Two-stage exchange is the standard in the treatment of PJI. High frequency of success has been reported with this strategy of management, with a rate of eradication of both total knee arthroplasty  and THA  infection of more than 80 percent. Although not evaluated in randomized controlled trials, the application of a temporary antibiotic-loaded bone cement spacer in the interim period before revision surgery is largely used in the two-stage exchange protocol because it enables preservation of the joint space after removal of the primary joint arthroplasty and ensures high local concentrations of antibiotics that can help to eradicate the infection. Despite the adoption of the two-stage exchange protocol, there are cases of persisting PJI where the cement spacer itself may act as a biomaterial surface that facilitates survival of microorganisms [3, 10]. Hence, it is of paramount importance to actively look for biofilm bacteria on the removed cement spacer to determine whether the infection has been cleared. Nevertheless, the limited sensitivity and specificity of standard microbiological culture techniques may limit their ability to detect the adherent bacteria responsible for PJI [11, 12]. The application of ultrasound before culturing may disrupt the biofilm from the surface of orthopaedic implants and free the bacteria, enhancing the sensitivity of traditional cultures [5, 6]. In the present study we used the sonication fluid culture of cement spacers to confirm the eradication of the infection obtained by a two-stage exchange protocol. We are aware of only one recent study that applied sonication culture to antibiotic-loaded spacers to identify bacterial growth on the cement spacer surface . This study reported a 14.5% rate of positive sonication fluid cultures of the removed spacer and showed that an infection of the cement spacer is associated with poor clinical outcome. The sonicate fluid culture of the removed spacer in our study was positive in six cases (29%). In three of these positive cases, the traditional culture of periprosthetic tissue failed to identify the pathogen. One notable difference with our results is that most sonication fluid cultures of the spacer in the above cited study were discordant with the culture of both the prosthesis (first-stage surgery) and peri-prosthetic tissue specimens (second-stage surgery). If a positive culture is obtained from the sonication process that reveals the same bacteria that caused the initial PJI as in the present study, it should be aggressively treated. The idea would be that in these cases a one stage exchange had been performed and a regimen of prolonged antibiotic therapy should be adopted.
Our results agree with the findings of other authors who reported MSSA and MRSA on gentamicin-and gentamicin-vancomycin-loaded cement beads, respectively, using traditional cultures [3, 14]. For patients in whom pathogens are isolated on the cement spacer the outcome should be guarded, particularly when the infection is sustained by highly resistant microorganisms. Indeed, the patient with MRSA in this study developed a recurrent infection even though a cement spacer containing vancomycin had been used. The emergence of resistance in bacterial strains is a concern about the long-term exposure to antibiotics at the site of a previous infection . In the case of MRSA infections, the prolonged implantation of gentamicin/vancomycin-loaded spacers may be associated with bacterial survival, and hence with the persistence of infection , particularly when an insufficient debridement has been carried out at the time of explantation of the primary prosthesis. Other studies have reported high failure rates after two-stage reimplantation when the infection is sustained by methicillin-resistant staphylococci [16–19]. In cases of PJI sustained by highly virulent microorganisms for which there is limited medical therapy, the arthrodesis or resection arthroplasty may be a viable option as the second stage surgery . We performed a knee arthrodesis as the second surgical stage in three patients who either failed to respond or were poorly compliant to the antibiotic therapy. This proved to have been a fair therapeutic choice because of the culture of both the sonication fluid and the periprosthetic tissue on samples collected at the time of the second surgical stage which showed persistence of the infection in these three patients.
Some reservations must be stated with regard to the interpretation of our results. One doubt emerging from the present study is the practical usefulness of a diagnostic test on the cement spacer carried out after the reimplantation has already been performed. Even though one previous in vivo study showed bacterial biofilm adhesion to orthopaedic metals in the first 48 hours after implantation , the reinfection of the reimplanted prosthesis should be regarded and treated as an early infection with a retention strategy, because bacterial adhesion to the new implant and the biofilm formation have not yet been fully realized. Indeed, our results and those of previous studies [22–24] have demonstrated that patients with a short-term infection with a known pathogen and a stable implant may be successfully treated by early debridement, with retention of the prosthesis and the use of a standardized regimen of antimicrobial treatment. The major limitation of this study is the small sample size, which implies low statistical power. Moreover, although the mean duration of implantation of the cement spacer between the two surgical procedures in a two-stage protocol is highly variable in the literature, in our study it was long and it could be questioned that the long time that had elapsed between the two surgical stages may have affected the bactericidal power of the cement spacer and caused the growth of bacteria on its surface . Actually, the ability of cement spacers to ensure long-lasting bactericidal levels of antibiotics is controversial. After one day, subinhibitory concentrations of antibiotics on Staphylococcus aureus strains from different cements have been detected in vitro and a declining trend over time in the antibiotic release from cement spacers has been reported in vivo[26, 27]. Conversely, other in vivo studies have shown that antibiotics released from the spacers result in long lasting bactericidal concentrations in the peri-prosthetic tissue [15, 28]. Individual variations in the local perfusion, the different characteristics of the bone cement (porosity, roughness, total surface area), and the method and doses adopted to prepare the antibiotic-impregnated cement may lead to a variable in vivo concentration and bioactivity of the antibiotics released from the cement spacers and help to explain the discrepancy between the published data [15, 27]. However, independent of the bactericidal power of the cement spacer, positive clinical results have been reported with long cycles of combined oral antibiotics plus a delayed reimplantation in the two-stage exchange protocol, even in infections sustained by highly resistant bacteria .