Effects of rhBMP-2 Loaded Hydroxyapatite Granules/beta-Tricalcium Phosphate Hydrogel (HA/ β-TCP hydrogel) Composite on a Rat Model of Coccygeal Intervertebral Fusion

Shinichi Nakagawa Osaka University Graduate School of Medicine Rintaro Okada Minoh Municipal Hospital Junichi Kushioka Osaka University Graduate School of Medicine Joe Kodama Kansai Rosai Hospital Hiroyuki Tsukazaki Kansai Rosai Hospital Zeynep Bal Osaka University Graduate School of Medicine Daisuke Tateiwa Osaka University Graduate School of Medicine Yuichiro Ukon Osaka University Graduate School of Medicine Hiromasa Hirai Osaka University Graduate School of Medicine Takahiro Makino Osaka University Graduate School of Medicine Shota Takenaka Osaka University Graduate School of Medicine Seiji Okada Osaka University Graduate School of Medicine Takashi Kaito (  takashikaito@ort.med.osaka-u.ac.jp ) Osaka University Graduate School of Medicine


Introduction
After the approval of recombinant human bone morphogenetic protein-2 (rhBMP-2) as a bone graft substitute for anterior lumbar intervertebral fusion by the Food and Drug Administration in 2002 1,2 , the use of rhBMP-2, including off-label use, has been adopted for approximately 30% of lumbar spine fusion surgeries in the United States 2 . However, adverse events related to the use of the supraphysiological dose (mg order) of rhBMP-2, such as soft tissue swelling, local in ammation, osteolysis, ectopic bone formation, retrograde ejaculation, and radiculitis, prevent its widespread use 3,4,5 . Thus, there is a need for an e cient drug delivery system for rhBMP-2 to mitigate these adverse events by enabling e cient bone formation with low-dose rhBMP-2.
The purpose of this study was to elucidate the effects of NP as a carrier for rhBMP-2 in bone formation and compare the adverse events of NP with those of a collagen sponge (CS) carrier using a rat coccygeal intervertebral fusion model.
Spinal segments harvested at 6 weeks postoperatively were scanned using high-resolution ex vivo microcomputed tomography (micro-CT). Intervertebral fusion was de ned as bridging of bone formation on both the coronal and sagittal images of the intervertebral disc space. The fusion rates at the intervertebral disc space were 25% in the IB group, 50% in the CS-low BMP group, 62.5% in the CS-high BMP group, and 87.5% in both NP-low and -high BMP groups, respectively, as determined using micro-CT. The fusion rate of both the NP groups was signi cantly higher than that of the IB group (p = 0.04). In contrast, the fusion rates of the CS group did not differ from those of the IB group (p= 0.21) (Fig. 1).
Representative coronal micro-CT images of treated intervertebral discs in each group. In the IB group, the grafted bone almost disappeared, and no bridging new bone formation was observed (a). In the CS-low and -high BMP groups, new bone formation that partly bridged the intervertebral disc space was observed, but the bone volume and density were low (b, c). In the NP-low and NP-high groups, the intervertebral disc space was occupied by dense new bone formation with a small amount of HA particles (d, e).
Time-dependent bone changes during the pre-mortal period determined using micro-CT In vivo micro-CT was performed immediately after surgery, 2 days postoperatively, and every week after surgery until euthanasia (6 weeks). In the IB group, the deisity of the implanted iliac bone decreased with time and the bone almost disappeared at 6 weeks post-operation. In both the CS groups, newly formed bone appeared outside the intervertebral disc space at 2 weeks post-operation, and new bone formation toward the intervertebral disc space was observed; however, the amount of bone formation at the disc space was small even at 6 weeks post-operation. In contrast, in both the NP groups, new bone formation was observed only in the intervertebral space, and the intensity of the new bone increased (Fig. 2).
In the IB group, the grafted bone was absorbed over time and almost disappeared at 6 weeks postoperation (a). In the CS group, new bone formation outside the disc space was observed at 2 weeks postoperation. The amount of new bone formation at the intervertebral disc space was limited until 6 weeks post-operation (b, c). In the NP groups, the density of HA and β-TCP decreased over time. New bone formation limited to the disc space was observed at 2 weeks post-operation, and the density of the new bone increased over time until 6 weeks post-operation (d, e).

Evaluation of adverse events at the surgical sites
The incidence of adverse events, including delayed wound healing, soft tissue swelling, ectopic bone formation, and osteolysis, was observed as follows. Ectopic bone formation was diagnosed when bone formation outside the intervertebral disc space greater than 2 mm was observed using in vivo micro-CT (Fig. 3a). Soft tissue swelling, the distance between the skin surface and plate measured using in vivo micro-CT, decreased to less than 1 mm on day 2 post-operation (Fig. 3b). Delayed wound healing was diagnosed when the operative wound did not heal within 2 weeks (Fig. 3c). Osteolysis of endplates was diagnosed when erosion of the bony endplate > 3mm was observed using in vivo micro-CT (Fig.  3d). The total adverse event score (AES) was calculated by assigning one point for each adverse event.
The adverse events at the surgical site are summarized in Table 1. The incidence of ectopic bone formation in the CS-high BMP group (87.5%) was signi cantly greater than that in the IB group (0%; p= 0.01) and in the NP-high BMP group (25%; p = 0.04).
Table1. Summary of the adverse events.
IB; iliac bone, BMP; rhBMP-2, CS; collagen sponge, NP; Novosis putty, AES; adverse event score, *; p < 0.05; (vs. IB group), **; p < 0.001; (vs. IB group), #; p < .05; (vs. CS-high BMP group) The incidence of soft tissue swelling was 62.5% in the CS-high BMP group (62.5%), 37.5% in the NP-high BMP group, 25% in both the CS-low BMP and NP-low BMP groups, and 12.5% in the IB group. The incidence of soft tissue swelling in the CS-high BMP group was signi cantly higher than that in the IB group (p = 0.03). The incidence of delayed wound healing was not different among the ve groups. Osteolysis of the endplates was not observed in this study. AES in the CS-high BMP group (1.88 ± 0.30) was higher than that in the NP-high BMP group (0.75 ± 0.25; p= 0.04).
Operative site (c), and in vivo micro-CT images (a, b, and d) as representative images of adverse events. Swelling ratio at the surgical sites The ratio of soft tissue swelling was calculated by dividing the soft tissue volume (TV) of the surgical site on day 2 post-operation by the TV on one day before surgery ( g.4a). The swelling ratio of the CS-high BMP group (164.9 ± 3.4%) was signi cantly higher than that of all the other groups. The swelling ratio of the NP groups was not affected by the BMP-2 dose, in contrast to the CS groups, in which the swelling ratio was increased when combined with high-dose BMP-2 (Fig. 4b).
(a) Region of interest for measurement of tissue volume. The dotted line denotes the middle of the intervertebral disc space. (b) Comparison of the swelling ratio among groups. The swelling ratio in the CShigh group was signi cantly greater than that in the other groups. *; p< 0.05, **; p< 0.001.

Histological analysis
In the IB group, the intervertebral disc space was predominantly composed of brocartilage tissue. In the CS-low BMP group, only a small amount of new bone formation was observed in the intervertebral disc space. In the CS-high BMP group, bridging of new bone formation between the endplates was observed, but the new bone tissue was predominantly composed of adipose tissue. In contrast, in both the NP groups, new bone formation bridging between the endplates was composed of thick trabecular bone. In the NP-low and NP-high groups, a small amount of HA granules remained in the intervertebral disc space, but the majority of β-TCP disappeared (Fig. 5).
Hematoxylin & eosin (H&E) staining of treated segments. Whole intervertebral disc space (a-e), Magni ed central parts of the disc space (f-j). Fibrous tissue was present between adjacent endplates in the IB group (a, f) and in the CS-low BMP group (b, g). In the CS-high group, new bone formation that bridged the endplates was observed, but the new bone was predominantly composed of adipose tissue (c, h). Thick trabecular bone formation between the endplates was observed in both the NP groups (d, e). A small amount of HA granules remained inside the newly formed bone (i, j). NB, new bone; HA, hydroxyapatite.
Quanti cation of the newly formed bone area in the interbody space The percentage of new bone area in the region of interest (ROI) was signi cantly higher in the NP groups (NP-low BMP group, 28.4%; NP-high group, 30.4%) than in the CS group (CS-low BMP group, 10.1%; CS-high group, 15.3%; *p< 0.03; Fig. 6b).

Discussion
In this study, the use of HA/β-TCP hydrogel composite as a carrier for rhBMP-2 resulted in a superior bone fusion rate and a lower incidence of side effects compared to those following the use of CS in a rat model of coccygeal interbody fusion. In addition, new bone formation in the HA/β-TCP hydrogel composite group was localized in the intervertebral disc space; in contrast, ectopic bone formation outside the intervertebral disc space was observed in the group treated with CS. Thus, the use of the HA/β-TCP hydrogel composite enabled e cient and spatially controlled new bone formation by rhBMP-2.
The use of rhBMP-2 remains controversial because the e cacy of BMP-2 for high fusion rate and short operative time may be counteracted by potential side effects such as ectopic bone formation, soft tissue swelling, local in ammation, osteolysis, and retrograde ejaculation 4,5 . In this study, using the HA/β-TCP hydrogel as a carrier material for BMP-2, e cient and spatially controlled bone regeneration was achieved with fewer in ammation-related side effects. The following characteristics of the HA/β-TCP hydrogel composites are expected to corroborate the results.
The rst characteristic of the composite material is the sustained release of rhBMP-2. The BMP-2-related side effects have been reported to be dose dependent [20][21][22] . Therefore, the decrease in local rhBMP-2 concentration due to sustained release of rhBMP-2 can reduce the side effects. The β-TCP hydrogel used in this study sustainably releases 20% of the total rhBMP-2 within 7 days, in contrast to the CS, which releases almost 100% of the total rhBMP-2 within one day 17 . Another study using the same HA/β-TCP hydrogel composite demonstrated that rhBMP-2 release within 7 days was 19.5% from the composite and 98.3% from the collagen carrier 19 . This sustained release of rhBMP-2 is supposed to contribute to the mitigation of BMP-2-related side effects and improvement in fusion rate by controlling the spatial spread of BMP-2 and inducing new bone formation for a long time. In fact, soft tissue swelling did not change between low dose (25%) and high dose (37.5%) of rhBMP-2 in the NP groups, in contrast to the CS groups, in which a signi cant increase in soft tissue swelling was observed in high-dose BMP group (62.5%) compared to that in low-dose BMP group (25%). Ectopic bone formation outside the intervertebral space was infrequent in the NP group (low BMP/high BMP, 12.5%/25%) than in the CS group (low BMP/high BMP, 50%/87.5%). In addition, the quality of newly formed bone was superior in the NP group, in which abundant thick trabecular bone formation was observed, than that in the CS group, in which fatty bone marrow was predominant.
Another advantageous characteristic of the HA/β-TCP hydrogel composite is the combination of two calcium phosphate scaffolds (HA and β-TCP) with different biodegradabilities. HA and β-TCP can provide biomechanical strength and moldability of the composite, and the low-biodegradable HA provides a longterm scaffold for bone formation 12 , and the highly biodegradable β-TCP microspheres can provide a space for new bone formation in addition to enhancing osteogenic cell differentiation [23][24][25] . Histological evaluation demonstrated that a small amount of HA granules remained in the intervertebral disc space at 6 weeks post-operation, but the majority of β-TCP was remodeled and regenerated into new bone (Fig. 5i, j).
This study has several limitations. First, the results for rodent models cannot be directly extrapolated to humans because of the different biomechanics of the spine between quadrupeds and bipeds. Second, no intervertebral fusion cages were applied to the intervertebral disc space in this study. The use of an intervertebral cage might increase the fusion rate in the IB group or rhBMP-2-loaded CS group.
In conclusion, the HA/β-TCP hydrogel composite enabled superior bone induction with a low dose of rhBMP-2 and reduced the incidence of side effects caused by high doses of rhBMP-2 compared with the collagen carrier in a rat model of coccygeal intervertebral fusion. The HA/β-TCP hydrogel composite is a novel biomaterial for e cient bone regeneration using rhBMP-2.

Characterization of HA and β-TCP in HA/β-TCP hydrogel composite
The HA/β-TCP hydrogel biomaterial adopted in this study, which contained 40% HA and 60% β-TCP, was manufactured by CGBio Co., Ltd., Seongnam, Republic of Korea. In addition to their general appearance, scanning electron microscopy (SEM; Hitachi S-4800, Japan) was used to determine the detailed microstructures of β-TCP hydrogel and HA (Supplementary Fig. 1).
The HA granules, β-TCP microspheres, and hydrogel were manufactured by CGBio Co., Ltd., Seongnam, Republic of Korea. The size of HA granules ranged from 3.0 mm to 6.0 mm, characterized by approximately 70% of porosity and 99% of interconnectivity. The β-TCP microspheres produced using the spray-drying method were ~45-75 μm in size, with approximately 68% porosity. The hydrogel was composed of a polyethylene glycol (PEG)/ polypropylene glycol (PPG)/PEG block copolymer and hydroxypropyl methylcellulose (HPMC) composite. The PEG/PPG/PEG block copolymer is thermosensitive, and is null at low temperature and can be mixed homogeneously, while it is gel at body temperature and can be shaped into a desired form. The HPMC composite is viscoelastic, which is helpful for setting a certain shape for the nal injected form, with increased resistance against external stress.

Animals and experimental groups
The  Table 1). The dosage of rhBMP-2 was decided based on previous reports in which the same coccygeal intervertebral fusion model was used 27 . Volumetric comparison of rat and human intervertebral disc space revealed that 5 μg of rhBMP-2 in rats corresponds to 1.5 mg of rhBMP-2 in humans.

Preparation of grafting materials
Allogenic iliac cancellous bone Allogenic iliac cancellous bone was harvested from a donor SD rat immediately before surgery. The volume of iliac bone grafting for each disc space was set to 60 mm 3 (0.05 mg).
HA/β-TCP hydrogel composite NP consisting of HA/β-TCP microsphere/poloxamer 407 hydrogel was kindly provided by CG Bio Co., Ltd. 28 . The hydrogel contained poloxamer 407, which is a biodegradable and biocompatible polymer that has been applied in various elds as a biomaterial 29,30 . The hydrogel was mixed with β-TCP microspheres in a 1:1 weight ratio. Next, HA granules soaked with rhBMP-2 were mixed with the β-TCP hydrogel in a 3:2 weight ratio in a special mixing syringe to form homogeneous clay-like composites (Fig. 7a). The composites were molded to a volume of 60 mm 3 (0.05 mg) before implantation in a container (Fig. 7b).
Subsequently, a 10-mm dorsal midline incision was made to expose the intervertebral disc. The posterior one-third of the annulus brosus and total nucleus pulposus were removed, and then the cartilaginous endplates were detached and excised using a small raspatory and rongeur to avoid damage to the bony endplates 27 . After irrigation with saline, grafting materials (60 mm 3 of IB, CS, or NP) were implanted into the intervertebral disc space (Fig. 7d).
The screw positions and preservation of the bony endplate were con rmed via micro-CT immediately after surgery (Fig. 7e).
Postoperative sagittal (e, upper) and coronal (e, lower) views of micro-CT, with yellow arrowheads indicating the surface of the bony endplate in the sagittal view.
In vivo micro-CT analysis In vivo micro-CT was performed immediately after surgery, on postoperative day 2, and every week after surgery until euthanasia (6 weeks). The treated coccygeal vertebrae were scanned using micro-CT (R_mCT; Rigaku Mechatronics, Tokyo, Japan) at a resolution of 59 μmvoxel in vivo, and the data were collected at 90 kV and 160 μA. Visualization and data reconstruction were conducted using TRI/3D-BON (Ratoc System Engineering, Tokyo, Japan).
To quantify the soft tissue swelling at the surgical sites, the TV was calculated by setting a ROI on micro-CT images as follows: a rectangle including the skin surfaces in the axial width and 5 mm in the longitudinal length that centers the intervertebral disc space (threshold [L = 15500], software; TRI/3D-BON, Ratoc System Engineering, Tokyo, Japan) (Fig. 4a). The ratio of soft tissue swelling was calculated by dividing the TV of the surgical site on postoperative day 2 by the TV one day before surgery.

Histological analysis
Dissected and formalin-xed coccygeal segments were xed with 10% formic acid, dehydrated in a graded ethanol series, decalci ed with K-CX (Falma, Tokyo, Japan), and embedded in para n wax. Serial sagittal sections (3-µm thickness) were cut and stained with H&E. A 1.5 × 2-mm 2 ROI (interbody space) was extracted from the newly formed fusion mass. The new bone area (red) was color coded using ImageJ software (version 1.52q, U. S. National Institutes of Health, Bethesda, Maryland, USA) 31 , and the percentage of the newly formed bone area in the ROI was calculated (Fig. 6a).

Statistical analysis
All statistical analyses were performed using the JMP 15 Statistics software (SAS Institute Inc., Cary, NC, USA). Results are presented as means ± standard deviation. Fisher's exact test was used to compare the success of fusion and the incidence of adverse events in each group. Differences in the measured variables between multiple groups were analyzed using one-way analysis of variance followed by Dunnett's test or Tukey-Kramer test. Statistical signi cance was set at p < 0.05. Figure 1 Ex vivo micro-CT images of treated segments Figure 2 Temporal in vivo micro-CT images of the treated segments Representative images of each event.

Figure 4
Quanti cation of surgical site swelling.
Page 16/18 Figure 5 Histological images of treated segments at 6 weeks post-operation.

Figure 6
Quanti cation of the new bone area in the interbody space. (a) Comparison of new bone area inside the spinal fusion mass in histological sections. A 1.5 × 2-mm2 ROI (interbody space) was extracted from the newly formed fusion mass. The new bone area (red) was color coded using ImageJ software (version 1.52q, U. S. National Institutes of Health; https://imagej.nih.gov/ij/). (B) The percentage of the newly formed bone area in the ROI was signi cantly higher in the NP groups than in the CS groups (n = 8, *p < 0.03) (IB; 9.4%, CS-low BMP group; 10.1%, NP-low BMP group; 28.4%, CS-high group; 15.3%, NP-high group; 30.4%; data represent mean ± SD; *p < 0.03 by Turkey-Krammer test).