Have Recent Vertebroplasty Trials Changed the Indications for Vertebroplasty?
- First Online:
- Cite this article as:
- Gangi, A. & Clark, W.A. Cardiovasc Intervent Radiol (2010) 33: 677. doi:10.1007/s00270-010-9901-3
- 199 Views
Two different investigators in the New England Journal of Medicine recently published two randomized controlled trials (RCTs) regarding the efficacy of vertebroplasty for painful osteoporotic vertebral compression fractures. In their results, both investigators concluded that there was no significant difference in pain relief between the vertebroplasty group and control group 1 month after treatment. The trials described a different patient cohort from the one we treat with vertebroplasty. Both enrolled patients had back pain for ≤12 months. This duration of pain was far too long for a vertebroplasty trial, resulting in parallel trials of vertebroplasty on healed fractures. Where a study is needed, it should be comprised of patients with acute osteoporotic compression fractures, particularly those who are hospitalized or bedridden because of the pain of such fractures. Magnetic resonance imaging was not systematically performed before vertebroplasty, and inpatients were excluded. Inpatients with acute fracture pain are the group most likely to respond well to vertebroplasty. Enrolment was a problem in both trials. Randomization in both RCTs took >4 years for completion. We advise that vertebroplasty be offered to patients with recent fractures <8 weeks old who have uncontrolled pain as well as patients progressing to osteonecrosis and the intravertebral vacuum phenomenon (Kummels disease). The availability of recent MRI scanning is also critical to proper patient selection.
KeywordsVertebroplastyOsteoporosisVertebral compression fractureVertebroplasty trials
Osteoporotic spinal fractures are one of the more common skeletal fractures  and are increasing in frequency in our ageing population due to the growing prevalence of osteoporosis. During the last 15 years, vertebroplasty has been increasingly used in the management of the most painful vertebral compression fractures (VCFs). It is acknowledged that the normal evolution of spinal VCF is spontaneous consolidation with progressive pain decrease during a 6- to 8-week period. Rest and analgesia have been advocated and are still used in the majority of patients. There is, however, a subgroup of patients with severe acute fracture pain that is not well controlled by oral analgesics. There is also a subgroup whose fractures do not heal normally but progress to osteonecrosis and non-union, resulting in “Kummel’s disease,” including ongoing pain. It is in these two groups of patients that we have seen dramatic clinical benefit from vertebroplasty.
The recent publication of two placebo-controlled vertebroplasty randomised controlled trials (RCTs) [2, 3] has incited great debate about the merits of vertebroplasty. The published trials were accompanied by an editorial  in the New England Journal of Medicine proclaiming them as “the best available scientific evidence for an informed choice” on vertebroplasty. Both trials compared vertebroplasty to a “sham procedure” or placebo using a similar model to that used to evaluate pharmaceuticals. Both found that vertebroplasty offered no more benefit than would a sham procedure.
Analysis of Randomised Studies
The intent of this analysis is not to in any way defame the published studies or to question the integrity of the investigators. We thank the investigators of these two trials for their important contributions. However, close analysis of both recent RCTs shows that similar and fundamental flaws in both study design and patient recruitment affect both trials. We will argue that these flaws decrease the practical importance of these studies and that their findings cannot be generalised to all patients with acute osteoporotic fractures.
A major problem for both RCTs is inappropriate patient selection. Both trials enrolled patients with back pain for ≤12 months. This duration of pain is far too long for a vertebroplasty trial and resulted in parallel trials of vertebroplasty on healed fractures. A more appropriate patient selection for a trial is fracture pain lasting >6 weeks as in the Vertos2 trial , which still awaits publication at the time of writing. Vertebroplasty is an internal fixation technique that unites the fragments of an acute vertebral body fracture to prevent the pain of fracture fragment motion. This effect of fracture fixation in ameliorating acute fracture pain is well accepted in fractures elsewhere in the skeleton. It is not possible for the technique to be of benefit in healed fractures. Vertebral body fractures have usually healed by 8 weeks. The average duration of pain in the two trials was 18 weeks  and 9.5 weeks , respectively. This timeline indicates that the majority of patients had healed fractures; thus, the pain may be have been arising from causes other than fracture fragment motion. Both studies included 26 patients with fractures of <6 weeks’ duration. This is insufficient to allow for separate statistical analysis of this acute fracture subgroup by the investigators’ own previous calculations .
The Vertos2 trial enrolled 202 patients with acute fractures of <6 weeks’ duration. The patients were randomised to vertebroplasty or to usual care. The trial results as presented at CIRSE 2009 (Lisbon, Spain) demonstrated excellent analgesia in the vertebroplasty group, which remained statistically significant at ≤12-month follow-up. Given that this is the first large RCT of acute fractures of <6 weeks’ duration, and that the Kallmes et al.  and Buchbinder et al.  trials cannot draw statistically meaningful conclusions in this group of patients, then the VERTOS2 trial will provide the best evidence for the application of vertebroplasty in the acute osteoporotic fracture group.
Another disturbing aspect of patient selection in the Kallmes et al.  trial was the inclusion of patients who had not undergone magnetic resonance imaging (MRI). Only plain radiograph was required to demonstrate a fracture. During recent years, a crucial lesson we have learned is the necessity of recent MRI for proper patient selection. We require evidence of marrow oedema on MRI before performing vertebroplasty. MRI also shows the actual number of recent fractures and detects fractures with no sign of vertebral collapse on initial radiographs but marrow oedema on MRI . When MRI is contraindicated, a bone scan should be performed to confirm fracture activity and the number of vertebral bodies involved . Without a recent MRI or bone scan, one cannot confirm that the cause of pain was vertebral compression fracture. We conclude that patient selection was improper and that a number of patients were included who had no indication for vertebroplasty.
The Kallmes et al.  trial was also substantially weakened by its exclusion of inpatients . This important fact was not shown in the trial publication nor in previous protocol publication . This is unfortunate because inpatients with acute fracture pain are the group most likely to respond well to vertebroplasty. There is also no mention of any inpatients in the Buchbinder et al. trial , which also failed to include them. Neither study makes any comment on the utility of vertebroplasty for inpatients with acute vertebral fractures. This is an important oversight.
Enrolment was a problem in both trials. Randomization in both RCTs took >4 years to complete. The Kallmes et al.  study screened 1813 patients, and 431 were considered eligible for the study. Of these, 70% (300 of 431) declined to participate. The Buchbinder et al.  trial screened 468 patients, and 220 were considered eligible for enrolment. Of these, 65% (141 of 220) declined to participate. Thus, in both studies, only 30 to 35% of eligible patients entered the study, and only half of those underwent vertebroplasty, making patient recruitment extremely poor. The pain severity and functional compromise of the patients who refused participation were not reported. Thus, there exists an unquantifiable selection bias in the final patient group. Recruitment problems in the Buchbinder et al.  trial were exacerbated by half of the hospitals dropping out of the study early.
The control groups in both of these studies underwent sham procedures. However, this was not a sham procedure so much as an alternative treatment. Injection of anaesthetic into the facet capsule and/or periosteum may have had a plausible mechanism of pain relief, albeit not fracture pain relief, in this patient population. Although it was stated in both studies that patients had back pain, it is unclear if the origin of the back pain was the osteoporotic VCF or other common reasons for back pain in the elderly, such as arthritis facet pain. By the nature of the patient population studied, “sham” facet injections may have led to decreased facet pain. Perhaps a sham procedure in which a dry needle was inserted might have been a more appropriate control .
The Buchbinder et al.  trial excluded the interventional radiologist from patient selection. We doubt whether a surgeon would agree to perform surgery without being involved in patient selection. The interventional radiologist’s role in the trial was confined to the performance of vertebroplasty, and they “strictly adhered to a detailed, standardized protocol”  This protocol mandated the use of 13G needles and hand-injected cement by way of 1-cc syringes. This is an inappropriate technique for lumbar fractures, which require either 11G needles or more modern high pressure–injecting systems to obtain adequate filling of the large vertebral body. The average cement volume injected in the study (2.8 cc) is less than what we inject into acute fractures and also significantly less than the average volume injected in the Vertos2 trial (4.1 cc). Fracture stabilisation requires cement be distributed to every quadrant of the vertebral body. It is also unclear who was involved in the interpretation of the MRI examinations. The trial was exposed to technical bias because it was more like a single-centre study. Two of the four hospitals withdrew early from the study after five patients had been enrolled at each. The third centre enrolled 15 patients during 4.4 years. A total of 68% of the procedures were performed in one hospital by one radiologist.
Selection bias is an important issue when interpreting these two trials. The Kallmes et al.  trial reports: “There was a trend toward a higher rate of clinically meaningful improvement in pain in the vertebroplasty group than in the control group (64 vs. 48%, p = 0.06).” It would be interesting to know how the patient group who refused enrolment were treated. The majority of eligible patients refused enrolment in both studies. Enrolling patients in an RCT is a difficult task. It is reasonable to think that patients in severe pain would more often opt to decline the study and proceed with vertebroplasty. It would have been useful to see the outcome of this group of patients. Was there any follow-up of the patients who dropped out?
Both studies had poor efficacy of their vertebroplasty interventions, with a mean pain reduction score of only 2.7 points in the Kallmes et al.  study and 1.5 points in the Buchbinder et al.  study (scored using a 10-point visual analogue scale during early assessment after vertebroplasty). This contrasts with our clinical experience and also with the literature (mean pain score reduction of 5.7 points in the short term after vertebroplasty) [12–14]. The above-mentioned information could have been due to poor patient selection, improper inclusion and exclusion criteria, or even the interventional technique itself and requires further consideration.
The findings of the recent RCTs cannot be regarded as definitive evidence on the efficacy of vertebroplasty for patients with acute osteoporotic fractures. We believe that the assessment published in the New England Journal of Medicine , i.e., that these trials represent “the best available scientific evidence” for vertebroplasty, is a misleading generalisation. A trial can only provide evidence regarding the patient group that it samples. The trials had no inpatients and insufficient numbers of patients with fractures <6 weeks old to provide any statistical conclusion in these most important patient groups. According to a clinician who deals every day with osteoporotic patients, Bolster  reported that the highest priority after a fracture is to treat osteoporosis and that vertebroplasty may still benefit correctly selected patients. He added, “Having considered the results, conclusions, and limitations of these two RCTs, particularly in terms of recruitment, I cannot say that my practice has changed in terms of referring patients who have a vertebral compression fracture to an interventionalist.” The only deduction that can confidently be drawn from these trials is the futility of applying vertebroplasty to patients with healed vertebral fractures.
We advise that vertebroplasty be offered to patients with recent fractures <8 weeks old with uncontrolled pain and also for those who have progressed to osteonecrosis and the intravertebral vacuum phenomenon (Kummels disease). The availability of recent MRI results is also critical to proper patient selection. The best scientific evidence to support the utility of vertebroplasty in acute vertebral fractures <6 weeks old will be provided by the results of the Vertos2 trial.