Characterization of Degradation Kinetics of Additively Manufactured PLGA Under Variable Mechanical Loading Paradigms

Abstract

Controlled degradation of biodegradable poly-lactic-co-glycolic acid (PLGA) trauma implants may increase interfragmentary loading which is known to accelerate fracture healing. Additive manufacturing allows us to tune the mechanical properties of PLGA scaffolds; however, little is known about this novel approach. The purpose of this study was to use in vitro and in vivo models to determine the degradative kinetics of additively manufactured test coupons fabricated with PLGA. We hypothesized that 1) increases in infill density would lead to improved initial mechanical properties, and 2) loss of mechanical properties would be constant as a function of time, regardless of implant design. Porous and solid test coupons were fabricated using 85:15 PLGA filament. Coupons were either incubated in serum or implanted subcutaneously in rats for up to 16 weeks. Samples were tested in tension, compression, torsion, and bending on a universal test frame. Variables of interest included, but were not limited to: stiffness, and ultimate force for each unique test. Infill density was the driving factor in test coupon mechanical properties, whereas differences in lattice architecture led to minimal changes. We observed moderate levels of degradation after 8 weeks, and significant decreases for all specimens after 16 weeks. Results from this study suggest substantial degradation of 3-D printed PLGA implants occurs during the 8- to 16-week window, which may be desirable for bone fracture repair applications. This study represents initial findings that will help us better understand the complicated interactions between overall implant design, porosity, and implant biodegradation.

1. Introduction

Biodegradable materials for tissue engineering scaffolds and trauma implants continue to evolve in terms of form and function. Poly-lactic acid (PLA), Poly-glycolic acid (PGA), and the copolymer Poly(lactic-co-glycolic acid) (PLGA) are biocompatible and biodegradable materials that have been used to create a wide array of devices and implants. PLGA has effectively been utilized in bone healing applications, ranging from artificial bone substitutes (both with and without bioactive additives), bioresorbable pin fixation, and internal fixation of craniofacial fractures (Zhao et al., 2020). These devices are designed to provide initial mechanical stability during the healing process and then gradually degrade as the bone heals. As the degradation of a PLGA implant occurs, it loses mechanical strength and transfers an increasing mechanical load to the forming bone callus, which improves and accelerates bone healing (Boerckel et al., 2012; Goodship et al., 1998; Heye et al., 2022; Kenwright et al., 1991; Kenwright and Goodship, 1989; McDermott et al., 2016). Unlike traditional implants, PLGA fixation devices are biodegradable and do not require a secondary surgery to be removed (Heye et al., 2022).

The mechanical properties of PLGA are tunable and can be affected by manufacturing processes and polymer chemistry. In many orthopaedic applications, common design goals for PLGA scaffolds and trauma implants include high modulus and ultimate tensile strength, the ability to absorb energy without deforming (yield point) or fracturing (toughness), and resistance to fatigue due to cyclic loading. Manufacturing techniques for PLGA include injection molding, compression molding, casting, extrusion, and fused deposition additive manufacturing (de Melo et al., 2017; Guo et al., 2017; “Orthopaedic applications for PLA-PGA biodegradable polymers,” 1998). In one study, injection molded PLGA implants exhibited elastic moduli ranging between 1.9 – 2.0 GPa (de Melo et al., 2017), which is similar to trabecular bone (Ciarelli et al., 1991; Morgan and Keaveny, 2001). PLGA mechanical properties can also be altered by changing the molecular weight and the lactic acid:glycolic acid ratio of the copolymer (Danhier et al., 2012; Park et al., 2013). PLGA polymers with higher molecular weight and lactic acid to glycolic acid ratio (e.g. 85:15) have higher tensile strength, whereas lower molecular weights and relative amounts of glycolic acid (e.g. 50:50) result in decreased strength (Hasirci et al., 2000).

Many factors affect the degradation of PLGA, including mechanical loading, molecular weight, co-polymer composition, end-group functionalization, geometry of the material, and chemistry of the surrounding medium (Guo et al., 2016; Li et al., 2010; Yoshioka et al., 2008). Previous studies have demonstrated that degradation can take anywhere from 7 days to 8 weeks to occur (Guo et al., 2017; Silva et al., 2015; You et al., 2005). As the copolymer hydrolyzes, it becomes glycolic acid and lactic acid, both of which are readily metabolized within the body (Su et al., n.d.). Higher amounts of lactic acid:glycolic acid (85:15) lead to significantly longer degradation profiles compared to lower ratios (65:35, 50:50) (Guo et al., 2017). Low pH causes degradation to occur faster relative to neutral pH, and a higher level of relative humidity will also hasten degradation (Zolnik and Burgess, 2007). With this in mind, it is not surprising that separate in vivo and in vitro studies result in different degradation profiles (Guo et al., 2016, 2017; Ma et al., 2011; Silva et al., 2015).

Additive manufacturing (AM) provides new and unique opportunities to revolutionize PLGA-based orthopaedic implant designs. Layer-by-layer deposition of material allows for production of complex geometries on a patient-specific basis. A distinct advantage of AM is the ability to modulate the infill architecture and density of those internal structures without changing its overall shape (Sudarmadji et al., 2011). A previous study has shown that alterations to chemical composition and acid end caps changes the ability to accurately print parts. Although this experiment determined changes in degradation kinetics due to alterations in chemistry, it did not explore the influence of changes to infill or lattice architecture (Guo et al., 2017). Finite element analyses can provide some insight into initial mechanical properties of an implant design, but it is currently unknown how lattice architecture affects longitudinal degradation profiles and how that relates to mechanical properties (Alemayehu and Jeng, 2021). Before PLGA-based AM implants can be used for load-bearing bone healing applications, rigorous experiments must be performed to determine the relationships between print parameters, mechanical properties, and degradation kinetics.

The aim of this study was to quantify the mechanical properties of test coupons printed with different infill patterns and porosities at clinically relevant time points (between 0 – 16 weeks). We sought to perform this experiment with both in vitro and in vivo degradation models. It was hypothesized that: 1) increases in infill density would lead to higher initial mechanical properties, and 2) loss of mechanical properties would be constant as a function of time, regardless of implant design.

2. Methods

2.1. Overall study design

This study consisted of in vitro and in vivo experiments to model implant degradation. Pictorial overviews of the experiments are shown in Fig. 1. More details regarding the methods are found in the sections below.

Figure 1:

In vitro experiment: Degradation was modeled in an incubator with fetal bovine serum (FBS) solution and timepoints at 0, 8, and 16 weeks. Test coupons had 3 unique geometries: dog bones, cubes, and rectangular prisms. Print parameters for the parts were adjusted to create implants with different infill geometries (rectilinear, triangular, and gyroidal) at various infill densities (40%, 80%, and 100%). After incubation, tensile, compression, torsion, and bending tests were performed to quantify mechanical properties of the specimens.

In vivo experiment: Degradation was modeled with a subcutaneous rat model with time points at 4, 8, and 16 weeks. Dog bone samples with rectilinear infill patterns were used in the experiment. Test coupon infill densities were varied between 40%, 80%, and 100% and all samples were tested in tension.

2.1.1. Test coupon fabrication and design

Test coupons (n=558) were fabricated using a fused deposition 3-D printer (i3 MK3, Prusa Research a.s., Prague, Czech Republic) using 1.75 mm diameter 85:15 PLGA filament (Lattice Services, Loos, France). The filament was extruded at 210°C, with a layer height of 0.2 mm, an extrusion width of 0.45 mm, and speeds of 25 mm/sec for the perimeter, 40 mm/sec for the infill, and 15 mm/sec for the first layer. Because bone healing implants require smooth faces and edges and must resist loading, we included a 3-layer perimeter on each test coupon, with each layer measuring 0.33 mm in width (Fig. 2). 1000 grit sandpaper was attached to the build plate surface to improve first layer adhesion. After fabrication, the control group implants were photographed under a stereomicroscope. Implants were stored in vacuum sealed bags with silica gel desiccant packs and frozen at −20°C until the day of testing. All implants were soaked in 70% ethanol for 30 minutes prior to incubation, implantation, or mechanical testing (week 0).

Fig. 2:

Photographs captured with a stereomicroscope showing the (A) rectilinear, (B) triangular, and (C) gyroidal print patterns for 40% infills. Computer renderings and photographs show (D) tensile and torsional dog bones, (E) bending bars, and (F) compression cubes at various infill densities.

Test coupons were fabricated in 3 different shapes: dog bones (30 × 10 × 2.5 mm), cubes (10 × 10 × 10 mm), and rectangular prisms (30 × 3 × 3 mm, Figs. 1 and 2). Dog bones were used for tensile and torsional testing, cubes were used for compression, and prisms were used for 4-point bed tests. Seven unique lattice architectures were created by varying infill pattern (rectilinear, gyroid, triangle), and lattice density (40%, 80%, 100%).

In the in vitro experiment, samples (n=168) were created for each time point (0, 8, and 16 weeks). At each time point 84 tensile/torsional dog bones, 42 compression cubes, and 42 bending prisms were used. In the in vivo tests, 54 tensile dog bones with rectilinear infills were used. 18 samples were tested at the 4-week, 8-week, and 16-week timepoints. Several relevant ASTM standards indicate a sample size of 5 is appropriate for mechanical tests performed in this study (“ASTM Standard D695 – 02a Standard Test Method for Compressive Properties of Rigid Plastics,” 2002; “ASTM Standard D638–14 Standard Test Method for Tensile Properties of Plastics,” 2014). We used a sample size of 6 per lattice density for each infill pattern, test geometry and timepoint to improve statistical power of the experiments.

2.1.1. Degradation of implants

The in vitro test coupons were incubated at 37°C on a rocker in a serum solution composed of 30% fetal bovine serum, 69% PBS, and 1% v/v Penicillin-Streptomycin-Fungizone. Tensile bars were incubated in 6-well plates with n=2 samples per well in 5 mL of the serum solution. Compression cubes and 4-point-bend bars were incubated in 50 mL cups, with 6 cubes and 6 bars per cup with 25 mL of serum solution per cup. Media changes were completed every 3–4 days.

Female Sprague Dawley rats (n=15) were used for the in vivo portion of the experiment. All procedures were performed in accordance with Institutional Animal Care and Use Committee guidelines. Rats (n=5) were assigned to the 4-, 8-, and 16-week time points. Four rats received 4 subcutaneous implants and one rat received 2 implants. Briefly, rats were anesthetized with an inhalant anesthetic (isoflurane) and given mass-based doses of Buprenorphine, Bupivacaine, and Meloxicam to manage pain and inflammation. 10 mm incisions were made along the dorsal aspect of the rat, and fascia was gently released from the subcutaneous layer. In each cavity, a dog bone test coupon was implanted. Incisions were closed using a two-layer closure with resorbable suture. After surgery, animals were allowed to have unrestricted cage activity, while water and food was provided ad libitum. At each respective time point, animals were euthanized, and implants were harvested and frozen at −20°C until time of tensile testing.

2.2. Mechanical testing

Mechanical tests were performed on a universal test frame (TA Instruments, Electroforce 3550, Fig. 3). Tensile testing was completed at a rate of 50 mm/min (0.833 mm/sec) with serrated vice grips on each end of the tensile bar and a 1100N/14 Nm load/torque cell. Compression testing was conducted with a specialized fixture that ensured the loading plates were parallel to each other. Tests were performed at 1.3 mm/min (0.02167 mm/sec) using a 15kN load cell. Torsional testing used the same grips as tensile testing and was completed at a rate of 1 deg/sec up to 90 degrees. 4-point-bend testing was performed with supports at 6 and 24 mm along the length of the bar and loading anvils at 12 and 18 mm. The crosshead moved at a rate of 1.3 mm/min (0.02167 mm/sec) and forces were measured with the 1,100N/14 Nm load/torque cell.

Fig. 3:

Photographs of the mechanical test procedures for (A) tension, (B) compression, (C) torsion, and (D) 4-point bending

2.3. Calculations and Statistical Analysis

A custom MATLAB code (Mathworks, Natick, MA, USA) was created to quantify mechanical properties of the coupons. Non-normalized specimen properties (e.g. stiffness, maximum force) of test coupons were quantified rather than the material properties (e.g. modulus, maximum stress). This choice was made because normalizing forces by cross sectional area would eliminate the differences in behavior caused by alterations to infill density. Therefore, the variables of interest in this study were constrained to stiffness (slope of the linear region of the stress-strain curve) and maximum force. Two-way ANOVA tests were run with post-hoc Dunnett’s tests and Bonferroni corrections to determine significant differences from the control 0-week timepoint and interactions between implant design and time. Significance was set to p < 0.05.

3. Results

3.1. Week 0

We observed significant changes in stiffness and maximum force/torque caused by alterations in infill density with fewer differences caused by changes in infill architecture (Fig. 4). Results from 2-way ANOVA analyses quantified the percentages of variation that were caused by changes in lattice architecture, infill density, and the interaction between these two variables (Table 1). This calculation demonstrated how these variables, as well as their interaction, contributed to the variability in stiffness and maximum force/torque. Overall, the percent of total variation caused by changes in infill density ranged between 74.2 – 98.1%, except for maximum bending force measures (49.2%). On the other hand, percent of total variation caused by alterations to infill architecture ranged between 0.1 – 2.5%, except for maximum bending force (24.4%).

Fig. 4:

Box and whisker plots of stiffnesses (A-D) and maximum forces/torques (E-H) of coupons tested at Day 0. There were few significant differences caused by changes in architecture. * p<0.05; ** p<0.01; *** p<0.001; **** p<0.0001.

Table 1:

Sources of Total Variation for Stiffness and Max Force/Torque Measures

	Stiffness
	Infill Density		Infill Architecture		Interaction (Density × Architecture)
	% of total variation	p-value	% of total variation	p-value	% of total variation	p-value
Tension	81.6%	<0.0001	2.5%	0.01	4.7%	0.0027
Compression	96.9%	<0.0001	0.7%	<0.0001	1.2%	<0.0001
Torsion	74.2%	<0.0001	2.0%	0.103	4.5%	0.0456
Bending	87.8%	<0.0001	1.7%	0.0063	3.5%	0.0008
	Max Force/Torque
	Infill Density		Infill Architecture		Interaction (Density × Architecture)
	% of total variation	p-value	% of total variation	p-value	% of total variation	p-value
Tension	91.1%	<0.0001	1.1%	0.0155	2.3%	0.0027
Compression	98.1%	<0.0001	0.1%	0.5068	0.1%	0.5882
Torsion	81.9%	<0.0001	1.9%	0.031	5.1%	0.0017
Bending	49.2%	<0.0001	24.4%	<0.0001	14.8%	<0.0001

Overall, alterations in lattice architecture did not consistently lead to changes in stiffness or maximum force/torque. This was the case for tensile, compressive, and torsional tests (Fig. 4A–C&E–G, Table 2). For example, the rectilinear 40% infill group was 24.3% less stiff than the rectilinear 80% infill group (p < 0.0001) and 43.6% less stiff than the control (100% infill) group (p < 0.0001). The maximum tensile force for the 40% infill rectilinear group was 45.5% less than the rectilinear 80% infill group (p < 0.0001) and 61.5% less than the control (100% infill) group (p < 0.0001). These results were similar within the gyroid and triangle infill patterns within the tensile tests and across different loading protocols. When comparing across infill architectures, 40% gyroid infill was significantly stiffer than triangle (Fig. 4A), but not at 80% infill density. Similarly, maximum tensile force sustained by 40% gyroid designs was higher than triangle (Fig. 4E), but there were no significant differences at 80% infill. It should be noted that gyroid implants had significantly smaller maximum bending forces than the rectangle at 40% and triangle coupons at 40% and 80% infills (Fig. 4H). Because the influence of infill architecture was limited, only results from the rectilinear specimens will be shown for the in vitro degradation experiment. Sample force-displacement graphs can be found in Supplemental Figure 1 and complete results for gyroid and triangle latticed coupons can be found in Appendix A and B.

Table 2:

Summary of Day 0 Results

	Tensile Stiffness (N/mm)						Tensile Max Force (N)
	40% Infill		80% Infill		100% Infill		40% Infill		80% Infill		100% Infill
	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.
RECT	931.4	108.2	1189.2	70.0	1511.7	84.2	390.7	78.3	716.9	46.9	1015.8	84.4
GYR	1086.5	85.0	1273.1	117.1			548.3	28.8	681.1	40.7
TRI	808.4	77.5	1272.6	123.4			412.4	20.0	631.6	48.4
	Compressive Stiffness (N/mm)						Compressive Max Force N)
	40% Infill		80% Infill		100% Infill		40% Infill		80% Infill		100% Infill
	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.
RECT	5525.4	128.0	8134.4	181.0	15279.0	347.5	2165.8	62.6	3509.9	91.0	8270.4	133.2
GYR	5483.9	209.9	6110.6	1280.3			2183.9	25.9	3392.3	96.5
TRI	5536.2	158.2	8397.1	492.5			2174.9	66.5	3826.8	1101.1
	Torsional Stiffness (Nm/deg)						Maximum Torque (Nm)
	40% Infill		80% Infill		100% Infill		40% Infill		80% Infill		100% Infill
	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.
RECT	0.041	0.003	0.048	0.002	0.058	0.007	0.581	0.064	0.744	0.043	1.104	0.101
GYR	0.034	0.002	0.045	0.004			0.763	0.093	0.748	0.065
TRI	0.035	0.003	0.053	0.003			0.633	0.034	0.863	0.036
	Bending Stiffness (N/mm)						Bending Max Force (N)
	40% Infill		80% Infill		100% Infill		40% Infill		80% Infill		100% Infill
	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.
RECT	76.4	4.7	95.0	4.1	111.3	2.5	165.3	5.3	177.8	6.7	188.7	4.3
GYR	80.6	3.8	85.8	5.8			109.8	3.6	145.6	24.1
TRI	83.6	2.1	96.0	5.5			156.9	5.7	172.6	9.9

3.2. In vitro degradation

In vitro degradation of rectilinear coupons led to significant losses in mechanical properties between the 8-week and 16-week time points. By 16 weeks, 8 of the 12 test groups demonstrated significant decreases in stiffness (Fig. 5 A–C, Table 3). Bending stiffness did not decrease as a function of time, regardless of infill density (Fig. 5D). 100% infill (solid) test coupons did not lose tensile or bending stiffness. In terms of maximum force/torque, a similar pattern was observed, where 10 of 12 tests resulted in significant decreases after 16 weeks (Fig. 5 E–H). The maximum torque measures did not follow these generalized degradation patterns, where only the 100% infill had significant decreases by 16 weeks (Fig. 5G). Results from gyroid and triangle infill designs can be found in the Supplemental Appendix.

Fig. 5:

Box and whisker plots of stiffnesses (A-D) and maximum forces/torques (E-H) of rectilinear infill coupons tested after 8 and 16 weeks of degradation. Significant decreases in mechanical properties were most often observed after 16 weeks of degradation. * p<0.05; ** p<0.01; *** p<0.001; **** p<0.0001.

Table 3:

Summary of In Vitro Experiment Results

	Tensile Stiffness (N/mm)						Tensile Max Force (N)
	40% Infill		80% Infill		100% Infill		40% Infill		80% Infill		100% Infill
	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.
Week 0	931.4	108.2	1189.2	70.0	1511.7	84.2	390.7	78.3	716.9	46.9	1015.8	84.4
Week 8	910.8	25.5	1074.2	148.3	1454.9	161.1	393.8	95.5	742.7	96.3	1004.0	75.4
Week 16	111.2	21.5	95.8	6.4	1343.8	195.4	156.1	13.3	146.3	16.7	218.4	47.8
	Compressive Stiffness (N/mm)						Compression Max Force (N)
	40% Infill		80% Infill		100% Infill		40% Infill		80% Infill		100% Infill
	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.
Week 0	5525.4	128.0	8134.4	181.0	15279.0	347.5	2165.8	62.6	3509.9	91.0	8270.4	133.2
Week 8	3494.4	308.7	6056.1	1097.3	8944.9	620.1	976.3	85.8	2656.1	333.0	5866.6	2251.6
Week 16	1861.6	112.7	2149.5	555.3	6989.4	858.6	389.8	61.4	438.8	44.6	2161.1	861.8
	Torsional Stiffness (Nm/deg)						Torsion Max Torque (Nm)
	40% Infill		80% Infill		100% Infill		40% Infill		80% Infill		100% Infill
	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.
Week 0	0.041	0.003	0.048	0.002	0.058	0.007	0.581	0.064	0.744	0.043	1.104	0.101
Week 8	0.017	0.006	0.050	0.016	0.058	0.005	0.399	0.050	0.687	0.285	0.762	0.288
Week 16	0.020	0.002	0.016	0.001	0.031	0.002	0.641	0.024	0.671	0.024	0.638	0.025
	Bending Stiffness (N/mm)						Bending Max Force (N)
	40% Infill		80% Infill		100% Infill		40% Infill		80% Infill		100% Infill
	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.
Week 0	76.4	4.7	95.0	4.1	111.3	2.5	165.3	5.3	177.8	6.7	188.7	4.3
Week 8	111.6	15.0	124.3	11.8	123.7	15.5	202.7	10.0	238.8	12.9	186.6	12.2
Week 16	97.0	9.5	125.7	11.8	101.3	17.1	35.9	8.3	60.0	3.1	36.2	9.0

3.3. In vivo degradation

Because changes in lattice architecture had minimal impact on coupon mechanics, only rectilinear coupons were implanted into rats. Tensile coupons were mechanically tested after 4, 8, and 16 weeks of subcutaneous implantation. There were no significant losses of mechanical properties between the 4- and 8-week timepoints. At 16 weeks there were significant losses in stiffness and maximum force, regardless of the infill density (Fig. 6, Table 4). Sample force-displacement graphs from in vivo testing can be found in Supplemental Figure 2.

Fig. 6:

Box and whisker plots of stiffness (A) and maximum force (B) of tensile coupons that were implanted subcutaneously in rats for 4, 8, and 16 weeks. Like in vitro tests, loss of mechanical properties did not occur until 16 weeks.

Table 4:

Summary of In Vivo Experiment Results

	Tensile Stiffness (N/mm)						Tensile Max Force (N)
	40% Infill		80% Infill		100% Infill		40% Infill		80% Infill		100% Infill
	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.
Week 4	764.8	192.4	1170.0	167.4	1528.9	74.8	349.7	161.8	549.2	143.6	1070.0	139.8
Week 8	724.7	455.1	1244.8	129.9	1542.6	120.3	279.1	125.2	485.7	254.5	874.8	161.6
Week 16	42.8	12.3	38.9	14.4	230.8	111.5	48.6	7.7	59.1	7.0	57.5	12.2

4. Discussion

This study systematically assessed initial mechanical properties and degradation kinetics of 3D printed PLGA scaffolds by applying various loading paradigms to uniquely designed coupons. Results from the Day 0 experiment indicated that initial mechanical properties of PLGA test coupons were influenced by infill density, rather than lattice architecture. In particular, the bending bars had the highest ratio of perimeter material to infill material (Fig. 2A) and the perimeter walls absorbed the maximum bending stresses sustained during testing. Together, this may explain the relatively minor changes in bending stiffness and maximum bending load between 40%, 80% and 100% groups (Fig 5D, H). In the case of bending bars, the gyroidal architecture exhibited significantly lower maximum bending forces than the rectilinear and triangular architectures (Fig. 4H). We believe this occurred because the gyroidal designs had a lattice architecture with translating anchor points on a layer-to-layer basis, as viewed from the top plane. This led to the creation of small, isolated struts (Fig 2C), as opposed to larger ones that were fused directly on top of each other during layer-by-layer deposition of material (Fig. 2A, B). Despite this finding, changes in infill density were still the major the source of variation in bending (Table 1), Overall, our Day 0 findings generally agreed with previous studies (Fernandez-Vicente et al., 2016; Leung et al., 2008), and confirmed the first half of our hypothesis, that increases in infill density would lead to improved initial mechanical properties.

In vitro and in vivo experiments demonstrated that degradation kinetics varied between groups, and significant losses typically occurred between the 8- and 16-week timepoints. Broadly speaking, these experimental outcomes rejected the second half of our hypothesis, that loss of mechanical properties would be constant as a function of time. In the in vitro and in vivo tensile tests, the 40% and 80% infill specimens maintained stiffness and maximum force through 8 weeks, and only showed significant losses in mechanical integrity at the 16-week timepoint (Figs 5A, 7A). Interestingly, the 100% infill implants maintained stiffness throughout the entire in vitro study, but this was not the case during the subcutaneous in vivo experiment. For torsion, the 40% infill bars lost stiffness after 8 weeks, but 80% and 100% infill groups did not lose stiffness until 16 weeks. The maximum torque measures did not show consistent losses as a function of infill or time, and none of the measures approached zero - even at 16 weeks (Figs 5C&G). It is possible that the perimeter walls and lattice structures used in the experiment provided some inherent resistance to torsional loading (Wu et al., 2023), but a complete characterization of this phenomenon is beyond the scope of this paper. Bending tests demonstrated surprising behavior, as stiffness increased with time in several instances. The maximum bending force increased at 8 weeks for the 40% and 80% infills, but not the solid control group. In all cases, the maximum bending force was significantly decreased at 16 weeks (Figs 5D&H). The possible reasons for this will be discussed separately. Finally, our hypothesis was correct for the compression cubes, which systematically lost stiffness and maximum force across the 0, 8, and 16-week timepoints, which suggests that loading paradigms need to be considered when assessing degradation kinematics of AM PLGA devices (Figs 5B&F).

Results from our in vitro and in vivo experiments were similar, but not completely consistent. Specifically, in the in vivo experiment, we observed significant loss of stiffness and maximum force/torque for all designs between 8- and 16-weeks (Fig. 6). In contrast, there were no significant losses in stiffness for the 100% tensile samples in the in vitro experiment (Fig 5A). There were also larger amounts of variability in stiffness and maximum force measures in the in vivo experiment. There are several possible causes for these discrepancies. Variability of mechanical properties of additively manufactured parts can be attributed to a variety of factors, including changes in age and chemistry of raw material, print settings, temperature and humidity, and build orientation (Goh et al., 2020). Also, the in vitro aging process is more controlled than the in vivo experiment. For example, implant mechanics may have been altered by microdamage experienced during the implantation procedure. It is also possible that post-operative inflammation led to changes in pH levels in the subcutaneous microenvironment (Punnia-Moorthy, 1987). Finally, the in vivo samples may have experienced external loads, caused by movement of the animals during the post-operative time periods, which may have led to changes in the implants.

There have been several previous publications that have examined PLGA degradation, and results suggest that timelines for loss of mechanical strength are variable. For example, one short-term experiment evaluated the mechanical properties of in vitro and in vivo degradation of PLGA scaffolds and showed no change in maximum load and modulus at 1 and 2 weeks (Oh et al., 2006). Another study evaluated the efficacy of PLGA plates and screws for internal fracture fixation in a rabbit mandibular fracture model. This study found that the mandibular fractures were fully healed by approximately 10 weeks, with plate resorption beginning around 6 weeks (Park et al., 2013). A separate study showed no statistically significant differences in the tensile strength or modulus of PLGA test coupons from 12 weeks of hydrolytic degradation (Kobielarz et al., 2020). Our study did not replicate any of these timepoints, and instead focused on 4, 8, and 16 week timepoints. Interestingly, we found that the in vivo microenvironment does not necessarily hasten the degradation process, but instead decreases the stiffness of implants with high density infill. We believe that mechanical loading will accelerate the degradation of mechanical properties of PLGA scaffolds, but that experiment is outside of the scope of the current project.

In the bending experiment we observed temporary increases in stiffness and maximum load prior to losses (Fig. 5 D&H), which has been shown in several previous studies. For example, in one study PLGA scaffolds were seeded with nucleus pulposus cells and implanted into subcutaneous pouches in mice for 4 and 6-week timepoints. The results showed that peak-load hardness increased from 0 to 4 weeks and again from 4 to 6 weeks in the higher density scaffolds, while the lowest density scaffold increased from 0 to 4 weeks, but decreased from 4 to 6 weeks (Kim et al., 2017). Another experiment using a calcium phosphate cement PLGA composite showed that the compressive and flexural strength increased significantly after soaking in a saline bath for 1 week, followed by a steady decrease of mechanical properties (Dagang et al., 2007). In the current study, the perimeter walls may have swelled during the 8-week period, which may have caused an increase in bending stiffness. By 16 weeks, deterioration of the perimeter walls may have led subsequent losses of stiffness. This behavior would be most notable in the bending bars, because the beam-like perimeter walls of the bend samples sustained the maximum tensile and compressive bending loads during testing.

Previous experiments have demonstrated advances in additively manufactured orthopedic devices using PLGA, but typically these works have focused on scaffolds for drug or growth factor delivery, rather than focusing on the relationships between printing parameters and degradation kinetics. For example, one study showed that PLGA coatings on 3D printed, porous, titanium scaffolds increased angiogenesis at the interface between the bone and implant in an ovine bone defect model (Hu et al., 2018). This occurred, in part, because the PLGA was releasing lactic acid in the area as it degraded. Another study found that a melt composed of calcium sulfate (CaSO₄) and PLGA could be used to 3D print implants with customized geometry for critical defects in rabbits (Liu et al., 2022). In this experiment, it was found that the CaSO₄ improved the mechanical properties, and the osteo-induction of the PLGA. Another study incorporated β-tricalcium phosphate (β-TCP) and magnesium powder into PLGA, then 3D printed porous scaffolds with it (Lai et al., 2019). They found they could improve the mechanical properties of the PLGA while maintaining the usefulness of the magnesium and β-TCP for faster osteogenesis. While the current study is less complex than these examples, it adds much-needed context by elucidating relationships between print parameters and degradation kinetics of untreated PLGA, which is the foundation for a variety of scaffolds.

The test coupons used in this study included a 3-layer perimeter, which had some advantages and drawbacks. Perimeter walls are commonly used in additive manufacturing because they improve mechanical integrity, increase accuracy and quality of prints, provide improved print adhesion and stability, and create anchor points for lattice structures. In this study, inclusion of perimeter walls likely subdued mechanical changes caused by alterations in lattice design. It should be noted that this study was a first step towards designing functional, biodegradable implants for bone fracture repair. We further rationalized the use of perimeter walls to create smooth outward facing surfaces and to increase mechanical strength, both of which are preferred for bone plate implants. Although it was possible to fabricate test coupons without perimeter walls, we believed such specimens would have limited clinical relevance.

There are several other limitations associated with this study. None of the specimens experienced mechanical loads in the subcutaneous space during degradation, which is inherent to the clinical use of fracture reconstruction implants. Additionally, the subcutaneous implants tested in the study were not placed adjacent to bone, so it is unclear if differences in in vivo microenvironments would influence results. This study did not utilize computational modeling to determine the stresses that occur within the implants, and importantly, at the welds between layers, which may further explain differences in failure kinetics between loading paradigms.

3D printed biodegradable implants have the potential to shift the paradigms associated with surgical fixation of bone fractures. Here, we observed that many mechanical properties of PLGA test coupons were maintained from the 0- to 8-week timepoint but significantly decreased between the 8- and 16-week timepoints. The experiment provided new information that will help us better understand the complicated interactions between additive manufacturing techniques, lattice architecture, and biodegradation of 3D printed PLGA components. This information will guide the design and development of implants and scaffolds that will be used in future in vivo experiments.