Recombinant human bone morphogenetic protein-2 (rhBMP-2) was approved by the United States Food and Drug Administration in 20021. Many U.S. spine surgeons rapidly embraced this product, more often than not in an off-label fashion, without site-specific data on dosage, safety, or efficacy1. Spinal applications of rhBMP-2 are well attested to in the medical literature. Michielsen et al. make a worthy addition with their article. It is a well-designed and well-executed prospective, randomized, single-blinded study. It is sufficiently powered to detect clinically important differences in visual analog scale pain scores and Short Form-36 scores. Each group is one patient shy of the twenty necessary for detecting such a difference in the Oswestry Disability Index. Their data support the null hypothesis that there was no difference in the clinical outcomes between the study and control groups. The second hypothesis, that there is slower trabeculation with BMP, was substantiated. The use of strict inclusion and exclusion criteria, uniform implants and technique, along with a single surgeon yields high-quality evidence.

    Of course, these same factors limit generalizability. A spine surgeon performing a single-level posterior lumbar interbody arthrodesis in patients similar to those in the Michielsen study can probably use either BMP or autogenous iliac crest bone graft, and achieve a good clinical outcome and osseous fusion in the short term. But this assumes comparable surgical skills. If a different technique and different implants are used, the average surgeon may not be able to replicate Michielsen’s outcomes. These data may not apply to multilevel posterior lumbar interbody arthrodeses. They may not extrapolate to the related transforaminal lumbar interbody arthrodesis technique. As the spine surgeon considers other arthrodesis approaches and other regions of the spine, the Michielsen study offers no meaningful guidance concerning the use of rhBMP-2.

   There are other limitations. Faced with the type of patient treated by the authors, I would choose autogenous bone to perform this type of arthrodesis. But Michielsen did not describe the graft harvest technique in detail, which could explain the minimal donor-site morbidity reported. I suspect that they removed the small volume of cancellous bone through a cortical window, minimizing muscle stripping. Some have suggested that morbidity at the site of the bone-graft harvest may be overstated in some studies; however, in reality, technique and volume harvested are probably important factors2. The authors also did not report length of stay. Nor did they assess cost. Could the additional operating-room time for graft harvest partially offset the cost of BMP? Did the control group require transfusion, longer hospitalization, or more analgesia, adding to the expense of autograft? The graft dosage was effective, but was it optimal? Although the ectopic bone formation was asymptomatic, the authors only provide two-year outcomes. They expressed intent to follow the study group with computed tomography three years after surgery. This will provide important data about the ossification over time. Does it progress, stabilize, or regress? I would also urge them to obtain clinical data on all thirty-eight patients. It is possible the heterotopic ossification produces late symptoms. It is also possible that a difference in clinical outcome between groups develops over time. These questions and comments are not criticisms but rather reflect questions raised by a well-done study.