Comparative Clinical Effectiveness and Safety of Bone Morphogenetic Protein Versus Autologous Iliac Crest Bone Graft in Lumbar Fusion

Compared with iliac crest bone graft (ICBG), there is insufficient evidence demonstrating superiority of recombinant human bone morphogenetic protein (RhBMP). We conducted a comprehensive systematic literature review and meta-analysis to compare the differences between rhBMP and ICBG. We suggest rhBMP especially rhBMP-2 as a suitable alternative to ICBG.

Abstract

Study Design.

This is a systematic literature review and meta-analysis.

Objective.

We aimed to evaluate the efficacy and safety of recombinant human bone morphogenetic protein (RhBMP) and autologous iliac crest bone graft (ICBG) in lumbar fusion.

Summary of Background Data.

RhBMP has been emphasized in lumbar fusion due to high fusion success rate. However, ICBG remains the criterion standard graft approach for lumbar fusion. The safety and effectiveness of rhBMP are controversial.

Methods.

Prospective randomized controlled trials were searched from PubMed, EMBASE, and Cochrane Central Register of Controlled Trails by using Medical Subject Headings terms “bone morphogenetic protein," “bone transplantation," and “spinal fusion." Two independent investigators screened eligible studies, assessed the bias of original articles, extracted data including fusion success, Oswestry disability index improvement, improved short form 36 questionnaire scores, adverse events and re-operation, and a subgroup analysis. The GRADE approach was used to grade quality of evidence.

Results.

Twenty randomized controlled trials (2185 patients) met the inclusion criteria. There were higher fusion success rate (odds ratio [OR] 3.79, 95% confidence interval [CI] 1.88–7.63, P = 0.0002), better improvement of Oswestry Disability Index (mean difference 1.54, 95% CI 0.18–2.89, P = 0.03), and lower re-operation rate (OR 0.59, 95% CI 0.43–0.80, P = 0.0007) in rhBMP group. Heterogeneity was obvious in fusion success rate (I² = 58%); hence, a subgroup analysis, based on protein type (rhBMP-2 or rhBMP-7), was performed, which suggested that only rhBMP-2 was better than ICBG for lumbar fusion. There was no difference in the incidence of adverse events between rhBMP and ICBG (OR 0.91, 95% CI 0.70–1.18, P = 0.47).

Conclusion.

In lumbar fusion, rhBMP-2 exhibited a higher fusion success rate and reduced the risk of re-operation. No difference in complication rate is between rhBMP (rhBMP-2 and rhBMP-7) and ICBG. We suggest rhBMP especially rhBMP-2 as an effective substitute for ICBG for lumbar fusion.

Level of Evidence: 1

Chronic low back pain and leg pain are commonly caused or influenced by lumbar degenerative diseases, which in include lumbar spinal stenosis, lumbar spondylolisthesis, and lumbar disc herniation.¹ Some patients’ condition and symptoms may be controlled with conservative therapies. However, if symptoms or imaging indicate vertebral or spinal instability, lumbar fusion is recommended to re-establish stability.² Surgical options for interbody fusion include anterior lumbar interbody fusion (ALIF), posterior lumbar interbody fusion (PLIF), posterolateral lumbar fusion (PLF), or transforaminal lumbar interbody fusion.³

To improve success rate of intervertebral fusion, bone graft is considered necessary. For decades, autologous iliac bone graft (ICBG) has been recognized as the criterion standard for lumbar fusion.⁴ However, it has disadvantages including an elevated rate of bone nonunion, donor site complications, and relatively insufficient grafted bone in multisegment fusion.^5,6 As a result, various bone graft substitutes have been developed. One of the most widely used is bone morphogenetic protein (BMP), first reported by Urist et al,⁷ which can induce osteogenesis. Recombinant human bone morphogenetic protein (rhBMP) was developed on a large scale in mid-1990s using human recombinant genetic technology due to the limited yield of natural extracted and purified BMP.⁸ The US Food and Drug Administration approved clinical use of two rhBMP: rhBMP-2 and rhBMP-7 as alternatives to ICBG.⁹

Since Boden et al¹⁰ conducted the first RCT comparing rhBMP-2 with ICBG, >20 RCTs have been performed. RhBMP has demonstrated advantages in most studies, including higher fusion rate and shorter operation time than ICBG. However, the safety of rhBMP has been questioned due to reports linking the therapy with serious complications.^11–14 Noshchenko et al and Zhang et al suggested that rhBMP could be a good alternative approach to ICBG.^15,16 However, Ye et al¹⁷ indicated that using rhBMP-7 appeared to yield lower fusion rate in instrumented posterolateral fusion patients. A degree of uncertainty remains concerning efficacy and safety of rhBMP in lumbar fusion. In light of this, the purpose of present study is to further update this topic and attempt to clarify rhBMP safety and effects in lumbar fusion.

MATERIALS AND METHODS

Search Strategy

We comprehensively searched PubMed, EMBASE, and Cochrane Central Register of Controlled Trails databases for RCTs, through using MeSH terms “bone morphogenetic protein," “bone transplantation," “bone graft," and “spinal fusion." The retrieved results were last updated on May 29, 2019. To conduct a thorough search of all relevant literature, two independent investigators screened eligible studies. When consensus could not be reached, a third reviewer was consulted to resolve the disagreement.

Study Selection Criteria

Inclusion criteria for articles were: age 18 to 80 years, suffering from lumbar degenerative diseases requiring lumbar fusion, and RCT comparing rhBMP with ICBG.

Studies were excluded if patients presented with spinal deformities, fractures, tumors or infections, cases demonstrated spondylolisthesis classified higher than Meyerding Grade II, follow-up was <12 months, and there were incomplete follow-up data.

Data Extraction

Two investigators extracted data from forms containing relevant patient information. The data included study design, patient characteristics, sample size, intervention details, follow-up rate and time, outcomes. The primary outcomes included fusion success, improvement on the Oswestry disability index (ODI),¹⁸ improvement on short form 36 (SF-36),¹⁹ improvement on the Numeric Rating Scale (NRS) for back pain and leg pain,²⁰ adverse events, and reoperation. Fusion success was defined as an absence of radiolucent lines covering >50% of either implant, translation of 3 mm, and angulation <5° on flexion–extension radiographs.²¹ SF-36 includes Physical Component Summary (PCS) and Mental Component Summary (MCS). Improvement value was defined as absolute difference between preoperative and postoperative outcomes. Secondary outcomes included operation time, intraoperative blood loss, and duration of hospital stay.

Quality Assessment

Two investigators evaluated bias risk using the 12 criteria recommended by the Cochrane Back Review Group.²² The items were scored as “low risk," “high risk," or “unclear." If at least six of the criteria passed without serious potential flaws, items were considered to have an overall “low risk of bias." Our study utilized the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) group criteria to describe both quality of evidence and strength of recommendations.²³ The quality of evidence was classified as very low, low, moderate, or high.

Statistical Analysis

The results were expressed in terms of an odds ratio (OR) and a 95% confidence interval (95% CI) for dichotomous outcomes, and in terms of mean difference (MD) and 95% CI for continuous outcomes. The I² test was used to evaluate the heterogeneity of statistical results. When the I² value was <50%, statistical results were considered to have no heterogeneity. Also, a fixed-effect model was used. Otherwise, a random-effect model was used. Funnel plots were used to explore potential publication bias. Meta-analysis was performed by Review Manager Software (RevMan, Version 5.3).

RESULTS

Search Results and Study Characteristics

A total of 1118 related studies were initially identified from PubMed (n = 482), EMBASE (n = 599), and CENTRAL (n = 37) databases. Only 24 RCTs were included.^{10,14,24–45} One of these RCTs included follow-up of <12 months,²⁴ two were duplicate reports of the same set of patients,^34,40 and one reported missing rate >15%.³⁶ Ultimately, 20 RCTs with 2185 patients were included in meta-analysis. The process of identifying related reports is presented in Figure 1. The extracted data of the characteristics of 20 studies were recorded (Table 1).^{10,14,25–33,35,37–39,41-45}

Figure 1.

Inclusion and exclusion flow chart.

TABLE 1.

Characteristics of Included Studies.

Study	Study Type	Preoperative Diagnosis	Surgical Intervention	Comparisons	Sample Size		Mean Age, y		BMP Dose, mg			Follow-up, mo	Outcomes
BMP	ICBG	BMP	ICBG
Boden et al, 200010	RCT	Lumbar DDD, grade I spondylolisthesis	ALIF	rhBMP-2/ACS vs. ICBG	11	3	42.5	40.2	NR	24	FS, ODI, SF-36, AD, OT, LOS, BL
Boden et al, 200242	RCT	Lumbar DDD, grade I spondylolisthesis	ALIF	rhBMP-2/BCP vs. ICBG	11	5	57.6	52.9	20	17	FS, ODI, SF-36, reoperation, NRS, AD, OT, LOS, BL
Burkus et al, 200244	RCT	Lumbar DDD	ALIF	rhBMP-2/ACS vs. ICBG	143	136	43.3	42.3	4.2–8.4	24	FS, ODI, SF-36, reoperation, neurologic status, NRS, OT, LOS, BL
Burkus et al , 200220	RCT	Lumbar DDD	ALIF	rhBMP-2/ACS vs. ICBG	24	22	41.5	45.6	12–18	24	FS, ODI, SF-36, reoperation, AD, neurologic status, NRS, OT, LOS, BL
Johnsson et al, 200243	RCT	L5 spondylolysis and vertebral slip ≤50%	PLIF	rhBMP-7/Type 1 bone collagen vs. ICBG	10	10	42.9	40.4	7	12	FS, reoperation, AD
Burkus et al, 200341	RCT	Degenerative lumbar spondylosis	ALIF	rhBMP-2/ACS vs. ICBG	22	20	41.7	44.2	4.2–8.4	24	FS, reoperation
Haid et al, 200439	RCT	Lumbar DDD, grade I spondylolisthesis	PLIF	rhBMP-2/ACS vs. ICBG	34	33	46.3	46.1	4–8	24	FS, ODI, SF-36, reoperation, AD, neurologic status, NRS, OT, LOS, BL
Glassman et al, 200535	RCT	Lumbar DDD, grade I spondylolisthesis	PLF	rhBMP-2/CRM vs. ICBG	38	36	53	53	40	12	FS, reoperation, OT, BL
Burkus et al, 200537	RCT	Lumbar DDD	ALIF	rhBMP-2/ACS vs. ICBG	79	52	40.2	43.6	8.4–12	24	FS, ODI, SF-36, reoperation, NRS, OT, LOS, BL
Vaccaro et al, 200538	RCT	Degenerative spondylolisthesis and stenosis	PLIF	rhBMP-7/Type 1 bone collagen vs. ICBG	24	12	63	66	7	24	FS, ODI, SF-36, reoperation,AD, OT, LOS, BL
Dimar et al, 200614	RCT	Lumbar DDD, grade I spondylolisthesis	PLF	rhBMP-2/CRM vs. ICBG	53	45	50.9	52.7	40	24	FS, ODI, SF-36, reoperation, AD, NRS, OT, LOS, BL
Kanayama et al, 200633	RCT	Degenerative spondylolisthesis with stenosis	PLIF	rhBMP-7/Type 1 bone collagen vs. ICBG	9	10	70.3	58.7	7	16.4	FS
Glassman et al, 200831	RCT	Lumbar DDD or spondylolisthesis or stenosis;	PLF	rhBMP-2/ACS vs. ICBG	50	52	69.9	69.2	NR	24	FS, ODI, SF-36, reoperation, AD, NRS, OT, LOS, BL
Vaccaro et al, 200832	RCT	Degenerative spondylolisthesis and stenosis	PLIF	rhBMP-7/Type 1 bone collagen vs. ICBG	257	87	68	69	7	>48	FS, ODI, SF-36, reoperation, AD, VAS, OT, LOS, BL
Dimar et al, 200929	RCT	Lumbar DDD, grade I spondylolisthesis	PLF	rhBMP-2 matrix vs. ICBG	239	224	53.2	52.3	40	24	FS, ODI, SF-36, reoperation, AD, NRS, OT, LOS, BL
Dawson et al, 200930	RCT	lumbar DDD, grade I spondylolisthesis	PLF	rhBMP-2/ACS vs. ICBG	25	21	55.9	56.9	12	24	FS, ODI\, SF-36, reoperation, AD, NRS, OT, LOS, BL
Delawi et al, 201028	RCT	Grade II spondylolisthesis	PLIF	rhBMP-7/local autograft vs. ICBG	18	16	53	55	3.5	12	FS, ODI, AD, VAS, OT, LOS, BL
Michielsen et al, 201327	RCT	Lumbar DDD	PLIF	rhBMP-2/ACS vs. ICBG	19	19	43.2	42.2	8	24	FS, ODI, SF-36, reoperation, AD, VAS, OT, LOS, BL
Hurlbert 201326	RCT	Lumbar DDD	PLF	rhBMP-2/BCP vs. ICBG	98	99	53	53	42-63	48	FS, ODI, SF-36, reoperation, AD, NRS, OT, LOS, BL
Delawi et al, 201625	RCT	Degenerative or isthmic spondylolisthesis ≤ grade 2	PLIF	rhBMP-7/local autograft vs. ICBG	60	59	54	55	7	24	FS, ODI, reoperation, AD, OT, LOS, BL

ACS indicates absorbable collagen sponge carrier; AD, adverse events; ALIF, anterior lumbar interbody fusion; BCP, biphasic calcium phosphate; BL, blood loss; BMP, bone morphogenetic protein; CRM, compression resistant matrix; DDD, degenerative disc disease; FS, fusion success; ICBG, autogenous iliac crest bone graft ; LOS, length of hospital stay; NR, no report; NRS, numeric rating scale for back and/or leg pain; ODI, Oswestry disability index; OT, operative time; PLF, posterior or posterolateral lumbar fusion; PLIF, posterior lumbar interbody fusion.

Primary Outcomes

Fusion Success

Thirteen studies met the designated fusion success criteria.^{10,26,29,30,32,33,37–39,41,42,44,45} Significant differences were found between rhBMP and ICBG (odds ratio [OR] 3.79, 95% confidence interval [CI] 1.88–7.63, Figure 2). Subgroup analysis based on protein type was performed. The “rhBMP-2 group” comprised 10 articles including 1141 patients^{10,26,29,30,37,39,41,42,44,45} and the “rhBMP-7 group” comprised three articles including 245 patients.^32,33,38 In rhBMP-2 group, significant differences were observed between rhBMP-2 and ICBG (OR 5.57, 95% CI 2.95–10.52). In rhBMP-7 group, no significant difference was observed between rhBMP-7 and ICBG with regard to fusion success (OR 0.94, 95% CI 0.49–1.81).

Figure 2.

Fusion success—recombinant human bone morphogenetic protein versus autologous iliac bone graft.

Improvement on ODI

Eleven eligible studies including 1252 patients reported the improvement on ODI.^{10,25–32,42,45} Meta-analysis discovered significant difference between rhBMP and ICBG (MD 1.54, 95% CI 0.18–2.89, Figure 3) and there was significant heterogeneity between trials. Subgroup analysis based on protein type was performed. Significant difference was observed between rhBMP-2 and ICBG (MD 1.94, 95% CI 0.28–3.61). No significant difference was observed between rhBMP-7 and ICBG (MD 0.19, 95% CI −2.18 to 2.56).

Figure 3.

Improvement on ODI scores—recombinant human bone morphogenetic protein versus autologous iliac bone graft.

Improvement on SF-36 PCS

Details regarding improvement on SF-36 PCS on rhBMP-2 were available for four studies including 564 patients.^27,29,31,42 No significant difference was observed between rhBMP-2 and ICBG (MD 1.16, 95% CI −0.84 to 3.16). With regard to rhBMP-7, there were two studies^32,38 that reported data, but omitted the means and standard deviations. As a result, meta-analysis was not performed between rhBMP-7 and ICBG.

Improvement on NRS back Pain and NRS Leg Pain

Relevant improvement on NRS back pain and NRS leg pain data on rhBMP-2 were extracted from five eligible articles containing 755 patients.^{26,29,31,42,45} No significant difference in NRS back pain was identified between rhBMP-2 and ICBG (MD −0.05, 95% CI −1.15 to 1.06). With respect to NRS leg pain, a significant difference was observed between rhBMP-2 and ICBG (MD −0.84, 95% CI 0.02–1.65). Due to studies^{25,28,32,33,38,43} lacking corresponding data on rhBMP-7, meta-analysis was not performed between rhBMP-7 and ICBG.

Reoperation

Reoperation information was available from 17 studies. Thirteen^{14,26,27,29–31,35,37,39,41,42,44,45} of these studies including 1594 patients were about rhBMP-2, and four^25,32,38,43 more including 519 patients were about rhBMP-7. The combined result showed a significantly lower rate of reoperation in rhBMP compared to ICBG (OR 0.59, 95% CI 0.43–0.80, Figure 4). Subgroup analysis revealed a significant difference between rhBMP-2 and ICBG (OR 0.48, 95% CI 0.33–0.68), with no significant difference observed between rhBMP-7 compared with ICBG (OR 1.18, 95% CI 0.62–2.23).

Figure 4.

Reoperation—recombinant human bone morphogenetic protein versus autologous iliac bone graft.

Adverse Events

Fourteen studies provided specific data regarding adverse events. Nine^{10,26,27,29–31,39,42,44} of these studies were about rhBMP-2, and five^{25,28,32,38,43} more were about rhBMP-7. Statistical analysis did not reveal significant differences between rhBMP and ICBG (OR 0.91, 95% CI 0.70–1.18, Figure 5). Different surgical approaches may affect the incidence of complications; therefore, subgroup analysis based on surgical procedure (ALIF or PLIF/PLF) was performed. No significant differences in ALIF (OR 0.78, 95% CI 0.37–1.64)^10,42,44 or PLIF/PLF (OR 0.93, 95% CI 0.71–1.22)^{25–32,38,39,43} were observed when comparing rhBMP with ICBG. A funnel plot of the documented adverse events is presented in Figure 6. No evidence of publication bias was found. Besides, it is well-known that infection is an important complication following implant surgery. Therefore, we paid special attention to the data in this regard. Four studies reported surgical infections, involving 295 patients. Two^30,31 of these studies were about rhBMP-2, and two^25,28 more were about rhBMP-7. No significant difference was identified between rhBMP and ICBG (OR 0.76, 95% CI 0.29–2.00). Subgroup analysis identified no significant differences in rhBMP-2 (OR 0.37, 95% CI −0.07 to 1.99) or rhBMP-7 (OR 1.17, 95% CI 0.34–4.02) were observed when compared with ICBG.

Figure 5.

Adverse events—recombinant human bone morphogenetic protein versus autologous iliac bone graft.

Figure 6.

Funnel plot—adverse events.

Secondary Outcomes

Operation Time

Concerning operation time, 10 studies were analyzed. Seven^{10,26,27,29–31,42} of these studies included 876 patients and were about rhBMP-2, and three^25,28,38 more including 189 patients were about rhBMP-7. An overall reduced operation time was observed in rhBMP compared to ICBG (MD, −0.23, 95%CI, −0.44 to −0.02, Figure 7). Subgroup analysis revealed a significant difference between rhBMP-2 and ICBG (MD −0.27, 95% CI −0.54 to −0.01), with no significant difference observed between rhBMP-7 compared with ICBG (MD −0.13, 95% CI −0.38 to 0.11).

Figure 7.

Operation time—recombinant human bone morphogenetic protein versus autologous iliac bone graft.

Intraoperative Blood Loss

No significant difference in blood loss was identified between rhBMP and ICBG in nine studies (MD −27.21, 95% CI −80.53 to 26.12). Seven^{10,26,27,29–31,42} of these studies including 876 patients were about rhBMP-2, and two^25,28 more including 155 patients were about rhBMP-7. Subgroup analysis identified no significant differences in rhBMP-2 (MD −34.89, 95% CI −97.29 to 27.52) or rhBMP-7 (MD 6.32, 95% CI −84.41 to 97.05) when comparing with ICBG.

Duration of Hospital Stay

Nine studies provided specific data regarding hospital stay and reported no significant difference between rhBMP and ICBG (MD −0.52, 95% CI −1.02 to −0.01). Six^{10,26,29–31,42} of these studies including 838 patients were about rhBMP-2, and two^25,28,38 more including 187 patients were about rhBMP-7. Subgroup analysis revealed a significant difference between rhBMP-2 and the ICBG (MD −0.64, 95% CI −1.22 to −0.06), with no significant difference observed between rhBMP-7 compared with ICBG (MD, 0.07, 95% CI −0.98 to 1.12).

Risk of Bias in Included Studies

Following assessment of all original studies, we evaluated bias risk using the 12 criteria recommended by the Cochrane Back Review Group.²² The migration risk for each study is described in Table 2. The ratings across all included studies were summarized and presented in Figure 8.

TABLE 2.

Summary of Findings: Meta-analysis Comparison of Bone Morphogenic Protein and Iliac Crest in Degenerative Lumbar Disease

Figure 8.

Risk of bias summary. This risk of bias tool incorporates assessment of randomization (sequence generation and allocation concealment), blinding (participants, personnel and outcome assessors), completeness of outcome data, selection of outcomes reported, and other sources of bias. The items were scored with “yes,” “no,” or “unclear.”

Sensitivity Analysis and Quality Assessment

We performed a sensitivity analysis of the primary outcomes that exhibited higher heterogeneity. The primary method we utilized was to eliminate low-quality research on a study-by-study basis. When the studies of Hurlbert et al²⁶ were excluded, heterogeneity was significantly reduced. One explanation may be that the study's end point time was 4 years, with most other studies ending at 2 years. This indicates that the modifications in these indicators are related to the length of follow-up. GRADE quality assessment was used to rate the quality of evidence for all pooled outcomes. These results are shown in Table 2.

DISCUSSION

Lumbar fusion with ICBG is the criterion standard surgical procedure for discogenic pain refractory to conservative treatments.^46–48 However, due to some well-known complications, some bone graft substitutes including rhBMP have been developed and applied in clinical practice. Five systematic reviews^16,49–52 suggesting that rhBMP reduced risk of fusion failure compared with ICBG, but had no advantage over ICBG concerning pain relief and functional recovery. Fu et al and Simmonds et al suggested that there was some evidence that rhBMP-2 may cause serious complications.^49,50 Zhang et al¹⁵ included 19 RCTs for systematic review, with results concluding that rhBMP can reduce the reoperation rate and operation time. However, primary outcomes in that study such as fusion success rate and clinical success, were not clearly defined. Ye et al¹⁷ suggested that using rhBMP-7 instead of ICBG produced no any additional benefits in single level PLF. Han et al⁵³ reported no significant differences in fusion rate of no internal fixation at 24 months for rhBMP, but there was high risk of bias due to including pooling nonrandomized controlled trials and lacking subgroup analysis. Recently Mariscal et al⁵⁴ found rhBMP-2 in PLF reduced surgical morbidity and had more beneficial effects on the fusion rate. However, multiple important RCTs^{10,14,27,35,37,39,41,44,45} were not included in the analysis. The quality of evidence was not assessed for the included study and pooled results, and sensitivity analysis was not performed. As a result, evidence is still lacking to support the superiority of rhBMP compared with ICBG. Our present study serves as an update the in systematic evaluation. We incorporated the latest two large-sample RCTs,^25,26 utilizing a systematic review method approved by Cochrane Collaboration. This included extracting data more carefully, performing sensitivity analyses for pooled outcomes, assessing risk of bias and quality of evidence according to the GRADE approach. Our attention was focused not only on subgroup analysis, which based on protein type. Data processing of complications was also a subgroup analysis emphasizing anterior and posterior surgery.

We determine that rhBMP is a fusion material that demonstrates better efficacy compared to ICBG. In terms of fusion success, rhBMP was approximately 2.8 times more effective than ICBG. Further analysis showed that the fusion success rate in rhBMP-2 was approximately 5.5 times higher than that observed in ICBG, whereas the fusion rate with rhBMP-7 was 5% less than ICBG. Similarly, the ratio of reoperation in rhBMP was about 60% of ICBG. Subgroup analysis showed that the rhBMP-2 had reduced incidence of reoperation by approximately 40% compared with ICBG, whereas rhBMP-7 had approximately 1.2 times the incidence than ICBG. These results indirectly suggest that rhBMP-2 is more effective than rhBMP-7 in inducing bone formation. This difference may be due to the different carriers used, and it is necessary to perform a RCT comparing rhBMP-2 and rhBMP-7 efficacy to confirm our conclusions. A summary of clinical treatment effects indicates a difference in ODI and NRS leg pain, and rhBMP-2 appears to be a better approach.

In terms of safety, whether using ALIF or PLIF/PLF, there was no significant difference in incidence of complications between rhBMP and ICBG. It is worth mentioning that surgical site infection should be taken seriously. Delawi et al²⁵ reported that four patients required reoperation due to surgical infection. This serves as a reminder that it is necessary to pay attention to the surgical incision in the early postoperative period. In addition, it is important to treat surgical site infection as early and aggressively as possible to prevent a more serious infection that requires antibiotic treatment.³⁰ Regarding surgical data, operation time and hospitalization days with rhBMP-2 treatment is reduced compared to ICBG. But, with respect to blood loss, rhBMP-2 does not show much advantage. What is different from previous study¹⁷ is that there was no significant difference in any surgical data including hospital stay between rhBMP-7 and ICBG.

Despite the complexity of our study, we recognize that it still has some limitations. First, although there were 20 studies in total, a large portion of the studies did not provide SD values, and the inability to extract valid data led to limited data accuracy. In addition, the quality of evidence in this meta-analysis is limited by the low quality of the original studies. Most evaluated studies did not report their randomization or allocation methods. Nearly all studies failed to use independent blinding. Finally, a no cost–benefit analysis was conducted. Aside from the cost of research by Glassman et al,³¹ other studies have not reported this parameter, which made analysis difficult or impossible. Therefore, subsequent RCT should pay attention to the application of more rigorous methods and indicators, including more accurate reporting of pre- and postoperative scores, follow-up of long-term complications, and costs of treatment.

Our review indicates that rhBMP-2 may be superior in terms of fusion success, ODI, reoperation, and duration of hospital stay. It might represent a suitable substitute for ICBG in lumbar fusion. Conversely, rhBMP-7 is not recommended for lumbar fusion. Further studies including cost-effective data analysis and RCT for the comparison of the efficacy of rhBMP-2 with rhBMP-7 is necessary to confirm the results observed in the present study.

Key Points

• RhBMP-2 seems to be a more effective fusion material than ICBG.

• RhBMP-2 may be superior in terms of fusion success, improvement of ODI, re-operation, operation time, and length of hospital stay, but rhBMP-7 has no significant advantage compared with ICBG.

• No significant differences were observed between rhBMP-2 and ICBG regarding improvement of SF-36 and NRS back pain, adverse events, and blood loss.