Spine surgeons debate the value of vertebroplasty and kyphoplasty

Introduction

Cement injection into acute or subacute vertebral compression fractures has increased in popularity in recent years. There are many anecdotal accounts of bedridden patients being able to walk off the procedure table following such injections. Prospective randomized studies, while hard to design properly and execute, are underway and beginning to be disseminated. Two recent studies in the New England Journal of Medicine have questioned the benefit of vertebroplasty and given reason to re-examine the appropriate indications, safety, and expected outcomes for these cement injection procedures.

I have assembled a group of physicians from radiology, physiatry and spine surgery to discuss these issues in a virtual Round Table format. A summary of the discussion follows.

Scott D. Boden, MD
Moderator

Round Table Participants Moderator Scott D. Boden, MD Professor of Orthopedic Surgery Director, Emory Orthopedics and Spine Center Emory Healthcare Atlanta, Ga. Steven R. Garfin, MD Professor and Chair, Department of Orthopaedics University of California San Diego San Diego, Calif. Isador H. Lieberman, MD, MBA, FRCSC Professor of Surgery Chairman, Medical Interventional and Surgical Spine Center Cleveland Clinic Spine Institute Weston, Fla. Andrew C. Hecht MD Co-Chief, Orthopaedic Spine Surgery Assistant Professor of Orthopaedic and Neurosurgery Mount Sinai Medical Center and School of Medicine New York, N.Y. Michael K. Schaufele, MD Assistant Professor of Orthopedics Assistant Professor of Physical Medicine and Rehabilitation Emory Healthcare Atlanta, Ga. John G. Heller, MD Professor of Orthopedic Surgery The Emory Clinic Atlanta, Ga. Frank C. Tong, MD Assistant Professor, Departments of Radiology and Neurosurgery Emory Healthcare Atlanta, Ga. Joseph M. Lane, MD Department of Orthopedic Surgery Hospital for Special Surgery New York, N.Y.

Scott D. Boden, MD: Do you believe kyphoplasty is safer than vertebroplasty?

Steven R. Garfin, MD: Yes. I have done vertebroplasties and, anecdotally, I had significantly more leakage, particularly into the canal, than I have ever noted with kyphoplasty. Therefore, I stopped performing them. More scientifically, there have been three meta-analyses comparing balloon kyphoplasty (BK) to vertebroplasty (VP). They have all shown less leakage, more radiographic height gain, and less subsequent fractures with kyphoplasty.

In 2005, Hadjipavlov and colleagues published a systematic review from papers published from 1983 to 2004. They determined BK had a lower cement leakage rate than VP (8.4% vs. 29%). This was true for leakage within the epidural, foraminal, intradiscal and paraspinal spaces, as well as intravenously.

Taylor and colleagues published in Spine in 2006 their findings that BK had a lower rate of cement extravasation than VP (8% compared to 40%, respectively). Although most were asymptomatic, they reported symptomatic leaks were 0% for BK and 3% for VP. Also, BK had a lower complication rate than VP, including pulmonary embolism (0.3% vs. 1.8%), spinal cord compression (0% vs. 0.5%) and nerve root pain/radiculopathy (0.3% vs. 2.5%).

In 2006, Hulme, and colleagues reported that pain relief was higher in BK than VP (92% vs. 87%); cement leaks were higher in VP (9% vs. 41%). Height restoration was noted as 6.6° overall in BK, but only reached that level in VP when clefts in the vertebral body were present pretreatment. They also noted that BK was better than VP in terms of pain relief (92% vs. 87%), clinical complications (1.3% vs. 2.6%), pulmonary embolism (0.01% vs. 0.6%) and neurologic complication (0.03% vs. 0.6%).

Overall these data confirm to me that BK is safer, the height restoration more predictable and cement leakage is less.

In an unpublished, ongoing randomized trial in Europe sponsored by Medtronic (Kaviar), preliminary data demonstrate that there is superior correction in spinal deformity with BK vs. VP.

Andrew C. Hecht, MD: I believe BK is safer than VP for two principle reasons: First, BK involves cement injection with a lower pressure and, second, it is in a more viscous state when injected. The cement is often injected in a less-viscous state and under high pressures for VP which often leads to cement extravasation outside of the vertebral body either into the disc space, outside the margins of the body or into the epidural space. Also, the creation of the cavity with the balloon during a kyphoplasty facilitates the safe introduction of the cement into this defined cavity. Both procedures do carry the risk of cement extravasation; however, there is significantly less risk for kyphoplasty.

John G. Heller, MD: No. In skilled hands and when performed with due regard for the relative contraindications, they ought to be equally safe and effective.

Joseph M. Lane, MD: Both VP and BK provide pain relief for vertebral fractures. In the literature the relief appears to be similar, but there has not been a large-scale randomized trial to define the relative advantages and disadvantages. Also, the literature notes a greater PMMA (polymethylmethacrylate) extrusion and there have been several acute deaths from VP secondary to cardiac/pulmonary compromise. In VP, the PMMA is more liquid and potentially can enter the vascular system easier carrying the PMMA and bone marrow products. However, with time, the physicians who use both BK and VP are well aware of the problems and think that both procedures are safe.

Isador H. Lieberman MD, MBA, FRCSC: Safety is a relative issue. VP and BK are different procedures that require different appreciations and technical skills which can be done equally well by qualified individuals and equally poorly by lesser-skilled individuals. The common components during both procedures of needle placement are safe provided one has an appreciation of the anatomy and understands the imaging. The application of an inflatable bone tamp during BK has unique aspects that in less-than-qualified individuals can certainly be dangerous. The literature would suggest that the cement leakage rate is greater in VP than BK, although the rate of symptomatic cement leakage in qualified hands I suspect is similar and low.

Michael K. Schaufele, MD: I do not believe that one procedure is safer then the other. However, both BKs and VPs have advantages and disadvantages in this area.

VP has an advantage for certain patients because it usually is a quicker procedure. Usually, it can be performed through a unipedicular approach, which reduces the overall procedure time. This is an important factor for elderly patients with multiple medical problems, who have an increased anesthesia risk, even with conscious sedation.

BK has the advantage of reduced cement leakage. The insertion of the bone tamp prior to PMMA injection significantly reduces cement leakage. This becomes an important safety factor in fractures with multiple fracture lines, retropulsion and when a fracture extends into the end plates and the posterior wall. In my experience, BK allows for a more controlled PMMA delivery into the fractured vertebra, therefore increasing the physician’s comfort level with the procedure.

Frank C. Tong, MD: Both procedures are safe given appropriately chosen patients, good fluoroscopic visualization and experienced operators with few symptomatic complications. A large systematic review of 69 VP and BK studies by Hulme and colleagues in 2006 found aggregate symptomatic complication rates of 2.6% and 1.3% of treated levels, respectively. In this retrospective analysis, all asymptomatic pulmonary emboli were counted as symptomatic clinical complications, which occurred with higher frequency of 0.6% in the VP group compared to 0.01% in the BK group. Conversely, if these asymptomatic pulmonary cement emboli are indeed counted as asymptomatic, then the rates for both procedures are essentially equal.

There is, however, a significantly higher rate of nontarget cement deposition with VP — 41% compared to 9% for BK. Our experience performing percutaneous vertebroplasty has been that it is unlikely for a patient to have symptoms from nontarget cement provided there is prompt recognition of cement leakage and cessation of injection, as the development of symptoms is likely related to the volume of nontarget cement. The lower nontarget cement leakage rate for BK is likely related to injection of more viscous cement into a balloon-tamped cavity.

Indications

Boden: What are your indications for vertebral cement injection procedures?

Garfin: Patients who have: painful or progressive vertebral compression fractures (VCFs), primarily from osteoporosis, with pain to percussion over the spinous processes that correlates with the level of fracture — a marker placed at the area of pain reproduction followed by AP and lateral radiographs that localize the marker to the fracture level; MRI that demonstrates edema in the vertebral body, and possibly other levels that do not show up on plain radiographs, who have failed nonoperative care — which can be less than a week of short-term severe pain in hospitalized or incapacitated patients, or more subacute pain that does not respond to a few weeks of nonoperative care in an at home patient; and/or radiographic evidence of progression.

Hecht: My indications for vertebral cement injections are in painful, osteoporotic compression fractures after a 6- to 12-week period of conservative care. I try to follow the degree of symptomatic (pain) improvement after 4 to 6 weeks to see if the patient’s pain has reduced.

I often use a supportive brace for lumbar fractures; however, I feel most elderly patients do not tolerate larger thoraco-lumbo-sacral-orthosis (TLSO)-type braces. Patients who are not improving in terms of their overall pain by at least 30% on Visual Analog Scales (VAS) or are failing to thrive often have BK earlier. However, patients who are improving despite some persistent pain are followed for additional time to see if the pain reduction continues. I try to encourage patients to continue with conservative care as long as their pain continues to improve, because the natural history suggests that most fractures will heal with conservative care.

Other indications for BK include: myeloma, metastasis and vertebral angioma with intractable pain and with no neurological symptoms.

Heller: My indications include: painful, vertebral insufficiency fractures that are not reasonably responding to medical management, such that they cause significant functional disruption in activities of daily living (ADL), undue burden on family or care providers, or require inpatient admission; painful nonunions of insufficiency fractures (Kummel’s disease); and osteoporotic burst fractures with concomitant cauda equina compression that require open decompression. A BK can obviate the need for segmental instrumentation and fusion for this unique subset of patients; and it, not VP, may also be occasionally indicated for traumatic burst fractures which are severely painful and prevent mobilization of patients. Obesity and inability to brace may be a contributing factor for this.

Lane: The purpose of vertebral augmentation is to provide pain relief and prevent further vertebral collapse. There is the potential to partially correct the deformity of the vertebra and improve local spine alignment. My indications are: pain greater than four on a 10-point VAS; evidence of vertebral collapse greater than 20%; at least 25% marrow edema on MRI; and failure of conservative therapy for at least 2 weeks, unless the pain is so strong that the patient has been admitted for pain control. The pain must be localized to the fracture site. A weak indication is a collapse of greater than 40% with a low pain level, such as two to four on the 10-point VAS.

Lieberman: Osteoporotic and osteolytic vertebral compression fractures represent a spectrum of clinical symptoms and biomechanical issues. Vertebral augmentation is indicated in the treatment of progressive and painful vertebral compression fractures in those with osteoporosis or tumor osteolysis. Vertebral augmentation may be considered in vertebral bodies at risk, ie, those between already collapsed vertebrae, those at biomechanically vulnerable junctions, and/or those at the apex of a deformity.

VP and BK should not be considered mutually exclusive in the spectrum of treatment. Both are distinct tools to be used at the appropriate time for the appropriate clinical situation and should be combined as needed. If simple augmentation is required, a VP technique may be best; if reconstruction is required, then a BK technique may be best.

Schaufele: My indications are: acute to subacute vertebral compression fractures due to osteoporosis or tumors; focal thoracic and lumbar pain over the fracture site, reproduced on prone physical examination, with correlating imaging findings; MRI of the thoracic/lumbar spine indicating a fracture with increased signal on T2 or short T1 inversion recovery (STIR) images; or evidence of nonunion (Kuemmel’s disease). Alternatively, significant uptake on a bone scan at the site of the fracture, or vertebra in nonunions; lateral flexion-extension X-rays or extension X-rays over a bolster indicating an unstable fracture; and significant pain (average VAS five out of 10) without signs of improvement with time, bracing and pain medications; and documented associated functional decline.

Tong: My indications for VP are: acute to subacute painful vertebral compression fractures which can arise from osteoporosis, benign tumor (eg, aggressive hemangioma), malignancy (eg, multiple myeloma or metastatic lesion), or cystic osteonecrosis; inadequate pain control with medical therapy or symptomatic contraindications to pain medication resulting in impaired functionality and ability to perform ADL; and occasionally we are asked to inject a severely osteopenic vertebral body in anticipation of spinal surgery with planned instrumentation of that vertebral body.

Adjacent fractures

Boden: Do vertebral cement injections increase the risk of adjacent fractures? Is there any difference in this between VP and BK?

Garfin: The natural history of new fractures, as reported by Lindsey, and colleagues, is 12% to 22% per year. Of note, one vertebral body fracture predicts a threefold chance of having another fracture; two fractures a tenfold increase; and with three fractures there is almost a 25% chance of another fracture in a year. With VP, the reported incidence of adjacent level fractures is 20% to 30%. With BK it is approximately 10% to 11%. Many of the VP fractures are adjacent and many of the BK fractures are one or two segments removed from the index fracture(s).

Hecht: It is clear in several studies that both BK and VP increased the risk of adjacent-segment fractures during the first 180 days after the procedure. The rates approach 20% to 25% in some reported series. Interestingly, once the patient reaches 180 days postoperative there is no different in new or adjacent fractures compared to age-matched controls. I have not seen a difference between VP and BK in this regard.

Heller: No and no.

Lane: The older literature suggests that there is a greater risk for adjacent-level fracture after VP. This has not been demonstrated in recent studies. Of note, the work by Bouxseim has demonstrated that parathyroid hormone (PTH 1-34) drug therapy decreased the fractures by 70%. Most of the literature did not include the treatment status of the patients. Considering the suspect quality of the early cohort studies and the clear prevention with drug intervention, the actual risk for adjacent fractures today should be quite low in both the augmented and conservatively treated patients. There is an old unmatched series suggesting that VP had greater fractures, but again this was often in the times when these patients were not fully treated for osteoporosis.

Lieberman: VP per se does not increase the risk of remote or adjacent-level fractures. The literature pertaining to spine biomechanics clearly documents that “kyphosis begets kyphosis.” The literature pertaining to remote and adjacent-level fractures seem to suggest a rate of fracturing similar to the natural history of vertebral collapse progression. There is a literature trend to suggest that individuals treated with VP may have a higher rate of subsequent fracture; however, I suspect that is more related to the overall sagittal balance (residual kyphosis) and the degree of osteoporosis than to the technique.

Schaufele: My experience is that there is an increased risk of fractures adjacent to the augmented level within the first couple of weeks after both VP and BK. To me, BK seems to carry a somewhat higher risk of adjacent fractures than VP. This is not necessarily substantiated by clinical studies, but has been a fairly consistent observation in my own practice. I can only speculate about the cause. In general, BKs are done with higher volumes of PMMA to fill the cavity created by the bone tamps. Because of the higher strength of PMMA compared to osteoporotic bone, higher loads may be transferred into the adjacent vertebra, thus resulting in a higher adjacent-vertebral-body fracture risk. Clearly, this theory needs to be substantiated by biomechanical studies. If there is an increased fracture risk, injectables that better mimic the biomechanical properties of osteoporotic bone may be desirable.

Tong: A review of 27 studies reporting subsequent fractures in VP patients by Trout and colleagues in 2006 showed that approximately 20% of patients treated with VP will have an additional fracture. Because no randomized datasets are available for direct comparison, a frequently quoted “control” population is that published by Lindsay and colleagues in 2001, who evaluated the fracture risk among placebo groups in several large series evaluating the clinical efficacy of risedronate (Actonel, Procter & Gamble Pharmaceuticals). They found an overall 19.2% incidence of additional fracture within 12 months of initial fracture presentation. This would suggest that the new fracture rates are similar, although the number of VP patients receiving bone density drugs is not precisely known.

Adjacent-level fractures have been reported for both VP and BK, but direct comparison is difficult given the variability within study designs. Biomechanical factors such as accentuation of angulation and focal loading may also explain fracture distribution clustering.

NEJM studies

Boden: Two studies in the New England Journal of Medicine (NEJM) recently concluded that VP was no better than placebo. Do you think these studies are flawed and are the conclusions different than your personal experience?

Garfin: They are terribly flawed. A small percentage of patients entered the study, which leads to potential selection bias because patients with more pain may have chosen the “real” treatment. There was a much higher rate of crossover from the control groups to the VP groups than vice versa, and they did an intent to treat analysis and did not include “as treated” or “observational” arms as was done for the SPORT trial, which also had a high crossover rate. There was no follow-up of patients as to why they were may have failed; standard clinical care would include new imaging studies, including MRI scans and/or an evaluation for other etiologies of pain following percutaneous cement augmentation, rather than just accepting the patients’ pain complaints as a failure of treatment. Perhaps there was a poor technique, leakage, new fractures, etc. The entry workup seems scant. The patients did not routinely have MRI scans, which were only obtained if they could not determine the “length of time of the fracture.” How can they determine that from a radiograph? By not obtaining an MRI they may have: missed other fractures not seen on plain films; treated old fractures (no edema on the MRI); or missed other etiologies for the pain. They did not describe the preoperative exam, which should have included correlation of the location of pain (to percussion) to the radiographic abnormalities.

The entry VAS score was low, three out of 10, perhaps lower than many would include in treatment, and because it was low, large improvement would be impossible to statistically determine. The outcome end points were altered during the study because of limited enrollments, leading to accepting a wider range of variation (standard deviation). This led to difficulty determining significant changes that may have shown up if the assessment of the new power analysis had not changed during the study. It is unclear how and why this occurred. A randomized controlled trial, per se, does not make a study useful or scientifically valid if: it is under-powered; the outcome endpoints change during the study; the workup is incomplete (pre- and/or post- treatment); the number of crossovers is significantly different between the groups; and the entry pain levels are low.

My experience is consistent with the literature: A 90% or more success rate with percutaneous cement augmentation for painful osteoporotic VCFs using BK. Their results, which are much poorer than that in the literature or the three meta-analyses, suggest: the entry criteria are poor; the techniques perhaps should be questioned; and their diligence in working up the patients and following them for other fractures or other disorders may not have been thorough or was limited compared to clinical practice.

Hecht: There are flaws in both of these trials. First the numbers for each trial are low, and this may be attributed to a selection bias. In one of the studies, more than 30% of the patients who met its inclusion criteria refused to participate in fear of receiving the sham procedure. This may have led to the exclusion of patients with more severe pain levels who feared exclusion from the trial. There were only 78 and 131 patients in each study, which may explain why it took many years to get sufficient patients to conduct these trials.

The experience worldwide with over 100,000 patients seems to run counter to the clinical findings in these studies. However, I found it most interesting that patients in the sham groups experienced some pain relief. This may force us to reinvestigate the mechanism by which these procedures relieve pain.

In my experience, BK and VP have been important tools in the care of patients with these osteoporotic compression fractures that fail conservative treatment. In patients with severe pain, using increasing doses of narcotics and failing to thrive they can provide substantial pain relief. These procedures often replaced much larger open operations that were significant more morbid. The biggest complaint I have seen in patients undergoing this procedure relates to new adjacent fractures that occur early on (first few months) after BK.

Heller: The studies were very badly flawed.

Cement injection for the right fractures represented a major step forward in what we have to offer patients who sustain these injuries. The issue at hand is to restrain the application of these procedures to those who fit the proper indications. Indiscriminate use will invite less effective results and degrade the measured value of the procedures for those who truly benefit from them.

Lane: The two studies demonstrated in randomized trials that there is no advantage for VP vs. conservatively treated patients. These studies were flawed, but do give some credence to caution in picking the correct patients. The investigators could only entice less than 20% of the candidates to enter the trial. These patients had failed conservative care and wanted surgery. Only the milder patients would have chosen to enter the trial. This same problem was seen in the spine fusion trial.

Also, the patients often went several months before treatment. Since VP leads to little height restoration, it is most effective early, 2 to 4 weeks, when postural correction is possible. Late VP is not successful after there is partial healing or if the collapse is greater than 30%.

In the crossover study there was further delay in the failed patients and, as expected, the VP did not work. Ideally VP is an early intervention and loses its efficacy if the marrow edema is less than 50%, if the collapse is greater than 30% or if the fractures are over 6 weeks old. The actual success of the study was only 40% far below the other cohort studies in which VP was performed within weeks of the injury.

My personal use of VP is only in patients with less than 30% collapse, within 4 weeks of fracture and more than 50% marrow edema on the MRI. Pain has to be more than six on the 10-point VAS.

Lieberman: I think the studies were flawed and the conclusion is different from my personal experience. The two recent NEJM sham-controlled VP studies are underpowered to detect differences in their primary endpoints. Also, the use of injected anesthetic in the Kallmes study is not a sham procedure; it is an unproven method of treatment. Furthermore the fact that patients crossed over to the VP arm from the sham treatment at a much higher and statistically significant rate suggests that there may be benefits from VP that this underpowered study is unable to detect. Other aspects, including the patient-entry criteria, specific VP techniques, the overall time to enroll patients, the number of centers involved and the number of patients from each center, are likely to influence the outcomes.

There are known limitations of the VP technique, including the ability of the cement to infiltrate the bone and the lack of predictable cement distribution. Also, if a cement leak occurs, the physician is likely to discontinue the fill, resulting in inadequate fill, which may in turn not provide adequate mechanical support, resulting in bone micro-motion and persistent pain.

Having said that, taken together the two studies do not support robust benefits from performing VP. However, we must not discount all the previous positive literature, including previous randomized studies, simply on the basis of these two sensationalistic, simultaneously published, flawed studies with small cohorts.

Schaufele: The studies are the highest quality research studies available so far in regards to VP. However, there are several criticisms that are important and may explain the reason for not showing any difference in outcomes between VP and sham control:

• The Kalmes study screened a large number of patients, but enrolled less than 10% of them into the trial. Enrollment for most randomized controlled trials is normally in the range of 20% to 25%. This could have produced a significant selection bias.
• The recruitment problem in the Kalmes study is underlined by the fact that it required softening of the inclusion criteria and lowering of the overall enrolled number of patients. After changing the inclusion criteria, investigators introduced patients with a pain score as low as three out of 10 into the trial. Most physicians I know would consider this procedure only for patients with pain scores of at least five or six out of 10 on the VAS. In my practice, I would estimate that the pain scores are closer to eight or nine on average. Enrolling patients with such low pain scores most likely resulted in a selection bias toward patients with relatively little pain. Likely, these patients would have improved without any intervention anyway.
• Most of the patients enrolled appeared to have chronic fractures, with the average time of pain 4 to 5 months in the Kalmes study. As we all know, most of these fractures heal over time. Many patients may have been at the tail end of their fracture pain and were about to improve anyway. Most of us feel that patients with acute fractures and pain onset less than 3 months benefit the most from the procedure.
• Several people argue that the so-called sham procedure was really not a sham procedure. Local anesthetic was introduced to the pedicle. This effectively is the same procedure as a medial branch block, which by itself may have a temporary pain-relieving effect from pain originating from the posterior elements.
• There was a trend in both studies that VP patients did better, and in the end, the studies may just have been underpowered.
• My biggest issue is that there was no clear correlation between the patient’s pain location, severity of pain, imaging findings and pain on physical examination in either study. The Kallmes study did not even require MRI/bone scan on every patient.

The outcomes presented do not mirror my personal clinical experience. Personally, I feel that these studies are a start and we have to better define those patients who actually benefit from the procedure.

In my experience, patients with the following criteria seem to benefit the most from the procedure: painful compression fractures that are less than 3 months old; clear correlation of imaging findings and physical examination findings; localized tenderness over the fracture site on prone examination; and in patients with chronic fractures, those who have fracture-nonunion or cleft formation in the vertebra with associated instability do well with a VP and BK. I believe that fracture morphology and appearance can predict improvement from the procedures.

Clearly, we know now that VP is not the panacea we thought it was. Having said this, I still believe that it is a useful procedure in well-selected patients as I described. Future studies will show if my “clinical” opinion holds up to strict research methodology.

Tong: Ideally, both of these studies would have been done nearly a decade ago. Being recently completed means they were designed and completed in an environment where there was presumed superiority of VP based upon numerous good but nonrandomized studies. The study concession to these widely held beliefs was to allow study participants to cross over from one modality to another after 1 month. This crossover makes evaluation of the data beyond 1 month significantly more complicated. One could point to the higher crossover rates from sham to VP as evidence of VP’s superiority. Alternatively, this could simply represent “steering” patients to treatment either by their unblinded physicians or secondary to unintentional unblinding of study patients themselves.

The take-home message from both studies is that patients get better with both VP and the injection of local anesthetic, not that VP doesn’t work. The statistical justification for the specific study conclusions is robust, although it is certainly possible that one treatment arm may be superior to another to a more subtle degree than the studies were powered to detect. Furthermore, it is important to view both of these studies and make informed clinical decisions not in a vacuum, but in the context of the other published work. We believe that both studies raise worthwhile questions and would indicate that further randomized study evaluation is certainly warranted.

In terms of my experience, we have performed hundreds of VPs in our practice and have witnessed firsthand the dramatic clinical improvement published in many previous papers. Our experience also correlates with the clinical and functional improvements described in the two recent NEJM studies. However, we have not performed any of the so-called sham procedures outlined in the studies where only local anesthetic was injected. For this reason we can’t directly compare the study conclusions with our experience.

Diagnostic criteria

Boden: What diagnostic criteria do you use to identify a VP candidate?

Garfin: We percuss patients’ backs and place a marker where they hurt and then obtain AP and lateral radiographs to see if the area of pain reproduction, the marker, overlies or is near the level of fracture. If possible, we obtain MRI scans to assess for marrow edema that would be consistent with an ongoing, potentially painful fracture or a cleft. We tend to follow patients with serial radiographs over the first few weeks of nonoperative treatment to see if there is progressive collapse. If a patient cannot have an MRI scan, we obtain a bone scan and a CAT scan, the latter looking for significant canal violation or any evidence of tumor and/or infection.

Hecht: My diagnostic criteria first involves a careful history to determine if the patient has “new” back pain that is different from any pre-existing back pain. Many elderly patients have long-standing or intermittent back pain from other causes. At times, it may be helpful to see if this can be differentiated from the baseline lumber degenerative complaints.

I also use plain radiographs acutely and usually an MRI to assess the acuity of the fracture, or edema on STIR sequences, and to ensure there is not significant retropulsion of bone into the canal. If a patient cannot have an MRI, I utilize a bone scan or SPECT/CT scan to ensure there is increased activity on the bone scan. I will use a lumbar-extension radiograph over a bolster in cases where I suspect there may be a compression fracture nonunion (Kummel’s disease) to see if there is any gapping or motion at the fracture site. Physical examination techniques are often nonspecific, but can help differentiate a thoracic or higher lumbar fracture if there is any point tenderness.

Heller: My criteria include a history and physical examination, especially when palpation, percussion or manipulation of the fracture reproduces the pain; X-rays, especially when old films are available to help age a fracture; MRI to investigate the marrow signal, evaluate the spinal canal involvement and identify relevant issues, such as myeloma, metastases, infection, etc.; CT scans to render bone anatomy in detail as an operative road map; and comparative upright and supine hyper-extension films are useful for diagnosing chronic nonunions. Bone scans are not helpful.

Lane: Criteria for vertebral augmentation includes a scoliosis radiograph series standing; MRI with STIR positive in at least 25% of the body; a collapse of more then 20% but less than 75%; pain greater than four out of 10 on a VAS that is localized to the vertebra with the fracture; and evidence that the posterior wall is intact. Because osteoporosis and osteomalacia are key to this disease, I include DXA, 25(OH) vit D, SMA 12 for calcium and albumin and kidney function, intact PTH level, CBC and if anemia myeloma workup.

Lieberman: I would recommend a vertebral augmentation procedure to a patient with a progressive and painful osteoporotic or osteolytic vertebral collapse whose DXA is -2.5 or worse with pain emanating from verifiable levels on T2 sagittal STIR MRI images. In the absence of an MRI, I rely on the combination of history, prior X-rays, recent CT scans and technetium bone scans.

Tong: Plain films and noncontrast CT scans are helpful for identification of cortical disruption as well as potentially identifying other possible sources of back pain such as malignancy. If there is any doubt about the acuity of a fracture, noncontrast MRI with STIR techniques can be useful for demonstrating bone marrow edema as well as potential canal compromise using T2-weighted images. MRI is also useful for clarifying nonfracture sources of pain such as spinal or foraminal stenosis. Bone scan with SPECT imaging is helpful for determining activity level in a fracture of unknown age as well as identifying posterior element fractures, additional vertebral body fractures, and even rib fractures. In our experience, even chronic fractures manifesting bone marrow edema or focal radionuclide activity on bone scan may respond to treatment.

The clinical criteria are perhaps most important in patient selection. Focal appropriate point tenderness has been demonstrated to be a very prognostic indicator that a patient will respond clinically to vertebroplasty. Additionally, the presence of pain that does not correlate well with an appropriate underlying fracture, such as the presence of radicular symptoms or rib pain, would suggest against treatment.

Contraindications

Boden: What are your contraindications to a cement injection procedure?

Garfin: Infection, tumor with significant destruction involving the posterior cortex that could lead to cement leakage, significant acute bone retropulsion that could lead to significant leakage and/or retropulsion of bone — although this has not been observed frequently in osteoporotic fractures, or a patient who has fractures at nearly every level in his or her spine.

Hecht: My contraindications are related to burst fracture with retropulsion of fractures into the spinal canal, unstable fractures, coagulopathies, infection/sepsis, cord compression, and radiculopathy. I also try to avoid cement injections in younger patients, due to the lack of knowledge about the long-term consequences of PMMA in the spine. However, newer cements that are bioabsorbable are facilitating including these younger patients.

Heller: Fractures with only modest pain that can be well managed with medication while they heal; healed fractures with residual axial pain; neurological deficits associated with eligible fractures, unless the procedure is combined with a suitable decompression technique; an inability to come off anti-coagulants; presence of an infection, with a pathologic fracture resulting from vertebral osteomyelitis; oncologic fractures with sufficient osteolysis to prevent safe containment of PMMA; and the inability to adequately visualize the anatomy in two planes for the procedure.

Lane: Contraindications for vertebral augmentation are: lack of new fracture, meaning MRI STIR with less than 25% edema; pain not localized to the vertebra in question, collapse of less than 20% or greater than 75%; solid metastatic tumor; fracture of the posterior wall; dementia and pain less than four out of 10 on a VAS.

Schaufele: In addition to the typical procedure contraindications — uncontrolled bleeding disorder, anticoagulation, infection etc. — my contraindications are: pain that does not correlate with physical examination and imaging findings; pain not severe enough to cause significant functional limitations; imaging findings indicating a healed fracture; severe burst fractures resulting in significant spinal stenosis or posterior wall disruption; and/or vertebra plana.

Tong: My VP contraindications are: asymptomatic compression fractures — nontarget pain source such as posterior element fracture, rib fracture, or pain arising from spinal stenosis; satisfactory pain control on medical therapy; retropulsed fracture fragments greater than 1/3 of the spinal canal diameter; incomplete vertebral body posterior wall or intracanalicular extension of tumor; ongoing infection (local or systemic); vertebra plana; hypersensitivity to bone cement or uncorrectable coagulopathy; and burst fracture.

Nonoperative treatment

Boden: How much time with nonoperative treatment would you give a compression fracture before intervening if it did not improve?

Lieberman: VCFs are a biomechanical and a clinical pain issue. I would not leave a distal radius fracture in a malreduced or biomechanically compromised position. I would splint or cast the limb immediately. With that in mind, I would recommend a vertebral augmentation to anyone who presents with an osteoporotic or osteolytic VCF whose DXA is -2.5 or worse. Why let the spine collapse and the pain persist?

I typically brace those whose DEXA is between -2.5 and -1.0. If the VCF progresses or the pain persists, I recommend a vertebral augmentation. Anyone with a DEXA better than -1.0 should be treated with contemporary fracture techniques, as these are not osteoporotic VCFs. Ambulatory status and medical condition are considerations that could sway the decision either way after considering the biomechanical and pain issues.

Shaufele: To me, it largely depends on the severity of the patient’s pain and the associated functional impairments.

I may treat an elderly patient with multiple medical problems, severe pain and significant functional impairments as soon as possible. In my opinion, vertebral augmentation procedures allow these patients to be mobilized and pain-free much earlier, therefore avoiding the well-known complications of prolonged inactivity.

On the other hand, if a patient presents with more moderate pain, I would wait at least 6 weeks before I would offer a vertebral augmentation procedure.

If the patient’s pain is not severe and the functional limitations are mild, I may not recommend a vertebral augmentation procedure at all.

For more information:

• Scott D. Boden, MD, can be reached at 59 Executive Park South, The Emory Spine Center, Ste. 3000, Atlanta, GA 30329; 404-778-7143; e-mail: sboden@emory.edu. He receives royalties and is a consultant for Medtronic Sofamor Danek and Osteotech, and receives institutional support from Synthes, Medtronic Sofamor Danek and DePuy.
• Steven R. Garfin, MD, can be reached at UCSD Dept of Orthopaedics, 350 Dickinson St. Ste. 121, San Diego, CA 92103; 619-543-2644; e-mail: sgarfin@ucsd.edu. He receives royalties from DePuy, is on the speakers board for Biomet, Blackstone Medical, DePuy and Nuvasive, and is a consultant for DePuy, Kyphon, and Nuvasive, and receives institutional support from Abbott, Arthrocare, Biomet, DePuy, Kyphon, Lippencott, NIH, Nuvasive, Stryker and Trimed.
• Andrew C. Hecht, MD, can be reached at 5 E 98th St., Box 1188, New York, NY 10029; 212-241-0735; e-mail: Andrew.Hecht@mountsinai.org.
• John G. Heller, MD, can be reached at 59 Executive Park South NE, Ste. 3000, Atlanta, GA 30329-2208; 404-778-7112; e-mail: jhell02@emory.edu. He receives royalties from Medtronic; is on the speakers bureau for Abbott and Medtronic; receives institutional support from Abbott, Medtronic and Synthes; and owns stock in Medtronic.
• Joseph M. Lane, MD, can be reached at Hospital for Special Surgery, 535 E. 70th St., New York, NY 10021; 212-606-1172; e-mail: lanej@hss.edu. He receives royalties from GlaxoSmithKline, Eli Lilly, Procter and Gamble, sanofi-aventis, Novartis and Roche, and is a paid consultant for Biomimetics, Orthovita, Osteotech, Zimmer, Innovative Clinical Solutions, D’Fine, Biomimetics, Soteria and Zelos Therapeutics.
• Isador H. Lieberman, MD, MBA, FRCSC, can be reached at Cleveland Clinic Florida Spine Institute, 2950 Cleveland Clinic Blvd., Weston, FL 33331; 954-659-5635; e-mail: lieberi@ccf.org. He receives royalties from AxioMed, MAZOR Surgical Technologies, Merlot Orthopedix, Pearl Diver, Stryker Spine and Crosstrees Medical; is a consultant for, has stock in and receives other support from AxioMed, Mazor Surgical Technologies, Merlot Orthopedix, Trans 1, and Crosstrees Medical; and receives institutional support from AxioMed, Mazor Surgical Technologies, Merlot Orthopedix, Trans 1, Crosstrees Medical, Medtronic, Orthovita, Stryker Spine and Depuy.
• Michael K. Schaufele, MD, can be reached at Emory Orthopaedics & Spine Center, 59 Executive Park South, Atlanta, GA 30329; 404-778-7000; e-mail: mschauf@emory.edu. He has no direct financial interest in any product or company mentioned in the article.
• Frank C. Tong, MD, can be reached at Emory University Hospital, 1364 Clifton Road, NE, Suite A121, Atlanta GA 30322; 404-712-4991; e-mail: ftong@emory.edu.