were searched from January 1980 to July 2011. Searches were performed from Medical Subject Headings (MeSH) used by the National Library of Medicine. Specifically, MeSH terms “vertebroplasty” and “compression fracture” were used in each database. Based on the National Library of Medicine, the term vertebroplasty includes “kyphoplasty” and “balloon vertebroplasty” in the MeSH tree structure. In addition, key word approach used “vertebroplasty AND compression fracture AND randomized control trial.” In PubMed, Clinical Queries filter was used on “therapy” category and “narrow” scope. Furthermore, a filter for “randomized control trials” was also separately searched using above key words in PubMed. No restrictions were imposed on language. Reference lists were also reviewed for any additional studies not identified by the search.

thumbnail image

Figure 1. Literature search methodology.

Download figure to PowerPoint

Quality assessment

Eight articles met criteria as being Level I or Level II studies based upon the Levels of Evidence for Primary Research as adopted by the North American Spine Society.6 Six of these articles represent unique studies, whereas two articles describe the same study at two different time points. Data were entered into Review Manager (RevMan) software used by Cochrane reviews and bias determined using Cochrane Risk of Bias table.7 In addition, the quality of this meta-analysis was assessed using the PRISMA checklist, which is a 27-point set of standards for publication of systematic reviews of randomized trials.8

Outcome variables

The outcomes of interest were pain, spine-specific function, and health-related quality of life (HRQOL). Further, the primary outcome of each study and time point was assessed. To evaluate adverse events, new fracture risk was analyzed. Other complications such as neurologic injury, embolism of cement, or infection were either rare or not systematically reported and were not amenable to meta-analysis but were recorded in evidentiary tables.

The time points vary among studies. To allow quantitative analysis and determination of time-dependent effects, the outcomes were analyzed as early (≤12 weeks) and late (≥26 weeks). In cases where multiple time points were assessed, we utilized those closest to 8 weeks for the early and 52 weeks for the late groups.

Meta-analysis methods

The studies were assumed to be heterogeneous and, therefore, a priori a random effects model was chosen. This was supported by large Cochrane Q values in all analyses. For continuous variables, the standardized mean difference (Hedges's g) for each study was calculated by comparing improvement from baseline of the experimental group to the control group using the “intention to treat” analysis when reported. The new vertebral fracture risk for each treatment was calculated by comparing rates in each group using log odds ratio. To calculate the variance between studies, the DerSimonian and Laird method was used. Pooling of data was performed with Comprehensive Meta Analysis, version 2.2.050 (Biostat, Englewood, NJ, USA). A p value of 0.05 was chosen as being statistically significant. All confidence intervals (CI) were reported at 95% levels.

Sensitivity analysis was performed to detect bias by sequential removal of each study from the analysis. Further, the correlation between preoperative and follow-up outcomes was adjusted, ranging from 0.1 to 0.9. Publication bias was assessed by visual inspection of funnel plots. If asymmetric, the classic and Orwin's fail-safe n were determined. The classic fail-safe n determines the number of null studies needed to change the p value to less 0.05. The Orwin's fail safe n determines the number of studies to change the standardized mean difference to a trivial level, which we defined as 0.1.

In one study, the error was not reported at time points of interest and was estimated by imputation using error reported from other studies. In these cases, sensitivity analysis was done using 50% greater and lesser error and repeating the analyses.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

Systematic review

A total of eight prospective randomized control trials met inclusion criteria after a complete systematic review was performed (Fig. 1). However, one study was excluded from outcome analysis because baseline data was missing but was included in new fracture risk assessment. Another study was an update of another investigation, leaving a total of six RCTs for meta-analysis.9 As a result, a total of six unique RCTs were ultimately used in our meta-analysis. The objectives, number of patients, methods, results, adverse events, and authors' conclusions from all the studies were tabulated (Table 1).

Table 1. Results of Systematic Review
 Kallmes et al.2Buchbinder et al.1

Wardlaw et al.3

and Boonen et al.9

Voormolen et al.10Farrokhi et al.4Klazen et al.5Rousing et al.12
InterventionVertebroplastyVertebroplastyKyphoplastyVertebroplastyVertebroplastyVertebroplastyVertebroplasty
Conservative treatmentSham, injection of local anesthetic agent adjacent to vertebral bodySham, needle insertion, gentle tappingNonsurgical careOptimal pain medicationOptimal medical therapyOptimum pain treatmentConservative treatment
Patients (experimental/control)131 (68/63)78 (38/40)300 (149/151)34 (18/16)82 (40/42)202 (101/101)50 (26/24)
Inclusion criteria1–3 painful VCFs, levels T4–L5, occurred <12 months before, tried medical therapy, VAS at least 3, >50 years of age, no recent surgery (within 60 days)1–2 painful VCFs, occurred <12 months before, confirmed by MRI1–3 acute VCFs, levels T5–L5, confirmed by MRI, height loss, included patients with fracture resulting from osteopenia from osteopetrosis, myeloma, and osteolytic metastatic tumorsPainful VCF refractive to medical therapy, 6 weeks to <6 months, confirmed by plain films and MRI, >50 years of agePainful osteoporotic VCF of at least 4 weeks and <1 year, height loss on X-ray, MRI and physical examination confirmationAcute VCFs confirmed on radiograph, level T5 or lower, back pain for 6 weeks or less, focal tendernessPainful acute (<2 weeks) or subacute (2–8 weeks) VCF
Exclusion criteriaTumor or malignancy, pedicle fracture, or cord compression>2 VCFs, malignant disease, neurological complications, unstable VCFsYounger than 21 years, fracture age >3 months, pedicle fracture, cord compression, neurological deficit, primary bone tumors, osteoblastic metastasesInfection, cord compression, radicular pain, poor cardiopulmonary condition, untreatable coagulopathyUncorrected coagulopathy, infection, secondary osteoporosis, posterior wall defects of vertebral body, spinal cancer, traumatic fracture, neurological complicationsUntreatable coagulopathy, severe cardiopulmonary comorbidity, infection, malignant disease, radicular syndrome, cord compressionYounger than 65 years, infection in spine or skin, malignant disease, bone metabolic disease
Primary outcomeRDQ, pain VAS at 1 monthPain VAS at 3 monthsSF-36 PCS score at 1 monthPain VAS at 2 weeksPain VAS and functional QOL at 1 week, 2, 6, 12, 24, and 36 monthsPain VAS at 1 month and 1 yearPain VAS at 3 months
Secondary outcomesPFI, PBI, SOF-ADL, EQ-5D, SF-36QUALEFFO, AQoL, EQ-5D, RDQEQ-5D, VAS, RDQQUALEFFO, RDQ EQ-5D, QUALEFFO, RDQSF-36 (PCS and MCS), EQ-5D, Barthel index
ResultsSimilar improvement in pain-related disability in both groups; trend toward higher rate of pain improvement in vertebroplasty group but not significantNo beneficial effect during 6 months of follow-upImproved quality of life, function, mobility, and pain more rapidly and significantly in kyphoplasty groupBetter pain relief with vertebroplasty treatmentSignificant improvement in pain in vertebroplasty groupSubgroup of patients have significantly greater pain relief after vertebroplasty, sustained for at least 1 yearReduction in pain comparable in the 2 groups at 3 months despite immediate reduction in pain for patients after vertebroplasty
NotesPatients allowed to cross over after 1 month; 8 crossed over from tx group versus 27 from control at 3 monthsInterventional radiologists performed procedures Intention of study to have 1 year follow-up, changed study design during trial, allowed crossover at 2 weeks, post patients in study initially had trial of conservative treatment50% of patients had multiple VCFs, 10 patients in control requested to cross over at 1 year2 patients required atropine because of pain-induced vasovagal reaction, 1 patient developed an acute asthma exacerbation, one patient had cement migration toward lungs but remained asymptomatic41 females, enrollment from 2001 to 2008, no crossover, missing baseline data in 25% of patients
Adverse events1 patient in VP group had an injury to the thecal sac during the procedure; 1 patient in control group hospitalized overnight with tachycardia and rigors2 patients in VP group and 1 in placebo group died during follow-up (unrelated to procedure); 1 VP patient developed post-op wound infection with osteomyelitis 2 weeks post-procedure; 7 patients with incident VCF with 6 months (3 in VP and 4 placebo)21 (14%) of kyphoplasty patients within 12 months had new clinical VCF and 9 (6%) underwent additional KP; 58 (38.9%) of KP patients had serious adverse event within 12 months; 54 (35.7%) in control group had serious adverse event within 12 months2 patients treated by VP developed adjacent fracture within 2 weeks post-VP; no significant discussion of adverse events1 patient sustained epidural cement leakage requiring laminectomy for severe radiculopathy; 1 (2.6%) patient developed adjacent fracture in VP and in 6 (15.4%) patients in the OMT group2 patients developed vasovagal reaction; 1 patient developed acute asthma exacerbation; 1 patient with cement migration toward lung but asymptomatic; 15/91 (16%) patients in VP group developed new fractures; 21/85 (24%) developed new fractures in conservatively treated patientsNo adverse events except for asymptomatic cemental leaks; in the VP group there were 3 new fractures after 3 months (2 were adjacent fractures); in the conservative group there was 1 new fracture not adjacent to the initial fracture

Study design for the various RCTs' inclusion criteria selected varied. Kallmes and colleagues and Buchbinder and colleagues defined 1 to 3 painful vertebral compression fractures (VCFs) within 12 months.1, 2 Wardlaw and colleagues did not define the length of time by which the fractures were painful.3 Voormolen and colleagues defined painful fractures of 6 weeks to less than 6 months duration.10 Farrohkhi and colleagues reported a range of painful fractures from 4 weeks to 1 year duration.11 Klazen and colleagues defined a 6 weeks or less time point for painful fractures.5 Finally, Rousing and colleagues subdivided painful fractures to acute (less than 2 weeks) and subacute (2 to 8 weeks) time points.12

Study quality

All articles in the study were initially determined to be Level I evidence before NASS Guideline application. In the Kallmes and colleagues article, there was a downgrade from Level 1 to Level II evidence based upon inclusion criteria and subsequent high crossover.2 In a similar manner, Buchbinder and colleagues' article was downgraded to Level II evidence based upon concerns of inclusion criteria.1 The Voormolen and colleagues and Farrokhi and colleagues studies were both downgraded for lack of power analysis.4, 10 The Rousing study was eliminated from the efficacy analysis because baseline data were missing but was included in adverse events assessment.

Selection bias had low risk given the use of computerized random number generator and opaque sealed envelopes in all studies (Fig. 2). Performance and detection bias is unclear because of the nature of the intervention and the difficulty of constructing a double-blinded study. Only two studies used a sham (Kallmes and colleagues and Buchbinder and colleagues) procedure for comparison with the treatment group.1, 2 Attrition bias owing to incomplete outcome data (>10%) was high in two included studies (Klazen and colleagues and Wardlaw and colleagues).3, 5 Besides the Voormolen and colleagues study, the other studies had low reporting bias when a protocol was available to evaluate their study design.10 The Voormolen and colleagues and Buchbinder and colleagues studies changed their study design during the trial as they intended originally to have a one-year follow-up.1, 10 In addition, Buchbinder and colleagues and Kallmes and colleagues enrolled fewer patients than their power analysis originally required to reach significance.1, 2 The Farrokhi and colleagues study used a translator for some of the English-based patient outcome measures; thus, this contribution to the risk of bias remains unclear.4 The Wardlaw and colleagues study had less stringent inclusion criteria and was funded by the Medtronic Spine LLC.3

thumbnail image

Figure 2. Risk of bias summary: review of authors' judgments about each risk of bias item for each included study. Plus sign indicates low risk of bias, question mark indicates unclear risk of bias, and minus sign indicates high risk of bias.

Download figure to PowerPoint

The quality of this systematic review and meta-analysis was rated as 25 of 27 using PRISMA criteria.8

Primary outcome variable

Pain visual analog scale (VAS) was evaluated as the primary outcome in four studies; RDQ and SF-36 PCS in one each (Table 2). The time points were early, ranging 2 to 12 weeks. The standardized mean difference of the primary outcome variable was 0.62 (CI 0.22, 1.03) (Fig. 3A). This was statistically significant in favor of the vertebroplasty group (p = 0.0024).

Table 2. Overall Meta-Analysis Results at Early and Late Time Points
 Early (2–12 weeks)Late (≥26 weeks)
Hedges's g95% confidence intervalHedges's g95% confidence interval
Lower limitUpper limitLower limitUpper limit
  • *

    Denotes statistical significance.

Primary outcome0.62*0.221.03   
Pain0.73*0.351.100.58*0.190.97
Functional outcome1.08*0.331.821.16*0.142.18
Health-related quality of life0.39*0.160.620.33*0.160.51
thumbnail image

Figure 3. Meta-analysis results of primary outcome variable. (A) Forest plot of Hedges's g standardized mean difference for primary outcome variable. The summary result (gray diamond) is statistically significant. (B) Funnel plot showing symmetry about mean result. Two studies, one of each side, have larger or smaller effect sizes than predicted.

Download figure to PowerPoint

Sensitivity analysis showed no change in statistical results. The funnel plots showed symmetry about the standard error (Fig. 3B). The classic and Orwin's fail-safe n showed that 91 and 31 missing null studies, respectively, were required to change the p value to be nonsignificant.

Pain (VAS)

The pain VAS standardized mean difference was 0.73 (CI 0.35, 1.10) for early and 0.58 (CI 0.19, 0.97) for late time points, favoring vertebroplasty, both of which were significant (p < 0.001) (Table 2). All studies favored vertebroplasty with three being statistically significant (Fig. 4). Sensitivity analysis did not result in any statistical changes to the results. No publication bias was identified because the funnel plots for VAS were symmetric.

thumbnail image

Figure 4. Forest plot of pain. VAS evaluated at early time points shows significantly better improvement in the vertebroplasty group. The summary result (gray diamond) is statistically significant.

Download figure to PowerPoint

Comparison of VAS between early and late time points was not statistically significant. However, this analysis has significant biases because studies reporting multiple time points have increased weighting and the analysis assumes independence, which is unlikely. To overcome these biases, a synthetic variable, the mean standardized difference between early and late time points, was used as recommended by Borenstein and colleagues.13 This was calculated as the difference between the mean standardized difference of early and late groups.

Only the four studies that reported multiple time points were used in this analysis. The overall standardized mean difference of the synthetic variable was −0.38 (CI −0.55, −0.21), indicating that vertebroplasty effect over control was greater at earlier than later time points (Table 3). This was statistically significant. No statistical or significant clinical differences were noted when varying the correlation coefficient from r = 0.1 to r = 0.9.

Table 3. New Fractures After Enrollment
 Time point (weeks)ControlVertebroplasty
PatientsNew fractures (%)PatientsNew fractures (%)
Buchbinder et al.126354 (11.4)353 (8.6)
Wardlaw et al.3529524 (25.2)11538 (33.0)
Rousing et al.1212230232 (8.7)
Voormolen et al.102160182 (11.1)
Farrokhi et al.4104396 (15.4)372 (5.4)
Klazen et al.5488521 (24.7)9115 (16.5)
Overall 29355 (18.8)31962 (19.4)

Functional outcome

Spine-specific functional outcome was measured using a variety of instruments. A priori, we selected the Roland Morris Disability Questionaire (RDQ) or Oswestry (OSW). The standardized mean difference for early and late time points was 1.08 (CI 0.33, 1.82) and 1.16 (CI 0.14, 1.18), respectively (Table 2). These both were statistically significant results in favor of vertebroplasty (Fig. 5).

thumbnail image

Figure 5. Forest plot of Hedges's g standardized mean difference for spine-specific functional outcome. The summary result (gray diamond) is statistically significant. RQD = Roland Morris Questionnaire; OSW = Oswestry disability index.

Download figure to PowerPoint

Publication bias was identified. Farrohki and colleagues had a large effect size, 5.50, which was far greater than any other study. Elimination of the study reduced the effect size of both time points, but the results remained statistically significant. The classic and Orwin's fail-safe n were 131 and 29 missing studies, indicating that missing studies were unlikely to change statistical results.

Health-related quality of life (HRQOL)

The HRQOL was assessed using the Quality of Life Questionnaire of the European Foundation for Osteoporosis (QUALEFFO) and the EQ5-D. The QUALEFFO is used to evaluate patients with osteoporosis and vertebral body fractures. The EQ-5D is a standardized instrument to measure health outcomes.

The overall standardized mean difference was 0.39 (CI 0.16, 0.62) at early time points, which decreased to 0.33 (CI 0.16, 0.51) at late time points (Fig. 6). At both time points, the effect was statistically significant in favor of vertebroplasty. No publication bias was noted on funnel plots and sensitivity analysis did not change results.

thumbnail image

Figure 6. Forest plot of Hedges's g standardized mean difference for health-related quality of life. The summary result (gray diamond) is statistically significant.

Download figure to PowerPoint

Adverse events

Medical complications varied across studies. These included tachycardia, acute asthma exacerbation, vasovagal reactions, and asymptomatic cement leaks. Two more serious complications included a postoperative osteomyelitis after vertebroplasty and a laminectomy for severe radiculopathy secondary to cement leakage. No deaths occurred directly attributable to either conservative or cement augmentation in any of the studies. There was no statistically significant difference in medical adverse events in either conservative or cement augmentation groups.

Six trials reported the incidence of secondary fractures between two weeks and two years (Table 3). Three studies favored the control group (fewer fractures) and three favored the vertebroplasty group (Fig. 7). The standardized mean effect was 0.064 (CI −0.57, 0.70), which was not significant. Single elimination of a study did not change statistical results.

thumbnail image

Figure 7. Forest plot of log odds ratio of new fractures after enrollment. The summary result (gray diamond) is not statistically significant.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

This meta-analysis showed cement augmentation to result in greater pain relief, functional recovery, and health-related quality of life than nonoperative or sham treatment. The results favoring vertebroplasty were present for the defined primary outcome and were significant for both early (<12 weeks) and late time points (6 to 12 months). The mean standardized difference was large (0.6 for VAS), indicating that the vertebroplasty effect is robust. Further, all studies had improvements in all outcomes greater in the vertebroplasty group, four of which were statistically significant.

For patients, pain relief is probably the most important outcome preference. The benefits of vertebroplasty over conservative treatment were seen at the very early times of 1 and 2 weeks and remained significant at 1 year. However, the effect size significantly decreased over time. The weighted mean difference in VAS improvement between groups was greater than 3.0, which exceeds the minimum clinical important difference reported for VAS in spine patients.11

The only adverse event that was reported routinely was new fracture. Vertebroplasty has been thought to cause higher rates of new fractures secondary to increased stiffness of the cemented motion segment compared with osteoporotic bone. This was not observed in this meta-analysis, as three studies had higher and three lower rates of new fractures after vertebroplasty compared with controls and the mean difference was close to zero. Other complications as discussed earlier were rare with only one neurologic complication being reported in the vertebroplasty treatment arms. This required a laminectomy for severe radiculopathy. Other medical adverse events were minor and did not require significant hospitalization as a result.

Publication bias was not observed. The funnel plots were symmetric about the mean standardized difference but had outliers, Farrokhi and colleagues having a large effect and Kallmes and colleagues a smaller one. Single elimination of these studies did not have an effect on any outcome assessed. The classic and Orwin's fail-safe n analysis indicated that large numbers of null studies in far excess of what has been published would be required to change a significant to a nonsignificant result.

The overall statistical results were insensitive to variation in any of the potential variables. We reported the most conservative assumptions so that the true effect of vertebroplasty may be larger. We also were concerned that the study with follow-up at only 2 weeks might be biased, but it had similar standardized mean differences as the one other study reporting 1 week and the three studies reporting outcomes at 4 weeks.10 Elimination of this study did change any of the statistical analyses.

Overall, based on our quality analysis, the included trials represent well-constructed studies comparing percutaneous vertebral body cement augmentation to conservative treatment. All three authors of this article independently agreed to eliminate the RCT by Rousing and colleagues because it was missing about 25% of baseline data on their patient group.12 The quality of the systematic review was high, meeting 25 of 27 criteria using the PRISMA criteria.8 Despite areas of potential bias in the studies, their results can be used to begin to elucidate the role of vertebroplasty or kyphoplasty for vertebral body compression fractures.

A random effects model was used to account for significant heterogeneity in study design. Most relevant differences were inclusion criteria, control type, surgical technique, and time points. Two of the negative studies (Buchbinder and colleagues and Kallmes and colleagues) utilized sham controls where local anesthetic was injected into the facet joints and spine periosteum.1 This is a rigorous study design that minimizes placebo effect and blinds patients to treatment. However, administration of local anesthetics may have long-term pain relief benefits through complex neurophysiologic mechanisms. Bogduck and colleagues hypothesized that facet subluxation resulting from kyphotic vertebral body deformity can generate posterior element mediated pain and proved this in six patients who had successful pain relief by medial branch block with local anesthetic.14 Radiofrequency rhizotomy resulted in long-term pain relief. Mitra and colleagues reported two cases of successful pain relief in vertebral compression by steroid injection into the facet capsules.15 Thus, the sham controls might be an effective therapeutic treatment, thereby resulting in small effect sizes of the experimental treatments. These sham treatments are under further investigation for treatment of painful vertebral compression fractures.

The studies by Buchbinder and colleagues and Kallmes and colleagues had negative results, which may have been because of changes in study design during the trial, as well as low power.1, 2 The study design of both trials was reported before publication of the results.16, 17 Both studies enrolled fewer patients and had shorter follow-up periods than stated in their power analyses and were further weakened by loss to follow-up and, in Kallmes and colleagues' case, crossover patients.2 Thus, both studies had study design changes and ultimately may have lacked statistical power.

The intention-to-treat principle maintains the integrity of the randomization process and avoids selection bias. However, compliance to treatment groups becomes problematic when a high percentage of patients cross over (usually from control to experimental treatment groups). Study designs in this meta-analysis differed in whether they allowed crossovers. Kallmes and colleagues and Farrokhi and colleagues had 43% and 24% crossovers, respectively.2, 4 Voormolen and colleagues allowed crossovers to vertebroplasty after their 2-week end point, which 88% of control patients requested.10 When utilizing aggregate data in studies having high crossover rates, bias is introduced in favor of conservative treatments, as potentially poor outcomes were changed to good after crossover but are statistically counted in the control group. This could have been a major factor in the negative study by Kallmes and colleagues and reduced the positive results by Farrokhi and colleagues. In our meta-analysis, we honored the intention-to-treat principle and did not account for crossover. An important observation is that patients who cross over to vertebroplasty appeared to benefit.4, 10

Limitations in the interpretation of our meta-analysis conclusions relate to the observed heterogeneity. Obvious confounding variables are differences in inclusion-exclusion criteria, treatment methods of both groups, and endpoints. To account for this heterogeneity, a random effects model was chosen. The conservative treatment groups varied widely, two of which used sham surgery controls. All but one study used vertebroplasty, whereas the other treatment was kyphoplasty. Our intent was not to compare the two procedures but to compare with nonoperative treatment. Because results may have been biased by including kyphoplasty, we performed sensitivity analysis by excluding the kyphoplasty study. No difference in statistical results was observed. Thus, both vertebroplasty and the combined studies of the two techniques had significantly greater standardized means effects than nonoperative treatments. Two recent meta-analyses show minimal long-term differences between the two cement techniques, although in one analysis the more severe fractures did better with kyphoplasty at early time points, whereas the other showed better short-term relief with vertebroplasty.18, 19 Other differences, such as unilateral versus bilateral injections and injection volume, were not standardized and could have an influence on outcomes and adverse events, but sensitivity results suggest that the true effect was insensitive to individual study variables. Finally, the training and experience of those providing both operative and nonoperative care may affect outcomes but could not be assessed.

The time from fracture to treatment is an important factor that likely influences outcome and was not standardized. Studies that included only short times between onset of pain and randomization tended to have greater effects favoring vertebroplasty. Similarly, the diagnostic criteria for enrollment varied, and studies that used MRI edema as a criterion had larger effect sizes in favor of vertebroplasty. Despite including studies that had these two less favorable inclusion criteria, the pooled results were still significant.

Several systematic reviews and meta-analyses have been reported on vertebroplasty and kyphoplasty compared with nonoperative treatment. Staples and colleagues reported pooled results of the Buchbinder and colleagues and Kallmes and colleagues studies, designating this as a meta-analysis.20 They concluded, as would be expected from the studies selected, that cement augmentation was not effective over sham. This study was not a meta-analysis because a systematic literature review was not performed and meta-analysis statistical methods were not utilized. The Cochrane Musculoskeletal group registered a systematic review on vertebroplasty in 2009, but results thus far are not available.21 Taylor and colleagues reported a systematic review of cohort studies comparing nonoperative to cement augmentation.22 Similar to our study, pain improvement was significantly greater in the cement group at all time points from 3 to 12 months, although the magnitude was less over time. No differences were seen between kyphoplasty and vertebroplasty. Two other systematic reviews comparing balloon kyphoplasty to vertebroplasty also did not show significant differences between groups.23, 24 Lee and colleagues reported a meta-analysis of complications after vertebroplasty and kyphoplasty. In prospective studies, the medical complications, symptomatic cement leakage, and mortality all occurred in less that 1% of patients each.25 Asymptomatic cement leakage in osteoporotic fractures was seen as high as 21% and 7% for vertebroplasty and kyphoplasty, respectively. New fracture risk was similar to that observed in this meta-analysis, ranging 16% to 18%.

This meta-analysis provides strong evidence for the use of cement augmentation in the treatment of refractory, painful, osteoporotic compression fractures. The optimal time between fracture and treatment could not be evaluated in this analysis, although results appeared less significant in studies that included longer times. Significant adverse events are rare and new fracture rates are similar in vertebroplasty and nonoperatively treated patients. Patients who are treated nonoperatively but later request vertebroplasty benefit from vertebroplasty. Further research into the effect on outcome based on cement augmentation technique and the use of adjunctive treatments of the associated metabolic bone disease are needed.

Disclosures

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

PAA has served as consultant for Aesculap, Pioneer Surgical, and Medtronic; has received royalties from Stryker and Pioneer Surgical; and owns stock in Pioneer Surgical, SI Bone, Spartec, Expanding Orthopedics, and Titan Surgical. ABF and WLT state that they have no conflicts of interest.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

No funds were used for this study.

Authors' roles: Figures, data analysis, data interpretation, writing, editing, study design: PAA. Study design, data collection, data analysis, writing, data interpretation, editing, figures: ABF. Study design, data collection, data analysis, writing, data interpretation, editing: WLTJr.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information
  • 1
    Buchbinder R, Osborne RH, Ebeling PR, Wark JD, Mitchell P, Wriedt C, Graves A, Staples MP, Murphy B. A randomized trial of vertebroplasty for painful osteoporotic vertebral fractures. N Engl J Med. 2009;361(6):55768.
  • 2
    Kallmes DF, Comstock BA, Heagerty PJ, et al. A randomized trial of vertebroplasty for osteoporotic spinal fractures. N Engl J Med. 2009;361(6):56979.
  • 3
    Wardlaw D, Cummings SR, Van Meirhaeghe J, et al. Efficacy and safety of balloon kyphoplasty compared with non-surgical care for vertebral compression fracture (FREE): a randomised controlled trial. Lancet. 2009;373(9668):101624.
  • 4
    Farrokhi MR, Alibai E, Maghami Z. Randomized controlled trial of percutaneous vertebroplasty versus optimal medical management for the relief of pain and disability in acute osteoporotic vertebral compression fractures. J Neurosurg Spine. 2011;14(5):5619.
  • 5
    Klazen CA, Lohle PN, de Vries J, et al. Vertebroplasty versus conservative treatment in acute osteoporotic vertebral compression fractures (Vertos II): an open-label randomised trial. Lancet. 2010;376(9746):108592.
  • 6
    Bono CM, Ghiselli G, Gilbert TJ, et al. An evidence-based clinical guideline for the diagnosis and treatment of cervical radiculopathy from degenerative disorders. Spine J. 2011;11(1):6472.
  • 7
    Bhandari MF, Kulkarni AV, Tornetta P 3rd. Meta-analyses in orthopedic surgery. A systematic review of their methodologies. J Bone Joint Surg Am. 2001;83-A:1524.
  • 8
    Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. BMJ. 2009;339:b2535.
  • 9
    Boonen S, Van Meirhaeghe J, Bastian L, et al. Balloon kyphoplasty for the treatment of acute vertebral compression fractures: 2-year results from a randomized trial. J Bone Miner Res. 2011;26(7):162737.
  • 10
    Voormolen MH, Mali WP, Lohle PN, et al. Percutaneous vertebroplasty compared with optimal pain medication treatment: short-term clinical outcome of patients with subacute or chronic painful osteoporotic vertebral compression fractures. The VERTOS study. AJNR. 2007;28(3):55560.
  • 11
    Farrar JT, Young JP Jr, LaMoreaux L, Werth JL, Poole RM. Clinical importance of changes in chronic pain intensity measured on an 11-point numerical pain rating scale. Pain. 2001;94(2):14958.
  • 12
    Rousing R, Andersen MO, Jespersen SM, Thomsen K, Lauritsen J. Percutaneous vertebroplasty compared to conservative treatment in patients with painful acute or subacute osteoporotic vertebral fractures: three-months follow-up in a clinical randomized study. Spine. 2009;34(13):134954.
  • 13
    Borenstein M, Hedges LV, Higgins JPT, Rothstein HR. Introduction to meta-analysis. Philadelphia, PA: John Wiley; 2009.
  • 14
    Bogduk N, MacVicar J, Borowczyk J., The pain of vertebral compression fractures can arise in the posterior elements. Pain Med. 2010;11(11):166673.
  • 15
    Mitra R, Do H, Alamin T, Cheng I. Facet pain in thoracic compression fractures. Pain Med. 2010;11(11):16747.
  • 16
    Gray LA, Jarvik JG, Heagerty PJ, et al. Investigational Vertebroplasty Efficacy and Safety Trial (INVEST): a randomized controlled trial of percutaneous vertebroplasty. BMC Musculoskelet Disord. 2007;8:126.
  • 17
    Buchbinder R, Osborne RH, Ebeling PR, et al. Efficacy and safety of vertebroplasty for treatment of painful osteoporotic vertebral fractures: a randomised controlled trial. BMC Musculoskelet Disord. 2008;9:156.
  • 18
    Ma XL, Xing D, Ma JX, Xu WG, Wang J, Chen Y. Balloon kyphoplasty versus percutaneous vertebroplasty in treating osteoporotic vertebral compression fracture: grading the evidence through a systematic review and meta-analysis. Eur Spine J. 2012 July 26 [Epub ahead of print].
  • 19
    Han S, Wan S, Ning L, Tong Y, Zhang J, Fan S. Percutaneous vertebroplasty versus balloon kyphoplasty for treatment of osteoporotic vertebral compression fracture: a meta-analysis of randomised and non-randomised controlled trials. Int Orthop. 2011;35(9):134958.
  • 20
    Staples MP, Kallmes DF, Comstock BA, et al. Effectiveness of vertebroplasty using individual patient data from two randomised placebo controlled trials: meta-analysis. BMJ. 2011;343:d3952.
  • 21
    Lambert RGW, Golmohammadi K, Majumdar SR, Jones A, Buchbinder R, Dhillon SS, Owen R, Homik J, Kallmes DF, Siminoski K. Percutaneous vertebroplasty for osteoporotic vertebral compression fracture. (Protocol) Cochrane Database of Systematic Reviews. 2007;1: CD006349.
  • 22
    Taylor RS, Fritzell P, Taylor RJ. Balloon kyphoplasty in the management of vertebral compression fractures: an updated systematic review and meta-analysis. Eur. Spine J. 2007;16(8):10850.
  • 23
    Eck JC, Nachtigall D, Humphreys SC, Hodges SD. Comparison of vertebroplasty and balloon kyphoplasty for treatment of vertebral compression fractures: a meta-analysis of the literature. Spine J. 2008;8(3):48897.
  • 24
    Gill JB, Kuper M, Chin PC, Zhang Y, Schutt R Jr. Comparing pain reduction following kyphoplasty and vertebroplasty for osteoporotic vertebral compression fractures. Pain Physician. 2007;10(4):58390.
  • 25
    Lee MJ, Dumonski M, Cahill P, Stanley T, Park D, Singh K. Percutaneous treatment of vertebral compression fractures: a meta-analysis of complications. Spine. 2009;34(11):122832.

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
jbmr_1762_sm_chinese_translation.pdf621KSupporting Information

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.