Sustaining Antibiotic Release from a Poly(methyl methacrylate) Bone-Spacer.

One of the challenges in using a bone-spacer to cure infection is the fabrication of a material that can continuously release required antibiotics at effective concentrations for at least 4-6 weeks. Poly(methyl methacrylate) (PMMA) impregnated with antibiotics is one of the popularly used bone-spacer materials. Currently, improved sustained release of hydrophobic and hydrophilic antibiotics is needed for this material. Here, hydrophilic vancomycin (VAN) was encapsulated into calcium citrate (CC) particles and natural rice granules, and hydrophobic erythromycin (ERY) was encapsulated into ethyl cellulose and poly(lactic-co-glycolic acid) particles. The four antibiotic-loaded particles were each incorporated into the PMMA cement. The two unencapsulated drugs and all four drug-loaded particles distributed well in the obtained composites. PMMA composited with VAN-loaded CC showed prolonged VAN release at an effective concentration for more than 40 days, but the composite possessed lesser compressive strength than the PMMA with no drug. PMMA composited with unencapsulated ERY showed a better sustainment of drug release than those composited with encapsulated ERY. VAN elution from the VAN-CC-PMMA did not significantly affect the compressive strength of the material, whereas ERY elution from the ERY-PMMA composite significantly decreased the material's mechanical strength.

Introduction

Poly(methyl methacrylate) (PMMA) has been used as a bone cement in joint replacement surgery since 1953.[1] At the beginning, the PMMA cement was first used at the junction between a metal implant and the bone to help securing the implant in place and lessening the stress at the connection between the implant and the bone; the material gave an added benefit on releasing antibiotics to prevent infection.[1] Later on, PMMA cement has also been used as a bone-spacer to temporarily take up the space where the infected implant has been removed. When used as a bone-spacer, local delivering of antibiotics becomes a primary role of the material in addition to providing temporary structure and securing the space.[2−7] As the bone-spacer is usually used temporarily and the spacer will be replaced with the metal implant once the infection has been eradicated, the spacer material can be less durable and less compression-resistant than the metal implant. Currently, antibiotic-impregnated PMMA bone-spacers are usually made in-house by a direct mixing of powdered antibiotics into the materials prior to the polymerization setting of the cement.[8−11] Numerous efforts to prolong the drug release from PMMA bone-spacers have been reported. Examples include finding antibiotics and cement formulations that are compatible.[9,11] Different PMMA cements have been compared for their ability to sustain the release of incorporated antibiotics; this is because cements of different formulations usually possess different degrees of cross-linking and homogeneity, which can affect the elution of the incorporated drug.[11] Different antibiotics such as tobramycin and vancomycin (VAN) usually show different elution rates from the same bone-spacers.[9,11] A surfactant has been used to help blend liquid antibiotics into the PMMA matrix, but usually comes with the cost of mechanical strength.[12] The use of drug carriers such as liposomes,[1,13] and sodium dodecyl sulfate–polyvinylpyrrolidone particles,[14] to sustain the release of different antibiotics from PMMA, have been reported. Other antibiotic particulates used with PMMA bone cement include silver nanoparticles capped with tiopronin,[15] gentamycin-loaded chitosan/chitosan derivative nanoparticles,[16,17] and gold nanoparticles.[18] Nevertheless, still one of the challenges of using a bone-spacer to cure infection caused by the bone replacement is the fabrication of a biocompatible bone-spacer that can continuously release required antibiotics at an effective concentration for at least 4–6 weeks.[10,19]

Here, we have investigated the uses of four different drug carriers to help in blending both hydrophilic and hydrophobic antibiotics into PMMA and prolonging the drug elution from the resulted composite cements. We show here the use of natural rice granule (RG) and calcium citrate (CC) particles to encapsulate the hydrophilic VAN, the fabrication of composite cements of PMMA with the two VAN-loaded particles, the investigation on the distribution of the VAN-loaded particles in the resulted composite matrixes, the 42-day monitor of drug releases from the composites, and the study on the effects of those added drugs over the mechanical strength of the composites. We also show similar studies on erythromycin (ERY), a hydrophobic antibiotic drug, with the use of ethyl cellulose (EC) and poly(lactic-co-glycolic acid) (PLGA) particles as carriers. This report gives a new perspective on the differences in strategies to prolong the elution of different antibiotics from the PMMA bone cement, and also some interesting observations on the differences in the releasing character between the hydrophilic VAN and the hydrophobic ERY.

Results and Discussion

The hydrophilic VAN was successfully encapsulated into two different carriers, natural RGs and CC particles. The preparation of VAN-loaded RG or VAN–RG was carried out using the previously reported heat expansion method in which RG was incubated in VAN aqueous solution at 83 °C.[21] On the basis of the quantification of the free drug left in the water medium, the encapsulation process gave encapsulation efficiency of 70.2 ± 1.5% and the obtained 4.58 ± 0.84 μm granules possessed a VAN loading of 84.9 ± 0.3% (Figure A). It should be noted here that the encapsulation process did not change the morphology of the RGs; the shape was still polyhedral of multipentagonal faces with a similar average diameter to the original granules (data not shown). Successful encapsulation has also been confirmed by thermogravimetric analysis (TGA) and IR analyzes of the product.

Figure 1

Scanning electron microscopic images of drug-loaded particles: (A) VAN-loaded RGs (VAN–RGs), (B) VAN-loaded CC particles (VAN–CC), (C) ERY-loaded EC particles (ERY–EC), and (D) ERY-loaded PLGA particles (ERY–PLGA).

As reported earlier that RG is thermoresponsive, the granules can expand in an aqueous medium at high temperature and shrink back to their original size when cool.[21] Unlike the medium, the VAN that had been taken up into the granules during the heat-induced granule expansion probably bound to the biopolymer matrix of the RG and could not move out from the granules during the cool down, resulting in an effective drug encapsulation into the RG. This speculation is supported by the disappearance of the endothermic peaks in the TGA profile of the RG (maximum at 298 °C) and that of the free VAN (broad peak with the maximum at 234 °C), and an appearance of a new broad peak with the peak maximum at 237 °C in the profile of VAN-RG (Figure S1A in the Supporting Information). The broad endothermic peak likely represents various interactions between VAN and RG such as multiple hydrogen bondings between the two as previously speculated and illustrated.[21] IR spectrum of the VAN–RG also shows characteristic peaks of VAN, for example, a peak at 1638 cm–1 which corresponds to N–H bending, the 1600 and 1498 cm–1 absorption which corresponds to C=C stretching (in ring), and the peak at 1229 cm–1 which corresponds to C–O–C stretching (Figure S2A in the Supporting Information).

Encapsulation of VAN into CC particles was carried out by growing the CC nanospheres in the presence of VAN. Approximately 554.4 ± 220.3 nm particles were obtained (Figure B) with the VAN loading content of only 4.9 ± 3.2%. The encapsulation efficiency of the process was 27.9 ± 18.3. Encapsulation of VAN into the CC was probably taking place through the entrapment of the VAN into the CC particles assembled from calcium and citrate ions. It was likely that ionic interactions between carboxylates in the VAN and Ca2+ took place. In addition, as the encapsulation was performed at a very basic pH of more than 8, the phenolic moieties in the VAN structure should be in the phenolate anion state. The phenolate could also form an ionic bond with the Ca2+. Nevertheless, VAN probably could not very well compete with the much smaller citrate ions during the particle growth; therefore, low encapsulation efficiency and low loading were observed. It should be noted here that the encapsulation process was optimized through 10 combinations of citrate, calcium, VAN ratios (data not shown). Also, pH optimization during the encapsulation was also carried out (data not shown). We speculate that the aggregation of the particles and the uneven particle growth probably could cause wide particle size distribution. Nevertheless, particle size is not very critical for the application on PMMA cement; the current product of 554.4 ± 220.3 nm was further used in the PMMA casting.

TGA and IR analyses of the VAN–CC indicate successful encapsulation of VAN into the particles (Figures S1A and S2B in the Supporting Information). The disappearance of the endothermic peaks in the TGA profile of CC at 381, 482, and 675 °C and the appearance of peaks at 372, 462, and 678 °C in the profile of VAN–CC, can be clearly observed, indicating some new molecular interactions in the VAN–CC particles. The inorganic nature of the CC correlates well to its endothermic peaks at higher temperature than 300 °C. With the entrapment of VAN into CC, the obtained product shows an endothermic peak starting from 270 °C. IR spectrum of the VAN–CC clearly shows characteristic peaks of VAN and CC, for example, the peak at 3051 cm–1, which corresponds to N–H stretching of the amine group in the VAN structure, the peak at 1662 cm–1, which corresponds to N–H bending of VAN, the 1602 and 1425 cm–1 absorptions, which correspond to the antisymmetrical and symmetrical vibrations of the COO– group of citrates in the CC.

To encapsulate hydrophobic drug ERY, EC and PLGA were experimented. Encapsulation of ERY into the EC particles was successfully carried out through the self-assembly of the EC and the drug molecules during the slow change from an ethanol-rich to a water-rich medium. The obtained ERY–EC of 312.5 ± 55.3 nm (Figure C) possesses ERY loading of 58.1 ± 5.3%. The process gave encapsulation efficiency of 62.8 ± 5.7%. TGA profile (Figure S1B in the Supporting Information) and IR spectrum (Figure S2C in the Supporting Information) of the particles indicate the presence of ERY in the particles. The shifts in endothermic peaks in the TGA profile from 343 °C of the EC and 300 °C of the free ERY to the broad peak with a maximum at 349 °C of ERY–EC (Figure S1 in the Supporting Information) imply some new molecular interactions in the particles. The new molecular interactions are stronger than the original molecular interactions in the EC and ERY. In addition, these TGA profiles indicate that the ERY likely distributed in the EC polymer matrix in the form of a solid solution, not the core–shell assembly. The IR spectrum of the ERY–EC (Figure S2 in the Supporting Information) shows characteristic peaks of ERY, for example, peaks at 1736 and 1701 cm–1, which correspond to C=O stretching of ester and C=O stretching of aliphatic ketone of the ERY structure.

Through the emulsion solvent evaporation technique, we have also successfully fabricated ERY–PLGA as spherical particles of 11.2 ± 8.1 μm (Figure D) with the drug loading content of 6.0 ± 0.9%. Encapsulation efficiency of 85.1 ± 13.2% was observed for the process. TGA profile of the particles (Figure S1 in the Supporting Information) supports the presence of ERY in the particles. The shift in the endothermic peaks from 336 °C of the PLGA and 300 °C of the free ERY to 260 and 556 °C imply some new molecular interactions, likely from the ERY and the polymer components. The small endothermic peak at 428 °C, which corresponds to the decomposition of the poly vinyl alcohol (PVA) component,[22,23] can also be observed (Figure S1B in the Supporting Information). With no absorption peaks of all the original starting materials’ presence in the profile of the product, it is very likely that the ERY–PLGA possesses the solid solution character of ERY within the PLGA–PVA polymer matrix. The IR spectrum of the ERY–PLGA (Figure S2D in the Supporting Information) also contains characteristic peaks of PVA, for example, peaks at 3331 cm–1, which corresponds to O–H stretching, peak at 2938 cm–1, which corresponds to C–H stretching, and the 1746 cm–1 absorption, which corresponds to C=O stretching.

Particle size, encapsulation efficiency of the preparation process, and the drug loading in the four obtained particles are summarized in Table . The four drug-loaded particles, VAN–RG, VAN–CC, ERY–EC, and ERY–PLGA, and also the two unencapsulated drugs, VAN and ERY, were each mixed with the PMMA powder and the mixed powder was processed into the PMMA cement by mixing with the methyl methacrylate (MMA) monomer liquid. The formulation was controlled to get the PMMA pieces at the same drug concentration, and the casting was carried out to get the product with the same 1 × 1 × 1 cm3 dimension. Under this condition, it should be noted here that, drug particles with low drug loading were incorporated into the PMMA at higher masses, comparing to drug particles with higher drug loading. Scanning electron microscopy (SEM) analyses indicate good distribution of VAN, ERY, VAN–RG, VAN–CC, ERY–EC, and ERY–PLGA in the corresponding PMMA composite matrixes (Figure , and also S3 and S4 in the Supporting Information). PMMA composites consist of mostly the 48.1 ± 21.2 μm PMMA preformed microparticles linked together through the PMMA formed by the polymerization of MMA. The 4.58 ± 0.84 μm VAN–RG particles account for approximately 6.2% (by wt) of the PMMA preformed particles, and the SEM image of the composite cement agrees well with its content (Figures row C, and S3 and S4 row C in the Supporting Information). The 554.4 ± 220.3 nm VAN–CC particles account for 40.8% (by wt) of the PMMA particles. The SEM image of VAN–CC–PMMA shows distribution of a lot of VAN–CC along with the PMMA particles (Figures row D, and S3 and S4 row D in the Supporting Information). The 312.5 ± 55.3 nm ERY–EC particles account for approximately 11.0% (by wt) of the PMMA particles and these drug particles can be observed in the SEM image of the cement (Figures row F, and S3 and S4 row F in the Supporting Information). The 11.2 ± 8.1 μm ERY–PLGA particles account for 8.9% (by wt) of the PMMA preformed particles; they can also be observed in the SEM image of the cement (Figures row G, and S3 and S4 row G in the Supporting Information).

Table 1

Particle Sizes, % EE, and % Loading of VAN-Loaded RGs (VAN–RG), VAN-Loaded CC (VAN–CC), ERY-Loaded EC (ERY–EC), and ERY-Loaded PLGA (ERY–PLGA) Particles

drug-loaded particles	particle sizes (average ± SD)	% EE (average ± SD)	% loading (average ± SD)
VAN–RG	4.6 ± 0.8 μm	70.2 ± 1.5	84.9 ± 0.3
VAN–CC	554.4 ± 220.3 nm	27.9 ± 18.3	4.9 ± 3.2
ERY–EC	312.5 ± 55.3 nm	56.6 ± 1.3	52.0 ± 5.0
ERY–PLGA	11.2 ± 8.1 μm	85.1 ± 13.2	6.0 ± 0.9

Figure 2

Scanning electron microscopic images of (A) PMMA, (B) PMMA impregnated with VAN (VAN–PMMA), (C) VAN-loaded RGs (VAN–RG–PMMA), (D) VAN-loaded CC particles (VAN–CC–PMMA), (E) ERY (ERY–PMMA), (F) ERY-loaded EC particles (ERY–EC–PMMA), and (G) ERY-loaded PLGA particles (ERY–PLGA–PMMA). The drug (green arrows) and drug-loaded particles (red arrows) can be observed together with the 48.1 ± 21.2 μm PMMA microparticles (blue arrows). The images taken before (two columns on the left) and after (two columns on the right) the 42-day drug release are shown. Images of the surface of the pieces (column denoted outside) and of the inside of the broken pieces (column denoted inside) are shown.

To mimic the real biological fluid which is continuously renewed in the body parts, we used phosphate buffer saline (PBS) as the release medium[6,11,16] and also changed the release medium into a fresh one every day during the drug release test. The cement pieces were submerged into the release medium with constant shaking, and the release medium was quantified for the released drug. PMMA composited with VAN–RG showed some improvement in prolonging the effective concentration of the released VAN when compared to the PMMA that was impregnated with unencapsulated VAN (Figure A and also S5A in the Supporting Information). It should be noted here that the minimal inhibitory concentration of VAN against Staphylococcus aureus is 2.0 μg/mL.[24] Using RG to encapsulate VAN prior to the incorporation into the cement could extend the effective concentration of VAN in the released medium from 16 days (observed for the PMMA doped with unencapsulated VAN) to 19 days. The PMMA composited with VAN–CC showed much higher VAN release every day, from day 1 to day 42, the last day of the test period (Figure A and also S5A in the Supporting Information). The VAN concentration in the release medium at day 42 was well above the minimal inhibition concentration (MIC) values. The result here agrees with the SEM images that shows many VAN–CC particles in the composite. Approximately 65% of the VAN in the VAN–CC–PMMA was released during the 42-day period, comparing to only 15% for the other two composts (Figure S5A). We speculate that the significantly higher concentrations of the released drug observed for the VAN–CC–PMMA were likely caused by the presence of many micro/nanochannels in the composite matrix, formed as a result of the alteration of the original PMMA packing because of the presence of 40.8% (by wt) of the VAN–CC particles in the composite formulation. These channels allowed the elution of VAN from deep inside the matrix.

Figure 3

Drug release profiles, shown as drug concentrations in the release medium, of: (A) PMMA cement loaded with VAN (VAN–PMMA), VAN-loaded RGs (VAN–RG–PMMA), and VAN-loaded CC particles (VAN–CC–PMMA); (B) PMMA cement loaded with ERY (ERY–PMMA), ERY-loaded EC particles (ERY–EC–PMMA), and ERY-loaded PLGA particles (ERY–PLGA–PMMA). Averages are shown with the error bar representing standard deviation from three independent data points.

In contrast to the hydrophilic VAN in which the CC carrier could significantly increase the concentration of the released VAN at later days, EC and PLGA carriers used for the encapsulation of the hydrophobic ERY induced more burst release at the beginning and produced lower drug release in later days (Figures B and S5B). Surprisingly, PMMA impregnated with unencapsulated ERY showed less burst released at the beginning and higher concentrations of the released drug at later days. Even at day 42, the ERY concentration in the release medium of the ERY–PMMA composite was four times higher than the MIC of this drug against S. aureus (8 μg/mL).[25] Although the total drug released during the 42-day period of the three composites was in close ranges of approximately 80–100% of the total drug presence in the materials at the beginning, the release from ERY–PMMA was the most steady (Figure S5B in the Supporting Information).

The result indicates that incorporating unencapsulated hydrophobic antibiotic ERY into the PMMA can produce bone cement that possesses good sustained release of the drug. It should be noted here that the limited solubility of ERY in the release medium (0.60 mg/mL at 37 °C) is not the cause of this sustained release. We speculate that the hydrophobic ERY probably formed into small particulates that packed well with the PMMA particles, and the slow erosion of some ERY particulates at the surface of the matrix then granted access of the release medium to the ERY further inside the cement pieces. This process kept going slowly and resulted in the sustained release of the ERY from the ERY–PMMA. In contrast, because of the water insolubility of EC and PLGA, release of ERY from the EC or PLGA particles situated at the surface of the cement did not dissolve the two particles and, therefore, did not generate further exposure of the drugs inside the cement piece to the release medium.

The compressive strengths shown as the maximum stress values of the tested bone cements are shown in Figure and Table . Adding either unencapsulated VAN or VAN–RG into the PMMA produced no significant difference of the compressive strength at the 0.05 level of significance. However, VAN–CC addition produced a significant decrease in the compressive strength. We speculate that the high amount of VAN–CC (40.8% of PMMA) was responsible for such effect, noting that the amounts of unencapsulated VAN and VAN–RG were 5.5 and 6.2% of the PMMA particles, respectively. More importantly, the VAN–CC–PMMA, which possessed an excellent drug release character, showed no change in the compressive strength after drug release. This observation implies that the CC particles were probably not eluted out along with the drug.

Figure 4

Compressive strengths of various PMMA composites measured before and after the 42-day-release test. Significant differences between the tested groups are denoted with *, **, and *** for level of certainty of 0.05, 0.01, and 0.001, respectively.

Table 2

Maximum Stress of the 1 × 1 × 1 cm3 PMMA Cements Incorporated with Various Forms of Antibiotics

	maximum stresses (average ± SD, MPa)
incorporated antibiotics	before 42-day drug release	after 42-day drug release
none	135.5 ± 10.8
VAN	92.9 ± 24.1	76.2 ± 6.5
VAN–RG	112.7 ± 4.5	76.2 ± 6.5
VAN–CC	73.4 ± 17.2	70.0 ± 7.6
ERY	120.9 ± 1.9	61.7 ± 7.3
ERY–EC	72.2 ± 12.0	66.0 ± 5.6
ERY–PLGA	83.6 ± 0.6	68.2 ± 7.6

Adding unencapsulated ERY into the PMMA produced no significant difference of the compressive strength as compared to the PMMA control piece (no drug). However, among the three ERY-containing composites, only ERY–PMMA, the composite with the best ERY release character, showed a significant decrease in compressive strength when the drug was eluted. Our explanation is as follows. As mentioned above, the ERY particulates distributed in the polymer matrix were eroded during the drug release and the erosion of the outer drug particulates likely granted contact of release medium to the ERY situated inside the piece, thus allowing further erosion of more drug particulates. These erosions probably resulted in void spaces in the polymer composite, which weakened the mechanical strength of the cement piece. In contrast, when the drug carrier EC or PLGA was used, the release of the drug did not affect the mechanical strength of the materials. This implies that during the drug release, the carriers were probably retained in the composites, serving as the filler materials. It should be noted here that the minimum compressive strength for implanted acrylic cement is around 70 MPa.[26] After 42-day release, the VAN–CC–PMMA cement possessed a compressive strength of around 70 MPa, whereas that of ERY–PMMA was around 62 MPa. Nevertheless, it should be kept in mind that the amount of drug added into the PMMA could be adjusted as the concentration in the released media at day 42 was still much higher than the MIC values of the two drugs against S. aureus.

Conclusions

In order to make bone cement with prolonged drug release character, two points need to be considered, (1) ability of the cement to sustain the drug release and (2) ability of the cement to allow the drug located inside the cement to move out. Here, we have observed that the hydrophobic antibiotic ERY does not require any encapsulation prior to its incorporation into the PMMA cement to sustain the drug release and to mediate the drug inside the cement to move out. Unencapsulated ERY-doped PMMA cement showed that 85% of the drug molecules could steadily move out during the 42-day period with minimal burst at the beginning. In contrast, the cements incorporated with either ERY–PLGA or ERY–EC produced larger burst release during the first week and much lower drug concentrations at later days. Although the unencapsulated ERY possesses good release character from PMMA on its own, the hydrophilic VAN requires encapsulation into the right carriers prior to the incorporation into the cement. The PMMA incorporated with unencapsulated VAN or the VAN encapsulation into RGs showed burst release during the first 2–3 days, and only 18% of the incorporated drug could be released out from the cements during the 42-day period. Although the PMMA incorporated with the VAN that had been encapsulated in CC produced burst release at the beginning, VAN release at the concentration above the MIC value against S. aureus could still be obtained during the later days of the 42-day period because of the ability of the drug to not be trapped inside the PMMA cement matrix, that is, 70% of the incorporated drug was released from the cement during the 42-day period when VAN–CC was used. Further investigations are needed to clarify whether the discrepancy between the hydrophobic ERY and the hydrophilic VAN will also hold for other hydrophobic and hydrophilic drugs. Both the VAN–CC PMMA and the ERY–PMMA showed a similar drop in compressive strength as compared to the undoped PMMA.

Materials and Methods

Drug Encapsulation

VAN-Loaded RGs

VAN (400 mg, Hunan HuiBaiShi Biotechnology Co., Ltd, Hunan, China) was dissolved in water (20 mL), and RG (50 mg, Thai Wah Public Company Limited, Thailand) was added. This mixture was stirred at room temperature (30 °C) for 30 min and then at 83 °C for 3 h. The mixture was cooled and kept at room temperature overnight. The resulting suspension was centrifuged at 2350g for 30 min. The pelleted down VAN-loaded RGs (VAN–RGs) were freeze-dried. Encapsulation efficiency of the process [% EE = (weight of drug being encapsulated/weight of drug originally used) × 100] and drug loading content of the obtained particles [% loading = (weight of drug being encapsulated/weight of drug-loaded particles) × 100] were determined by quantifying VAN in the supernatant (corresponded to unencapsulated drug), using an ultraviolet spectrophotometer at 283 nm with the aid of the calibration standards.

VAN-Loaded CC Particles

VAN solution (200 mg in 2 mL of water) and calcium chloride (Merck KGaA, Darmstadt, Germany) solution (760 mg in 2 mL of water) were mixed and trisodium citrate (2020 mg in 6 mL) was added and the mixture was vortexed for 5 min. The mixture was then continuously shaken (in an incubator shaker) at room temperature for 24 h. Then, sodium hydroxide (3.31 mg) was added into the suspension to reduce the solubility of VAN[20] and shaking was continued for another 24 h. The resulting suspension was then left at 4 °C for 5 days. The resulting suspension was centrifuged at 9400g for 30 min. The pelleted down VAN-loaded CC (VAN–CC) particles were freeze-dried. Encapsulation efficiency of the process and drug loading content of the obtained particle were determined as described above.

ERY-Loaded EC Particles

First, the polymer–drug solution was prepared by dissolving EC (40 mg, viscosity 300 cP for the 5% in toluene/ethanol 80:20, 48% ethoxyl content, Sigma-Aldrich) and ERY (40 mg, Greenway Biotech Co., Ltd., Suzhou, China) in ethanol (16 mL). Then, water (64 mL) was slowly added dropwise at a controlled rate (1.13 mL/min), under constant stirring. After that, the suspension was stirred under vacuum to evaporate ethanol. The resulting suspension was centrifuged at 15900g for 45 min. The pelleted down ERY-loaded EC (ERY–EC) particles were freeze-dried. Encapsulation efficiency of the process and drug loading content of the obtained particle were determined by (1) evaporating the supernatant, (2) dissolving ERY in the residue with dichloromethane, (3) quantifying ERY in the obtained solution using UV–visible absorption spectrophotometry at the maximum absorption wavelength of ERY of 296 nm with the aid of calibration ERY standards prepared in dichloromethane, and (4) calculating for the %EE and %drug loading content as defined above.

ERY-Loaded PLGA Particles

ERY (40 mg) and PLGA (40 mg, ratio of lactic acid to glycolic acid of 50:50, Mw of 13 000–23 000, 87–89% hydrolyzed, Sigma-Aldrich) were added to dichloromethane (5 mL) and stirred. An aqueous solution of PVA [20 mL of water and 1.0 g of PVA (Mw 31 000–50 000, 87–89% hydrolyzed, Sigma-Aldrich)] was added and the mixture was sonicated for 1 min. This emulsion was stirred under vacuum to evaporate dichloromethane. The resulting suspension was centrifuged at 15 900g for 45 min. The pelleted down ERY-loaded PLGA (ERY–PLGA) particles were freeze-dried. Encapsulation efficiency of the process and drug loading content of the obtained particles were determined by (1) evaporating the supernatant, (2) dissolving ERY in the residue with ethanol, (3) quantifying ERY in the obtained ethanolic solution using UV–visible absorption spectrophotometry at the maximum absorption wavelength of ERY of 289 nm with the aid of calibration ERY standards prepared in ethanol, and (4) calculating for the % EE and % drug loading content as defined above.

All four drug-loaded particles were subjected to scanning electron microscopic (JSM-7610, JEOL, Tokyo, Japan), infrared spectroscopic (ATR-IR Nicolet 6700, Thermo Electron Corporation, Madison, WI, USA), and thermal gravimetric (Netzsch STA 449 F1, Germany) analyses.

PMMA Cement Casting

The PMMA cement was prepared using the PALACOS bone cement kit (Richards, Memphis, Tennessee, USA). For the undoped cement, PMMA powder (1.0 g) was mixed with the MMA liquid monomer (0.5 mL). The mixture was poured into silicone molds with 1 × 1 × 1 cm3 dimension and left at room temperature overnight to obtain the control PMMA cement pieces. PMMA composite pieces were prepared similarly, except that unencapsulated drug or encapsulated drug was first mixed with the PMMA powder. Six PMMA composites were prepared at the same drug content in the cement.

The drug-incorporated PMMA cements were prepared with unencapsulated drugs and various encapsulated drugs at the same fixed amount of final drug concentration. The dry drug powder or the dry drug-loaded particles were each mixed with the PMMA powder homogeneously. Amounts of drug or drug-loaded particles mixed with 1.0 g of PMMA powder were as follows: 55 mg of VAN, 55 mg of ERY, 62 mg of VAN–RG (containing 55 mg VAN), 408 mg of VAN–CC (containing 55 mg VAN), 110 mg of ERY–EC (containing 55 mg ERY), and 89 mg of ERY–PLGA (containing 55 mg of ERY). Then, each mixture was mixed with the MMA liquid monomer (0.5 mL). The reaction was exothermic and the hot mixture (around 70–100 °C) was poured into the silicone molds with 1 × 1 × 1 cm3 dimension and left at room temperature overnight to obtain the antibiotics-impregnated bone cement pieces.

The obtained cement pieces were subjected to SEM analysis, compressive strength measurement (UTM, LR10K, LLOYD Instrument, England), and drug release tests.

Drug Release

In a plastic tube with a screw cap, the tested cement piece was submerged in 20 mL of the release medium, which is PBS (pH 7.4, 0.01 M phosphate, 0.137 M NaCl, 0.002 M KCl). The closed tube was incubated at 37 °C with continuous shaking using a WNE 14 Water Bath with an M00 Shaker. Every day, each bone-spacer was removed from the tube and put into another tube containing 20 mL of fresh PBS. The experiment was carried out in triplicate for 6 weeks. The release medium in the previous tube was subjected to drug quantification using ultraviolet–visible spectrophotometry using the maximum absorption of VAN and ERY (in PBS) at 283 and 210 nm, respectively. Calibration standards were prepared in PBS.

Mechanical Properties

Measurement was carried out using the universal testing machine (UTM, LR10K, LLOYD Instrument, England). Protocol according to the ASTM standard was followed with modification on PMMA cement dimension.[26] The casted PMMA composites were dried in a desiccator overnight and analyzed using the compressor with a diameter of 6 mm and the compressive force of 10 kN at the constant cross-head speed of 20.0 mm/min. Measurement was stopped when the sample was fractured or passed the upper yield point.

Statistical Evaluation

All results were expressed as the mean value ± standard deviation (SD) of at least three independent samples. One-way analysis of variance (ANOVA) was performed to evaluate the differences between the tested groups.