Injectable, Solvent Free Strontium Carbonate Poly(Allyl Glycidyl Ether Succinate) Composite Networks for Vertebral Augmentation

Abstract

Vertebral body compression fractures are a major cause of chronic back pain, particularly in older adults. The current standard of care is a percutaneous polymeric cement injection into the fractured vertebra of these patients to alleviate pain and stabilize the fracture. Augmentation is currently performed by injecting a poly(methyl methacrylate) (PMMA) slurry of polymer, monomer, and initiator mixed with barium sulfate (BaSO₄) into the vertebrae, which then polymerizes in vivo. PMMA is nondegradable, possesses a 100-fold higher ultimate compressive strength (⁓100 MPa) than native vertebra (⁓1 MPa), and the in vivo free radical polymerization is highly exothermic. These factors are thought to contribute to the development of new fractures in vertebral bodies above and/or below the augmented level during long-term recovery. Herein, a solvent-free polymer system using poly(allyl glycidyl ether succinate) (PAGES) is developed for vertebral augmentation. PAGES crosslinks in situ through thiol-ene click chemistry with a cure time at 37 °C ranging from 17 to 53 minutes based on degree of polymerization and crosslinker concentration. The addition of SrCO₃ increased the ultimate compressive strength (σ_max) of the PAGES composite to 4.4 ± 0.4 MPa. Furthermore, SrCO₃ increases osteoblast proliferation and differentiation of mesenchymal stem cells seeded onto the surface of PAGES composite. Finally, the compressive strength of fractured vertebra was increased in an ex vivo surrogate rabbit model when filled with injected PAGES composite demonstrating its potential as a bone augmentation material.

Graphical Abstract

Introduction

Vertebral compression fractures (VCFs) arise from normal loading on an abnormal vertebra or traumatic loading on a normal vertebra.[1] VCFs can be a source of chronic back pain and morbidity, particularly in older adults, and is estimated to cost the United States healthcare system more than $1 Billion per year.[2] VCFs most commonly result from either osteoporosis, a disease that weakens bones, or bone destruction caused by cancerous or benign masses.[1,3] Treatment for these fractures varies from medication and physical therapy to invasive intervention with open surgical fusion or percutaneous vertebral augmentation.[4] There are two different percutaneous augmentation strategies in widespread use: vertebroplasty and kyphoplasty.[2] Augmentation is accomplished through one or two small incisions in the back through which a self-tapping drill and cannula are introduced. Both surgical procedures use PMMA cement to reinforce the VCF; however, kyphoplasty utilizes a balloon to restore the height of the vertebra prior to injection.[5] Recently, kyphoplasty has been found to confer a survival benefit compared to conservative management.[5] Percutaneous augmentation procedures have a significantly lower rate of surgical complications compared to open spinal fusion procedures.[6] Additionally, for patients with fractures due to cancerous masses, the tumor can be ablated during the vertebral augmentation procedure, adding some localized cancer control.[7] Due to these clinical advantages, vertebral augmentation is an increasingly common procedure to address VCFs.

In current practice, a powdered solid of barium sulfate (BaSO₄), for radiopacity, and poly(methyl methacrylate) (PMMA) is mixed with a liquid portion containing methyl methacrylate (MMA) monomers, stabilizers, and a radical initiator. This polymeric slurry is then injected into the vertebral void space where it cures over time until fully solidified. Despite its widespread use as an augmentation material, PMMA has multiple drawbacks including compressive modulus and peak compressive strength that are approximately 100-fold greater than native vertebral bodies.[8,9] This mechanical property discrepancy is thought to contribute to the development of fractures in adjacent vertebral bodies following augmentation, occurring in 7–15% of patients.[10–12] Additionally, PMMA is not degradable, the polymerization reaction within the vertebral body is highly exothermic, and MMA monomers have been shown to induce neurotoxicity.[13,14] These factors are thought to contribute to an absence of new bone growth around and within the PMMA block.

Ceramics and metal salts are added to augmentation materials to alter biological effects and afford radiopacity. Radiopacity is essential so that the material can be easily identified using fluoroscopic guidance during the procedure. Barium sulfate (BaSO₄) is the current standard additive used to make PMMA radiopaque and is known to be tolerated by patients. Other ceramic additives have been shown to promote osteogenesis, including strontium carbonate (SrCO₃) and alpha tricalcium phosphate (α-TCP).[15,16] Additionally, SrCO₃ doped cements have been found to promote osteoblast attachment and osseointegration.[17,18] Finally, SrCO₃ containing hydroxyapatite cements have been used in vertebroplasty in patients with similar levels of pain relief to PMMA.[19]

Other materials currently under investigation for use in bone augmentation systems include calcium phosphate cements (CPCs) which show promise for improved biocompatibility.[16] Unfortunately, many CPCs have long setting times and can experience phase separation which has further hampered their clinical translation.[13,20] Methods have been explored to soften native PMMA bone cement through the incorporation of non-miscible phases during polymerization, such as linoleic acid or hyaluronic acid.[21–23]. More recently, Yang et al. incorporated microcapsules of paraffin into the cement to mitigate the effect of the exothermic polymerization on the surrounding vertebrae. 24 h after injection, PMMA cement with 20% microcapsules showed a significantly smaller thermal zone of necrosis.[24] Within PMMA cements, groups have looked to delivery active compounds and integrin-specific peptides in order to aid in bone regeneration.[25,26] Many reports to date look to augment the standard PMMA resin through additives or introduction of pores but alternative polymeric networks are limited.

An optimal bone augmentation material should have mechanical properties similar to native bone, solidify on a timeframe that is compatible with the current interventional procedure, afford earlier patient mobility, and ideally promote the deposition of new bone.[27] Herein, we explore the use of poly(allyl glycidyl ether succinate) (PAGES) crosslinked with trimethylolpropane tris(3-mercaptopropionate) via a thiol-ene mechanism for vertebral augmentation.[28] PAGES polymer possesses a low neat viscosity (36 Pa·S for DP₂₉ and 77 Pa·S for DP₄₈), affording injectability without the addition of solvents or diluents.[29] The absence of solvent increases the translational relevance of this work and avoids a major limitation of other possible polymer formulations. Additionally, the thiol-ene “click” crosslinking reaction does not generate residual by-products, and has been applied in many other biomaterial applications.[30–33]

Results

PAGES Synthesis:

PAGES was synthesized using ring opening copolymerization of succinic anhydride and allyl glycidyl ether with a propargyl alcohol initiator (Figure 1A), as described previously.[28] Polymers possessing various degrees of polymerization (DP₂₉ and DP₄₈) were synthesized by altering the ratio of monomer to initiator (Table 1, Figure S1–8). As the DP_n increases, the viscosity increases proportionally as well as the number of potential crosslinking handles, which are present on each monomer repeat unit.

Figure 1:

Polymer network formation via crosslinking of PAGES polymer. A) Ring opening copolymerization of succinic anhydride and allyl glycidyl ether with a propargyl alcohol initiator to produce PAGES polymer. B) PAGES polymer and crosslinker mixtures form covalent networks through thiol-ene addition reactions. Specific crosslinking percentages can be targeted by adjusting the stoichiometric ratio of allyl groups in the polymer backbone to thiol groups in the crosslinker. C) Fourier Transform Infrared (FTIR) spectroscopy data of a 60% crosslinked sample demonstrating decrease in the stretch of allyl group (1650 cm⁻¹) as crosslinking occurs over time. D) Evolution of normalized allyl stretch intensity (1650 cm⁻¹) as crosslinking occurs over time for samples with varying targeted crosslinking (20%, 30%, and 60% of allyl groups reacted). E) Change in normalized allyl stretch intensity (1650 cm⁻¹) between 15 minutes and 1 day after mixing polymer with crosslinker for samples with varying targeted crosslinking (20%, 30%, and 60% of allyl groups reacted) (n=5). Box plots show median, first quartile, third quartile, maximum, and minimum. Extent of statistical significance determined using Tukey’s multiple comparisons test with *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Table 1.

Characterization of each PAGES polymer.

[SA]:[AGE]:[I]	DPna	Mn (kDa; NMR)a	Mn (kDa;SEC)b	Mw (kDa;SEC)c	Ðm d	Tg (°C)e	Td (°C)f	Ƞ*(Pa.s)g
30:33:1	29	6.2	8.6	10.1	1.2	−35 to −32	318	36
48:52:1	48	10.3	10.8	12.7	1.2	−31 to −29	317	77

^a– Calculated using ¹H-NMR in CDCl₃.

^b– Calculated using M_w by MALS and Ð by RI detection using THF SEC.

^c– Calculated using MALS using THF SEC.

^d– Calculated using RI detection using THF SEC.

^e– 10°C min⁻¹, second heating cycle using DSC.

^f– 10% mass loss, 10°C min⁻¹ using TGA.

^g– 0.1 rad/s to 500 rad/s frequency sweep at 10% strain and 25 °C, values of complex viscosity are reported at 10 rad/s

Effect of Degree of Polymerization on Network Formation:

In a series of experiments, PAGES networks and composites were crosslinked under ambient conditions with trimethylolpropane tris(3-mercaptopropionate) via a thiol-ene addition mechanism (Figure 1B) with 20%, 30%, or 60% of the polymer allyl groups theoretically reacted due to exposure to ambient light during mixing and no added photo initiator.[34] Fourier Transform Infrared (FTIR) spectroscopy measurements were used to monitor the decrease of the peak associated with the allyl stretch (1650 cm⁻¹)[35] in the PAGES spectra as the network crosslinking progresses and allyl groups are consumed (Figure 1C, Table S1). Furthermore, the normalized decrease in the allyl stretching peak height in the 20%, 30%, and 60% crosslinked samples at 1 day was greater as additional crosslinker was added (Figure 1D&E). The ratio of the allyl stretching peak height percentage decrease between conditions was proportional to the amount of crosslinker added to the different samples, suggesting that the amount of allyl conversion has a nearly linear relationship with the amount of crosslinker added over the range tested.

The gelation time of different PAGES DP_ns and crosslinker percentages were determined using rheology. Time to gel is known to be longer for smaller DP_n related to increased requirements for interchain crosslinking, and larger DP_n, related to increased viscosity and thus decreased diffusion.[36] Based on this, experiments were performed with DP₂₉ or DP₄₈ PAGES polymers. The time to gel was significantly lower with increasing crosslinking and temperature for both polymers tested (Figure 2). The average time to gel of the DP₂₉ with 60% crosslinking was 25 ± 3 minutes at 25 °C and 17 ± 3 minutes at 37 °C compared to 49 ± 7 minutes at 25 °C and 33 ± 1 at 37 °C for DP₄₈ PAGES polymer networks. Gelation proceeded even in samples that were tested in dark conditions where they were shielded from ambient light during curing (Figure S11).

Figure 2:

Rheological investigation of gelation time of PAGES networks. A-B) Time to gelation point for PAGES DP₂₉ and DP₄₈ resins with different amounts of crosslinking (20%, 30%, and 60% of allyl groups reacted) at room temperature (25 °C) (n=3) (A) and physiological temperature (37 °C) (n=3) (B). C-F) Median sample storage modulus and loss modulus at 10 rad/s as crosslinking occurs over time for DP₂₉ and DP₄₈ PAGES samples with varying amounts of crosslinking (20%, 30%, and 60% of allyl groups reacted) at 25 °C and 37 °C. G-J) Enlarged plots from C-F showing the median gelation point when the median storage modulus curves and median loss modulus curves intersect (tan(δ)=1) for DP₂₉ and DP₄₈ PAGES samples with increased crosslinking (20%, 30%, and 60% of allyl groups reacted) at 25 °C and 37 °C. Bar charts show mean with error bars representing standard deviation. Extent of statistical significance was determined using Tukey’s multiple comparisons test with *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Degree of Polymerization and Crosslinking Impacts Mechanical Properties

Crosslinked samples were fabricated to determine the compressive properties of the networks. Samples were allowed to cure under ambient conditions for varying amounts of time and were then compressed until failure (Figure 3A, Figure S12). We have also applied ideas from rubber elasticity to relate stress to the shear modulus (G) of the networks through a quadratic function of the compression ratio via a Neo-Hookean fitting of the stress-strain curves (Figure S13 & Table S3). After one week, 30% and 60% crosslinked DP₄₈ PAGES resins were significantly stronger than 20% crosslinked DP₄₈ PAGES resins (1 ± 0.3 MPa for 30%, 1.1 ± 0.6 MPa for 60%, 0.4 ± 0.06 MPa for 20%) (Figure 3B). 60% crosslinked DP₄₈ PAGES resins exhibited stiffening over time with a strain at break of 55 ± 8% after 1 day and a strain at break of 28 ± 2% after 14 days (Figure 3C). Furthermore, DP₂₉ PAGES resins were found to be significantly less flexible (26 ± 3% compared to 35 ± 3%) and stronger (2.0 ± 0.2 MPa compared to 1.1 ± 0.6 MPa) than DP₄₈ at 60% crosslinking after 7 days (Figure 3B&C, Table S4). Since the ultimate compressive stress of human vertebra is reported to be between 1 and 6 MPa, additional testing was performed with the 60% crosslinked PAGES resins.[37]

Figure 3:

Compressive strength of PAGES DP₂₉ and DP₄₈ resins with different amounts of crosslinking (20%, 30%, and 60% of allyl groups reacted) allowed to cure under ambient conditions for varying amounts of time (1, 7, and 14 days). A) Median compressive stress-strain curves produced after resins were allowed to cure under ambient conditions for 1 day. B) Maximum compressive stress at break for samples allowed to cure under ambient conditions for varying amounts of time (n=5). C) Strain at break for resins allowed to cure under ambient conditions for varying amounts of time (n=5). Box plots show median, first quartile, third quartile, maximum, and minimum. Extent of statistical significance was determined using Tukey’s multiple comparisons test with *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Addition of Ceramics Increases Mechanical Strength without Impacting Gel Time

Ceramic nanoparticles (BaSO₄, SrCO₃, or α-TCP) were added to DP₂₉ PAGES resins prior to crosslinking to increase the mechanical strength of the resulting composites. BaSO₄ and SrCO₃ were each found to have better qualitative dispersal through the resulting composite polymer than αTCP (Figure S14) and allowed for crosslinking at higher mass fractions of composites. Microscopic dispersal of SrCO₃ in PAGES samples was further confirmed with scanning electron microscopy (SEM) and energy dispersive X-ray spectroscopy (EDS) (Figure S15). BaSO₄ and SrCO₃ composite samples exhibited increased strength compared to both α-TCP and additive-free networks at multiple time points for both DP₂₉ and DP₄₈ polymers (Figure 4, Figure S16–18). PAGES samples without additives had a maximum stress of 2.0 ± 0.3 MPa compared to 2.5 ± 0.2 MPa for 20% αTCP, 3.4 ± 0.4 MPa for 50% SrCO₃, and 3.0 ± 0.7 MPa for 50% BaSO₄ after 7 days of ambient curing (Table S5–7). There was some mass fraction dependency of the SrCO₃ composite strength with the 50% SrCO₃ samples demonstrating significantly higher ultimate stress than the 20% SrCO₃ samples after 14 days of ambient curing (4.3 ± 0.5 MPa versus 3.5 ± 0.4 MPa). Rheological analysis was used to confirm that the addition of ceramics did not impact the time to gelation (Figure S9 & 10).

Figure 4:

Addition of ceramics increases compressive strength of polymer networks. Compressive strength of PAGES DP₂₉ and DP₄₈ samples at 60% crosslinking with varying ceramic additives (alpha tricalcium phosphate (αTCP), strontium carbonate (SrCO₃), and barium sulfate (BaSO₄)) allowed to cure under ambient conditions for varying amounts of time (1, 7, 14, and 28 days). A-B) Median compressive stress-strain curves produced after PAGES DP₂₉ (A) and DP₄₈ (B) samples were allowed to cure under ambient conditions for 1 day. C-D) Maximum compressive stress at break for PAGES DP₂₉ (C) and DP₄₈ (D) samples allowed to cure under ambient conditions for varying amounts of time (n=6). E-F) Strain at break for PAGES DP₂₉ (E) and DP₄₈ (F) samples allowed to cure for varying amounts of time (n=6). Box plots show median, first quartile, third quartile, maximum, and minimum. Extent of statistical significance was determined using Tukey’s multiple comparisons test with *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Degradation of Crosslinked PAGES Networks and Composites

Crosslinked PAGES networks degrade through hydrolysis of ester bonds within the polymer backbone and the trimethylolpropane tris(3-mercaptopropionate) crosslinker.[28] Two degradation studies were performed by mass and recurring mechanical measurements to determine network degradation rate and changes in mechanical properties over time (Table S8 & 9). Samples were incubated at 37 °C with agitation for the duration of the study. Degradation in an acidic solution of 1M HCl was chosen to promote accelerated environmental degradation.[38] Crosslinked samples incubated in phosphate buffered saline (pH 7.4, PBS) demonstrated a maximum 5% swelling without significant mass loss over 26 weeks (Figure 5C). Samples in 1M HCl began to demonstrate mass loss at 4 weeks (Figure 5E). To compare our analysis to relevant clinical models on vertebrae fracture, we have calculated the values of the compressive modulus based on a linear fit of data between 5% and 7% strain.[39,40] All samples exhibited an increased compressive modulus over the first week likely due to continued crosslinking, as was also observed in the ambient condition samples (Figure 5D&F). A small decrease in compressive modulus was seen in all 1M HCl samples after 6 weeks, although it then stabilized through 18 weeks before further decreasing in the additive-free networks and 50% BaSO₄ composite networks (Figure 5F). The lack of degradation in the 50% SrCO₃ networks is thought to be related to a buffering effect of the CO₃²⁻ and ultimately HCO₃⁻ arising from the SrCO₃ containing polymer samples that slowed the accelerated acid mediated degradation. Mechanical strength of the samples in PBS was maintained throughout the duration of the study (Figure 5D). These findings demonstrate that PAGES maintains its strength even while degrading, suggesting that it will continue to provide augmentation support while being absorbed. The maintenance of mechanical properties during degradation is a desirable property as it may help to keep the augmented vertebral body from compressing further due to mechanical strength loss.

Figure 5:

PAGES DP₂₉ 60% crosslinked materials exhibit retained mechanical strength while degrading. A-B) Representative compressive loading and unloading curves of PAGES samples without any ceramic additive (A) and with 50% strontium carbonate (SrCO₃) additive (B) after degrading in 1M HCl for varying amounts of time (6 hours, 1 week, 10 weeks). C) Mass of PAGES samples without any ceramic additive, with 50% SrCO₃ additive, and with 50% BaSO₄ additive after degrading in PBS for up to 26 weeks (n=6). D) Compressive modulus of PAGES samples without any ceramic additive, with 50% SrCO₃ additive, and with 50% BaSO₄ additive after degrading in PBS for up to 26 weeks (n=6). E) Mass of PAGES samples without any ceramic additive, with 50% SrCO₃ additive, and with 50% BaSO₄ additive after degrading in 1M HCl for up to 26 weeks (n=6). F) Compressive modulus of PAGES samples without any ceramic additive, with 50% SrCO₃ additive, and with 50% BaSO₄ additive after degrading in 1M HCl for up to 26 weeks (n=6). Plotted data points show mean with error bars representing standard deviation.

Addition of SrCO₃ Increases Osteoblast Differentiation

Rat mesenchymal stem cells (MSCs) were cultured on thin-layers of PAGES networks with and without ceramic additives. To assess osteoblast differentiation, cells were stained for osteocalcin after 17 days in culture and the percentage of nuclei that colocalized with osteocalcin was determined. MSCs attached to and proliferated on all PAGES network coated plates demonstrating that PAGES is nontoxic and biocompatible (Figure 6). Adding SrCO₃ to the PAGES samples resulted in significantly increased osteoblast differentiation compared to any other PAGES network (53 ± 8% for PAGES + 50% SrCO₃ versus 26 ± 7% for PAGES) and was not significantly different to what was seen on the control plate surface (67 ± 8%) (Figure 6F, Table S10).

Figure 6:

In vitro analysis of DP₂₉ PAGES 60% crosslinked materials with ceramic additives. A-E) Representative images of rat mesenchymal stem cells (MSCs) after 17 days of culture on either tissue culture plates or a thin layers of PAGES network with or without additives (alpha tricalcium phosphate (αTCP), strontium carbonate (SrCO₃), and barium sulfate (BASO₄)). All scale bars represent 50 μm. Blue color represents cell nuclei stained using DAPI, and red color represents osteocalcin. F) Quantification of percentage of cells expressing osteocalcin (n=9). Box plots show median, maximum, and minimum. Extent of statistical significance determined using Tukey’s multiple comparisons test with *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Injected PAGES Networks Fill the Vertebral Body

Pre-fractured rabbit vertebra (Figure 7A) were filled with DP₂₉ PAGES augmentation material with and without SrCO₃ following a clinically relevant drilling procedure with a self-tapping kyphoplasty drill. Qualitatively, after filling the fracture, vertebrae were more stable and loose fragments were adhered to the bone. There was appropriate filling of the drilled vertebra with interdigitation of the polymer into the porous structure of the bone (Figure 7C–G, open arrows). Volumetric analysis through microcomputed tomography (μCT) demonstrated that PAGES with SrCO₃ augmentation material fills drilled spaces equally as well as PMMA does, with an average of 84% of the drilled space filled when using either material (Figure 7H). This analysis could not be performed for PAGES without ceramic because its radiographic density was too similar to the surrounding bone in the μCT scans, demonstrating that the addition of SrCO₃ did increase radiopacity (Figure 7B&C). Mechanical testing performed on the vertebra both before and after filling demonstrated an increase in the mechanical strength of the filled vertebra in comparison to the fractured vertebra, although the filled strength was below that of the native vertebra for all PAGES samples (Figure 7J&K). The ultimate strength of PMMA filled vertebra was higher than the native vertebra but did not reach statistical significance (Figure 7J&K, Table S11&12).

Figure 7:

Using DP₂₉ PAGES polymer at 60% crosslinking as a vertebral augmentation material in an ex vivo rabbit vertebral augmentation procedure. A-G) Microcomputed tomography (μCT) images of rabbit vertebra during ex vivo vertebral augmentation. All scale bars represent 3 mm. Solid arrows highlight the location of a drilled cavity. Hollow arrows represent porous bone tissue that can be filled with augmentation material. A) μCT image of broken rabbit vertebra. B) μCT image of broken rabbit vertebra filled with PAGES with no additive. C & F) μCT image of rabbit vertebra filled with PAGES with 50% strontium carbonate (SrCO₃). D) μCT image of rabbit vertebra filled with PAGES with 50% SrCO₃ with teal color displaying infill of PAGES into the porous bone structure. E & G) μCT image of rabbit vertebra filled with poly(methyl methacrylate) (PMMA), the current clinical standard for vertebral augmentation. H) Percentage of a drilled cavity in a rabbit vertebra that is filled with augmentation material using standard augmentation techniques (n=13 for PAGES + SrCO₃, n=6 for PMMA). J) Average compressive stress-strain curves for rabbit vertebra initially, after being fractured, and after being fractured and filled with augmentation material (n=8, n=6 for PMMA). K) Maximum compressive stress at break for rabbit vertebra initially, after being fractured, or after being fractured and filled with augmentation materials (n=8, n=6 for PMMA). Plotted data shows mean with error bars representing standard error of the mean. Box plots show median, first quartile, third quartile, maximum, and minimum. Extent of statistical significance determined using unpaired t-test or Tukey’s multiple comparisons test with *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Discussion

Crosslinked PAGES networks loaded with SrCO₃ meet many of the requirements for a percutaneously applied bone augmentation material. The addition of SrCO₃ makes the polymer radiopaque with a radiographic density higher than the surrounding bone, making it easily visible with μCT imaging (Figure 7). Crosslinked PAGES samples demonstrated a peak compressive strength in the range of a native human vertebral body (1–6 MPa) with increased strength following the addition of either BaSO₄ or SrCO₃ (Figure 4). There was some restoration of ex vivo rabbit vertebral body strength following filling with PAGES both with and without added SrCO₃ compared to the strength of the vertebra immediately following fracture (Figure 7). The addition of SrCO₃ also increased osteoblast differentiation of MSCs cultured on PAGES samples, demonstrating that the SrCO₃ maintains its expected biological effects when incorporated into the crosslinked PAGES networks (Figure 6). The average procedural time for vertebroplasty is 31 minutes and for kyphoplasty is 40 minutes.[41] The time to gelation of 60% crosslinked DP₂₉ PAGES samples is approximately 17 minutes at physiological temperature and occurs over an appropriate duration for a typical procedure (Figure 2). PAGES is a degradable polymer, but maintains strength during degradation and has a long degradation time. This property will help continuously support the bone throughout the healing process (Figure 5). This is particularly important in the vertebral augmentation population as these patients tend to be older and take longer to heal bony injuries.[42] Overall, crosslinked PAGES composites show promise as a bone augmentation material for percutaneous vertebral augmentation.

This work shows that crosslinked PAGES samples have tunable mechanical properties (Figure 5). One area of future study will focus on increasing the stiffness of the PAGES composites to more closely mimic bone. Currently, PAGES samples have a higher strain at break and a lower maximum compressive stress than native bone (Figure 7E), which could cause additional stress on the surrounding bones following implantation. Future efforts will explore changing the crosslinker to increase stiffness of the network.

The addition of SrCO₃ is well known to encourage bone deposition in vivo and osteoblastic differentiation in vitro. Future work will explore the effects of PAGES implantation in vivo and aim to determine if the addition of SrCO₃ improves bone deposition. Sr²⁺ releasing bone scaffolds have been previously studied in mice and have been found to promote osteogenesis, reduce osteoclast functionality, and help resolve inflammation.[43] Another area of additional investigation will explore the release of Sr²⁺ from the PAGES SrCO₃ composite samples to determine how much Sr²⁺ is released over time.

Conclusions

Current vertebral augmentation is limited by the lack of degradability and cell interaction with PMMA augmentation material as well as the mechanical property mismatch between PMMA and native bone. The reported solvent free, injectable PAGES polymer composite system, particularly the SrCO₃ composites, showcases an encouraging novel bone augmentation material with many advantages including appropriate bone filling, promotion of osteoblastic differentiation, crosslinking in a time frame consistent with current clinical practice, and maintenance of mechanical strength while degrading.

Experimental Methods

Rheology: Viscosity.

The complex viscosity of PAGES polymers was measured (n = 1) using a TA Instruments Discovery HR-3 Hybrid Rheometer. Samples were loaded on 20 mm parallel plate geometry with a 0.2 mm gap. A trim gap of 50 μm was used to ensure proper distribution of each sample between plates. The complex viscosity was collected over a frequency sweep from 0.1 rad/s to 500 rad/s at 10% strain and 25 °C. All measurements took place within the linear viscoelastic regime for these materials as determined using a prior stain sweep. The viscosity of samples was reported as the complex viscosity at 10 rad/s.

Rheology: Gel Point Determinations.

Gel points were determined for each composition (n = 3) and crosslinking percentage at both 25 °C and 37 °C using a TA Instruments Discovery HR-3 Hybrid Rheometer. Samples were prepared by mixing polymer and crosslinker according to the specified theoretical molar crosslink percentage, which is defined as the percentage of allyl groups in the polymer that react with thiol groups in the crosslinker, in a glass vial under ambient conditions using a spatula until homogenous. Samples were then loaded on 20 mm parallel plate geometry with a 0.2 mm gap. A trim gap of 50 μm was used to ensure proper distribution of each sample between plates. Samples were conditioned to the measurement temperature for 120 s prior to the experiment beginning. The storage modulus, loss modulus, and tan(δ) were recorded over frequency sweeps repeated every 6 minutes from 0.1 rad/s to 500 rad/s at 1% strain. All measurements took place within the linear viscoelastic regime for these materials as determined using a prior stain sweep. Data are reported as the storage and loss moduli as a function of time with 1 measurement of each moduli reported from each frequency sweep at 10 rad/s and time being the total elapsed time to halfway through each frequency sweep. The gel point was calculated in two ways: 1) the time of the intersection of the storage and loss modulus and 2) the time at which tan(δ) became independent of frequency, as determined by the most constant frequency sweep which was calculated as the frequency sweep with the lowest ratio of tan(δ) standard deviation to average tan(δ).

Compression Testing.

Compression measurements were completed for DP₂₉ and DP₄₈ polymers at each crosslinking percentage (n=5) and composite formulation (n = 6) using an Instron Model 5965 material testing system (Norwood, MA) equipped with parallel metal compression plates and a 1 kN load cell. Samples were prepared by mixing polymer and crosslinker according to the specified theoretical molar crosslink percentage, which is defined as the percentage of allyl groups in the polymer that would react with thiol groups in crosslinker, in a glass vial under ambient conditions using a spatula until homogenous. The respective ceramic fillers were added to the neat polymer prior to addition of crosslinker according to the specified wt% of the polymer mass. Samples were then placed in freeform silicone molds to cure overnight under ambient conditions. Samples were removed from molds and cut into 5 mm × 5 mm × 10 mm rectangular prisms using a razor blade. Samples were left under ambient conditions to continue to cure until being tested at specified timepoints of 1, 7, 14, and 28 days. Sample dimensions were confirmed by measurement using calipers prior to compression testing. Samples were subjected to uniaxial compressive forces at a strain rate of 1.2%/min until failure. Data is reported as compressive stress-strain curves. Maximum compressive stress is reported as the maximum stress recorded in the test. Strain at break is reported as the strain at the maximum stress recorded in the test. Compressive modulus was determined as the slope between 5–7% strain on the stress-strain plots.[39,40] Shear modulus was determined as the slope of the Neo-Hookean fit (engineering stress versus λ-λ⁻², where λ is the compression ratio) of the stress-strain plots.

Degradation of PAGES Polymers and Composites.

In Vitro degradation measurements of PAGES and PAGES composites (10 mm × 5 mm × 5 mm rectangular prisms) were performed in physiological conditions (PBS, pH 7.4) or under accelerated conditions (1 M HCl). The sample blocks were cast in a silicone mold and cut to size with a razor blade 4 hours after crosslinker addition. Sample mass and compressive strength were repeatedly measured over the study. Time points for measurement of mass changes were 6 hours, 1 day, 7 days, and then weekly, while compression time points were 6 hours, 1 day, 1 week, 2 weeks, 3 weeks, 4 weeks, 6 weeks, 8 weeks, 10 weeks, 12 weeks, 14 weeks, and 18 weeks. Prior to mechanical testing, samples were measured dimensionally with calipers and those numbers were used for stress and strain calculations. Testing was performed from 5–10% strain with a strain rate of 1%/min using a Discovery DMA 850 (TA Instruments). Data was collected both when loading and unloading. Compressive modulus was determined as the slope between 5–7% strain on the stress-strain plots.[39,40]

Rat Mesenchymal Stem Cell Culture.

Primary Rat Mesenchymal Stem Cells (MSCs) were maintained in adherent culture in T75 flasks (Thermo Fisher Scientific) and fed every 2–3 days with MSC media (minimum essential media alpha (MEMα) (Cytiva) plus 10% v/v Fetal Bovine Serum (Gibco) and 1x Antibiotic/Antimycotic(Gibco)). Cells were passaged using 0.25% trypsin when they reached 70–90% confluency and reseeded at 5,000 cells/cm².

Osteoblast Differentiation.

MSCs were seeded at 5,000 cells/cm² in 48-well plates (ThermoFisher Scientific). The wells were coated with a thin layer of PAGES with crosslinker prior to gelation. Plates were sterilized under UV light for 1 hour and left in the culture hood overnight prior to seeding. Plates were washed twice with MSC media prior to cell seeding. Media was replaced every 2–3 days while the MSCs were in culture.

Cell Staining and Imaging.

After 17 days in culture, plates were washed twice with sterile PBS. Cells were fixed with 4% paraformaldehyde (Boster) shaking at room temperature for 20 minutes. Cells were then permeabilized with 1% Triton-X 100 (Promega) for 15 min on a shaker at room temperatures. Cells were incubated with 3% BSA blocking buffer (ThermoFisher Scientific) for 1 hour on the shaker at room temperature or overnight at 4°C. Cells were incubated with mouse anti-hOsteocalcin (R&D biosystems) at 5 ug/ul in 1% BSA blocker buffer overnight at 4°C. Plates were washed 3 times with 1% BSA. Cells were then incubated in donkey anti-mouse antibody conjugated to NL557 (R&D Biosystems) diluted 1:200 in 1% BSA for 1 hour at room temperature in the dark. Plates were then washed 2x with 1% BSA and then incubated for 15 minutes with DAPI (AAT Bioquest). Plates were then imaged using a Keyence BZ-X710 fluorescent microscope (Itasca, IL). Images were then analyzed using Cell Profiler to determine the percentage of nuclei that co-localized with osteocalcin staining. Three images were acquired per well and they were averaged together for a well average. Three wells were used for each condition and three independent seedings were measured.

Rabbit Vertebral Testing.

Rabbit vertebrae were acquired from animals who were used for procedures outlined in Duke University Institutional Animal Care and Use Committee protocol (A122–20-06) that had already been sacrificed. All investigators complied with the National Research Council’s Guide for the Care and Use of Laboratory Animals. Vertebra were separated from one another and the posterior elements were removed with Rongeurs leaving only the vertebra body. One end of the vertebral body was then embedded in 2–4 cm of PC-clear epoxy resin. The vertebral was then drilled with a self-tapping bone drill with a 3 mm outer diameter (Merit Medical) along its longitudinal axis. Embedded bones were then imaged with micro-computed tomography as detailed below. Following imaging, bones were soaked for at least 24 hours in PBS at 4°C prior to mechanical testing.

Vertebra Compression Studies.

Rabbit vertebra both before and after filling with polymer were crushed on an Instron Model 5965 material testing system. Wet vertebra were preloaded with 5 N and were then tested at a 1.2%/min compression rate. Stress was calculated as Force/(bone volume/height). Bones were compressed to failure. Following initial fracture, the bone fragments (if any) were placed back into anatomic alignment and the fractured bone was compressed again to determine post-fracture strength. Post-fracture testing was performed without preload and was stopped when the resin block started to impact the force curve or 30% strain was reached. Filled vertebra compression was also performed without preload and was stopped at failure. Compression rate was unchanged for all tested samples.

Micro-Computed Tomography (μCT).

μCT scans were completed using a Nikon XTH 225 ST instrument. Voxel size was 15–18 μm³. X-ray projections were taken in a 360 degree scan. The x-ray energy was 155 kilovolts (kV). Beam power was determined by the voxel size whe e beam wattage was equal to the voxel value. Scans were reconstructed in Nikon software and volumetric analysis was performed in Avizo software. Avizo was used to determine the bone volume of the vertebral body and height which was used to calculate the average bone area for the sample. This average bone area was used to calculate the stress in mechanical testing. Volumetric analysis was also used to calculate the volume of the cavity created in the bone and how much augmentation material was filled into the cavity.

PMMA Resin Preparation.

PMMA crystals were milled using a cryomill (Retsch) with 2 5-minute cycles at 30/s yielding a fine white powder. Solid components were combined using the following weight ratios: 88% PMMA, 10% BaSO₄, and 2% benzoyl peroxide. Liquid components were combined using the following volumentric ratio: 98.5% methyl methacrylate and 1.5% n,n-dimethyl-p-toludine. Solid and liquid compounds were then combined in a 1:1 w/v ratio to yield PMMA. Clinical PMMA from Merit Medical was prepared using the instructions on the package insert.

Ex Vivo Kyphoplasty: Filling Rabbit Vertebra with PAGES or PMMA Augmentation Materials.

Once the vertebral bodies were fractured and the mechanical properties had been obtained, they were filled with PAGES both with and without SrCO₃ or with PMMA. PMMA used for rabbit vertebra was from a clinical kyphoplasty kit (Merit Medical). To fill the vertebra bodies, PAGES and crosslinker were mixed in a glass vial and then transferred to a syringe 15 minutes after the addition of crosslinker. The syringe was then connected to an 18 gauge needle which was placed into the drilled cavity in the vertebra body at multiple angles to attempt to get filling into the medullary space of the bone. PAGES augmentation material was added to the bone until it was visible at the top of the drilled hole and the polymer could no longer be injected through the needle. For vertebra that had multiple fragments, the fragments were placed in anatomic position prior to polymer filling. The filled vertebra were then stored at 37°C in PBS for 7 days prior to mechanical testing. The vertebra were imaged with μCT 1–2 days after being filled with polymer.

Statistical Analysis.

Statistical analysis was performed using Prism GraphPad software and Excel. Sample averages were compared using an ordinary one-way ANOVA, two-way ANOVA (for cellular data), or repeated measures ANOVA (for degradation and rabbit vertebra data). Statistical significance was determined as a calculated p-value < 0.05. Threshold values for extent of statistical significance are as follows: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Sample averages were compared to each other using a Tukey’s multiple comparisons test or paired t-test with statistical significance being determined as a calculated p-value < 0.05. Threshold values for extent of statistical significance are as follows: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Supplementary Material

SUPPORTING INFORMATION

Supporting Information outlining the synthesis and characterization of the PAGES polymers, gel point experiments, network formation, composite experiments, and in vitro testing are available free of charge.

DATA AVAILABILITY

Data will be made available upon reasonable request to the authors.