Recombinant human bone morphogenetic protein-2 (rhBMP-2) combined with an absorbable collagen sponge has been shown to be a safe and effective replacement for autogenous iliac crest bone graft in anterior lumbar interbody arthrodesis1. In a pooled analysis of 679 patients who had undergone lumbar interbody arthrodesis, the use of rhBMP-2 was associated with a significantly higher fusion rate than was the use of iliac crest bone graft (94.4% compared with 89.4%; p = 0.022)2. These fusion rates were consistent with those in other published reports on the use of rhBMP-2 in lumbar interbody arthrodesis1-7.
Reported fusion rates after posterolateral lumbar arthrodesis procedures done with iliac crest bone graft vary widely, from 73% to 95%8-13, and depend on numerous factors, such as the diagnosis and the methods used to assess the fusion. We are not aware of any published prospective randomized trials on the use of rhBMP-2 combined with an absorbable collagen sponge in posterolateral arthrodesis. When this combination was first investigated in nonhuman primate models to evaluate its ability to induce a posterolateral fusion, it was observed that the rhBMP-2-soaked absorbable collagen sponge alone did not lead to a robust fusion mass14. When the rhBMP-2-soaked sponge was combined with an osteoconductive bulking agent capable of providing resistance against compression and longer-term scaffolding, robust bone formation and high rates of fusion were reported15,16. These studies illustrated the role of chemical composition and showed that ceramics with a high hydroxyapatite content result in fusion masses containing residual ceramic material, whereas ceramics high in beta tricalcium phosphate have been found to lead to smaller fusion masses as a result of rapid remodeling17,18. A 15% hydroxyapatite/85% tricalcium phosphate composition (MASTERGRAFT Resorbable Ceramic Granules; Medtronic Sofamor Danek, Memphis, Tennessee) demonstrated an optimal resorption profile and led to high rates of fusion when used as an osteoconductive bulking agent with rhBMP-2 on an absorbable collagen sponge15-17.
Our purpose was to investigate the use of rhBMP-2 on an absorbable collagen sponge combined with 15% hydroxyapatite/85% tricalcium phosphate ceramic granules as a replacement for iliac crest bone graft in posterolateral arthrodesis with instrumentation.
Study Design
Institutional review board approval was obtained, and all patients gave their informed consent before participating in this study, which was conducted at four centers by seven surgeons. Fifty patients scheduled to undergo treatment for degenerative disc disease were assessed for inclusion in this prospective, randomized, multicenter, U.S. Food and Drug Administration (FDA)-approved, Investigational Device Exemption (IDE) pilot trial. Patients were randomly assigned to either the investigational group or the control group. Randomization was stratified by site with a fixed block size of four. After consent and randomization, two patients in each group elected not to participate in the study. Forty-six patients were enrolled from April 2003 to August 2004 (Fig. 1). Twenty-five patients in the investigational group received rhBMP-2 on an absorbable collagen sponge combined with a 15% hydroxyapatite/85% tricalcium phosphate ceramic bulking agent (INFUSE/MASTERGRAFT; Medtronic Sofamor Danek) and twenty-one patients in the control group received autogenous iliac crest bone graft as part of a single-level posterolateral arthrodesis with instrumentation. Preoperatively, all patients met the inclusion criteria of symptomatic single-level degenerative disc disease at a level from L1 to S1. Degenerative lumbar disc disease was confirmed by the patient's history; objective physical findings; and correlative neuroradiographic studies showing instability, osteophyte formation, decreased disc height, thickening of ligamentous tissue, disc degeneration or herniation, or facet joint degeneration. The clinical symptoms included low back pain and radicular leg pain. All patients had failed to respond to nonoperative treatment for a minimum of six months. Additional enrollment criteria were no spondylolisthesis or grade-I spondylolisthesis (a spinal condition that occurs when =25% of the vertebral body has slipped forward over the adjacent disc) and no previous arthrodesis at the index level. We did not identify any differences between the demographic profiles of the two treatment groups (Table I).
Surgical Technique
Device Preparation
The investigational group received implants of rhBMP-2 on an absorbable collagen sponge combined with ceramic granules. A single 10.2 × 7.6-cm sponge from a Large II Kit of INFUSE Bone Graft (Medtronic Sofamor Danek) was sectioned into two 5.1 × 7.6-cm pieces (Fig. 2). RhBMP-2 was reconstituted in the standard fashion at a concentration of 1.5 mg/mL, and 4 mL of this solution was applied to each of the two sponge strips. Thus, each patient in the investigational group received 12 mg of rhBMP-2 on 8 cm3 of collagen sponge. The ceramic granules used as the bulking agent were a blend of 15% hydroxyapatite and 85% tricalcium phosphate (MASTERGRAFT Resorbable Ceramic Granules). No rhBMP-2 was applied directly to the ceramic. The rhBMP-2 was allowed to bind to the collagen sponge for at least fifteen minutes, after which 5 cm3 of the ceramic granules were distributed on and rolled within each of the collagen sponge strips.
Surgical Procedure
Both patient groups underwent a posterolateral arthrodesis with instrumentation through a standard open midline approach; the Cotrel-Dubousset Horizon Spinal System (Medtronic Sofamor Danek) was used for all patients. Posterior decompression of the spinal canal was carried out as needed. The transverse processes were exposed bilaterally, and their dorsal surfaces were decorticated. The morselized autogenous iliac crest bone graft or the rhBMP-2-soaked absorbable collagen sponge wrapped around the ceramic matrix was then placed into the lateral gutters and along the pars interarticularis. Pedicle screws were placed and rods were attached. Importantly, all of the local bone removed as a result of the decompression was discarded and not used as graft in either group. Interbody arthrodesis was not performed in either group.
Radiographic Analysis
Radiographic assessment of fusion was performed at six, twelve, and twenty-four months after the surgery by two independent radiologists who were blinded to the treatment assignment. A third reviewer was available for adjudication. Plain lateral flexion and extension radiographs and thin-cut computed tomography scans with sagittal and coronal reconstructions were used to evaluate the fusion mass. A successful fusion was defined, on the basis of the IDE criteria, as bridging trabecular bone between the transverse processes, the absence of motion (with motion defined as =3 mm of translation and >5° of angulation on flexion-extension views), and the absence of radiolucent lines through the fusion mass (Fig. 3).
Clinical Outcome Measures
Clinical and functional outcomes were assessed preoperatively; at six weeks postoperatively; and at three, six, twelve, and twenty-four months postoperatively. Clinical outcomes were measured with use of the Oswestry Disability Index (ODI)19, components of the 36-Item Short Form Health Survey (SF-36)20, back and leg pain scores, and return-to-work status. Back and leg pain scores were determined with use of numeric rating scales ranging from 0 to 20 points, with a maximum of 10 points assigned for the frequency of the pain and a maximum of 10 points assigned for its intensity.
Overall Success
The primary outcome of this study was overall success, defined as a combination of successful fusion, an improvement in the ODI score of >15%, the absence of severe device-related adverse events, the absence of a second surgical procedure (revision, removal, or supplemental fixation) involving the index level, and maintenance or improvement of neurological status. All of those criteria had to be met for the outcome to be considered successful.
Statistical Methods
For comparisons of demographic characteristics, preoperative measurements, and surgical data between the treatment groups, analysis of variance was used for continuous variables and the Fisher exact test was used for categorical data. A paired t test was used to assess the significance of postoperative improvement in the outcome score, as compared with the preoperative score, within each treatment group. To compare clinical and radiographic outcomes between the treatment groups, analysis of covariance was performed with use of either a general linear model for continuous responses or a logistic model for binary responses to adjust for potential confounding effects of Workers' Compensation, litigation, and previous back surgery. Significance was defined as a p value of <0.05.
Source of Funding
Medtronic Spinal and Biologics, Memphis, Tennessee, sponsored this FDA-approved IDE clinical trial. The authors are consultants or clinical investigators, or both, for Medtronic Spinal and Biologics, which distributed the device that was studied.
Surgery
Surgical data for the two groups are shown in Table II. The operating times and hospital stays were similar for the two groups. The investigational group had a trend toward less blood loss. The mean volume (and standard deviation) of iliac crest bone graft used in the control group was 42.4 ± 29.9 cm3 (range, 10 to 80 cm3).
Patient Flow and Clinical Follow-up
At twenty-four months, 88% (twenty-two) of the twenty-five patients in the investigational group and 86% (eighteen) of the twenty-one patients in the control group were available for follow-up (Fig. 1). (According to the FDA-approved IDE protocol, patients who were seen at this interval, with the exception of those who had had a failure of treatment even if they had actually been seen for follow-up, were included. Therefore, a patient who had had a second surgical procedure [a failure] before a follow-up interval was not considered to have completed follow-up at that time interval; however, radiographic and clinical outcome data for that patient were included in the analysis at that time interval and his or her data were carried forward to twenty-four months.) Two patients had a second surgical procedure and one patient died in the investigational group, and two patients had a second surgical procedure and one patient did not return for a twenty-four-month follow-up examination in the control group. The two second surgical procedures in the control group occurred between twelve and twenty-four months; therefore, although these patients were not considered to have completed the twenty-four-month follow-up, outcome data at twenty-four months were included in the analysis. Data for the patient in the investigational group who died and the patient in the control group who was lost to follow-up were not included in the analysis at twenty-four months. Thus, twenty-four-month radiographic and clinical outcome data were available for twenty-three patients in the investigational group and twenty patients in the control group.
Clinical outcome data (ODI, SF-36, back pain, and leg pain scores and return-to-work status) were evaluated for twenty-three patients in the investigational group and twenty patients in the control group at twenty-four months. Fusion success was determined for the patients who had complete radiographic follow-up (all appropriate and readable radiographs) or had had a second surgical procedure (a failure), or both, at twenty-four months (nineteen patients in the investigational group and twenty patients in the control group). Twenty-one patients in the investigational group and twenty patients in the control group had radiographic, second-surgery, and clinical outcome data that allowed for evaluation of overall success.
Clinical outcome measures showed significant improvements compared with the preoperative scores in both groups. The investigational group exhibited a trend toward greater improvements in these outcome measures. This trend was observed both before and after we performed the adjustment for Workers' Compensation, litigation, and previous back surgery, with the general observation that p values increased following the adjustment. At twenty-four months, the mean ODI score had improved 28.2 points, as compared with the preoperative score, in the investigational group and had improved 23.0 points in the control group (p = 0.953) (Fig. 4). Ninety-one percent (twenty-one) of the twenty-three patients in the investigational group and 70% (fourteen) of the twenty patients in the control group had a =20% improvement in the ODI score (p = 0.532).
At twenty-four months, the investigational group had slightly greater mean improvements, as compared with the control group, in the physical component summary score of the SF-36 (13.0 compared with 9.9 points; p = 0.927) and the physical functioning score of the SF-36 (36.3 compared with 18.5 points; p = 0.200). At twenty-four months, the mean back-pain scores in the investigational and control groups had improved, as compared with the preoperative scores, by 9.6 and 7.2 points (p = 0.664), respectively, and the mean leg-pain scores had improved by 9.3 and 7.2 points (p = 0.892), respectively.
Fusion Assessment
At each time interval, there was a trend toward more successful fusions in the investigational group. The percentage of patients with evidence of fusion, as defined by the stringent IDE criteria, was 22% higher in the investigational group than in the control group at six months (eighteen of twenty-two compared with twelve of twenty; p = 0.160), 16% higher at twelve months (seventeen of twenty-one compared with thirteen of twenty; p = 0.359), and 25% higher at twenty-four months (eighteen of nineteen compared with fourteen of twenty; p = 0.174).
Computed tomography scans with thin-cut (1-mm) axial slices along with coronal and sagittal reconstructions were evaluated to assess the presence of bridging trabecular bone. At six, twelve, and twenty-four months, 89% to 95% of the patients in the investigational group had evidence of bridging bone. At the time of the last follow-up, nineteen of twenty patients in the investigational group and twelve of eighteen patients in the control group had bridging trabecular bone (p = 0.120; Table III). Thin-cut computed tomography scan reconstructions demonstrated the progressive formation of bridging bone across the transverse processes and showed incorporation and remodeling of the ceramic component (Fig. 5).
Additional Surgery or Other Treatment
Two patients (one in each group) sustained incidental durotomies intraoperatively, which were repaired. A wound infection developed at the surgical site in two patients (one in each group) and resolved after antibiotic treatment. One patient in the control group had an infection at the graft donor site. There were two second surgical procedures that were considered failures of the index procedures in the investigational group: a revision procedure was done one day after the initial surgery because of bilateral malpositioned pedicle screws, and hardware was removed from another patient at six months. Two revision procedures were performed because of pseudarthrosis in the control group.
Overall Success
At twenty-four months, the overall success rate was 81% (seventeen of twenty-one) in the investigational group compared with 55% (eleven of twenty) in the control group (p = 0.345). Ninety-six percent (twenty-two) of the twenty-three patients in the investigational group compared with 75% (fifteen) of the twenty patients in the control group had an improvement in the ODI score of >15% (p = 0.240).
Return to Work
By twenty-four months, 35% (eight) of the twenty-three patients in the investigational group were working, and all six patients who had been working before the surgery were working at twenty-four months. Of the twenty patients in the control group, 30% (six), including six of the nine who had been working before the surgery, had returned to work.