Introduction

Debridement and prosthesis retention, combined with a prolonged antibiotic regimen including rifampicin, is an accepted therapeutic approach when the duration of symptoms is less than 4 weeks and there are no radiological signs of loosening. The outcome of patients managed with this strategy has been previously assessed in several articles with success rates between 60% and 90% [1–6]. The best results have been reported when rifampicin was associated with fluoroquinolones [4, 5]; however, the rate of multi-resistant staphylococci, including fluoroquinolones, is high and therefore, oral antibiotic alternatives are necessary. Linezolid has a 100% oral bioavailability and reaches high concentrations in musculoskeletal tissues (skin, synovial fluid and bone) [7–9]; therefore, it is an attractive oral alternative and some data from experimental foreign-body infection model showed good results [10]. Recently, two studies performed in healthy volunteers have analyzed the interaction between linezolid and rifampicin after 3 days of combined therapy [11, 12]. Both articles support the interaction and found a reduction of about 30% in the area under the concentration–time curve (AUC) of linezolid. In addition, 2 cases of orthopedic implant infections where this combination was associated with low linezolid serum concentrations and clinical failure have been described [13]. However, the clinical experience with this combination is still scarce.

The aim of the present study was to retrospectively review the clinical experience with linezolid in 3 different hospitals from Spain and France in a particular group of patients with a prosthetic joint infection (PJI), who underwent open debridement with retention of the implant, whilst being treated with linezolid with or without rifampicin.

Methods

Study Design

A retrospective observational study was performed in 3 hospitals from Barcelona, Tours and Lille between 2005 and 2011. All patients included had an acute PJI, were treated with an open debridement with implant retention and received linezolid for more than 7 days. Relevant information about demographics, co-morbidity, type of implant, surgical treatment, microorganism isolated, antimicrobial therapy, adverse events (AEs) and outcomes were recorded. Linezolid dose was 600 mg/12 h. When rifampicin was added, the dose varied from 600 mg/24 h to 10 mg/kg/12 (not exceeding 900 mg/12 h). In case of polymicrobial infection, ciprofloxacin or a β-lactam were added according to the Gram-negative antibiogram.

Compliance with Ethics Guidelines

This study was approved by the Ethics Committee of our institution.

This article does not involve any new studies of human or animal subjects performed by any of the authors.

Definitions

PJIs were defined by the presence of local inflammation, macroscopic evidence of extension of the infection through the capsule during open debridement, and isolation of significant microorganisms from deep samples. In the case of coagulase-negative staphylococci, ≥2 positive deep samples were required for considering this microorganism a true pathogen. The infection was considered acute when it occurred within 4 weeks after the arthroplasty (early post-operative infection) or after 4 weeks with acute onset of symptoms (late acute infection).

Hematological toxicity was defined as a >2 g/L decrease in the basal hemoglobin concentration without another plausible explanation.

Outcome was classified according to the following definitions: (1) remission, when the patient had no symptoms of infection, the C-reactive protein (CRP) was <1 mg/dl and the prosthesis was retained after at least 1 year of follow-up; or (2) failure, when inflammatory signs and high CRP reappear during or after treatment. Failure was divided into relapsed or new infection according to the isolated microorganism. If the isolated microorganism was the same it was considered as relapsed, and when the microorganism was different, it was considered as reinfection. It was not considered failure when the patient developed an aseptic loosening that required the prosthesis to be exchanged and deep samples taken during surgery were negative.

Statistical Analysis

Categorical variables were described as percentage and continuous variables as median and interquartile range (IQR). Categorical variables were compared by Chi-square test or Fisher’s exact test when necessary and continuous variables by Mann–Whitney U test. The Kaplan–Meier survival method was used to estimate the cumulative probability of being in remission in the last visit in those patients receiving or not receiving rifampicin. The Log-Rank test was applied to evaluate the influence of rifampicin. Statistical significance was defined as a two-tailed P < 0.05. The analysis was performed using SPSS, version 20.0 (SPSS, Inc., Chicago, IL, USA).

Results

A total of 39 patients were retrospectively reviewed. The mean age (SD) was 70.5 (8.8) years, 21 were females (54%) and 9 patients had diabetes mellitus (23%). There were 25 (64%) knee prostheses, 13 (33%) hips and 1 shoulder (3%). Only 4 (10%) were late acute infections. The median (IQR) days from arthroplasty to infection diagnosis was 17 (19–48) and 33 (85%) cases were diagnosed within the first 60 days. Infections were monomicrobial in 24 (62%) cases and polymicrobial in 15 (38%), and the isolated microorganisms are described in Table 1. The median (IQR) number of days on linezolid treatment was 44.5 (30–81) and the median (IQR) duration of all antibiotic treatment was 70.5 (34–96) days, including treatment for microorganisms not covered by linezolid in polymicrobial infections. AEs were observed in 15 patients (38%), with gastrointestinal complaints (nausea, vomiting or diarrhea) in 10 cases and hematological toxicity in 5 cases the most frequent. There were 11 failures (28%) including 8 (21%) relapses and 3 new infections (8%). Therefore, 28 patients (72%) were in remission after a median (IQR) follow-up of 2.5 (1.8–3.6) years from stop** antibiotic treatment.

Table 1 Isolated microorganisms in 39 cases of prosthetic joint infection
Full size table

In 22 patients (56%) rifampicin was given concomitantly with linezolid and 17 did not receive rifampicin because of resistance or drug interaction (i.e., acenocoumarol). The characteristics of patients according to whether or not they received rifampicin are shown in Table 2. Although no difference between both groups was statistically significant, patients receiving rifampicin had a higher rate of diabetes mellitus (27% vs. 18%), a longer duration of symptoms before open debridement (9 vs. 2 days), and all MRSA infections were recorded in the rifampicin group (5 vs. 0). The remission rate was lower in the rifampicin group (64% vs. 82%, P = 0.28) due to a higher relapse rate (27% vs. 12%). There were 9 infections due to Staphylococcus aureus, 8 cases (including the 5 MRSA infections in the rifampicin group) were considered in remission (89%) and 1 patient had a new infection. In contrast, 15 out of 26 infections were due to coagulase-negative staphylococci.

Table 2 Characteristics of patients receiving or not rifampicin concomitantly with linezolid
Full size table

CoNS were in remission (58%) and 7 (27%) were due to infection relapse. The cumulative probability of being in remission in the last visit in patients receiving or not rifampicin is shown in Fig. 1 (Log-Rank test, P = 0.25). There were no differences in the total number of AEs between both groups; however, gastrointestinal complains were more frequent in the rifampicin group (32% vs. 18%) while hematological toxicity was more frequent in the group without rifampicin (24% vs. 5%).

Fig. 1
figure 1

The cumulative probability of being in remission according to whether the patient received concomitant rifampicin or not (Log-Rank test, P = 0.25)

Full size image

Discussion

An alternative agent for treating PJIs due to fluoroquinolone-resistant staphylococci is necessary [14]. In the present study, acute PJIs were managed with debridement, retention of the implant and linezolid with a remission rate of 72% and when considering only relapses (isolation of the same species), it was 80%. These results are similar to those presented by Bassetti et al. [15] using the same surgical strategy and linezolid alone in 20 PJIs with a remission rate of 80% and 20% of relapsing infections. Monotherapy with linezolid was also evaluated by Rao et al. [16] in 11 cases with a remission rate of 95%. Although the experience is limited, these results are in contrast to the 23% remission rate described using intravenous vancomycin in MRSA PJI treated with retention of the implant [17] and it suggests that linezolid could be an alternative for infections due to multi-resistant staphylococci.

The addition of rifampicin to linezolid would be reasonable [18, 19], particularly when the foreign-body is not removed, due to the potent activity of rifampicin against biofilm bacteria [4, 20]. It has been demonstrated that rifampicin reduces about 30% the AUC of linezolid [11, 12]; however, the clinical implication of this interaction is not well established. This combination was assessed in a retrospective study that reviewed 28 osteomyelitis and orthopedic implant infections [21]. The success rate was 89.2%, however, only 4 cases were managed without removing the implant. In contrast, Gomez et al. [22] showed a success rate of 69% but, in this series, all patients were managed with implant retention and rifampicin. In our cohort, no statistically significant difference was observed in the success rate between those patients receiving or not receiving rifampicin but slightly worse results among those receiving rifampicin were observed. This finding could be explained, at least in part, because these patients had a higher rate of diabetes mellitus (32% vs. 18%), and a longer duration of symptoms before open debridement (9 days vs. 2 days), variables that have been associated with a higher failure rates [23]. MRSA infections were all in the group of rifampicin and all achieved remission; therefore, this difference cannot explain the difference between the 2 groups. In addition, it is not possible to rule out a low linezolid concentration in the rifampicin group as an additional explanation. Linezolid is a time-dependent antibiotic [24]; therefore, the pharmacodynamic target is to maintain a trough serum concentration around 2 times over the minimum inhibitory concentration (MIC). Since the MIC90 for Gram-positive staphylococci is 2 mg/L [25], the optimal trough level will be 4 mg/L, a concentration that also it has been associated with low risk of toxicity [26], which is the major concern when linezolid is administered for prolonged time. These results suggest that monitoring trough serum concentration could be useful for improving the outcome, most especially when linezolid is combined with rifampicin, and for avoiding toxicity in patients that require prolonged treatment [27]. Indeed, hematological toxicity was more frequent in the monotherapy group (24% vs. 5%) probably due to the higher linezolid concentrations.

The main drawbacks of this study are the low number of patients, the retrospective design, that clonal relationship between microorganism isolated in primary and relapse episodes was not performed in order to confirm the relapse rate and the fact that linezolid concentrations were not measured; however, the information reported is useful to improve the results in PJIs due to resistant staphylococci.

Conclusion

Acute PJIs managed with debridement and retention of the implant linezolid, with or without rifampicin, are associated with a high remission rate and this is therefore an alternative therapy for infections due to fluoroquinolone and/or rifampicin-resistant staphylococci. However, prolonged linezolid may have important AEs that require close monitoring by infectious diseases physicians.