Role of Rifampin for Treatment of Orthopedic Implant–Related Staphylococcal Infections :  A Randomized Controlled Trial FREE

ORTHOPEDIC DEVICES are used for bone fixation or joint replacement. Orthopedic device–related infections are rare, but they carry a high morbidity for the patient and are costly.^1- 6 Traditionally, the management of such infections includes resection arthroplasty or removal of fixation devices.^1,3- 4,7 Observational studies showed that despite prolonged (4-6 weeks) intravenous treatment with β-lactam antibiotics and subsequent long-term oral therapy, the failure rate with retention of the device is between 32% and 82%.^8- 16 Among the best results are those of Tsukayama et al,⁸ who reported a failure rate of only 13 (32%) of 41 devices in patients with early postoperative and hematogenous infection after hip arthroplasty. However, in this study the polyethylene insert of the acetabular component was replaced in all patients. High failure rates (69%-77%) were reported in 3 series.^10,14,16 In these studies, risk factors for failure were a long history of infection and a delayed debridement.

Osteomyelitis associated with fracture fixation devices occurs more frequently than infection after joint replacement.^{5- 6,17- 18} The incidence of infection after internal fixation of closed fractures should not exceed 1% to 2%, whereas the infection of open fractures can be higher than 30% depending on the type of fracture.^5- 6,19 The treatment of infected bone fixation devices usually requires device removal and stabilization with an external fixation device. The success rate of Staphylococcus aureus device–related infection with a quinolone was only 20% (1/5) despite treatment for 6 months.²⁰

Hitherto, there is not a single controlled, randomized clinical trial evaluating the value of the antibiotic treatment of orthopedic device–related infection. In most studies, only surgical procedures, not antimicrobial therapies, are described.^4- 18,21 Results from our animal model for implant-associated infection demonstrated the clear superiority of rifampin combinations.^22- 25 In addition, a prospective pilot study showed the success rate of rifampin combinations in orthopedic implant–associated staphylococcal infections to be 82% (9/11).²⁶ These results were confirmed in a larger series showing a success rate with ofloxacin plus rifampin of 62% (13/21) without removal of the device.²⁷

Our study question was to estimate the cure rate of a conservative approach with a controlled trial. Therefore, we conducted this double-blind, randomized clinical trial evaluating the role of rifampin in patients with a stable orthopedic implant infected with S aureus or coagulase-negative staphylococci.

Eligible for this study were patients who had a diagnosis of orthopedic device–related infection due to S aureus or coagulase-negative staphylococci established by arthrocentesis or surgical revision. Only subjects in whom the stable implant was kept in place were included. All consecutive patients treated by the infectious diseases consultants of the study centers were asked to participate in the study. The following exclusion criteria were applied: lack of written informed consent, symptoms for more than 1 year before randomization, age younger than 16 years, less than 2 years of expected survival, predictable inability to comply with the treatment and follow-up visits, known or suspected allergy to quinolones and/or rifampin, mixed infection with microorganisms other than staphylococci, staphylococci resistant to ciprofloxacin and/or rifampin, removal of the implant before randomization, clinical or radiological signs of implant loosening, refusal to discontinue wearing soft contact lenses during treatment period, refusal to discontinue hormonal contraception during treatment period, and an antimicrobial treatment of more than 2 weeks after the microbiologically established diagnosis. In addition, patients who took less than 85% of the study medication were excluded a posteriori (poor compliance) and were regarded as dropouts. Patients were randomly assigned to antimicrobial combination treatment (see below) either with rifampin or with placebo, using a computer-generated list distributed to the study centers in sealed envelopes. Patients were randomized by blocks of 4, stratified into groups of patients with knee protheses, hip protheses, or fixation devices. The study was approved by the local ethics committees of the University Hospitals, Basel and Geneva, Switzerland. Written informed consent was required for all study patients.

The patients were clinically assessed at enrollment in the study, at weeks 1 and 2, then once monthly to month 6, then at 9, 12, and 24 months, or until failure of the treatment or death of the patient. All patients including dropouts were scheduled for a final evaluation at the end of the study. Radiological evaluation was performed at study enrollment, after 1 and 2 years, or on removal of the device. During treatment, the following laboratory parameters were determined at least twice a month: C-reactive protein, hemoglobin, erythrocytes, platelets, differential white blood cell count, and liver enzymes.

At study entry, any type of revision surgery was encouraged except the removal of the device (exclusion criteria). Only cases with radiological (n=33) and intraoperative (n=29) evidence of stability of the implant or the prosthesis were included in the study. Fifteen cases in the rifampin group and 14 in the placebo group had revision surgery for infection. At the revision, implants were left in place (inclusion criteria). Thorough debridement was followed by suction irrigation drainage or drainage alone. In osteosynthesis cases, open-wound therapy was allowed as an alternative. There was no case where gentamicin beads were used.

During the initial 2 weeks, patients were treated with flucloxacillin (2 g every 6 hours intravenously) or, in case of methicillin resistance or an allergy to penicillin, with vancomycin (1 g every 12 hours intravenously) plus either rifampin (1 coated 450-mg tablet every 12 hours) or placebo (1 matched coated tablet every 12 hours). This initial 2-week course with a standard intravenous treatment was chosen to minimize the risk of emergence of ciprofloxacin resistance. An oral form of rifampin and an identical placebo was provided by Ciba-Geigy Ltd, Basel, Switzerland. Patients were informed that body fluids can turn orange with placebo or drug. After 2 weeks, flucloxacillin or vancomycin was replaced by ciprofloxacin (750 mg every 12 hours by mouth), whereas the rifampin or placebo was continued. Patients with hip prostheses and internal fixation devices were treated for 3 months, those with knee prostheses for 6 months. After this time, antimicrobial treatment had to be stopped if the patient had no clinical signs and symptoms of infection and the C-reactive protein level was below 5 mg/L for at least 6 weeks. The surgeon was encouraged to remove osteosynthesis material after stopping antimicrobial therapy for at least 1 week, provided that the material was no longer required for stability. The whole implant was sent to the microbiology laboratory. Identification of the same microorganism was considered as failure.

Cure was defined as the lack of clinical signs and symptoms of infection (fever, local pain, redness, warmth, sinus tract infection, fever), a C-reactive protein level below 5 mg/L, and the absence of radiological signs of loosening, pseudoarthrosis (in case of fixation device), or dislocation of the artificial joint at the final follow-up visit 24 months after start of the treatment. In case of the removal of the internal fixation device because of sufficient stability, cure was defined as the absence of the infecting agent of the cultured implant. For this purpose, the entire device was cultured in trypticase soy broth and sonicated in case of no growth after 48 hours of incubation. Sonication was delayed since gram-negative microorganisms could be killed by sonication and superinfection could therefore be missed. Confirmed failure was defined as isolation of the initial microorganism (persistence or relapse) in the culture of intraoperative tissue specimens, synovia, or device. A probable failure was defined as clinical signs and symptoms of local infection, or an otherwise unexplained high C-reactive protein level (>50 mg/L) without microbiological documentation of infection.

The isolates were identified by standard techniques.²⁸ Minimal inhibitory concentrations of the study drugs were determined by E-test (AB BIODISKA, Solna, Sweden) at the principal study center. All isolates were collected and stored in skim milk at −70°C for molecular typing in case of treatment failure. For this purpose, pairs of isolates were characterized by the contour-clamped homogenous electrical fields modification of pulsed-field gel electrophoresis, after digestion of chromosomal DNA with low-frequency cutting enzymes (SmaI and Eag I).²⁹

Cure rates of orthopedic device–related infections with the standard regimen range from 20% to 30% without removal of the device.^10,12,14,16 Experimental data and uncontrolled case series with rifampin combinations indicated a cure rate between 70% and 90%.^26- 27 The sample size was calculated with the following assumptions: α was set at .05, the power at 80%, the probability of cure with standard treatment at 20%, and the probability of cure with study treatment at 75%. A sample size of more than 30 subjects was calculated, with an estimated dropout rate of 20%.³⁰

Time to failure was estimated with the Kaplan-Meier method, and compared between groups by the log-rank test.³¹ Categorical variables were compared by the χ²test or the Fisher exact test. The independent safety monitor was the only one who was aware of the type of blinded study drug. A P value of <.05 (2-tailed) was considered significant.

Patients enrollment started in May 1992, and the last follow-up visit was performed in November 1997, or at the removal of the device if failure occurred. After randomization of 33 patients, the investigators were asked by the safety advisor to stop randomization because all failures occurred in the same group.

The 2 groups were similar in demographic characteristics, type of devices, and infecting agents (Table 1). In the rifampin combination group, 12 of the 18 infections occurred within 2 months after implantation of the device, compared with 7 of the 15 in the placebo combination group. However, the duration of infectious signs and symptoms was short and similar in both groups, ie, all infections occurred either early after intraoperative contamination (<2 months) or late as a consequence of hematogenous seeding (Table 1). Twenty-four patients fully completed the trial and 9 dropped out for various reasons but received further follow-up (see below).

Figure 1 shows the Kaplan-Meier plot of disease-free survival in the 24 patients who completed the study according to the protocol. The cure rate was 12 (100%) of 12 in the rifampin combination arm, and 7 (58%) of 12 in the placebo combination arm, with a median follow-up of 35 (range, 24-46 months) and 33 (range, 15-41 months) months, respectively. The definite proof for cure of a device-related infection is the negative broth culture of the whole explanted foreign body after antimicrobial therapy.^23- 24 This unambiguous test was performed in 8 of the 10 patients with fixation devices in the ciprofloxacin-rifampin group and in 2 of 6 patients in the ciprofloxacin-placebo group. In the former group, all 8 implant cultures were negative; in the latter group, 1 of 2 implant cultures showed growth of the initial pathogen. All 5 failures were microbiologically confirmed (see below); all had flucloxacillin as the initial intravenous therapy.

We also performed an intention-to-treat analysis. Sixteen (89%) of the 18 patients of the rifampin combination arm and 9 (60%) of 15 patients of the placebo combination arm were cured without removal of the implant before the end of the antimicrobial therapy (P=.10). In the 2 patients from the former group in whom treatment failed, rifampin therapy was stopped after 8 weeks, in 1 patient due to an exanthema; in the other patient, the dose of rifampin was reduced after 3 weeks because of nausea.

As expected for a long-term study, a considerable percentage of patients dropped out. All dropouts had an identical follow-up observation as did the treated patients. In the rifampin combination arm 6 of the 18 patients dropped out, and in the placebo combination arm 3 of the 15 (P=.45, Table 2). In the rifampin combination arm, in 3 patients rifampin therapy had to be temporarily discontinued due to severe nausea, but could be continued within a few days with a reduced dose (300 mg every 12 hours). In 2 other patients, rifampin therapy was definitely stopped due to an allergic exanthema. One patient dropped out due to protocol violation because the orthopedic surgeon did not agree to stop therapy at the required time point according to the study protocol.

In the placebo combination arm, reasons for dropping out were an adverse event (nausea), noncompliance, and a protocol violation. Seven of 9 dropout patients were subsequently treated withrifampin combinations, 3 of them with a reduced dose. The success rate among the rifampin-treated dropout patients of both groups together was 5 (71%) of 7 without removal of the device during antimicrobial therapy (Table 2).

Four methicillin-sensitive S aureus and 1 methicillin-sensitive S epidermidis isolates were cultured from the 5 failures in the ciprofloxacin-placebo group. The isolates at failure were compared with the initial isolates by susceptibility testing and molecular subtyping. In 4 of these isolates (3 S aureus and 1 S epidermidis), the minimal inhibitory concentration of ciprofloxacin increased 3- to 8-fold to 4 mg/L, and in 1 failure S aureus remained susceptible to ciprofloxacin. The analysis of the chromosomal DNA of the ciprofloxacin-resistant strains by pulsed-field gel electrophoresis showed identity of all bands, indicating emergence of quinolone resistance during prolonged ciprofloxacin monotherapy of the pathogen isolated at randomization (Figure 2).

We found a superiority of the rifampin combination regimen compared with the ciprofloxacin monotherapy. Foreign bodies are favorite sites for bacterial persistence due to a local host-defense defect.³² In addition, the biofilm and the low growth rate of surface-adherent microorganisms render many antimicrobial agents ineffective.^{23- 24,33- 34} Therefore, most orthopedic surgeons remove all foreign material in case of an infected arthroplasty.^1,3- 4,7 The superiority of rifampin in the animal model could be explained by its high efficacy on adherent and stationary-phase staphylococci.^23- 25 The excellent results of the rifampin combination in this study confirm previous data from animal models and observational studies.^22- 27 In addition, this combination has also been shown to be efficacious in right-sided endocarditis due to S aureus.^35- 36 However, there are no controlled studies on the role of rifampin in nonmycobacterial infection.

Our results confirm the high risk of emergence of resistance of staphylococci to quinolones when used as monotherapy.³⁷ Emergence of resistance to rifampin was not observed in a single case in our study. Recurrent infections were exclusively caused by the original strain, as confirmed by molecular typing of both the original and consecutive isolates. This indicates that bacteria may persist despite initial clinical response to antibiotic treatment. Therefore, surface adhering staphylococci seem to be highly resistant even to prolonged quinolone monotherapy, as previously suggested by animal experiments and observational studies.^20,23- 24 The present study shows that the rifampin-quinolone combination was highly efficacious, not only in eliminating device-associated staphylococci, but also in preventing the emergence of ciprofloxacin resistance.

Our results cannot be generalized to every type of orthopedic implant–associated infection. According to the inclusion criteria, all devices were stable. Loosening of the infected device precludes an antimicrobial therapy without removing or exchanging the implant.⁴ Despite the fact that we allowed the inclusion of patients with up to 1 year of infection, the median duration of signs and symptoms of infection was only 4 and 5 days, respectively, with a maximum of 21 days. According to Schoifet and Morrey,¹⁰ the long duration of the infection before debridement is the major cause of treatment failure. However, in these studies, the antibiotic therapy was not standardized and did not include rifampin. Nevertheless, it is conceivable that the treatment with retention is only successful in patients with a short interval before therapy. In our study, only patients with early (<2 months after surgery) or acute hematogenous infection were treated. Therefore, we can only speculate that patients with chronic orthopedic device–related infections can also be successfully treated with ciprofloxacin plus rifampin with retention of the device.

Dropouts were mainly observed in the combination treatment group. All dropout patients had an identical follow-up as the other cases. Seven dropout patients were subsequently treated with rifampin combinations, 3 of them with a reduced dose due to nausea as the main adverse event. Six of 7 rifampin-treated patients had a successful outcome, 5 of them without removal of the device—supporting the main conclusion. Compliance was excellent in both groups as checked by pill counting. Only 1 patient dropped out due to insufficient compliance (<85% of the study medication).

This is the first randomized, controlled clinical trial evaluating a conservative treatment approach to staphylococcal device–related infection with rifampin. It confirms the in vitro and experimental animal data,^23- 25 as well as the results of observational clinical studies.^26- 27 In conclusion, orthopedic device–related infections due to rifampin- and ciprofloxacin-susceptible staphylococci can be cured without removal of the device, given the implant is stable, the duration of infection is short, an initial debridement is performed, and the patient tolerates long-term therapy.