Impact of rifampicin dose in bone and joint prosthetic device infections due to Staphylococcus spp: a retrospective single-center study in France

Background Prosthetic joint infections (PJI) are a major cause of morbidity and mortality burden worldwide. While surgical management is well defined, rifampicin (RIF) dose remains controversial. The aim of our study was to determine whether Rifampicin dose impact infection outcomes in PJI due to Staphylococcus spp. Methods single-center retrospective study including 411 patients with PJI due to Rifampicin-sensitive Staphylococcus spp. Rifampicine dose was categorized as follow: < 10 mg/kg/day, 10–20 mg/kg/day or > 20 mg/kg/day. The primary endpoint was patient recovery, defined as being free of infection during 12 months after the end of the initial antibiotic course. Results 321 (78%) received RIF for the full antibiotic course. RIF dose didn’t affect patients recovery rate with 67, 76 and 69% in the < 10, 10–20 and > 20 mg/kg/day groups, respectively (p = 0.083). In univariate analysis, recovery rate was significantly associated with gender (p = 0.012) but not to RIF dose, or Staphylococcus phenotype (aureus or coagulase-negative). In multivariate analysis, age (p = 0.01) and treatment duration (p <  0.01) were significantly associated with recovery rate. Conclusion These data suggest that lower doses of RIF are as efficient and safe as the recommended high-dose French regimen in the treatment of PJI.

Without clear data on outcomes according to RIF dose, clinicians are often guided by experiences of poor tolerance of high RIF dose [11,16]. However, up today all pharmacokinetics studies have failed to demonstrate a correlation between RIF serum concentrations and occurrence of adverse events (AEs) [16,17]. The rate of AEs linked to RIF in PJI varies between 4.3 and 31.2% depending on the study [12,[18][19][20], often attributed to variable inter-individual susceptibilities [21,22]. Because some studies, although with low sample sizes, suggest that lowerdose RIF could remain effective while improving the drug's tolerance [12,20], tendency of low-dose RIF has become common without guidelines changes.
The aim of our study was to compare recovery rate in PJI due to Staphylococcus spp. between low-dose (< 10 mg/kg/day), intermediate dose (10-20 mg/kg/day) and high-dose (> 20 mg/kg/day) RIF. We also aimed to compare AEs occurrence in both groups.

Settings
This retrospective monocentric study was conducted in the Orthopedic Surgery Unit of a University Hospital in Amiens, France between January 2008 and December 2018. Our common practice in Staphylococci PJI is immediate surgery, and probabilistic IV therapy. Then after 5-7 days, oral switch to RIF plus another antibiotic is the rule. During surgery, local antimicrobials in bone cement (containing gentamicin) can be used in hip and knee infections.
Medical records were screened with the assistance of the institution's Medical Information Department. We recorded data using the Electronic Medical Records system (EMR).
Inclusion criteria were as follows: adults over 18 years of age, hospitalized for a PJI due to RIF-sensitive Staphylococcus spp., treated by antibiotics regimen which included RIF for the all course of treatment.
We excluded patients with simultaneous fungal or mycobacterial infection, patients with missing data for the primary endpoint and patients with unknown RIF dose.   Clinical variables gathered included age, sex, weight, height, body-mass index (BMI), underlying comorbidities such as diabetes, active excessive alcohol consumption, immune deficiency factors (neoplasia in the current year, immunosuppressive treatment including systemic steroids (> 5 mg/kg/day), immunotherapy or chemotherapy), chronic kidney disease of grade IIIB or more and hepatic insufficiency. Surgical data included type of prosthetic device, type of surgery (one-step, two-step, arthrodesis …), infection site and microbiological data (bacterial species, number of positive samples). We classified infections as "early" or "late" when they occurred less than 3 months or more than 12 months after prosthetic device implantation, respectively [18]. We retrieved antibiotics course details including the duration (in days) and dosage of RIF, as well as the occurrence of AEs, their severity using the Common Terminology Criteria for Adverse Events (CTCAE) grades [23] and their consequences on patient management.
Staphylococcus aureus (SA) PJI were defined as the isolation of a SA strain (alone or not) in a surgical periprosthetic sampling. Coagulase negative Staphylococcus (CNS) PJI diagnosis required at least one positive intraoperative sample with compatible clinical characteristics, such as joint pain, fistula, local inflammation or presence of pus during surgery. Polymicrobial infections were defined by the co-identification of two or more bacterial species including Staphylococcus spp.
The primary endpoint was patient recovery, defined as being free of infection during 12 months after the end of the initial antibiotic course.
Secondary endpoints were adverse effects (AEs) and treatment failure (i.e recurrence, relapse, loss to follow-up and death). Recurrence was defined as a new PJI due to the same germ(s) as the initial infection, less than 12 months after the end of the antibiotic course. Relapse was defined as the occurrence of a new PJI due to different germ than initially, in the same interval. When 12 months clinical follow-up data were not available, patients were deemed lost to follow-up. Finally, death occurring * One or more criteria among the following: active neoplasia dating less than 1 year; immunosuppressive treatment including systemic steroids, chemotherapy or immunomodulatory drugs ** Shoulder, humerus, elbow, forearm, hand

Results
From January 2008 until December 2018, 837 patients were treated for PJI due to Staphylococcus spp. (Fig. 1) Fig. 1). However only 321 patients (78%) received RIF for the full length of treatment, while in 90 patients (22%) it was discontinued early because of AEs. The characteristics of the patients are detailed in Table 1. Recovery rate was not statistically different between different groups of RIF dose: 55,6%, 76,2 and 67.3% in low-dose, intermediate-dose and high-dose groups respectively (p = 0.083) ( Table 2). Results were similar while grouping low and intermediate dose-groups: the recovery rate was 74.8% in the < 20 mg/kg/day group and 67.3% in the > 20 mg/kg/day group. This was true when including patients who discontinued RIF before the end of treatment: 62.6% of recovery rate in the < 20 mg/kg/day group, and 62.9% in the > 20 mg/kg/day group (p = 0.310).
In a univariate analysis, being a female was the only variable associated with a lower recovery rate (p = 0.012), no association was found between recovery and RIF dosage or Staphylococcus phenotype (SA or CNS) or the number of germs on intraoperative sampling (Table 3). In a multivariate analysis, older age (p = 0.01) and shorter treatment duration (< 60 days, p < 0.01) were significantly associated with treatment failure (Table 4).
Seventy percent of PJI were due to SA (n = 289) with 82.4% of Meticillin-sensitive SA (MSSA), while only 54.9% (n = 67) of CNS were Meticillin-sensitive. The most frequently identified CNS species was Staphylococcus epidermidis (67.8%). Infections were polymicrobial in 26% of cases (n = 107). Recovery rate was not statistically different according to staphylococcal resistance profile or in polymicrobial infections.

Discussion
In our study, there was no difference in recovery rate between low-dose and high-dose RIF containing regimens for PJI due to Staphylococcus spp., both in patients who continued treatment for the recommended duration (p = 0.083) and in those who discontinued RIF early (p = 0.31). These findings are in line with previously published data in favor of using lower doses of RIF, without risking treatment failure [12,20]. However, the absence of statistical difference could be attributed to the low power of our study (1-β = 0.221, α risk = 0.05), despite our relatively large sample size. Treatment failure was significantly associated with duration of RIF treatment (< 60 days), but not with the dose. This finding is consistent with DATIPO study results, which found a higher risk of treatment failure with shorter courses of antibiotics (< 45 days) [24]. Another randomized multi-center clinical trial also demonstrated the non-inferiority of a combined 8-week fluoroquinolone-RIF regimen, versus 3 or 6 months for hip or knee prostheses respectively, using debridement and implant retention [25].
The rate of AEs attributable to RIF was higher in our study population compared to previously published series, in which the percentage varies from 2 to 15% [10,13,26,27]. One could blame our relatively high RIF dose compared to these studies where doses ranged from 600 mg [28][29][30] to 900 mg daily [26,31] regardless of patient weight. However AE occurrence was the highest in the 10-20 mg/kg/day group and the lowest in the > 20 mg/ kg/day (5.2%).Several studies have similarly reported that high dose of RIF (> 30/mg/kg/day) could be used safely over a long period for the treatment of tuberculosis [32,33]. Conversely, Hagihara et al. suggested that the toxicity associated with prolonged combined antibiotics therapy could be limited by lowering RIF dose [34] [18]. On the other hand, Randuineau et al. more recently suggested a two-fold increase in AEs in patients treated with RIF at > 12 versus < 12 mg/kg/day, leading to more frequent treatment discontinuation but without any significant reduction in recovery [20]. The poor tolerance of RIF could thus be solely attributable to inter-individual susceptibilities. Yet, multiple studies failed to show any influence of sex, age, height or weight on pharmacokinetic parameters of RIF [22,35,36]. Furthermore, studies by Roblot et al. and Dupouey et al. did not demonstrate any association between RIF serum concentrations and the occurrence of AEs neither digestive nor else [16,17]. Besides, RIF usage should not be limited to non-comorbid patients, given the safety data of this study and previous ones [37]. While the value of pharmacological monitoring during RIF containing regimen remains unproven pharmacogenetics studies could be of help in the future [17]. Biological and clinical surveillance remain the pillar of PJI followup to prevent a potential harmful RIF discontinuation due to the use of less powerful antibiotics [11,19,27,[38][39][40].
In addition, we did not find lower recovery rates in patients with methicillin-resistant Staphylococci, in line with previous studies [28,41,42], or in patients with polymicrobial infections, in contrast to a study by Senneville et al. [19].
This study has some limitations. Missing data are inherent to the retrospective nature of our work, cutting off 102 patients from our initial sample. The low number of patients receiving low-dose RIF has probably contributed to our low study power.

Conclusion
In conclusion, our data show the feasibility of using lower doses of RIF than these currently advised by national guidelines, which is consistent with the results of several observational studies [12,20,28,29,31]. A national French randomized controlled trial (PHRC EVRIOS -NTC 02599493), programmed for 2021, aims to include 460 patients, with the goal of determining optimal RIF dosage (< 10 or > 20 mg/kg/day) in PJI due to Staphylococcus spp., and will probably definitely answer this question.