With this chronological background, rhBMP-2 has been adopted as a clinical alternative to autologous bone grafting. It is widely recognized that the amount of new bone formation correlates with the dose of rhBMP-2 administered up to a certain threshold [66]. Studies have shown a linear relationship between dose and bone formation up to a dose threshold, beyond which it reaches a plateau with no additional bone formation [66]. In nonhuman primates, the dose-response curve is much steeper compared to other experimental models, necessitating higher doses to achieve successful fusion [42]. For example, 0.2-0.4 mg/mL rhBMP-2 would be sufficiently osteoinductive in rodents, while sheep or other primates need higher concentrations of 0.43 and 0.75-1.5 mg/mL, respectively [3,5]. The selection of the 1.5 mg/mL concentration for the initial human clinical trial of ALIF was based on the therapeutic range of 0.4-1.5 mg/mL rhBMP-2 in nonhuman primates for this indication. This concentration is currently approved by the FDA for human use (Infuse^TM Bone Graft) [18].

rhBMP-2 has a short half-life in the body when applied alone, is quickly eliminated, and loses its biological activity rapidly in vivo. To assess the extent and duration of systemic exposure to rhBMP-2, studies were conducted to characterize the pharmacokinetics of rhBMP-2 protein in the blood of rats and monkeys [18]. Following intravenous administration, rhBMP-2 was rapidly cleared from the systemic circulation via renal excretion in rats and nonhuman primates (t½ = 16 minutes in rats and t½ = 6.7 minutes in nonhuman primates) [18]. The lack of reports of systemic toxicity may be related to this rapid systemic clearance of rhBMP-2. Therefore, a delivery carrier system is essential to modulate the pharmacokinetic properties of rhBMP-2. Carriers for rhBMP-2 can increase protein retention at the treatment site for a sufficient period to allow regenerative, tissue-forming cells to migrate to the injured area, proliferate and differentiate. The carrier can also serve as a matrix for cell infiltration and maintain a space or volume for new bone formation. For example, applying rhBMP-2 with an ACS to subcutaneous or orthopedic sites in rats and rabbits extended the mean local residence time to up to 8 days [18]. The representative rhBMP-2-containing product approved by the Ministry of Food and Drug Safety (MFDS) in Korea is Novosis^TM (CG Bio Co., Ltd.). In a promotional brochure for Novosis^TM, CG Bio Co., Ltd., highlighted the rapid systemic clearance of rhBMP-2, similar to that of the Infuse^TM Bone Graft, as shown in Fig. 1.

Optimizing the therapeutic application of rhBMP-2 depends on localized, sustained release, which relies on a safe and well-characterized carrier system. The development of effective carrier systems is indispensable not only for enhancing the therapeutic efficacy of rhBMP-2 but also for enabling the safe and effective application of low-dose BMP-2, thereby minimizing dose-dependent adverse effects. Various types of carriers have been developed and recommended for optimal rhBMP-2 delivery [1]. Scaffolds for bone regeneration must meet essential requirements related to their ability to bind and release growth factors. To be clinically relevant, scaffolds should ideally occupy the intended space, be composed of biocompatible and osteoconductive materials and have mechanical properties suitable for the implantation site. They should also have a three-dimensional structure that supports cell metabolism, be manufactured efficiently, allow for secure fixation in the surgical site, degrade over a suitable time frame, and promote cell migration into the construct [35]. Previously studied rhBMP-2 carriers can be divided into three categories [1] : polymers (natural or synthetic), ceramics and ceramic/polymer composites. Each category possesses structural features that facilitate rhBMP-2 loading and delivery (Fig. 2). Currently, there is no ideal carrier that fulfils all the requirements necessary for clinical applications.

Despite its suboptimal biomechanical properties, ACS is the only carrier approved for use with rhBMP-2 in the United States owing to its high biocompatibility, biodegradability and low immunogenicity [19,20,29,49]. The primary clinical administrations of rhBMP-2 involved the use of ACS made from bovine type I collagen obtained from the deep flexor (Achilles) tendon. Previous studies have shown that ACS scaffolds facilitate effective cell infiltration, which may be a crucial factor in inducing new bone formation [19]. Release of rhBMP-2 incorporated within ACS is believed to occur through two primary mechanisms : first, desorption of proteins bound to the scaffold, and second, diffusion and/or convection of proteins that do not interact with the scaffold. However, rhBMP-2 easily detaches from the binding site because of its limited affinity for collagen. In addition, most of the unbound rhBMP-2 present within the pores of the collagen sponge is released in the early implantation stage [28,36]. Furthermore, collagen sponges generally degrade quickly in vivo, resulting in disruption of the structural integrity needed for the retention and release of rhBMP-2 before bone is fully regenerated. Pharmacokinetic studies using a rat ectopic assay investigated the in vivo retention and release of rhBMP-2 from ACS. The results showed an initial rapid loss of rhBMP-2 within hours of implantation, followed by a more gradual loss [28]. In preclinical studies and clinical practice, spinal fusion using rhBMP-2 with ACS has achieved high fusion rates but has been associated with several complications, including inflammatory reactions, spinal neuropathy, osteolysis and urogenital disorders [2,30,31,58]. These are thought to be caused by an initial uncontrolled, burst-like release of rhBMP-2, which may exceed physiological levels by more than 106 times [17,22,68]. Half of the rhBMP-2 injected over the ACS was released within the first couple of days [39]. Due to its limited mechanical integrity, the ACS carrier cannot withstand compressive forces, which may exacerbate the uncontrolled, burst-like release of rhBMP-2, especially when sponge-wrapped rhBMP-2 is subjected to compressive deformation by external pressure. Moreover, collagen sponges degrade rapidly, raising skepticism about their effectiveness as an ideal carrier, particularly in providing the structured 3D scaffold at the target sites for osteogenesis. Consequently, strategies to suppress the initial burst release while extending/prolonging the sustained release period have been proposed and widely recognized. Such approaches aim to reduce clinical complications and promote the recruitment of stem cells to the bone healing site [6,12,58].

Several materials with robust biomechanical properties, including resistance to compression and high affinity for rhBMP-2, have been investigated for applications in bone healing. Calcium phosphate ceramics like bone minerals (notably hydroxyapatite, HA) are used as rhBMP-2 carriers because of their osteoconductive properties and environmentally sensitive binding capabilities. rhBMP-2 is not only ‘ab’sorbed by the scaffold, but also ‘ad’sorbed through a noncovalent bond on HA surfaces. HA typically exists as granules and has porosity with interconnected pores, similar to cancellous bone. It also has a large surface area for rhBMP-2 adsorption. The increased surface area enables increased rhBMP-2 loading efficiencies. Molecular simulations and experimental results have shown that rhBMP-2 adsorption onto the HA surface involves electrostatic attraction and hydrogen bonding, which are time-dependent [16]. In one study, over 80% of the total rhBMP-2 adsorbed rapidly during the first minutes of exposure to HA [59,69]. Additionally, HA facilitates new bone formation by gradually releasing ions such as calcium and phosphorus when introduced into the body. rhBMP-2 also has a high affinity for HA, which enables sustained release [25,45]. The affinity of rhBMP-2 for HA has been measured to be 2.4-230×105 M-1 [7]. Novosis^TM, which was approved for use in posterolateral fusion by Korea’s MFDS in 2017, is a representative product that utilizes HA granules as a carrier for rhBMP-2.

Recently, ceramic and polymer composites have gained considerable attention as delivery systems for rhBMP-2. The final formulation of the composite carrier is designed as a gel or putty, addressing the structural limitations of ACS and granule-type ceramic carriers and providing the flexibility required to meet various clinical needs. Numerous composites have been developed for in vivo delivery of rhBMP-2 for bone regeneration [1]. For example, a putty carrier combining HA granules with β-tricalcium phosphate (TCP) poloxamer hydrogel combines the advantages of both polymer and ceramic carriers. This composite carrier could easily fill the space within the intervertebral cage without leaving a gap, enabling sustained release of rhBMP-2. When in vitro elease kinetics were compared using an enzyme-linked immunosorbent assay, most (78.4% of the initial load) of the rhBMP-2 loaded on ACS was released by day 24, and the majority of this release occurred during the 1-7 days interval, whereas in the composite carrier group, significantly greater amounts of rhBMP-2 were released during the 7-14 and 14-24 days intervals. When in vivo release kinetics were compared using fluorescently labelled rhBMP-2, the biological half-life of rhBMP-2 delivered through the composite carrier was longer than that of ACS. In addition, rhBMP-2 delivered through the composite carrier induced a higher level of osteogenic gene expression than that of ACS, and the quantity and quality of new bone formation were also superior in a rat spinal fusion model [57]. Composite carriers were able to control the spatial release of rhBMP-2 more efficiently than ACS. In a rat caudal intervertebral fusion model, two doses of rhBMP-2 were loaded into ACS and a composite carrier : a high dose of 10 μg and low dose of 3 μg. Micro-CT results after 6 weeks showed that the interbody fusion rate was significantly higher in the composite carrier group (both doses, 87.5%) compared to the ACS group (low dose, 50%; high dose, 62.5%). In the composite carrier group, new bone formation was confined to the intervertebral disc spaces, whereas in the ACS group, ectopic bone formation was observed outside the intervertebral disc spaces. The composite carrier group demonstrated superior new bone quality, with abundant thick trabecular bone, compared to the ACS group, which primarily exhibited fatty bone marrow. In the composite carrier group, soft tissue swelling remained unchanged with varying doses. In contrast, the ACS group showed a significant increase in swelling in the high-dose group compared to the low-dose group [44]. Another study reported on collagen/HA composite carriers used to deliver rhBMP- 2. The injectable carrier system composed of polylactic acid (PLA)-polyethylene glycol (PEG) block copolymer (7 : 3 ratio) demonstrated improved bone mineral density compared to an identical carrier with the same polymer but a different molecular weight at an equivalent ratio [51]. In another approach, calcium- HA bioceramics are combined with a PLA : PEG block copolymer (51 : 49 ratio) to promote rhBMP-2 activity. It has been found that 5 mg rhBMP-2 is sufficient to induce complete healing of critical-sized radial defects in rabbit [32]. Although many carrier systems have demonstrated acceptable limits of protein delivery, further investigation is actively underway to identify the ideal carrier for BMP-2 with the lowest required dose and fewest adverse reactions to maximize bone repair. Ceramic or ceramic/polymer composite carriers that are osteoconductive and allow sustained release of rhBMP-2 rather than an initial burst release are clinically utilized as alternative carriers to replace ACS. Table 1 summarizes the differences in the characteristics of ACS and ceramic or ceramic composite carriers.

Industry-sponsored trials of Infuse^TM before approval showed a trend of increasing rhBMP-2 dose over time, reflecting the prevailing philosophy that higher doses lead to better outcomes. rhBMP-2/ACS enabled volume-dependent dose control of rhBMP-2, contingent upon the degree of underfilling or overfilling. In the early Infuse^TM ALIF studies, the total dose of rhBMP-2 per level progressively increased from 3.9-7.8 mg to 4.2-12 mg [64]. Studies highlighting the adverse reactions of rhBMP-2, which had not been reported in earlier research, began to emerge in 2006 [10,43,48,53].

However, despite involving approximately 600 patients, previous industry-sponsored trial publications consistently reported adverse events associated with the investigational device. In 2011, Carragee et al. [12] published a critical review indicating that the complication rate for spinal surgery using rhBMP-2/ACS ranged from 20% to 70%. Accordingly, United States Senate Finance Committee staff investigated Medtronic’s involvement in the Infuse^TM study. They found that Medtronic significantly contributed to drafting and editing articles written by its physician consultants, paid the authors and advised against publishing a complete list of adverse events linked to Infuse^TM [61]. This reporting omission likely contributed to the extensive use of high-dose rhBMP-2. Amidst the controversy, the Yale Open Data Access (YODA) project conducted two systematic reviews and meta-analyses to comprehensively assess rhBMP-2/ACS’s safety. The analysis results from the YODA project showed that rhBMP-2 had a higher complication rate than ICBG in anterior cervical fusion surgery, whereas ICBG and rhBMP-2 were equally effective in thoracolumbar surgeries, such as ALIF and PLF. There was a trend toward more adverse events associated with rhBMP-2, yet the data were insufficient to draw definitive conclusions. While the use of rhBMP-2 has been associated with an increased risk of cancer, the incidence and absolute risk remain small. The heterogeneity of cancer types raises questions about the validity of this association. In addition, data on retrograde ejaculation and heterotopic bone formation were not collected [21,54]. The response to these findings precipitated two major shifts in clinical practice: reduction in use and reduction in dose of rhBMP-2.

A report investigating rhBMP-2 utilization trends using the largest inpatient administrative database in the United States revealed that the use of rhBMP-2 in spinal fusion surgeries increased from 0.7% in 2002 to 29.5% in 2010 then gradually declined to 14.7% of all fusion surgeries in 2015, which paralleled and reflected growing concerns about clinical safety [33]. In another retrospective database study, the use of rhBMP-2 in thoracolumbar fusion surgeries across the United States increased from 28.3% in 2004 to 47.0% in 2008 before steadily declining to 23.6% in 2014 [4]. There was also a marked change in the type of fusion using rhBMP-2. The rate of rhBMP-2 use decreased from 23.1% in 2006 to 12.0% in 2015, with the largest decreases observed in anterior cervical fusions (from 7.5% to 3.1%), posterior cervical fusions (from 17.0% to 8.3%) and PLFs (from 31.5% to 15.8%) [65]. Also, the use of rhBMP-2 in patient groups with a low risk-benefit ratio, such as younger patients and primary surgery patients, has also tended to decrease. Despite a decline in lumbar fusion surgeries following the COVID-19 pandemic, the use of rhBMP-2 in all spinal fusion surgeries was 16% in 2019 and is projected to increase to 20% by 2022 [64]. Although these inconsistent decremental trends may be reversed in the future, these cumulative data and statistical analyses provide insights into the correlation between the trends and the technology’s safety and efficacy profile in contemporary clinical practice.

Despite anecdotal reports of dose-dependent adverse reactions in previous studies, rhBMP-2 has been widely used off-label without a standardized dosage. Upon its initial development and commercial release, the rhBMP-2 dose was intended to be volume-dependent, with its volume equalling the internal volume of the cage [15]. As noted above, controlling the availability of rhBMP-2 signals at the cell surface is crucial for progressive and delayed peak expression during human bone healing, as well as for preventing dysregulated cell signalling [23]. Overfilling graft sites with high doses of rhBMP-2/ACS was associated with significant short-term osteoclastic activity, leading to bone resorption. A hypothesized mechanism for the promotion of osteolytic activity by rhBMP-2 confined within the interbody cage is that the geometry of the cage window limits rhBMP-2 diffusion, sustaining a higher concentration, or the mechanical pressure placed over the endplates by the cage induces bone remodelling through Wolff’s law [2,58]. However, the absence of guidelines or recommendations has led to substantial dose variability in the published literature, ranging from 4.2 mg to 40 mg per level [5,15,23,55]. Meanwhile, the designated concentration of commercially available rhBMP-2 has been standardized at 1.5 mg/mL, regardless of bioavailability and package size, thus adjustment of the total applied dose may have to rely on the number of carrier blocks used. Although the standardized concentration remains the same at 1.5 mg/mL, recent commercial formulations of rhBMP-2 for ACS also provide a lower dose of 1.05 mg. In an effort to reduce adverse reactions, clinicians have gradually decreased the total implant volume, which in turn has led to reduction in the total rhBMP-2 dose per level over time. A current clinical trial (ClinicalTrials.gov identifier : NCT04073563) is evaluating the efficacy of 2.1 or 4.2 mg rhBMP-2 per level compared to autologous bone in TLIF for one-and two-level fusions.

A longitudinal study in 1209 patients who received rhBMP-2/ACS between 2006 and 2020 demonstrated a significant decrease in the rhBMP-2 dose used per level over time. Of these, posterior lumbar interbody fusion (PLIF)/transforaminal lumbar interbody fusion (TLIF) and PLF demonstrated a significant decrease in the rhBMP-2 dose used per level, with major transitions seen in 2018, 2011, and 2013, respectively. In particular, PLIF/TLIF demonstrated the most significant dose reduction, with the average dose per level decreasing from 5.97 mg in 2009 to 4.09 mg in 2011, 2.36 mg in 2012, and 1.35 mg in 2020. In addition, the rhBMP-2 dose per level was identified as a significant predictor of complications following spinal fusion [14].

Research has focused on identifying the minimum effective dose (MED) of rhBMP-2 that optimizes fusion rates while minimizing complications. A quantitative exploratory meta-analysis of 48 studies involving 5890 patients analysed the dose-dependent effects of rhBMP-2 (i.e., fusion rate) and morbidity across various spinal arthrodesis procedures. This meta-analysis revealed considerable variability in the total rhBMP-2 dose used across studies and spinal arthrodesis procedures. This variability was likely influenced by differences in the fusion cage sizes employed in various surgical approaches. For ALIF, the use of rhBMP-2 significantly increased fusion rates (from 79.1% to 96.9%), and complication rates showed a positive correlation with rhBMP-2 dose (Pearson correlation coefficient 0.98, p<0.05). For PLF, the use of rhBMP-2 at doses ≥8.5 mg per level significantly increased fusion rates (from 75.3% to 95.2%), while BMP did not affect complication rates. In PLIF/TLIF procedures, the use of BMP resulted in a slightly higher fusion rate compared to the control group, but the difference was not statistically significant (95% vs. 93%). The dose of rhBMP-2 applied to the disc space varied between studies, ranging from 1.4 mg to 12.0 mg, with an average of 4 mg per level. The fusion rate in the lowest dose group (≤4 mg/level) was slightly higher than in the highest dose group (8.5-12.0 mg/level; 95.5% and 94.3%, respectively), but the difference was not statistically significant. The complication rate, which focused on the most commonly reported issues—including radiculopathy, heterotopic bone formation, wound infections and seromas—was 7.1%. This rate was slightly higher in the groups receiving rhBMP-2 compared to the control group. Furthermore, subgroup analyses indicated that the complication rates did not show a positive correlation with the rhBMP-2 dose used [26]. A logistic regression analysis conducted by Lytle et al. [41] investigating the rhBMP-2 MED in 690 patients who underwent TLIF between 2009 and 2014 identified an average dose of 1.28 mg rhBMP-2 per level as optimal, with the subsequent overall fusion rate of 95.2%. The rhBMP-2 per level was a significant predictor for fusion (2.35 odds ratio). The odds of achieving fusion significantly increased by 2.01 when the rhBMP-2 dose range increased from 0.16-1 mg/level to 1.01-2 mg/level. However, no significant increase in fusion odds was observed when the dose further increased from 1.01-2 mg/level to >2 mg/level. In addition, there was a logarithmic dose-dependent relationship between the fusion rate and BMP dose per level, with the fusion rate plateauing when the dose increased to ≥2 mg per level [41]. The same authors extended their investigation with a systematic review and meta-analysis, expanded to 2729 patients, exclusively with PLIF/TLIF. They reported that the average dose of rhBMP-2 used was 1.28-12 mg per level, with the lowest MED for fusion being 1.28 mg per level. Complication rates did not differ significantly between the various rhBMP-2 doses, and the rhBMP-2 placement site did not significantly affect fusion or complication rates [40].

However, given the heterogeneity of patient populations within the dose range groups and the lack of comprehensive data on the lower dose groups, this analysis required further investigation.

The recent global demographic shift toward an aging population has underscored the significance of rhBMP-2 as a potent osteoinductive agent. rhBMP-2 has demonstrated comparable fusion rates to autologous bone grafts, even in physically compromised individuals with multiple comorbidities. Despite its superior osteoinductive properties, the clinical application of rhBMP-2 has faced challenges due to the occurrence of unexpected adverse reactions not encountered with endogenous BMP-2 during bone healing. And despite efforts by surgeons to mitigate risks by adjusting the rhBMP-2 dose downward from the higher doses used in the early years of clinical adoption, the potential for adverse reactions remains a concern. Recent commercial preparations of rhBMP-2 delivered on ACS or osteoconductive carriers provide a standard dose of 1.05 mg, which can be segmented and distributed over multiple fusion segments. Although there is widespread availability of smallerdose rhBMP-2 preparations, there remains a need for further research to investigate the effect of these lower doses at each fusion level. Furthermore, there still lacks definitive data regarding the MED of rhBMP-2 that will achieve optimal fusion rates while minimizing associated complications.

Recent, ongoing research has demonstrated the efficacy of low-dose rhBMP-2 in promoting bone healing while minimizing complications. In the case of additional one-side PLF performed with interbody fusion, applying 1 mg rhBMP-2 per segment combined with HA granules resulted in fusion rates and patient-reported outcomes comparable to those achieved with ICBG. Notably, no device-related adverse events were observed for up to 1 year of follow-up [56]. In PLIF procedures, 0.5 mg rhBMP-2 was used for single-level fusion and 1 mg for 2-3 level fusions with a composite carrier (HA, β-TCP and poloxamer hydrogel). Average doses of 0.3-0.5 mg per segment promoted effective bone fusion in PLIF. Screw loosening was more common in the control group, but no significant differences were seen for other complications [47]. In TLIF for 1-2 level fusion, 0.5-1 mg rhBMP-2 was used intra-cage with the same composite carrier, demonstrating the safety and efficacy of E. coli-derived rhBMP-2 with minimal postoperative complications [37]. A recent retrospective study by Ryu et al. [50] examined the use of low-dose rhBMP-2 (0.5 mg/level) combined with HA granules in TLIF procedures for osteoporotic patients. Fusion outcomes were comparable between patients receiving rhBMP-2 alone and those additionally treated with denosumab, while the latter group showed a significantly lower incidence of osteolysis (p=0.013). These findings reinforce the feasibility of low-dose rhBMP-2 use in high-risk populations with adjunctive therapies [50]. In a pilot study aimed at preventing proximal junctional kyphosis (PJK) and proximal junctional failure (PJF) after spinal deformity correction, 0.5 mg rhBMP-2 was injected into the uppermost instrumented vertebra with a β-TCP poloxamer hydrogel carrier. This technique improved cancellous bone formation around the pedicle screws, resulting in fewer incidents of PJK and PJF, greater resistance to compression fractures and enhanced screw pull-out strength. Patient-reported outcomes were excellent during the first 6 months after surgery, with no revision cases among patients in the study group [38]. A clinical trial is underway to compare the safety and effectiveness of 0.5-1.0 mg rhBMP-2 per level (maximum dose of 2 mg per patient) with HA granules against the use of autogenous bone in TLIF for 1-2 level fusions. This trial is registered with the Korea Clinical Research Information Service (https://cris.nih.go.kr; identifier number : KCT0005610) [13]. These studies highlight the potential of low-dose formulations to achieve significant therapeutic outcomes, suggesting an optimized balance between efficacy and safety in clinical applications.