Abstract

Background: Percutaneousvertebroplasty (PVP) and balloon kyphoplasty (BKP) are minimally invasive andeffective vertebral augmentation techniques for managing osteoporotic vertebralcompression fractures (OVCFs). Recent meta-analyses have compared the incidenceof secondary vertebral fractures between patients treated with vertebralaugmentation techniques or conservative treatment; however, the inclusions werenot thorough and rigorous enough, and the effects of each technique on the incidenceof secondary vertebral fractures remain unclear.

Purpose:To perform an updated systematic review and meta-analysis of the studies withmore rigorous inclusion criteria on the effects of vertebral augmentationtechniques and conservative treatment for OVCF on the incidence of secondaryvertebral fractures.

Material and Methods:PubMed, MEDLINE, EMBASE, SpringerLink, Web of Science and the Cochrane Librarydatabase were searched for relevant original articles comparing the incidenceof secondary vertebral fractures between vertebral augmentation techniques and conservativetreatment for patients with OVCFs. Randomized controlled trials (RCTs) and prospectivenon-randomized controlled trials (NRCTs) were identified. The methodologicalqualities of the studies were evaluated, relevant data were extracted andrecorded, and an appropriate meta-analysis was conducted.

Results:A total of 13 articles were included. The pooled results from included studies showedno statistically significant differences in the incidence of secondaryvertebral fractures between patients treated with vertebral augmentationtechniques and conservative treatment. Subgroup analysis comparing different studydesigns, durations of symptoms, follow-up times, races of patients and techniqueswere conducted, and no significant differences in the incidence of secondaryfractures were identified (p> 0.05). Noobvious publication bias was detected by either Begg’s test (p = 0.360 > 0.05) or Egger’s test (p= 0.373> 0.05).

Conclusion:Despite current thinking in the field that vertebral augmentation proceduresmay increase the incidence of secondary fractures, we found no differences in theincidence of secondary fractures between vertebral augmentation techniques andconservative treatment for patients with OVCFs.

Keywords:Vertebral augmentation technique· Percutaneous vertebroplasty·Balloon kyphoplasty· Secondary vertebral fracture· Cement leakage

Introduction
As the population continues to age, thenumber of people with low bone mass and osteoporosis has risen, leading toskeletal fractures and the associated high morbidity and mortality.Osteoporotic vertebral compression fractures (OVCFs) are among the most commonskeletal fractures, approximately 1.4 million new fractures occurring annually.These fractures are associated with pain and disability and increased mortality.The traditional management of OVCF has consisted of anti-osteoporotic medicaltreatment with a combination of calcium, vitamin D, miacalcin andbisphosphonates and pain management with analgesics, bed rest and bracing. Althoughmost patients with OVCFs respond well to conservative management, the pain canlast for weeks or months; if conservative management fails, long-term care andhospitalization or even surgical interventions can be required. However,conventional open surgery for OVCFs is not optimal because of associated trauma,greater blood loss and difficulties achieving solid stabilization due toosteoporosis.

Percutaneous Vertebroplasty (PVP), a minimally invasive vertebral augmentationtechnique, has been widely applied to treat OVCFs; this technique consists of injectingpolymethylmethacrylate (PMMA) bonecement into the fractured vertebral body to intensify the vertebra and thereby reducepain. Balloon kyphoplasty (BKP), an improvementover PVP, can create a cavity in the vertebral body via an inflatable balloontamp prior to the bone cement injection to correct the deformity, thus allowingfor cement injection at a relatively low pressure aiming to reduce theincidence of cement leakages.

Although both procedures have appeared encouragingfor providing immediate pain relief and improved function in patients with OVCFs,complications still occur. The occurrence of secondary fractures followingvertebral augmentation is one of the post-procedure risks that can influencetherapeutic outcomes. However, it remains unclear whether vertebralaugmentation techniques incur a higher risk of secondary vertebral fracturethan conservative treatment. High-quality, large-scale, randomized controlled trialsof patients with OVCF treated by either vertebral augmentation techniques orconservative treatment are rare. A recent meta-analysis compared the incidenceof secondary vertebral fractures between patients treated with vertebralaugmentation techniques with those who underwent conservative treatment; however,this meta-analysis included a study with malignancy-related vertebral fractures.Therefore, the purpose of the present study was to perform an updatedsystematic review and meta-analysis of randomized and prospective non-randomizedcontrolled trials with more rigorous inclusion criteria to compare theincidence of secondary vertebral fractures between patients with OVCF treatedwith vertebral augmentation techniques and those who underwent conservativetreatment.

Material and Methods

LiteratureSearch

All of the listed authors contributed tothe PRISMA-compliant literature search. Articles describing randomized controlledtrials (RCTs) or prospective non-randomized controlled trials (NRCTs) comparingeither PVP or BKP to conservative treatment or sham treatment for OVCF wereidentified and reviewed. Because the first study describing the use ofvertebroplasty was published in 1987, we searched PubMed, MEDLINE, EMBASE, SpringerLink,Web of Science and Cochrane Library databases, for articles dating from 1987 toDecember 2013, using the combinations of the following terms: vertebroplasty,kyphoplasty, vertebral augmentation, secondary fracture, refracture, subsequentfracture, newly developed fracture, versus conservative treatment, non-surgicaland osteoporotic vertebral fracture. We also screened the references of theidentified articles and reviews on this topic to avoid omitting any relatedarticles. No limitations regarding the language of the article were imposed.

Selectionof studies

Two of the authors (DS, BM) independently evaluated the potentially eligiblestudies. To be included, the studies were required to: 1) describe original RCTsor NRCTs; 2) compare vertebral augmentation techniques (either PVP or BKP orboth) to conservative treatment in patients with OVCF; and 3) reported thenumber of patients in both groups and the number of patients who experiencedsecondary vertebral fractures during the final follow-upof both groups. When multiple reports from the same center or trial were found,we selected the most thorough publication. The following studies were excluded:1) retrospective studies, observational studies and studies with noconservative treatment group or lacking key information; 2) studies with noclear inclusion or exclusion criteria or with no clear description of thedesign; and 3) studies including patients with severecardiopulmonary comorbidities, untreatable coagulopathies, systemic or localspine infections, suspected underlying malignant diseases, spinal-cordcompression syndromes or contraindications for MRI. Full articles were obtainedwhen the studies qualified or when it was difficult to judge whether thearticle would qualify based on the title and abstract. Any disagreements duringthe evaluations were resolved by discussion.

Dataextraction

After selecting the studies, the sametwo authors extracted the relevant data from each qualified article. Thefollowing information was extracted independently: thefirst author’s last name, the year of publication, the country in which thestudy was conducted, the study design, the length of the patient enrollmentperiod and follow-up period, the duration of symptoms before inclusion, the stage ofsymptoms, specific techniques, the numbersof cases and controls and the mean ages of the patients in each group. A thirdreviewer was invited to resolve disagreements if agreement was not reached, andthe opinion of the majority was applied to the further analysis.

Assessmentof methodological quality

The same two authorsindependently assessed the methodological quality of the included studies. ForRCTs, the “assessing the risk of bias” table, which is recommended by CochraneHandbook for Systematic Review of Interventions, version 5.0, was applied. For NRCTs, the Methodological Index forNon-randomized Studies (MINORS), which consists of a list of 12 potentialitems, was applied for quality assessment. Each item on the MINORS has twoscores, resulting in a total score of 24; a trial is considered high qualitywhen the score is ≥ 16 points, whereas low quality is indicatedby a score < 16 points. Any disagreements were resolved by discussion, andif no consensus was reached, a third reviewer served as an adjudicator.

Statisticalanalysis

STATA software, version 11.0 (STATA Corporation, College Station, TX, USA) and RevMansoftware version 5.0 (The Cochrane Collaboration, Copenhagen, Denmark) were usedfor data analysis. Meta-analysis was performed in accordance with therecommendations of the Cochrane Collaboration. STATA software was used tocompute the pooled RRs and 95% confidence intervals (CIs), generate forestplots, determine statistical associations, assess heterogeneity, performsubgroup and sensitivity analyses and inspect for publication bias. A p value <0.05 was considered to be statistically significant.RevMan software was applied to generate the “assessing the risk of bias” tablefor RCTs.

The pooled risks ratio (RR) and the 95%confidence intervals (CIs) werecalculated and compared using a random-effects model. Heterogeneity, which wasassessed using Cochrane’s Q statistic and the I2 statistic ,was considered statistically significant with a P value <0.10. The I2 statisticwas used to evaluate the heterogeneity of the studies as follows: I2  < 25% indicates low heterogeneity;I2 value ranging from 25%to 50% indicates moderate heterogeneity;  I2 > 50% indicates high heterogeneity.

Sensitivity analyses were performed toinvestigate the influence of a single study on the overall risk estimate. BothEgger’s test and Begg’s test were performed to test for publication bias, with p < 0.05 indicating potential publication bias.Subgroup analyses by study design (RCT and NRCT), geographic region (westerncountries and eastern countries, as the ethnicity of patients was not clearlystated in every study), duration of symptoms before enrollment (acute and notacute) and specific techniques (PVP, BKP and PVP+BKP) were also applied. Acumulative meta-analysis was also applied to investigate the influence of the publicationdate and sample size.

Results

LiteratureSearch

Fig. 1 showsa flow diagram describing the literature search and selection process. Weinitially identified a total of 80 relevant studies after a computerizedsearch. Eventually, 13 articles, including 8 RCTs and 5 NRCTs,fulfilled all of the inclusion criteria. A total of 1459 patients were analyzedin all of these studies.



Characteristicsand quality assessment

Table 1 summarizes the characteristicsof the included studies. Three studies from Australia, two from China and one eachfrom Taiwan, Denmark, Iran, the Republic of Korea, The Netherlands, both The Netherlandsand Belgium, Slovenia and Spain were included. We defined Iran as a westerncountry because Iranians are different from the populations of eastern Asia.Therefore, 9 studies were from western countries, 4 studies were from easterncountries. Ten studies focused on PVP alone, two studies focused on BKP aloneand one study investigated both PVP and BKP. The follow-up time ranged from 3months to 80 months. Three studies had a mean follow-up time of less than 12months, 7 studies had a follow-up time of exactly 12 months, and 3 studies hada mean follow-up time of more than 12 months. Six studies stated that they includedpatients with OVCF in acute stages, 5 of which clearly stated that they enrolledpatients who had experienced symptoms for less than 6 to 8 weeks beforeenrollment; however, 1 study  wasexcluded from the subgroup analysis because the time stage for less than 12months was inaccurate. Three studies enrolled patients who had experiencedsymptoms at least 6 to 12 weeks prior to enrollment (i.e., not acute); because theremaining four studies did not clearly specify the duration of symptoms, weremoved them from this subgroup analysis. Only 8 studies were included in thesubgroup analysis related to the duration of symptoms. The assessment of the methodologicalquality of the studies is illustrated in Fig. 2. TheRCTs were evaluated using the “assessing the risk of bias” table, and the NRCTswere evaluated with the MINORS (Table 2).



Resultsof the Meta-Analysis

Fig. 3 showsthe RRs and 95% CIs comparing the incidence of secondary vertebral fracturesbetween patients treated with vertebral augmentation and those who underwent conservativetreatment. The summary RR of the 8 RCTs studies was 0.841 (95% CIs =0.521 to1.357, p for heterogeneity= 0.133, I2 = 37.1%) and the summary RR of5 NRCTs was 0.961 (95% CIs =0.529 to 1.745, p for heterogeneity = 0.256, I2= 24.8%) the pooled RR for the RCTs and NRCTs was 0.879 (95% CIs =0.617 to1.252, p for heterogeneity= 0.161, I2 = 28.2%). No statisticallysignificant differences in RR were observed among the different study designs. TheI2 was between 25% and 50%, indicating moderate heterogeneity. Nostatistically significant differences in RR were found among the following subgroups: study design, geographic region,follow-up time, specific technique and stage of symptoms (Table 3).



Sensitivityanalysis and publication bias

The results of Egger’s test (p = 0.373> 0.05) were consistent with Begg’stest (p = 0.360 > 0.05), indicating no statisticallysignificant evidence of publication bias (Fig.4). The results from the sensitivity analysis are shown in Fig. 5. The cumulative meta-analysis revealed no trend in RR with publication year or samplesize



Discussion

PVP and BKP provide more rapid relief of painand improvement of function than conservative treatment. However, complicationssuch as cement leakage or secondary vertebral fractures can reduce the therapeuticeffects of these techniques. Because most of the previous studies were retrospectivelydesigned, it was difficult to distinguish clinical sequelae resulting from the procedureitself from the natural consequences of osteoporosis, which could alsocontribute to the future fractures. Concomitant osteoporosis treatment regimens,the number and severity of prevalent fractures and clustering effects could alsohave influenced the results, unless these factors were accounted for inwell-designed, prospective, randomized, controlled trials comparing vertebralaugmentation techniques and conservative treatment. The purpose of the currentstudy was to update and supplement previous studies on this topic with morerigorous inclusion criteria. Our meta-analysis showed that there were nostatistically significant differences in the incidence of secondary vertebralfractures among patients treated with vertebral augmentation techniques andpatients treated with conservative treatment. These findings were consistentwith most of the included studies and with those of previous studies indicatingthat vertebral augmentation techniques were preferred over conservativemanagement, as the former can achieve immediate pain relief and functionalimprovement. In contrast, Blasco et al. found that PVP was associated with ahigher incidence of vertebral fractures than conservative treatment. They foundthat PVP resulted in a 2.78-fold increased risk of radiological vertebralfractures relative to conservative treatment, and they attributed thisdifference to the following explanations: 1) more vertebrae were treated withPVP in their study (mean of 2.46 vertebrae); 2) patients experienced increasedmobility after the PVP; 3) cement leakages into discs after PVP increased therisk of secondary fractures. As previously reported, because patients whoundergo PVP or BKP experience a rapid improvement in symptoms, they are proneto becoming more physically active and therefore more likely to sustain trauma.An intervertebral cement leakage, a procedure-related complication, can alsocontribute to the incidence of fractures. These 2 factors were related, asmultiple authors have reported that secondary vertebral fractures could resultfrom the increased stiffness and ultimate failure load of the treated vertebralbody or from the weakness of the surrounding local spinal segment of the treatedvertebrae due to osteoporosis. When cement leaks into the disc, the hard cementcreates a strong mechanical pressure that can cause adjacent endplate fracture.Because most patients who undergo PVP or BKP experience clear pain relief andfunctional improvement, they tend to increase their daily physical activities,which can place additional stress on the vertebrae.

With the exception of a meta-analysisdirectly comparing PVP and BKP, meta-analyses comparing the effects ofvertebral augmentation and conservative treatment on the incidence of secondaryvertebral fractures are relatively rare. Even though PVP has been consideredassociated with a higher incidence of secondary fractures than conservativetreatment, Anderson et al. also observed no significant differences in theincidence of secondary vertebral fractures among patients treated with the 2treatments. We excluded a study by Wardlaw et al. from our meta-analysis, asthis study enrolled patients with OVCF and patients with multiple myeloma orosteolytic metastatic tumors, which might have biased the conclusion. Ourmeta-analysis also included 3 new RCTs; however, the results remained similar.In a previous meta-analysis, Zou et al. studied the long-term incidence ofsubsequent vertebral body fractures following vertebral augmentation orconservative treatment and found no evidence of an increased incidence offractures of adjacent vertebral bodies after the procedure. However, only 2RCTs were included in their study, which limited the power of the analysis.Nevertheless, their results were similar to ours, even when we conductedsubgroup analyses with different follow-up times.

Limitations of this present study werestated as follows: 1) The demographics and comorbidities of all of the includedstudy participants were not reported. 2) Data on the specific segments of thesecondary vertebral fracture sites were not well recorded in all of the includedstudies; thus, a comparison of the incidence of fractures at different levels wasnot possible. 3) Few RCTs and NRCTs comparing OVCF patients who underwent BKPalone with OVCF patients undergoing conservative treatment have been conducted,we therefore included only 2 trials investigating BKP among the analyzedstudies. 4) Although we detected no statistically significant evidence of publicationbias in the current meta-analysis, publication bias may have influenced our findingsnevertheless.

In conclusion, no differences in theincidence of secondary fractures in patients who underwent vertebralaugmentation techniques and patients who underwent conservative treatment wereobserved. Additional randomized controlled trials, especially trials comparing theincidence of secondary vertebral fractures at different levels, trialscomparing BKP and conservative treatment and trials identifying risk factors forsecondary fractures, should be conducted.


Figurelegends

Fig 1. Flow diagram ofthe study selection process

Fig 2. Methodologicalquality assessment of the RCTs. “+” indicates a low risk of bias, "-"indicates a high risk of bias and “?” indicates unclear.

Fig 3. Forest plotdepicting the incidence of secondary vertebral fractures in RCTs and NRCTs.

Fig 4. Plots of Begg’s test andEgger’s test.

Fig 5. Sensitivityanalysis comparing the incidence of secondary vertebral fractures amongpatients undergoing vertebral augmentation and patients undergoing conservativetreatment.

Table 1.Characteristics of all the included studies in the meta-analysis

RCT, randomizedcontrolled trails; NRCT, non-randomized controlled trails; N/A, data notavailable; PVP, percutaneous vertebroplasty; BKP, balloon kyphoplasty; Case,patients underwent PKP or BKP; Control, patietns underwent conservativetreatment or sham-operated.

Table2. The MINORS appraisal scores for the studies of NRCT

The numbersrepresent the following items from MINORS methodological criteria as: (1) aclearly stated aim; (2) inclusion of consecutive patients; (3) prospective datacollection; (4) endpoints appropriate to the aim of the study; (5) unbiasedassessment of the study endpoint; (6) a follow-up period appropriate to theaims of the study; (7) loss to follow-up less than 5 %; (8) prospectivecalculation of the sample size; (9) an adequate control group; (10)contemporary groups; (11) baseline equivalence of groups; and (12) adequatestatistical analyses

The items are scored as follows: 0 (not reported); 1(reported but inadequate); or 2 (reported and adequate). The total score is 24;high quality is indicated when the score is ≥16 points,while low quality is indicated when the score is <16 points.

Table 3.Subgroup analysis between vertebral augmentation techniques and conservativetreatment