Mid-term efficacy and safety of Escherichia coli-derived rhBMP-2/hydroxyapatite carrier in lumbar posterolateral fusion: a randomized, multicenter study

This study aimed to evaluate the mid-term efficacy and safety of Escherichia coli-derived bone morphogenetic protein-2 (E.BMP-2)/hydroxyapatite (HA) in lumbar posterolateral fusion (PLF). This multicenter, evaluator-blinded, observational study utilized prospectively collected clinical data. We enrolled 74 patients who underwent lumbar PLF and had previously participated in the BA06-CP01 clinical study, which compared the short-term outcomes of E.BMP-2 with an auto-iliac bone graft (AIBG). Radiographs and CT scans were analyzed to evaluate fusion grade at 12, 24, and 36 months. Visual analog scale (VAS), Oswestry disability index (ODI), and Short Form-36 (SF-36) scores were measured preoperatively and at 36 months after surgery. All adverse events in this study were assessed for its relationship with E.BMP-2. The fusion grade of the E.BMP-2 group (4.91 ± 0.41) was superior to that of the AIBG group (4.25 ± 1.26) in CT scans at 36 months after surgery (p = 0.007). Non-union cases were 4.3% in the E.BMP-2 and 16.7% in the AIBG. Both groups showed improvement in pain VAS, ODI, and SF-36 scores when compared to the baseline values, and there were no statistically significant differences between the two groups. No treatment-related serious adverse reactions were observed in either group. No neoplasm-related adverse events occurred in the E.BMP-2 group. The fusion quality of E.BMP-2/HA was superior to that of AIBG. E.BMP-2/HA showed comparable mid-term outcomes to that of AIBG in terms of efficacy and safety in one-level lumbar PLF surgery.


Introduction
In lumbar fusion surgery, achieving solid bone fusion is essential to accomplish good clinical outcomes because it affects postoperative functional outcome, low back pain and postoperative radiating pain [1,2]. However, fusion rate of posterior lumbar fusion has stagnated to 88-95%, and approximately 10% of patients still experience nonunion [3].
Although autologous iliac bone graft (AIBG) is the gold standard, approximately 18% of patients reported graft site pain or discomfort 2 years after surgery [4]. In an effort to enhance the fusion process, application of recombinant human bone morphogenetic protein-2 (rhBMP-2) resulted in satisfactory fusion rates in anterior lumbar interbody fusion and posterolateral fusion (PLF) [4][5][6].
Currently, most rhBMP-2 is being produced in Chinese hamster ovary (CHO) cells which are mammalian cells. As CHO cell BMP-2 (C.BMP-2) is produced by purifying the secreted product through post-translational modification of mammalian cell, its production efficacy is very low, making mass production difficult [7,8]. To overcome these shortcomings, Escherichia coli-derived rhBMP-2 (E.BMP-2) has been introduced as a new source of rhBMP-2 [9,10]. Several in vitro and in vivo studies have shown that the biological activity of E.BMP-2 is comparable to that of C.BMP-2 [8,[11][12][13]. Previously the authors reported the efficacy and safety outcomes of E.BMP-2 with hydroxyapatite (Ca 10 (PO 4 ) 6 (OH) 2 , HA) [14] carrier in single-level PLF and the result was comparable to that of AIBG at 6 months. [15] However, the follow-up period was a short time to confirm the safety of the graft. Also, there was limitation to previous study because the fusion evaluation was done before HA carrier was absorbed. Furthermore, follow-up studies for more than 3 years after using E.BMP-2 in lumbar PLF have not been reported.
Therefore, this study aimed to report the mid-term efficacy and safety outcomes of E.BMP-2/HA compared with those of AIBG in patients who underwent 1 level PLF.

Study design
The present study is a multicenter, evaluator-blinded, prospective, and retrospective observational study that aims to establish the mid-term safety and efficacy in patients who had received E.BMP-2/HA or AIBG through lumbar PLF in a previously performed clinical trial(BA06-CP01) [15].
Among the subjects who completed the previous study conducted from March 2013 to 2016, patients who were followed-up for > 36 months with available efficacy data were enrolled and analyzed.
The inclusion criteria of the BA06-CP01trial included patients ≤ 80 years old and who required 1-level posterior lumbar fusion because of spinal stenosis, grade 1 spondylolisthesis or spondylolysis. The exclusion criteria were as follows: average spine T-score < − 3.0 on dual-energy X-ray absorptiometry, history of cancer, hypocalcemia, and specific conditions with internal, endocrine, or psychiatric disorders that makes lumbar fusion difficult or impossible. [15] Basic information, medical history/surgical history, clinical questionnaire results (Oswestry disability index [ODI], SF-36, and VAS), radiological outcomes (CT and X-ray), group allocation information, and ongoing adverse events were collected before surgery from the BA06-CP01 clinical trial. Hospital records of subjects were retrospectively collected at 12, 24, and 36 months. If patients in the BA06 CP01 clinical trial voluntarily visited the hospital after 36 months, written consent was obtained and they were included as participants. Retrospective study data were collected, and prospective visiting participants were recruited until July 31, 2020. This study was approved by the institutional review board of each institution and was conducted according to the Declaration of Helsinki and the guidelines for Good Clinical Practice.

Intervention
Lumbar PLF was performed routinely. Following the posterior midline approach, decompression with laminectomy and flavectomy was performed. Pedicle screw fixation was performed at the involved level and the allocated bone graft materials were applied between the two transverse processes [15]. In the E.BMP-2 group, we used Novosis (CG Bio Co., Ltd., Seoul, Korea), an E. coli-derived rhBMP-2 with an HA carrier. Approximately 3 g (8 cc) of HA was soaked in one vial (3.0 mg) of E.BMP-2 and carefully applied in the inter-transverse space. This process was repeated on the contralateral side. In the AIBG group, approximately 8 cc of auto-iliac bone graft was applied on each side. Bone grafts obtained via laminectomy were not used in either group.

Radiologic outcome
The fusion grade was mainly assessed by bone bridging in coronal reconstruction images of 36-month three-dimensional CT (3D CT, thin cut, 2 mm or less interval) scans. The fusion grade was additionally and independently evaluated by radiographs at 12/24/36 months. The fusion status was assessed by two blinded independent orthopedic surgeons who did not participate in the clinical trial. The fusion grade was defined as follows: grade 1, no fusion; grade 2, partial or limited unilateral; grade 3, partial or limited bilateral; grade 4, solid unilateral; and grade 5, solid bilateral [16]. To ensure stricter fusion rate evaluation than in the previous study, only fusion grades 4 and 5 were defined as successful fusion. If the evaluators' opinions differed, a final consensus was obtained. The width of the fusion mass was measured using a CT scan at ≥ 36 months, which was then summed and recorded. To evaluate the maturation of the fusion mass, trabeculation of the fusion mass was evaluated. Ectopic bone formation near the intervertebral foramen was also evaluated.

Clinical outcome
Pain VAS (lower back, right leg, and left leg), ODI and SF-36 scores were evaluated preoperatively and at 36 months postoperatively. Changes in the ODI, SF-36 and VAS scores at ≥ 36 months from baseline were assessed. Moreover, the incidence rate of additional interventional surgeries was also recorded.

Safety evaluation
All adverse events were standardized as "System Organ Class" and "Preferred Term" using MedDRA 23.0. [17] Adverse events (subjective or objective symptoms) occurring after surgery were evaluated at each follow-up visit of each body part. Moreover, the relationship between each event was assessed for its relationship with E.BMP-2. Death, life threat, prolonged hospitalization, severe disability, or continued incapacity were assessed as severe complications.

Statistical analysis
Statistical analyses were performed using the Statistical Analysis System (SAS Institute Inc.), and two-sided tests were performed at a significance level of 0.05. Comparative analysis between the groups was performed using the two-sample t-test or Wilcoxon rank-sum test for continuous variables to determine whether the data were normally distributed. The chi-square test or Fisher's exact test was performed for categorical variables. Differences between groups in percent change in ODI, SF-36 and VAS scores at ≥ 36 months compared with those at baseline were analyzed using analysis of covariance.

Study population and demographic data
Of the 78 patients who completed the 6-month postoperative visit in the previous BA06-CP01 clinical trial, 74 patients (94%) were enrolled in this study (32 patients in the E.BMP-2 group and 42 patients in the AIBG group), and four patients were excluded because of loss of follow-up. Seventy-three patients were included in the efficacy analysis, and one patient with no efficacy results was excluded. All 74 patients were included in the safety evaluation (Fig. 1). The average ages of the E.BMP-2 and AIBG groups were 63.19 (± 8.18) and 60.83 (± 8.95) years, respectively, and there were no significant differences between the two groups in age, sex ratio, height, weight, body mass index, smoking history, drinking history, BMD, and preoperative radiologic findings (all p > 0.05) ( Table 1).

Efficacy outcomes
(1) Radiologic outcome The CT-based bone fusion grade was 4.91 (± 0.42) in the E.BMP-2 group and 4.25 (± 1.26) in the AIBG group at 36 months, which was significantly higher in the E.BMP-2 group (p = 0.007). The bone fusion rates of the E.BMP-2 and AIBG groups over 36 months were 95.7% (22/23) and 83.3% (20/24), respectively; however, there was no significant difference between the two groups (p = 0.348). Moreover, the proportion of grade 5 fusion was higher in the E.BMP-2 than in the AIBG group (96.6% vs. 62.5%). The fusion mass in the 36-month CT scan was significantly wider in the E.BMP-2 group (31.1 mm) than in the AIBG group (22.9 mm) (p = 0.006), and trabeculation of the fusion mass was 100% in both groups ( Table 2, Fig. 2). The plain radiograph-based fusion grades at 12 and ≥ 36 months were significantly higher in the E.BMP-2 group than in the AIBG group (p < 0.05, Table 3). No ectopic bone formation was observed at the intervertebral foramen which could cause radiculopathy.
(2) Clinical outcome The clinical parameters showed postoperative improvement in both groups. The changes in the VAS (lumbar, right leg, and left leg), ODI, and SF-36 scores at ≥ 36 months are shown in Table 4. There was no significant difference between the two groups in terms of VAS scores for lumbar and leg pain (lumbar p = 0.309; left side, p = 0.1456; right side, p = 0.9463). In both groups, the ODI score decreased at 36 months after surgery, and the SF-36 score improved compared to that at baseline, but there was no significant difference between the two groups (ODI, p = 0.4105; SF-36, p = 0.7229) (Tables 4, 5). No additional

Discussion
Unlike C.BMP-2, which has been used for almost two decades, reports on efficacy and safety of E.BMP-2 for spine fusion are deficient. To our knowledge, clinical reports of E.BMP-2 for over 3 years are absent. Moreover, the fate of the HA granule identified at 6 months of E.BMP-2/HA has not been elucidated. Therefore, we attempted to evaluate mid-term efficacy and safety by analyzing the 3-year followup results of patients treated with E.BMP-2. At 36 months postoperatively, the fusion grade of the E.BMP-2 group was significantly higher than that of the AIBG group (E.BMP-2, 4.91 vs. AIBG, 4.25); however, there was no significant difference in fusion rate, clinical and safety outcomes. In the present study, stricter criteria (grades 4 and 5) were applied than those used in the BA06-CP01 study (grade 2, 3, 4, and 5) to confirm fusion. For this reason, it seems that the fusion rate of both groups was lower than those observed in the previous BA06-CP01 study (100.0% and 90.2%) [15] even  with a longer follow-up period. However, using the same fusion criteria, the fusion rates in this study would be 100% and 91.6%, which are consistent with the previous study.
Considering the higher ratio of the fusion grade 5 and wider fusion mass in the BMP-2 group, the fusion quality of the BMP-2 group was considered to be higher than the AIBG group. An important finding of this study was that a larger fusion mass was formed even with the same volume of graft inserted. Sheehan et al. also reported that the fusion volume increased compared to AIBG when CHO-BMP-2/ collagen was used in a canine model [18]. Therefore, we suggest that the E.BMP-2 with HA forms early but immature spinal fusion at 6 months, and then the HA granules are converted into a solid fusion mass over time without reduction in volume (Fig. 2).
C.BMP-2 has been reported to have good clinical outcomes in spinal surgery. [19] Achieving an early solid bone fusion can provide patients with faster postoperative recovery and lower reoperation rates [20]. Dimar et al. reported good results not only in ODI but also in functional aspects such as, return to work, when using BMP-2 [21]. However, it has been reported that CHO cell-derived BMP-2 (C.BMP-2) can increase the cost of surgery by 10-14% compared to conventional surgery [20]. This is closely related to the production efficiency of BMP-2 and the dose of C.BMP-2 per fusion level; however, there is still no consensus on the ideal dose of BMP-2. An excessive BMP-2 dose not only increases cost but also causes BMP-2 is a water-soluble protein that inherently requires a carrier. An ideal carrier should effectively absorb BMP-2, allow slow controlled release of BMP-2, provide scaffold, and should gradually degrade. In bio-ceramics such as HA or tricalcium phosphate, rhBMP-2 is not only absorbed into the scaffold but is also adsorbed as a non-covalent bond on the HA surface. Because collagen carriers are degraded or absorbed within a few weeks in the human body, BMP-2 absorbed in collagen can only be scattered by collagen degradation [22]. In preclinical experiments using Rabbit and minipig, when HA was used as an E.BMP-2 carrier, it showed successful fusion in PLF [10,12]. In this study, retention of BMP-2 was maintained by remaining on the HA granule surface for over 6 months, which may have a positive effect on fusion. Several studies have reported better results using a compression-resistant carrier with C.BMP-2 than with AIBG in posterolateral fusion, where mechanical muscle compression is expected [23], 24. In this respect, the HA carrier is considered as one of the appropriate carrier options in the lumbar PLF.
This study had several limitations. First, the number of patients in each group was relatively small (32 and 41, respectively). However, it was sufficient to compare fusion grade and fusion quality. Second, not all patients underwent CT scans which may have caused selection bias in the evaluation of fusion rate. Nevertheless, the result of this study is important as the first mid-term follow-up study using E.BMP-2 for lumbar fusion surgery.

Conclusion
In conclusion, E.BMP-2/HA showed a comparable efficacy and safety in the mid-term follow-up compared with AIBG in single-segment posterolateral fusion. E.BMP-2/HA can be a satisfactory alternative to AIBG in patients who require fast and solid fusion.