The three major options for reconstruction of the disk space during ACDF are iliac crest autograft, structural allograft, and synthetic cages, usually made from PEEK or titanium. While autograft has the highest fusion rate, it is associated with significant morbidity and increased operating room time, so surgeons use it relatively infrequently. Additionally, acceptably high union rates have been reported using allograft and an anterior plate. Allograft is widely used in North America, though it is used less frequently in other parts of the world where it is less widely available or not used for cultural or regulatory reasons. Device companies developed synthetic cages as an alternative to autograft or allograft and have the advantage of being more resilient and not at risk for resorption. However, they are not osteoconductive and potentially limit the volume in the disk space where fusion bone could form. No sufficiently powered RCT has been performed comparing fusion rates between allograft and synthetic cages, and such a study would require a high number of patients given that nonunion is a relatively uncommon event. In order to attempt to answer this question, Pirkle and colleagues from Chicago used the PearlDiver database to compare nonunion rates between over 4,000 ACDF patients treated with allograft and over 2,000 treated with a synthetic cage. The two groups were similar at baseline in terms of tobacco use, diabetes, and number of levels fused. Overall, the allograft group had a nonunion rate of 2% compared to 5.3% in the cage group, a highly statistically significant difference. The authors also performed subgroup analyses stratified by tobacco use, diabetes and number of levels fused. They found a lower nonunion rate for the allograft group in all subgroup analyses, and the differences were statistically significant in 25/26 of the subgroups.
The authors have done a nice job using an administrative database to attempt to answer a question that would be very difficult to answer in a prospective RCT. It is hard to imagine that any group would fund such a study that would likely require thousands of patients to be sufficiently powered, especially for subgroup analyses. This study has all of the limitations inherent in a retrospective, administrative database review. All such studies are affected by potential coding inaccuracies, though with this many patients, that seems unlikely to be a major problem. The biggest limitation is likely the subjectivity of the main outcome, namely a provider coding nonunion. The most serious potential confounder is the possibility that surgeons who use cages are more likely to code for nonunion. Using a synthetic cage results in higher reimbursement than structural allograft, and it is possible that surgeons more attuned to reimbursement issues may also be more likely to diagnose a nonunion and perform a revision procedure. It would have been interesting if the authors had also included an analysis of reoperation rate, which would have theoretically been associated with nonunion rate. Given that a Level 1 study is probably not going to provide the answer to this question, surgeons need to rely on the lower level data available and biological theory when making the decision about graft choice. There is no data suggesting that synthetic cages lead to better outcomes, and they generally cost more than allograft. Combining that with the results of this study would suggest that structural allograft is likely the preferred graft choice. The main motivation to use a synthetic cage may be the irrationally higher reimbursement for a procedure that takes the same amount of time and effort.
Please read Mr. Pirkle's article on this topic. Does this change your thoughts about graft/spacer choice in ACDF? Let us know by leaving a comment on The Spine Blog.
Adam Pearson, MD, MS
Associate Web Editor