Risk of reinfection after two‐ or multiple‐stage knee revision surgery using superficial vancomycin coating and conventional spacers

Abstract This study investigates the effect of superficial vancomycin coating (SVC) in two‐ or more‐stage exchange procedures of prosthetic knee joint infections. We hypothesized that spacer treatment with SVC result in lower reinfection rates than conventional spacers after prosthetic reimplantation. Our secondary aim was to determine the demographic and treatment factors associated with reinfection rates. This retrospective cohort study compromised 96 cases with prosthetic knee infections. Twenty‐four cases were treated with a temporary SVC spacer and 72 cases with conventional spacers. Prosthetic reinfection occurred after a median observation period of 1.7 ± 4.0 years in 24 cases (25%). The prevalence of having a reinfection was not significantly different between the two treatment groups (13% [3 cases] in the SVC group vs. 29% [21 cases] in the conventional spacer group [p = .104]). In seven cases (7.3%), two in the SVC group (8.3%) and five (6.9%) in the conventional spacer group (p ≥ .999), histological, respectively microbiological evaluations from the intraoperative specimens revealed persistent infection at the second stage. Nevertheless, in all seven cases no significant higher risk of periprosthetic reinfection was observed during follow‐up (p = .750). Our secondary investigation of cofactors revealed that spacers additionally stabilized by nails were significantly associated with a 3.9‐fold higher hazard ratio of sustaining a reinfection of revision prosthesis (p = .005).

antibiotic-loaded bone cement (ALBC) spacer. The ALBC is supplied as a two-component system. The two-components, one powder (polymethylmethacrylate [PMMA]) and one liquid monomer (methylmethacrylate) are mixed and polymerization is started by an activator (dimethyl-para-toluidine). 11 Usually, the spacers are augmented with glycopeptides or in combination with aminoglycosides and act as a local antibiotic delivery device during the following 2-8 weeks. [12][13][14][15] One major complication during the spacer interval is a persistent infection, which may lead to a spacer exchange and a multiple-stage procedure.
To reduce the risk of a persistent infection, a combination of aminoglycoside antibiotics like gentamicin with the glycopeptide vancomycin may be used, showing synergetic effects. Hence, this antibiotic combination shows higher elution rates from PMMA spacers than when added in the same concentration alone. 16 Furthermore, to enhance the local antibiotic effect, a new surgical technique has been established by pressing 2 g of vancomycin powder manually onto the cement surface (superficial vancomycin coating [SVC]). 17 Augmented cement spacers release antibiotics from the spacer surface and in relation to their water absorption properties. 18,19 In a previous longitudinal case series, this technique resulted in favorable exceptional high local vancomycin concentrations without risks of systemic side effects. 17 Besides ototoxicity, nephrotoxicity was reported as a systemic side effect, especially when vancomycin is administered together with aminoglycosides. 20 Adverse effects have been indirectly described as an increase in serum creatinine of greater than 0.5 mg/dL or greater than 50% over baseline levels. 21 During the spacer period, another important function of the bone cement spacer is to prevent soft tissue retraction. High mechanical spacer stability is important to avoid complications, such as loosening, dislocation and fracturing until the revision prosthesis can be implanted at Stage 2.
In this follow-up study, we aimed to investigate the effect of SVCspacers in comparison to conventional spacers. Our primary aim was to investigate the reinfection rate after septic revision endoprosthesis in two-or multiple-staged procedures. We hypothesized that treatment of spacers with SVC will be of good compatibility and result in lower reinfection rates after reimplantation than conventional spacers. Our secondary aim was to determine further demographic and treatment factors associated with reinfection rates.  (Table 1) were not met at the time of implant removal (n = 13), reimplantation did not take place to prove or disprove persistent infection (n = 8), a mega prosthesis due to tumor resection was primarily implanted (n = 3), the patient was lost of follow-up (n = 3) or amputation of the limb was performed instead of prosthesis reimplantation (n = 1). Therefore N = 96 cases were included in the final analysis ( Figure 1). The demographics of the study population are displayed in Table 2.

| METHODS
In the SVC group (n = 24) the spacer implantation was performed using the new superficial vancomycin coating ( Figure 2). A compound of 80-120 g of Palacos R + G (Heraeus) was mixed with 1 g of vancomycin powder to every 40 g of bone cement to form an augmented bone cement according to the manufactural protocol. The cement was placed in the correct position to form a static spacer (Figure 2A,B). To prepare the SVC-spacer, two additional grams of vancomycin powder was pressed manually onto the surface of the bone cement ( Figure 2C-F).
After hardening a wound drain was inserted but clamped for 2 h postoperatively to avoid premature washout of vancomycin. The conventional spacer (n = 72) was prepared in the same way, but without adding 2 g of superficial vancomycin coating. In 19 cases, a nail was inserted additionally due to the surgeon's preference or for increased stabilization after massive bone loss ( Figure 3).
Daily vancomycin concentrations in wound drainage fluids and serum samples were analyzed until the drainage was removed. The analyses of the vancomycin levels were performed using a cobas c 311 Analyzer (Roche) with a detection limit of less than 2.0 μg/ml. Serum creatinine levels were analyzed preoperatively and postoperatively to indirectly detect potential nephrotoxicity. If the serum creatinine concentration increased bygreater than 0.5 mg/dl or greater than 50% above baseline level, a nephrotoxic effect was assumed. 21 All patients received systemic intravenous antibiotic therapy during the hospital stay, followed by oral antimicrobial treatment up to the time point of their reoperation at Stage 2. The antibiotic agent was selected according to the microorganism and susceptibility. In cases of inconclusive culture, empiric antibiotic therapy was given in

| RESULTS
The study population (N = 96) was on average 69 years old (range:

33-88) and almost equally distributed in 52 females and 44 males.
Diabetes was present in 29%, smoking in 6.3%, and immunodeficiency in 4.2%. A substantial number of patients had already undergone one or more revision surgeries in the past (70%). The isolated microorganisms in our study sample were mainly Staphylococci (26%) and Streptococci (11%). In 55% of cases, no specific pathogen could be detected preoperatively or intraoperatively at the time of spacer implantation ( Table 2).
The analysis showed a significant difference in age between the SVC and the conventional group (mean difference = 5.9 years; p = .014).
No other significant differences between the two treatment groups were identified in patient characteristics, operation, or microorganism (Table 2).  (Table 3). At the time of discharge (range: 6-22 days), five cases (5.4%) revealed a creatinine increase above 0.5 mg/dl-one in the SVC group (4.8%) and four in the conventional group (5.6%).
The spacer interval until reimplantation was on average 11 weeks (range: 1.9-55) with no significant differences between the treatment groups. In seven cases (7.3%), two in the SVC group (8.3%) and five (6.9%) in the conventional spacer group (p ≥ .999), histological, respectively microbiological analysis from the specimens revealed persistent infection at the second stage. In those cases, no significant higher risk of periprosthetic reinfection was observed during follow-up (p = .750). Spacer dislocation (n = 2) and peri-spacer fracture (n = 1) showed no statistically significant difference between the treatment groups ( Table 3).
The follow-up time of all reimplantations was in median 2.8 years (±3.4 years). The total observation period comprised 343 person-years.
In that period 24 reinfections (25%) occurred, leading to an incidence rate per person-year of 0.07. In the observed population, 25% showed a prosthetic reinfection within 4.6 years, 50% within 10 years, and 75% within 13 years (median reinfection period = 1.7 ± 4.0 years). The prevalence of having a reinfection was 13% (3 cases) in the SVC group and 29% (21 cases) in the conventional spacer group. However, these results did not reach any statistical significance (p = .104) as the observation period was significantly shorter in the SVC group (1.9 ± 3.4 years) compared to the conventional control group (3.0 ± 4.6; p = .040) ( Table 3). As a consequence, the analysis of the survival distribution revealed no significant difference between the SVC and conventional spacer group (Table 4, Figure 5).
In contrast, spacer with nails were associated with a higher reinfection rate of the revision prosthesis during follow-up (

| DISCUSSION
The aim of our study was to investigate the risk of reinfection after septic revision endoprosthesis in two-or multiple staged procedures, using SVC spacer during first stage. We hypothesized that treatment of spacers with SVC will be of good compatibility and result in lower reinfection rates after reimplantation than conventional spacers. We observed a trend towards a lower prevalence of reinfections when SVC spacers were used, however, this finding remained without any statistically significance. Our secondary aim was to determine demographic and treatment factors associated with reinfection rates. Our data showed, that using a nail for   ances are yet limited. To increase the power of detecting a statistical significance, multicenter, randomized controlled trials would be preferable for further investigations. Third, a very high rate of culture negative cases resulted due to poor intraoperative specimen collection in the past. Interpretations of culture specific treatment outcomes were therefore not available in the current study. Newer guidelines have been defined to increase the detection rate and may enable studies on local antibiotic specific targeting in future studies.

| CONCLUSION
In our retrospective cohort study, superficial vancomycin coating of spacers in a two-or multiple stage-exchange did achieve high local vancomycin concentrations without systemically side effects. A trend towards a lower prevalence of revision prosthetic reinfection with SVC spacers was observed but without statistical significance. Our data showed, that the rate of endoprosthesis reinfection was significantly higher when nails were used in combination with PMMA bone cement spacers.