Degradable Poly-Lactic-Co-Glycolic Acid and Non-Degradable Polymer Implants Result in Similar Fracture Healing at Early Timepoints

Abstract

Background:

Although rigid interfragmentary fixation is required for fracture repair, overly stiff implants are known to cause stress shielding which ultimately inhibits healing. While gradual dynamization of the fracture site both accelerates and improves osteogenesis, this approach requires external fixators or secondary surgeries. This study leverages biodegradable implants as mechanisms of gradual, passive dynamization during fracture healing.

Methods:

Using a rat femoral osteotomy model, additively manufactured poly-lactic-co-glycolic acid implants were compared to geometrically matched non-degradable biocompatible resin devices. Bone healing was assessed at 3 and 6 weeks via micro-computed tomography, histology, and mechanical testing. Implant degradation kinetics were assessed through testing of plates that were used in the rat model and with an unloaded in vitro degradation model.

Findings:

Quantitative bone measures made with micro-computed tomography, histology, and mechanical testing of the healing femora revealed no differences between degradable and non-degradable implants at 3 or 6 weeks. Degradable implants caused significant increases in bone volume to total volume mean density (p<0.0001) and callus to cortical volume (p<0.05) ratios between 3 and 6 weeks. Poly-lactic-co-glycolic acid devices were significantly stiffer than resin at week 0, but the two groups were equivalent by week 6 due to in vivo degradation. In vivo ambulatory loading caused significant losses of degradable implant stiffness at both 3 (p<0.05) and 6 (p<0.01) weeks, but this was not observed in the unloaded in vitro model.

Interpretation:

The results from this early timepoint study demonstrate the feasibility of passive, internal fracture dynamization driven by implant material mechanics.

1. Introduction

Bone fractures are common injuries which can result in significant physical and psychosocial disabilities, increased healthcare costs, chronic pain, and an overall lower quality of life [1]. These injuries often require surgical fixation with combinations of metallic plates and screws which are stiffer than bone and can stress-shield the fracture, potentially leading to nonunion. Each year in the U.S., over 100,000 fractures result in nonunion [2], often requiring multiple surgeries over several years to address [3], [4], [5]. It has been established that small amounts of axial interfragmentary strain at the fracture site improves healing outcomes and can prevent nonunion events [2], [3], [4], [5]. As such, there is interest in developing surgical techniques and implants that strategically allow load transfer across the healing fracture callus at appropriate times. In some applications, secondary surgeries can be performed to remove screws and purposefully elicit desirable loading across the healing callus. However, these surgeries can cause sudden or premature increases in strain and excessive shear known to impede healing [6], [7]. As such, it is desirable to achieve a mechanism of passive mechanotransduction that is not reliant on secondary surgical procedures.

Bioresorbable implants are an attractive solution to reduce the effects of stress shielding and excessive loading by providing gradual load transfer across the callus during the healing process as they lose mechanical strength [8]. Additionally, they can eliminate the need for secondary surgeries, thereby minimizing additional risks associated with infection and time under anesthesia [9], [10]. Resorbable devices have particular utility in pediatric populations which often require removal surgeries to prevent interference with bone growth [9], [11]. To date, biodegradable materials have been used to create a variety of orthopaedic devices including screws, plates, and pins [12], [13], [14], [15]. Prior work primarily focuses on biodegradable metals such as magnesium and zinc [16], [17]. However, pure metallic implants pose challenges including rapid corrosion and hydrogen gas buildup due to impurities [18], so there is a need for alternative resorbable implant materials which mitigate these drawbacks.

Poly-lactic-co-glycolic acid (PLGA) is an attractive biomaterial for use in trauma implant design. PLGA is a synthetic co-polymer of lactic and glycolic acid with highly tunable degradation kinetics based on the ratio of lactic to glycolic acid. 85:15 PLGA has been found to have a slower degradation timeline compared to 75:25 and 50:50 PLGA due to its lower hydrophilicity. [19], [20], [21]. This property makes 85:15 PLGA the most suitable for fracture repair, as it is more likely to retain its mechanical strength in earlier phases of healing when the callus cannot sustain high loads. Several studies have found PLGA to exhibit similar mechanical properties to human trabecular bone [22], [23]. PLGA is also well known for its biocompatibility, as its degradation products (lactic and glycolic acid) are highly bioavailable compounds. Additionally, lactic acid has been shown to promote angiogenesis and osteogenesis, making it suitable for bone healing applications [24].

PLGA has been used in a variety of forms in orthopaedics, and some recent efforts have focused on creating additively manufactured (AM) constructs [25], [26], [27]. Additive manufacturing allows for adaptations to infill architecture and density which can be leveraged to alter the overall mechanical properties of a device [28]. This ability to leverage mechanical properties is attractive in the design of compliant fracture fixation constructs, providing opportunities to maximize early mechanical strength followed by a controlled decrease in rigidity as the bone’s load-bearing capacity increases throughout healing. A previous study used in vitro and in vivo experiments to characterize degradation kinetics of AM PLGA test coupons, and found that mechanical properties decrease between 8–16 weeks after implantation [19]. However, the design and development of bioresorbable trauma implants is still in its nascency, and there is still much to be learned about the utility and efficacy of various materials, designs, and applications.

The objective of this study was to investigate the utility of AM PLGA implants in a pre-clinical fracture model of controlled re-loading at early healing timepoints. A rat femoral osteotomy model was used to compare the healing effects and degradation kinetics of biodegradable and non-degradable implants. We hypothesized that AM PLGA implants would have decreased mechanical strength in comparison to non-degradable control implants over time, and that the bone healing response between groups would be similar at early timepoints.

2. Methods

2.1. Implant fabrication

PLGA implants (19×5×5 mm) were fabricated on a fused deposition 3-D printer (i3 MK3, Prusa Research a.s., Prague, Czech Republic) with 85:15 PLGA filament (Lattice Services, Loos, France) (Fig. 1) (220,000–260,000 g/mol, degradation temperature >250°C). The following print settings were used: layer height of 0.2 mm, print speed of 10 mm/sec, 100% infill density with a rectilinear pattern, 25% infill/perimeter overlap, and nozzle temperature of 215°C. Resin implants of the same geometry were synthesized via photocuring (BioMed Clear Resin, Formlabs Form 3, Somerville, MA, USA) according to manufacturer recommendations. Because PLGA could not be autoclaved due to its melting temperature (140–160 °C), PLGA implants were soaked in 70% ethanol for 30 minutes and washed with sterile 1x phosphate-buffered saline (PBS) prior to testing and implantation.

Figure 1.

A) Computer-rendering of implant model. B) Manufactured resin (top) and PLGA (bottom) femoral implants.

2.2. In vitro Implant Degradation

To assess differences between in vitro and in vivo degradation of PLGA implants, additional PLGA and resin implants (n=10) were incubated at 37°C and 5% CO₂ on a rocker in a solution of 30% fetal bovine serum, 69% 1x phosphate-buffered saline, and 1% v/v Penicillin-Streptomycin-Fungizone. Serum changes were completed every 3–4 days [19]. Implants were collected for mechanical testing at 3 and 6 weeks.

2.3. Surgical Protocol

Nineteen 12–18-week-old male SASCO Sprague Dawley rats were used in this study. All animal work was approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Pennsylvania (IACIC #807123). [5], [29], [30]. A bilateral fracture model used in prior studies [5], [29], [30] was chosen to prevent preferential loading in the intact limb, ensuring that both implant groups were being loaded equally post-surgery. Briefly, animals were anesthetized with isoflurane and received a subcutaneous injection of buprenorphine XR (1 mg/kg) for slow-release analgesia and bupivacaine (2 mg/kg) for local analgesia at both incision sites. A 3 cm curvilinear skin incision was made with a #10 scalpel blade the scalpel from the hip to the stifle, exposing the fascia latae which was then dissected. The biceps femoris was separated from the vastus lateralis, exposing the lateral aspect of the femur. A PLGA bone plate was secured to the lateral femur using 4 stainless steel screws (0–42 × 3/8”) (Fig. 2). A mid-diaphyseal transverse osteotomy was then created using a micro-oscillating saw (Fig. 2A). The muscles were closed over the internal fixation plate with 4–0 Vicryl suture in a simple continuous pattern, followed by subcuticular simple continuous closure using the same suture material. The skin was closed with wound clips and liquid skin bandage, and metronidazole powder was applied over the incision for infective prophylaxis and to prevent self-mutilation. The rat was then turned to the other side, and the procedure was conducted on the contralateral hindlimb using a bone plate made from resin of the same dimensions.

Figure 2.

A) Implant fixation surgical procedure. B) PLGA implant fixed to the femur post-mortem. C) Resin implant fixed to the femur post-mortem.

Post-operatively, animals were allowed to ambulate freely and were provided with hydrogel and food on the cage floor for the first week to ensure that they received appropriate nutrition if unable to stand on their hindlimbs. The animals were kept in single housing conditions, which included a 12:12 hour light cycle and a constant temperature of approximately 25 deg C at 40–50% relative humidity. Standard chow and water were provided ad libitum. 10-days post-operatively, rats were moved to double housing and transparent red tubes were provided to improve enrichment and comfort. Animals were monitored at least twice daily for the first 3 days post-operatively, daily until 1-week post-operatively, and at least 3 times weekly until the endpoint. Fluoroscopy was performed weekly to check for implant failure. Rotational deformities of the femora were observed in three rats less than 1-week post-surgery. These rats were sacrificed and excluded from the remainder of the study.

2.4. Micro-Computed Tomography

In vivo micro-computed tomography (μCT) was conducted at weeks 3 and 6 to monitor bone healing (VivaCT 80, SCANCO Medical, Wangen-Brüttisellen, Switzerland). Nine rats were scanned at week 3, and seven rats were scanned at week 6. Rats were anesthetized with isoflurane during the scanning process. Scanning parameters were as follows: 55 kVp, 60 μm slice thickness, 300 ms duration. Regions of interest (RoI) were selected in the areas between screw holes on the fixation plates (Fig. 3G). For each sample, six 1 mm diameter circular contours were drawn on uniform sections of the axial slices to create cylindrical cores of 10–20 slices thick through the RoI. Using custom software, metrics of bone volume (BV) mean density, total volume (TV) mean density, and the ratio of bone volume to total volume (BV/TV) were calculated. Relative amounts of callus and cortical bone from each specimen were analyzed in 3D Slicer (slicer.org). Manual segmentation of callus and cortical bone was performed to quantify the callus and cortical bone volumes, respectively, and the ratio of callus to cortical volume within each RoI.

Figure 3.

Quantitative assessment of bone volume mean density (A), total volume mean density (B), bone volume/total volume fraction (C), fracture callus volume (D), cortical bone volume (E), and the ratio of callus/cortical volume (F) of PLGA and resin groups at weeks 3 and 6. *p<0.05, **p<0.01, ****p<0.0001. Volumetric rendering represents the RoI utilized for analysis (G). White arrows indicate regions identified as fracture callus in the axial (H), coronal (I), and sagittal (J) planes.

2.5. Histology

Rat femora were harvested for histological analysis at 3 and 6 weeks (n=6). Upon sacrifice, the femora were dissected, and the implants removed. Samples were fixed in formalin for 14 days, then decalcified in formic acid for 3 weeks with solution changes every 7 days. The callus region of each sample was isolated and embedded in paraffin. Following embedding, samples were cut into 7 μm-thick sections in the mid-sagittal plane and stained with Safranin-O/Fast Green, Hematoxylin and Eosin (H&E), and Picrosirius Red. Brightfield images were taken at 10x magnification (Eclipse 90i, Nikon Instruments Inc., Tokyo, Japan). Picrosirius Red-stained samples were also imaged under polarized light to distinguish organized lamellar and woven bone based on the birefringence of the collagen fibers [31], [32], [33]. Polarized light images (DMLP polarized light microscope, Leica Camera AG, Wetzler, Germany) were taken with the polarizer set to 270° and analyzer set to 180°. Samples were imaged with collagen fibers oriented at 45° such that mature lamellar bone appeared yellow-green and immature woven bone appeared red-orange [32], [33], [34]. In each image, the RoI containing the fracture region was identified as the area between the two inner screw holes. Quantitative analysis of tissue areas of interest was completed using the color thresholding tool in ImageJ (NIH, Bethesda, MD, USA) to obtain the relative percentage of cartilage area (Safranin-O/Fast Green slides) and the ratio of immature:mature bone (Picrosirius Red).

2.6. Mechanical Testing

Implant and bone mechanics were measured with torsional test methods for several reasons. First, torsional loading is the most physiologically-relevant type of loading for long bones [35]. Second, torsional testing is not affected by the orientation of asymmetric fracture calluses, unlike bending tests [36]. The torsional stiffness measured is also independent of the testing direction, making torsional testing more reliable and reproducible [37]. Finally, torsional testing allows for the evaluation of the whole bone specimen, rather than a localized region like in three-point bending [36]. This approach provides a comprehensive biomechanical characterization of the specimen.

PLGA and resin implants from in vitro and in vivo experiments underwent torsional testing (90° rotation at 1°/sec under displacement control) at weeks 0, 3, and 6 on a universal test frame (Electroforce 3550; TA Instruments, Eden Prairie, MN) equipped with a 15kN/140 Nm load/torque cell recording at 100 Hz. Implants from the in vivo condition were removed from the bone prior to testing. Polysulfone implants utilized in previous studies [5], [29], [30], were also tested at week 0 to provide additional context to the findings of the current study.

Rat femora were harvested for mechanical testing at 6 weeks (n=7). After the sacrifice of a rat, femora were gross dissected from the animal, and implants were removed, ensuring that screw removal did not damage or alter the state of the bone. Bone samples were frozen in the muscle belly at −20°C, and implants were stored at 4°C with silica gel desiccant packs until the time of testing. Prior to testing, femora were thawed to room temperature and dissected from the muscle belly. The proximal and distal ends of the femora were potted in fields metal and mounted to a universal test frame (TA Instruments, Electroforce 3550, Eden Prairie, MN). Tests consisted of cyclic torsion to ±2° from neutral at 0.75 Hz for 3 cycles, followed by a 10 second hold and a final 90° internal rotation at 3°/sec to failure. Additional intact control femora (n=6) were also tested under the same protocol. To account for geometric differences between implants and bones, the torsional rigidity (Nm²/°) was calculated as the torsional stiffness (Nm/°) multiplied by the free length of the sample [38]. Healing femora were not tested at 3 weeks, as there was not enough callus tissue present at that time point to warrant meaningful mechanical comparison. Data processing was completed in MATLAB (R2023b; Mathworks, Natick, MA, USA) using custom scripts.

2.7. Statistical Methods

Implant mechanical properties across weeks 0, 3, and 6 were compared using a one-way ANOVA with post-hoc tests using a Bonferroni correction. Outcome measures for both implant groups from in vivo mechanical testing, μCT, and histology were compared at each timepoint using paired t-tests to control for inter-specimen variability. When tests for normality and equal variances failed, Kruskal-Wallis tests with Dunn’s post hoc and Mann-Whitney Rank Sum tests were performed. All significance levels were set to p<0.05. Reported confidence intervals represent the differences between groups. Statistical analysis was conducted using GraphPad Prism (v.10, GraphPad Software, LLC, Boston, MA, USA).

3. Results

3.1. μCT

Analysis of μCT scans revealed that PLGA implants significantly increased BV density by 20.5% and TV density by 63.7% from 3 to 6 weeks. Resin implants increased TV density by 39.4% but did not have a significant effect on BV density over time (Fig. 3A & 3B). The BV density for both groups was significantly lower than healthy bone (p<0.0001) at both 3 and 6 weeks. The PLGA group had a 65.1% increase in BV/TV from 3 to 6 weeks (p<0.0001) while the resin group demonstrated a 33.6% increase (p=0.0311) (Fig. 3C). There were no changes in callus or cortical volumes for either implant group between timepoints (Fig. 3D & 3E). However, the PLGA group demonstrated a significant 179.4% increase in callus/cortical volume fraction from 3 to 6 weeks (Fig. 3F). There were no significant differences between the resin and the PLGA groups at 3 or 6 weeks for any μCT outcome measure. The significant p-values and confidence intervals for differences in μct outcomes are summarized in Table 1.

Table 1.

Significant pairwise comparisons (p-values and confidence intervals) for differences in μct outcome measures between healthy, PLGA-fixed, and resin-fixed femora.

	Resin-fixed	PLGA-fixed
	Bone Volume Mean Density (mg HA/ccm)
Week 3 vs week 6	n.s.	p=0.0035 (39.84,165.9)
Heathy vs Week 3	p<0.0001 (374.8,628.4)	p<0.0001 (411.2,664.8)
Healthy vs week 6	p<0.0001(263.9,531.6)	p<0.0001 (301.3,569.0)
	Total Volume Mean Density (mg HA/ccm)
Week 3 vs week 6	p=0.0331 (14.95,308.9)	p<0.0001 (143.7,266.9)
	Bone Volume/Total Volume
Week 3 vs week 6	p=0.0311 (0.02592,0.5147)	p<0.0001 (0.2443,0.4292)
	Callus Volume (mm3)
Week 3 vs week 6	n.s.	n.s.
	Cortical Volume (mm3)
Week 3 vs week 6	n.s.	n.s.
	Callus Volume/Cortical Volume
Week 3 vs week 6	n.s.	p=0.0158 (0.5071,4.124)

3.2. Histology

Quantitative histological analysis revealed no significant differences between groups. Sections stained with Safranin-O/Fast Green revealed that the percentage of cartilage area in femora fixed with PLGA and resin implants decreased from week 3 (PLGA: 3.97±6.14%, Resin: 3.85±4.96%) to week 6 (PLGA: 0.542±0.471%, Resin: 1.16±1.17%), but this difference was not statistically significant (Fig. 4). Analysis of collagen fiber maturity was obtained from sections stained with Picrosirius Red and analyzed with polarized light microscopy (Fig. 5). Again, no differences were found in the ratio of immature:mature collagen between PLGA and resin from week 3 (PLGA: 0.3289±0.241, Resin: 0.324±0.290) to week 6 (PLGA: 0.243±0.268, Resin: 0.222±0.0778) (Fig. 5). Finally, H&E staining revealed no differences between groups, but demonstrated increased osteocyte-populated woven bone in the fracture region (white arrows, Fig. 6) from week 3 to 6. Woven bone was identified based on the random orientation of collagen fibers compared to the aligned fibers within the mature lamellar bone, visible in the non-fractured regions of cortical bone near the osteotomy site.

Figure 4.

Representative images of Safranin-O/Fast Green stains of healing femora treated with resin and PLGA implants at 3 and 6 weeks used to quantify cartilage area. Red stains represent cartilage in the bone callus. Scale bars=2 mm. Although there were visible differences from week 3 to 6, the quantified changes were not statistically significant.

Figure 5.

Representative images of healing femora sections stained with Picrosirius Red and imaged with polarized light fixed with resin and PLGA implants at 3 and 6 weeks. Immature collagen is bright red/orange, while mature collagen is yellow/green. Scale bars=1 mm. Quantification of the ratio of immature:mature collagen showed no significant changes.

Figure 6.

H&E stains of representative callus sections of PLGA and resin samples at weeks 3 and 6. Scale bars=500 μm. White arrows indicate newly formed woven bone within the fracture. gap.

3.3. Mechanical Testing

Mechanical testing of femora at week 6 revealed no differences in torsional rigidity between those fixed with PLGA and resin implants (Fig. 7). Healing femora had 71.3% lower torsional rigidity than healthy bones (p<0.05). Exact p-values and confidence intervals for significant differences from bone mechanical testing are shown in Table 2. All femora experienced simple fractures during testing within the callus region during torsional testing, and two femora experienced spiral fractures.

Figure 7.

Torsional rigidity of healing femora at 6 weeks in PLGA and resin groups compared to healthy samples. *p<0.05.

Table 2.

Significant pairwise comparisons (p-values and confidence intervals) for differences in torsional rigidity between healthy, PLGA-fixed, and resin-fixed femora.

	Resin-fixed Torsional Rigidity (Nm2/deg)	PLGA-fixed Torsional Rigidity (Nm2/deg)
Healthy Torsional Rigidity (Nm2/deg)	p=0.0161 (7.462×10−5, 8.338×10−4)	p=0.0115 (8.862×10−5, 8.560×10−4)

Degradation of PLGA implant mechanical properties was measured for both in vivo and in vitro experiments. Compared to week 0, in vivo PLGA implants lost 30.2% torsional rigidity at week 3 and 43.9% at week 6 (Fig. 8). Additionally, PLGA implants degrading in vivo were 28.8% less rigid than those in the in vitro condition at week 3 and 41.1% less rigid than in vitro samples from week 6. We also compared mechanical properties of non-degradable implants to PLGA implants. At week 0, both PLGA and resin implants were significantly more rigid than the polysulfone (p<0.0001) (Fig. 8). Resin implants from the in vivo study were 14.9% less rigid than the implants from the in vitro study only at week 3. In vitro degradation resulted in significantly more rigid PLGA implants than resin at 0, 3, and 6 weeks (p<0.0001) (Fig. 8). There were no differences between the PLGA and resin implants at 3 and 6 weeks within the in vivo experiment. The pairwise comparisons between PLGA and resin implants from mechanical testing are also summarized in Table 3.

Figure 8.

Torsional rigidity of PLGA, resin, and polysulfone implants at week 0, and PLGA and resin implants at weeks 3 and 6 degrading in vivo and in vitro. *p<0.05, **p<0.01, ***p<0.001, ***p<0.0001.

Table 3.

Significant differences (p-values and confidence intervals) for implant torsional rigidity from in vivo and in vitro testing conditions. Results reflect ANOVA post-hoc pairwise comparisons between weeks 0, 3, and 6, as well as t-tests between conditions at each time point.

			in vitro Torsional Rigidity (Nm2/deg)				in vivo Torsional Rigidity (Nm2/deg)
		PLGA Week 0	PLGA Week 3	PLGA Week 6	Resin Week 3	Resin Week 6		PLGA Week 3	PLGA Week 6	Resin Week 3	Resin Week 6
	Resin Week	p<0.0001 (2.544×10−4, 3.834×10−4)	-	-	n.s.	n.s.		-	-	n.s.	n.s.
in vitro Torsional Rigidity (Nm 2 /deg)	PLGA week 3	n.s.
	PLGA Week 6	n.s.	n.s.
	Resin Week 3	-	p<0.0001 (2.052×10−4, 3.624×10−4)	-
	Resin Week 6	-	-	p<0.0001 (1.695×10−4, 2.955×10−4)	n.s.
in vivo- Torsional Rigidity (Nm 2 /deg)	PLGA Week 3	p=0.0310 (1.080×10−5, 4.551×10−4)	p=0.0036 (8.521×10−5, 3.556×10−4)	-	-	-
	PLGA Week 6	p=0.0007 (1.429×10−4, 5.443×10−4)	-	p=0.0015 (1.358×10−4, 4.756×10−4)	-	-		n.s.
	Resin Week 3	-	-	-	p=0.042 (4.000×10−6, 1.590×10−4)	-		n.s.	-
	Resin Week 6	-	-	-	-	-		-	n.s.	n.s.

3.4. Implant Failure Mechanisms

Torsional testing of the implants revealed differences in failure mechanisms between implant types. For both in vivo and in vitro degraded specimens, the resin implants were more brittle than the PLGA, breaking cleanly in the region bridging the fracture callus. There was no visible degradation or large defects in either implant group upon extraction from the in vitro condition at either timepoint. At 6 weeks, the resin implants shattered rather than breaking cleanly. In contrast, the AM PLGA implants primarily failed due to delamination between print layers (Fig. S1). These failure mechanisms were similar in both the in vivo and in vitro conditions. No delamination was observed in the implants upon extraction from the rats; instead, these failures only occurred under high torsional loads during post-mortem testing.

4. Discussion

To our knowledge, this is the first study to investigate the effects of AM PLGA implants for weight-bearing fracture repair. This experiment demonstrated that biodegradable PLGA implants promote bone healing at early timepoints. At both 3 and 6 weeks, there was no significant difference in healing outcomes when comparing PLGA and non-degradable resin implants. These findings support our initial hypothesis that we would observe similar healing outcomes between the two groups at early timepoints.

There was clear quantitative and qualitative evidence of the early stages of secondary bone healing in both the PLGA and resin implant groups. μCT revealed an increase in BV/TV ratio for both groups, and it was observed that only the PLGA group experienced increases in BV mean density from 3 to 6 weeks. Evidence of secondary fracture healing progression was further supported by the overall decrease in cartilage area (Fig. 4) and the increase in woven bone (Fig. 6) for both groups. These histological findings are also consistent with the decrease in the ratio of immature to mature collagen from weeks 3 to 6 (Fig. 5), indicating lamellar bone formation and lower amounts of type I cartilage within the callus region [33], [39], [40]. Although not statistically significant, these quantitative changes signify progressive fracture healing.

As this experiment focused on early timepoints, the incomplete healing process and at 6 weeks was to be expected and is consistent with previous findings. A previous study investigating the effects of mechanical loading on fracture healing did not begin to see differences in bone mineralization until weeks 8–12 of healing [5]. Another study investigating the effects of age and mechanical stimulation on sub-critical rat femoral defects found that bones fixed with flexible fixation devices had torsional failure moments of only 60% of healthy controls at 6 weeks [41]. Similarly, another study of rat tibial fractures found significantly lower torsional stiffness at 6 weeks compared the healthy contralateral bone [42]. While preclinical studies of biodegradable polymer implants are limited, our results also show similar fracture healing progression compared to prior research around biodegradable metallic implants. For example, a study of Mg/ZA/CaP intramedullary pins in a rat osteoporosis model found BV/TV fraction to increase by approximately 66% [43]; similarly, our study found that femora fixed with PLGA implants experienced a 65.1% increase in BV/TV from 3 to 6 weeks [45]. Overall, the healing progression evident at these early timepoints provides promising evidence that PLGA implants have utility in fracture healing models beyond 6 weeks. To better understand the long-term performance of PLGA implants on fracture healing, future work should extend to 10–12 weeks post-fracture to coincide with the expected healing time in rat models [44], [45].

There were differences in the baseline mechanical properties of implants within this study. Specifically, PLGA implants were significantly more rigid than the resin implants at week 0. We made the decision to use the BioMed clear resin rather than polysulfone because it was autoclavable, biocompatible, and we could easily fabricate implants with the exact same geometry using AM techniques. It should be noted that the modulus of rigidity of the PLGA and resin implants were within the same order of magnitude from each other and were also similar to polysulfone implants utilized in previous studies (approximately 2–12 MPa) [5], [29], [30]. The modulus of these materials is around 3 orders of magnitude lower than other common implant materials such as stainless steel and titanium [46], so the difference in mechanical properties between the PLGA and resin used in this study was comparatively small. Additionally, the similar healing outcomes in femora fixed with PLGA and resin implants suggest that the differences in baseline mechanical properties did not produce different stress-shielding effects at these early timepoints. Moreover, by week 3, there were no differences in torsional rigidity between PLGA and resin implants, so for the majority of the in vivo healing period, PLGA and resin had comparable mechanical properties, likely contributing to similar healing outcomes. However, future work should investigate the effects of such differences in implant material properties on fracture healing, particularly in the early stages of callus formation.

Mechanical loading of PLGA implants accelerated degradation of the devices. This conclusion is evidenced by the significant decreases in PLGA torsional rigidity in the in vivo condition at weeks 3 and 6 compared to week 0 while PLGA in the in vitro condition did not experience such changes. (Fig. 8). There were differences in microenvironment chemistry between in vivo and in vitro experiments, but we do not believe this was the primary driver of implant degradation. A recent study investigating in vitro PLGA degradation did not find significant differences in degradation kinetics when comparing in vitro test coupons to those implanted subcutaneously [19]. Specifically, this study did not observe consistent loss of mechanical properties until after 8 weeks, a finding which is consistent with our observations finding no significant changes in PLGA mechanical properties after 6 weeks degrading unloaded in vitro. Based on the results from our study, mechanical stimuli caused by ambulatory loading by the animals in the in vivo condition Instead, mechanical loading is likely the cause of accelerated degradation. This conclusion is supported by prior work investigating PLGA scaffolds for tibial bone defect repair, where degradation of devices was observed at 4 weeks [47]. Similarly, another study found that PLGA membranes degrading in vitro under constant tensile stresses had consistent decreases in elastic modulus over 8 weeks [21]. Clinically, this is an important finding, as patient non-compliance with weight-bearing restrictions after lower-limb fracture can be as high as 72.5% [48], and patients are also shown to exert 35.3% more body weight than prescribed during fracture healing [49]. As such, future work investigating PLGA for use in orthopaedic implants must do so under mechanically loaded conditions to mimic behavior in load-bearing fracture healing conditions.

Implants fabricated entirely from PLGA have limitations which must be managed carefully. It is known that PLGA lacks the mechanical strength for use in high load-bearing applications [8], [20], [25]. Based on the implant failures observed in this small animal model, it is unlikely that PLGA fracture plates can be scaled up for clinical use in high load-bearing applications. Despite this limitation, we remain confident in the utility of AM PLGA as a model for controlled loading of a healing callus for several reasons. First, no delamination was observed between the print layers in vivo, demonstrating the feasibility of AM PLGA in lower load bearing settings. Additionally, results showed clear evidence of fracture healing which was similar to prior studies using PLGA devices [50], [51]. As such, PLGA still poses utility in elucidating optimal AM device designs to induce effective, controlled re-loading in preclinical fracture models. However, future work is still needed investigating these implants at longer timepoints and in large animal models to understand the behavior of PLGA under higher loads for extended periods of time. Such work would further elucidate the clinical utility of PLGA implants.

This study has several limitations. First, our analysis was conducted at early fracture healing timepoints. Future work is needed to understand in vivo PLGA degradation kinetics over longer timescales. Additionally, although the PLGA implants were biodegradable, the screws were not, so this model does not entirely bypass the need for secondary surgeries to remove non-degradable devices. We also note that the geometry of the implants in the present study do not mimic that of clinical trauma implants; however, such design choices were necessary to accommodate the small size of the model and screw dimensions. Future work may investigate the effect of both geometry and material on fracture healing to inform future pre-clinical implant models and increase their clinical relevance. Preclinical small animal models are also limited in translation to clinical applications compared to large animal models [52]. The bilateral fracture model was chosen to effectively compare the degradable and non-degradable implants, but a unilateral model may be necessary in the future to evaluate the behavior of PLGA implants under more realistic asymmetrical loading during healing.

5. Conclusions

To improve clinical outcomes, the next generation of fracture implants should be designed to harness the well-established principles of mechanotransduction [3], [4]. This study demonstrated the feasibility of biodegradable PLGA trauma implants as models of controlled re-loading during fracture repair. Compared to non-degradable controls, PLGA implants elicited similar healing responses at early timepoints and showed faster degradation rates under in vivo mechanical loading. Future experiments should seek to strategically incorporate biodegradable materials within more traditional metallic plates to introduce gradual increases of loads across the callus while preventing structural failures of the device to enhance the clinical utility of PLGA.

Supplementary Material

1

Figure S1. PLGA (top) and resin (bottom) implants extracted from the in vivo condition pre-and post-torsional testing. PLGA implants commonly failed due to delamination, whereas resin implants commonly failed by shattering.

Highlights.

• Bioresorbable implants successfully promoted fracture healing at early time points

• Degradable and nondegradable implants led to similar healing responses

• Mechanical loading accelerated implant degradation after just 3 weeks