Bone formation with rhBMP-2 is robust and beneficial, but HO-related complications are significant, and identifiable on assessment of radiculopathy symptoms and CT characteristics.
DISCUSSION
We identified 38 patients with a recognizable and consistent pattern of complications of off-label use of rhBMP-2 in TLIF performed at our institution between 2002 and 2015. This pattern included consistent radiculopathy symptoms with corresponding HO at the annulotomy site in continuity with bone in the disk space or ectopic bone forming a distinctive shell around the thecal sac or nerve roots, as well as showing a distinct pseudo-pedicle pattern encompassing nerve roots and the thecal sac. Our finding differs from other findings of similar complication characteristics, but with much larger variations without consistency within the patient population.19,20,22,24 Specifically, previous studies found an association between off-label rhBMP-2 use in the posterior spine and radiculopathy with and without neuroforaminal HO. However, our study found consistent radiculopathy symptoms with pseudo-pedicle-like HO complications in all its 38 patients a mean (SD) of 3.8 (1.0) months after surgery.
In this study, consistent radiculopathy symptoms with pseudo-pedicle-like HO complications were independent of the amount of rhBMP-2 used, as some complications occurred with use of small pack rhBMP-2 with TLIF. It is well understood that high doses of rhBMP-2 may be required to improve fusion rates, but to our knowledge an optimal dosing strategy for TLIF has not been reported, particularly with respect to potential complications.8,20,31-33 For anterior lumbar interbody fusion surgery, the FDA-approved use of rhBMP-2 appears to have a significantly decreased risk of neuroforaminal HO complications. This may be attributable to the protective presence of the intact posterior annulus and longitudinal ligament for this procedure.20,33 For TLIF, it has been suggested that rhBMP-2 should be placed only along the anterior annulus with a posterior strut and morselized bone allograft barricade,33 and that fibrin glue should be used to limit BMP diffusion through the annulotomy site31 to prevent this complication.
Our study results suggest that radiculopathy symptoms with pseudo-pedicle-like HO complications appear to be caused by leakage of rhBMP-2 from the disk space through the annulotomy site. This was often initially interpreted incorrectly on MRI in the first year after surgery as being fibrous or granulation tissue, or even postoperative changes that the heterotopic tissue was bone was obvious only on CT. Even then the tissue may be incorrectly identified, as the encasing nerve roots in bone are similar to the scar tissue having no compressive effect. HO may compress, but it also has an inflammatory component that the scars lack. Additionally, the HO from the disk space, caused by leakage of the BMP placed in or around the fusion cage, can create a pseudo-pedicle of varying size and extent. This was present in all 38 of our cases.
This retrospective case series had its limitations. Its clinical and radiographic findings were not blinded. Confounding variables cannot be isolated for causal relationships, if any, to the complication in a case series such as this.
Bone formation with rhBMP-2 is robust and beneficial, but HO-related complications are significant, and identifiable on assessment of radiculopathy symptoms and CT characteristics.