Cylinders that had a diameter of 6 mm and a height of 12 mm were either 3D-printed with a photoresin or molded out of PMMA-based bone cement (PALACOS^® R), as shown in Figure 1. The 3D-printed samples were created via stereolithography (SLA), which allowed for antibiotics to be incorporated into photoresin. A computer-aided design (CAD) model of the cylinder was sliced into thin 100 μm 2D layers and uploaded into the Form 2 3D printer (Formlabs). Biomed clear (BMC), a biocompatible liquid photoresin from Formlabs was used as the printing material. The commercially available resin that has been evaluated in accordance with ISO 10993–1:2018, ISO 7405:2018, ISO 18562–1:2017 and have passed the requirements associated with being not cytotoxic, not an irritant, not a sensitizer, not toxic (subacute/subchronic), not mutagenic, non-pyrogenic, and not systemically toxic (BioMed Clear Resin, 2024). The BMC was then doped with varying amounts of gentamicin sulfate (VWR) or vancomycin hydrochloride (VWR). The measured amount of antibiotic powder was added to a beaker filled with the printing resin. An immersion blender was used to create a homogeneous mixture, which was then poured into the printing vat. The printing process began shortly afterward to ensure the antibiotic powder remained evenly suspended in the solution. The concentration of gentamicin ranged from 0%–16% (W/W) and the concentration of vancomycin ranged from 0%–10% (W/W). A build plate was lowered into the vat where a laser selectively polymerizes the liquid resin into a solid. The build plate was lifted, and the process was repeated in a layer-wise fashion until the print was complete (Pagac et al., 2021). After the samples were printed, they were washed in fresh Isopropyl Alcohol (VWR) for 20 min, and cured at 60°C for 60 min. Both gentamicin and vancomycin were chosen because they have been clinically used in antibiotic PMMA spacers where they have been found to be heat stable in the exothermic curing process that can reach 115°C (Carli et al., 2018; Chen et al., 2019). Additionally, preliminary studies showed that the eluted gentamicin and vancomycin were still functional enough to kill bacteria plated on a Petri dish following the extensive processing required to make a 3D print.

PMMA samples were made with commercial bone cement containing PMMA copolymer (84%), zirconium dioxide (15%) and benzoyl peroxide (1%) in the powder component, and methyl methacrylate (98%) and N, N-dimethyl-p-toluidine (2%) in the liquid component. Either gentamicin or vancomycin were thoroughly mixed by hand into 2000 mg of the dry powder so the W/W percentage of antibiotics and PMMA bone cement (dry + liquid mass) was equivalent to BMC samples when 1 mL of liquid monomer was added. The samples were mixed until uniform and was transferred to a 3 mm slip tip syringe. The bone cement was then injected into cylindrical molds that were machined out of Delrin blocks (Mcmaster Carr). After 20 min, the cylindrical samples were tapped out of the molds using a brass punch. The flat face of the BMC and PMMA cylinders were polished with sandpaper to a final height of 12 mm.

The composites were referred to by their base material and the weight percentage of antibiotic. For example, BMC 2G represents a BMC composite doped with 2% gentamicin, and PMMA 10V represents a PMMA composite doped with 10% vancomycin.

The thermodynamic properties of BMC-G and PMMA composite were assessed using Differential Scanning Calorimetry (DSC) (Gill et al., 2010). 4 pieces of each sample (5–10 mg) were cut from a rectangular bar of printed BMC or molded PMMA. The samples were sealed in an aluminum hermetic pan and placed in a TA Discovery 2500 DSC machine. The samples were cycled twice between −60°C and +150° at a rate of 10°C/min. The glass transition point T_g for DSC was defined as the inflection point in the DSC curve.

The temperature dependent properties of BMC-G and PMMA composites were examined using Dynamic Mechanical Analysis (DMA) (Patra et al., 2020). DMA was performed on 3 samples from each of the groups. A bar (2 mm × 3 mm x 15 mm) was placed into a 3-point bending fixture of a RSA G2 DMA machine (TA Instruments). A small oscillating displacement with a peak of 15 μm was applied at a frequency of 15 Hz as the sample was heated from 0°C to 150°C. The storage modulus (ability to store energy elastically), loss modulus (ability to dissipate energy), and tan delta (ratio of loss and storage modulus) were calculated at each temperature.

All of the constructs were placed in 48-well plates filled with 1 mL of normal saline warmed to 37°C to replicate body conditions. The concentration of the gentamicin or vancomycin eluted into the saline was sampled at days 1, 2, 3, 4, 5, 7, 10, 15, 20 and 30. At each time point, the saline surrounding the reservoir was collected and replaced with fresh saline to maintain sink diffusion conditions. All samples were stored at −20°C until all time points had been collected. The samples were thawed at 4°C for a disc diffusion assay (Balouiri et al., 2016). An overnight culture of Escherichia coli (EC) (ATCC 25922) for the gentamicin samples and Staphylococcus Aureus (SA) (ATCC 29213) for the vancomycin samples was prepared in 3 mL of sterile Muller Hinton II Broth (MilliporeSigma) (Minogue et al., 2014; Soni et al., 2015). The bacterial solution was diluted to have an optical density of 0.50 at 600 nm measured with an Ultraspec 10 cell density meter (Biochrom). This corresponded to 5 × 10⁷ Colony Forming Units (CFU)/mL for EC and 1 × 10⁹ CFU/mL for SA. 40 μ L of EC or 20 μ L of SA were spread evenly with an inoculation loop on 90 mm diameter Petri dishes filled with 30 mL of Muller Hinton agar (Difco). Next, five 6 mm Whatman antibiotic assay discs were placed onto each agar plate and loaded with 20 μL of thawed recovered saline. The agar plates were incubated overnight at 37°C and imaged after 16 h. Using ImageJ software (FIJI), the Zone of Inhibition (ZOI) or the diameter of the circle around the disc where bacteria was not able to grow, was measured. Saline samples with known concentrations of antibiotics were tested to produce a standard curve and allow for the concentration of antibiotic to be calculated from the ZOI. The cumulative release was calculated by multiplying the measured concentration of antibiotic by the volume of the recovered saline (1 mL) and adding to the cumulative release from the prior timepoint. The 30-day time point was chosen after preliminary experiments with PMMA mixed with clinically used amounts of antibiotics demonstrated that the drug concentration in those samples was too low to produce a measurable ZOI using the method described above.

The compressive properties of the BMC and PMMA composite were evaluated by compressing test cylinders with a height of 12 mm and a diameter of 6 mm from each group in accordance with ISO 5833 (ISO - ISO 5833, 2022). Five samples from each composite group were tested before and after the 30-day elution. Each test cylinder was polished and measured with calipers before being placed between compression platens of a TR nano uniaxial test frame (Test Resources). The samples were compressed at a rate of 20 mm/min. Compressive stress was calculated from the load that was measured using a 25 kN load cell. Compressive strain was calculated from the vertical displacement of the crosshead. The yield strength was calculated using the 0.2% offset method (Morgan and Keaveny, 2001).

The surface features of the composite samples after the 30-day elution were examined using scanning electron microscopy (SEM). The samples were dried and placed inside a Hitachi TM3030Plus Tabletop so the long axis of the cylinder faced the electron source. Electrons were accelerated at 15 kV at the samples and secondary electrons were detected to produce an image at ×100 magnification.

Results were reported as averages with one standard error. One way ANOVA was used to compare the average T_g values. R² coefficients were calculated to assess the quality of the standard curves relating the size of the ZOI to the concentration of antibiotics. The standard deviation for the cumulative release plots was calculated by summing the variance measured at the time point and all the previous time points, and then taking the square root of the cumulative variance. A multivariate linear regression model was used to assess the independent effects of drug concentration and the 30-day elution on the compressive properties of the BMC and PMMA composites. All statistical analysis was conducted with RStudio.

The doped resin was still able to selectively polymerize the resin with the same print settings to form complex structures like a gyroid lattice. (Figure 2A). As gentamicin or vancomycin was added to the BMC resin, the printed samples went from a translucent appearance to a white opaque appearance (Figure 2B). Up to 16% gentamicin or 10% vancomycin could be added to the BMC resin before the prints started to fail.

DSC was used to measure the T_g of the BMC (Figure 3A) and PMMA (Figure 3B) composites. The BMC composites had T_g values of 65.0 ± 1.5°C and the PMMA composite had T_g values of 102.3 ± 2.2°C. The addition of gentamicin did not alter the T_g of the BMC composites (p = 0.941) or the PMMA composites (p = 0.251).

The storage modulus, E, for each of the BMC-G groups and PMMA bone cement was plotted from 0°C – 150°C on a logarithmic scale (Figure 3C). There is overlap between the BMC-G concentrations, suggesting that the addition of antibiotics does not change the mechanical properties at small displacements over a wide range of temperatures. In all the BMC-G plots, the storage modulus decreases modestly at 60°C and plateaus at 150°C. In contrast, the storage modulus of PMMA bone cement starts to decrease significantly at 80°C, and continues to decrease at higher temperatures. The BMC plots are consistent with a thermoset while the PMMA plots are consistent with a thermoplastic.

The representative stress-strain curves for the BMC and PMMA composites are shown in Figure 4. None of the samples fractured before 10% strain, but they underwent barreling. In general, the stress-strain curves for BMC composites did not significantly vary with the increased percentage of antibiotics. A notable exception included BMC 16G post-elution samples that were significantly less stiff and had a lower strength than the rest of the BMC G composite post-elution. Another exception was that the addition of any amount of vancomycin significantly lowered the compressive strength of the samples. The different groups of PMMA composite varied more in general compared to the BMC curves. Trends were further analyzed by plotting the modulus, yield strength, and ultimate strength as a function of drug concentration for each type of composite (Figure 5). The BMC lines (black) did not change as much with increasing drug concentration as the PMMA lines (gray). Most of the samples got stronger and stiffer after the 30-day elution (dashed lines) compared to before the 30-day elution (solid lines) with the exception of the PMMA samples with over 8% antibiotics. The effect of drug concentration and the 30-day elution were quantified with a multivariate linear regression model (Table 1). There was a statistically significant decrease in modulus, yield strength, and ultimate strength associated with doping PMMA composites with additional antibiotics. In contrast, the addition of antibiotics did not statistically alter the modulus or yield strength for the BMC composites. There was a 0.25 MPa decrease in ultimate strength associated with each percent increase of gentamicin. The same increase in gentamicin was associated with a 1.17 MPa decrease in ultimate strength in the PMMA composites.

The standard curve demonstrated that ZOI measurements could be used to accurately measure the concentration of the antibiotics from 8 μg/mL to 10,000 μg/mL (Supplementary Figure S1). The R² values were 0.99 for both the gentamicin and vancomycin standard curve. The concentration of antibiotic that eluted between each timepoint is shown in Figure 6 on a logarithmic scale. There was a measurable ZOI from most samples over the 30-day elution with the exception of BMC-G composites with less than 8% gentamicin and PMMA-G composite with less than 4% gentamicin. Overall, samples released more antibiotics at the early time points. Some curves have an increased measured concentration after day 5 (Figure 6). This is due to the time points being spread out, giving the samples more time to elute antibiotics. There was more antibiotic eluted from the PMMA composites and from samples with a higher concentration of antibiotics (Figure 7). Both PMMA and BMC samples doped with gentamicin were still bactericidal when placed directly on a bacterial Petri dish after the 30-day elution. (Figure 8). The undoped controls were not able to prevent bacteria from growing up to the surface. The BMC 1G samples prevented bacteria growth within a small area around the cylinder. All other composites produced an oval-shaped zone of inhibition around the cylinder.

SEM imaging of the samples after the 30-day elution highlight differences in the surface morphology. (Figure 9). Dried salt crystals from the saline can be seen on the surface. The BMC samples are relatively free from defects. In contrast, there are 50–100 μm diameter craters on the PMMA samples.