The article in this issue of the AJNR by Lin et al represents an important addition to the growing literature on percutaneous vertebroplasty. The authors have studied the effect of vertebroplasty on the risk of subsequent fracture in a subpopulation of treated patients, specifically those in whom cement was noted to extravasate into the intervertebral disk space. Based on a sample of 14 incident fractures in 34 patients, the authors conclude that the presence of cement within the intervertebral disk increases the risk of subsequent fracture and, thus recommend that such extravasation should be avoided, if possible.
Since inception of the technique, practitioners of percutaneous vertebroplasty have questioned whether the presence of a rigid, acrylic polymer within a vertebral body would increase the risk of fracture of neighboring vertebra. The answer to this question is highly relevant, since it would impact not only the overall safety of the procedure but also would address the wisdom of prophylactic vertebroplasty. Numerous biomechanical and clinical studies have addressed the question, yet no consensus has been reached. Several daunting challenges face those researchers attempting to answer the question, “Does vertebroplasty increase the risk of subsequent fracture?” Perhaps the most difficult challenge is identification of an appropriate control group. The development of subsequent fractures following vertebroplasty is a complicated issue made increasingly difficult by a lack of understanding of the natural history of osteoporotic vertebral collapse. It is well documented that patients presenting with a new osteoporotic compression fracture have a fivefold increased incidence of developing a new compression fracture within a year. In addition, certain vertebral bodies; eg, those located at the thoracolumbar junction, are more susceptible to fracture than others. What has not been described in affected, untreated individuals is the order in which osteoporotic vertebral bodies collapse and the degree of fracture susceptibility at individual levels. If one is to prove that vertebroplasty increases the risk of adjacent fractures, then the natural progression of the disease, including location of subsequent fractures, must first be identified.
Lin et al have circumvented the difficulty finding an appropriate control group by markedly narrowing the scope of their study to include only patients with or without endplate extravasation. While this focus limits the overall applicability of their findings, it does offer the ability to perform statistical comparisons between groups. If one accepts the authors conclusions that endplate extravasation increases the risk of subsequent fracture, then one might be tempted to alter one’s practice to minimize or eliminate such extravasation, as suggested by the authors. This would include, as noted by the authors, such maneuvers as placing the needle away from the center of the vertebral body, altering the consistency of the cement, and diminshing the volume of cement injected.
We would readily accept a common sense approach to avoid large amounts of cement deposition into the disk; however, before one accepts the authors’ recommendations to systematically avoid the endplate region of the vertebral body, a critical assessment of their data would be warranted. In our opinion, two specific issues merit further scrutiny. First, proof of association does not necessarily indicate causation. In other words, even though endplate extravasation may be associated with increased fracture risk, it is not yet proved that the extravasation caused the fractures. Second, it remains unclear whether avoidance of peri-endplate deposition of cement would limit the efficacy of vertebroplasty, and thus modification of technique should be done cautiously.
1) Does the association between endplate extravasation and subsequent fracture prove that such extravasation caused the fractures? Some doubt regarding this conclusion is raised by the lack of correlation between “small” and “large” amounts of extravasation in the authors’ own study. Further, existing biomechanical data suggest that, rather than the disk space itself, it is the bowing of the vertebral endplate with loading that allows cushioning of an axial load. If one aims to fill a fracture line in the subendplate region, then perhaps increased fracture risk will prevail even in the absence of disk space extravasation. Alternatively, the biomechanical alteration of the involved disk space by the mere presence of an endplate fracture may predispose a patient to fracture of the opposing endplate. Additional doubt regarding the causative effect of disk space dysfunction may be raised by the fact that other disk space abnormalities, such as dessication and degenerative disk disease, have not been associated with increased fracture risk, to our knowledge. Finally, it remains possible that significant selection bias may explain the apparent association between endplate extravasation and fracture. For example, the presence of large, nonhealing subendplate fracture lines, which likely increase the incidence of disk space extravasation, may be a marker for a more “aggressive” osteoporosis with higher likelihood of additional fracture, irrespective of therapy.
2) Would systematic avoidance of peri-endplate cement deposition diminish the efficacy of the procedure? The first decade of widespread application of vertebroplasty has failed to give definitive insight into the procedure’s mechanism of action. It remains possible that pain relief is afforded by filling of micro- and macro-fractures within the vertebra. If one accepts the authors’ recommendation to avoid peri-endplate cement deposition, then the procedure may fail to relieve pain. There is no clinical data in the study under consideration to suggest whether pain relief was affected by cement deposition in and around the endplate. Two small clinical studies have addressed the question with diametrically opposed results. Substantially larger studies are needed to show this effect. Thus, when one identifies on preprocedural imaging a fracture line in a subendplate location, it seems counterintuitive to purposefully avoid depositing cement in the fracture line, irrespective of the perceived risk of disk space extravasation.
In summary, the study by Lin et al is a welcome addition to a growing but still immature literature on percutaneous vertebroplasty. Their data are intriguing and point to the need for further study of the effect of vertebroplasty on the natural history of osteoporosis, specifically regarding risk of subsequent fracture. The study may however raise more questions than it answers, and systematic modification of vertebroplasty technique should await the availability of larger, prospective studies.