Double-layered antibiotic-loaded cement spacer as a novel alternative for managing periprosthetic joint infection: an in vitro study

Periprosthetic joint infection (PJI) is one of the most serious postoperative complications of total joint arthroplasty, occurring in 1–2% of joint replacement surgeries [1‐3]. Patients should be considered for debridement and retention of the prosthesis when PJI occurs within approximately 30 days of prosthesis implantation or fewer than 3 weeks of the onset of infectious symptoms. In contrast, patients who do not meet these criteria are usually treated using a two-stage exchange strategy that includes placement of an antibiotic-loaded cement spacer after removal of the prosthesis and thorough debridement [4]. Use of an antibiotic-loaded cement spacer is considered the standard procedure for local delivery of antibiotics. Staphylococcus aureus is the most common pathogen in PJI [5], with a ratio of methicillin-resistant S. aureus (MRSA) of approximately 50%. Despite its decreasing percentage in recent years [6], a spacer with vancomycin (VCM) (1 to 4 g per 40-g package of cement) is recommended for PJI [7]. Previous studies have reported that antibiotic-loaded calcium phosphate cement (CPC) can release a large amount of antibiotics over a long period [8, 9] and is effective for the treatment of PJI [9]. We previously used polymethylmethacrylate (PMMA)-only spacers for treatment of PJI, but because CPC was reported to have more sustained release of VCM than PMMA, we used CPC-only spacers for better treatment of PJI. However, spacers composed of only CPC are mechanically weak. We experienced breakage of the spacer during the waiting period, and we experienced difficulty removing scattered fragments of fractured CPC. Therefore, the double-layered, PMMA-coated CPC spacer (D-L spacer) was developed to achieve high-concentration and long-lasting elution of the CPC spacer with the mechanical strength of the PMMA spacer. The D-L spacer has been used in our hospital since 2008 [10]. It comprises a CPC core coated with PMMA and a PMMA stem (Fig. 1a, b). When CPC is encapsulated with PMMA, antibiotics are not released from CPC. Therefore, we attempted to drill the surface of PMMA. However, biomechanical stability decreases when holes are drilled. Fortunately, we have not experienced D-L spacer head fractures during clinical use. Therefore, we believe that this spacer might be advantageous in contrast to PMMA and CPC alone because it comprises a combination of the individual strengths of these materials. However, the strength and antibiotic elution of the D-L spacer is not known. This study aimed to compare antibiotic release, bioactivity, and mechanical strength of D-L and PMMA spacers.

D-L spacers used in the clinical setting are not spherical, and those are with stems. Spherical spacers without stems may have different mechanical strengths compared to spacers with stems. However, the transition from the head to the stem of D-L spacer used in the clinical setting was used of only PMMA. Therefore, we considered that past data regarding strength of the transition from the head to the stem of PMMA spacers [11] can be applied to D-L spacers. In this study, we only used spherical spacer models without stems to compare the amount of VCM elution and mechanical strength of the two types of spacers.

The D-L spacer was developed to achieve high-concentration and long-term elution of antibiotics. We made the following important observations regarding the D-L spacer in this study: (1) it released more antibiotics, (2) it was able to maintain bioactivity against MRSA, and (3) it was able to maintain high mechanical strength.

Regarding our first observation, we found that VCM concentrations of D-L spacer eluents significantly exceeded those of PMMA spacer eluents. This is consistent with the results of previous studies that showed that although PMMA allows only partial release of loaded antibiotics [15, 16], CPC allows the most loaded antibiotics to be released [8, 17, 18].

Sasaki et al. reported that the VCM elution volume from CPC on day 7 was 62.6 times that from PMMA, but that the rate thereafter decreased to 6.7 times on the day 13 [9]. However, in this study, the VCM elution volume from the D-L spacer on day 7 was 2.3 times that from the PMMA spacer, but the rate continued to increase to 11.9 times until day 84. This indicated that VCM could be released slowly and over a long period by covering CPC with PMMA. It has been reported that this antibiotic release is initially controlled to some extent by surface phenomena, whereas long-term elution depends on penetration depth, as determined by the bulk porosity of the cement used [19]. As such, we drilled multiple holes into CPC cores from the spacer surface to further facilitate VCM elution. Cement porosity, and thus the elution of antibiotics, can be increased with non-antibiotic fillers such as xylitol [20, 21]. However, fillers may degrade the mechanical properties of spacers, and the ideal amount of filler has not been established.

We also found that both D-L spacers and PMMA spacers maintained bioactivity throughout our study. Before the test, we expected that polymerization in these materials might have affected VCM activity. Specifically, because CPC does not generate the substantial amount of heat that is observed with PMMA polymerization, it does not cause the heat-induced antibiotic denaturation that may occur. It has also been reported that some organic solvents affect the stability of VCM complexes and peptide ligands [22], meaning the organic solvent contained in PMMA may have had some effect on VCM bioactivity. Considering these points, we expected to observe decreased bioactivity in the PMMA spacer. In contrast, inactivation or denaturation of loaded antibiotics, which was indicated by the presence of different peaks in HPLC analysis, was not observed. However, eluted antibiotics that show the same peak as the original loaded antibiotic do not necessarily have the same bioactivity. Therefore, we compared HPLC and bioactivity test results and found that they were comparable, and antibiotic inactivation/denaturation could be ruled out in our study.

VCM bioactivity can be affected by other factors not replicated in our study. First, biofilms that are formed by bacteria on implant surfaces can prevent the penetration of antibiotics. Nishimura et al. reported that even though antibiotics are effective against planktonic bacteria, the minimum bactericidal concentrations for biofilm bacteria of all antibiotics are high [23]. In addition, it has been reported that the protein non-binding rate of VCM in serum is only approximately 45–50% [24]. Although we could not obtain data regarding its protein-binding rate in joint fluid, the results of these studies suggest that the effect of VCM might be lost earlier in the living body than it was in our study.

It has also been reported that the MIC of VCM against MRSA is gradually increasing in several countries [25‐27], which is a phenomenon known as MIC creep. We found that the minimum VCM bioactivity in PMMA barely exceeded 2 μg/mL until day 28, and it was lost by at least day 56. However, the minimum VCM bioactivity in the D-L spacer was higher than 2 μg/mL until at least day 84. If MIC creep with MRSA is considered (i.e., higher methicillin resistance levels, such as MIC of 2 μg/mL), then VCM concentrations in PMMA spacer eluents would only be effective until day 28 and are not guaranteed afterward; however, those in D-L spacer eluents would remain effective until at least day 84. This demonstrated the unreliability of the PMMA spacer and the possibility that it may be disadvantageous, especially when used for longer durations.

Finally, we found that the D-L spacer retained high mechanical strength throughout the study period, confirming our hypothesis that coating the mechanically weak CPC [28] with the stronger PMMA would result in a spacer with better mechanical properties than only CPC. This is important because the mechanical properties of cement are a primary clinical problem. Patients who are scheduled to undergo two-exchange arthroplasty for PJI are instructed to not place mechanical loads on the spacer-containing extremity while the spacer remains in the body (usually for 2 weeks to several months). However, it is possible that loads may be unconsciously placed on the affected extremity. When the patient stumbles, there could be a hip joint force of 7.2 times the body weight [29]. Although we found that maximum compressive loads were significantly lower for D-L spacers than for PMMA spacers, even the weakest spacer could withstand 5.64 kN, or a load of 575 kg. This is the value that can withstand the hip joint force when an 80-kg patient stumbles. However, even if the spherical D-L spacer can withstand an 80-kg person stumbling, there is no guarantee that the strength of the D-L spacer will be sufficient when applied clinically. Future studies are necessary.

Despite the insights provided by this study, there were some limitations. First, because we changed the solution around the spacers every day, the experimental fluid dynamics differed from actual synovial fluid dynamics in the living body. Therefore, the possibility that the dissolution kinetics in our study differed from that in a real-life environment is high. Dissolution kinetics also vary with the amount of solvent present, which we did not consider during our comparison of D-L and PMMA spacers. Furthermore, because the spacer models used in this study were smaller than spacers used clinically, it is expected that the elution amount and period are different. Second, D-L spacers that are actually used in the clinical setting are not spherical; they also have a stem. Although cement spacer fractures sometimes occur, these tend to be localized mainly in the spacer stem and neck [30], and not on the spacer head. However, in this study, we have only studied the biomechanical properties of spherical femoral head spacer models without a stem. Third, it has been reported that the amount of antibiotics, the type of cement, and the method used to create the cement are correlated with the strength of the cement [31]. Drilling several large holes can have a negative impact because these holes can act as stress concentration sites under loading. However, this was not investigated in our study. Fourth, we tested load failure once. We did not investigate the influence of repeated biomechanical loading as it happens during the clinical application. Therefore, future studies to further verify spacer strength are necessary.

We compared D-L and PMMA spacer properties in this study and found that the D-L spacer was superior to the PMMA spacer in terms of elution volume and maintenance of VCM bioactivity. The D-L spacer also maintained high mechanical strength during our study duration. Therefore, we concluded that due to the beneficial biomechanical and drug-eluting properties of the D-L spacer, it might be a promising biomaterial that could potentially be used for managing PJI.