A 52-year-old man underwent right primary total knee arthroplasty for osteoarthritis at our hospital in March 2008. The patient's medical history included pulmonary tuberculosis treated and cured with medication 18 years ago, and hypertension being treated with medication. The patient had no medical history of immune suppression. On preoperative physical examination, soft tissue of the proximal and anterolateral compartments of the right lower leg showed atrophy with scar change compared to contralateral leg, which was related to the history of car accident injury requiring skin graft at the age of seven. Patella baja and osteoarthritic changes were seen on radiology evaluation. No abnormal findings were recorded at the time of the surgical procedure; the skin incision was performed on the medial side to maintain a distance from the skin scar site. However, four weeks after the total knee arthroplasty, the patient complained of pain with redness and local heat on the distal portion of the incision line in his right knee. The patient was admitted to the hospital for suspicious surgical wound infection. We regarded it as a superficial infection and did not aspirate the joint fluid at that time for fear of spreading infective microorganism into the knee joint. C-reactive protein (CRP) level was 0.54 mg/dl (normal: below 0.3). He was treated with intravenous antibiotic therapy with cefazolin 2 g three times a day and then ciprofloxacin was added as 400 mg intravenous every 12 hours after observing inadequate improvement in skin color and tenderness to initial cefazolin treatment. Blood culture result, which was the only possible microbiologic laboratory workup, showed no identifiable microorganism. After the 15 days course of antibiotic treatment, symptoms improved and the patient was discharged switching to oral antibiotic therapy with ciprofloxacin 1,000 mg/d and fusidate Na 1,000 mg/d with an intent to prevent any further deep site infection focusing on gram positive biofilm making microorganism. Total of 3 months course of oral antibiotic therapy was completed and the infection seemed to be cured.

A year after discharge the patient presented with aggravation of symptoms showing fluctuation of laboratory findings such as elevated CRP levels of 0.67 to 4.86 mg/dl. Joint fluid aspiration was done and synovial leukocyte count was 1,125/mm³, with differential count of 91% neutrophils. During follow-up period after the first arthrocentesis, joint fluid was aspirated on four occasions repeatedly for a microbiologic culture study. But the culture studies including bacterial, fungal and Mycobacteria tuberculosis were all negative and the polymerase chain reaction (PCR) for M. tuberculosis showed negative results. Antibiotics were prescribed intermittently (levofloxacin 200 mg/d), and his signs were closely monitored.

After 18 months from the total knee arthroplasty, despite the early attempt of eradication of probable superficial surgical infection with antibiotics, the patient developed a sinus tract in the right knee with purulent discharge. Repeated joint fluid analysis showed a synovial leukocyte count of 14,004/mm³ with 92% neutrophils. Culture studies showed negative results again. Under the impression of deep prosthesis infection, a two-stage reimplantation operation was recommended but the patient absolutely refused. Debridement and irrigation with soft tissue repair was planned as the patient hardly accepted the recommendation of medical staffs. The patient was hospitalized and underwent only surgical debridement of the sinus tract and empirical treatment with antibiotic therapy. Tissue and synovial fluid specimens on the operation were sent for cultures at the time of debridement. Both tissue and synovial fluid cultures from the operative specimen revealed A. xylosoxidans as the causative organism. The patient then received intravenous antibiotics (imipenem 500 mg every eight hours) according to the results of the antibiotic sensitivity testing. After treatment with imipenem, pain and wound discharge seemed to improve. However, three weeks later the symptoms were again aggravated; there was purulent discharge from the right knee joint. An X-ray showed loosening of the femoral component (Fig. 1). The patient was then scheduled for a two-stage revision of the infected total knee arthroplasty. Aspiration of the knee joint was performed before implant removal surgery and culture identified repeatedly the growth of A. xylosoxidans.

At the first-stage revision surgery for removal of implant, there was pus within the joint space and loosening of both the femoral and tibial prosthetic components. A thorough washout, debridement, and complete synovectomy were performed.

A temporary articulating knee spacer with antibiotic (vancomycin 1 g per batch) mixed cement was used in attempt to prevent further gram positive microorganism co-infection. After this, a five-week course of intravenous imipenem was started. The second-stage of the revision was performed seven weeks after the first stage with two weeks of observation without imipenem to confirm eradication of previous infectious microorganism. Clinically, the knee was cool, non-tender, and showed no external signs of infection. The CRP was decreasing and was 0.36 mg/dl (normal: below 0.3) before the second stage procedure. The microbiology and histology samples from the second-stage revision for new implant insertion demonstrated no growth on tissue and synovial fluid cultures including the bacterial, fungal and M. tuberculosis cultures; in addition, the M. tuberculosis PCR was negative. Twelve months after the second-stage surgery, the patient was clinically well and reported no pain in the knee. Physical examination revealed no signs of infection or instability during routine follow-up.

A. xylosoxidans is an aerobic, oxidase- and catalase-positive, non-lactose fermenting, gram-negative bacillus widely distributed in the natural environment.3) It is an opportunistic human pathogen capable of causing a variety of infections, including bacteremia, meningitis, pneumonia, peritonitis, endocarditis, and pyelonephritis. Those who are immune suppressed or have underlying diseases related to immune compromised status are especially vulnerable.5) Despite the high volume of arthroplasty procedures, periprosthetic joint infection due to A. xylosoxidans has rarely been reported. Review of the literature showed only one case report of septic arthritis caused by this organism after a total knee arthroplasty. The patient in that case had rheumatoid arthritis and was receiving high doses of prednisone, in which both conditions contributed to immune compromisation.4) There have been no previously reported cases of infection with A. xylosoxidans following total knee arthroplasty in patients without underlying immune compromised status or immune suppression.

In this case, four weeks after initial total knee arthroplasty, signs of infection developed in the right knee and antibiotic therapy was started under the impression of superficial wound infection. It is not evident yet that the joint fluid aspiration should be always performed under the suspicious superficial infection after prosthesis surgery, which is usually indistinguishable from a hidden deep sited infection. The physician should make a balance between the benefit and the risk of arthrocentesis under that condition. When the pain and redness reappeared after substantial period, 6 months from the initial response to the empirical antibiotic treatment, four repeated joint fluid aspiration procedures with cultures workup were performed but reported as negative results. In this case, because antibiotics were used prior to the joint aspiration, the synovial fluid culture might have been falsely negative. In addition, the value of microbiological cultures for the diagnosis of late periprosthetic infections is limited; standard methods may fail to detect biofilm-forming sessile or other fastidious slow growing bacteria.6)

A. strains are frequently resistant to aminoglycosides, ampicillin, first- and second-generation cephalosporins, and chloramphenicol. However, they typically respond to anti-Pseudomonas third-generation cephalosporins, carbapenems, and trimethoprim-sulfamethoxazole. Carbapenems are the most active antimicrobial drugs, especially meropenem (MIC₅₀, 0.25 µg/ml; 88% susceptible), which is four-times more potent than imipenem (MIC₅₀, 2 µg/ml; 84% susceptible).7) In this case, various empirical antibiotic therapy for the most common causative microorganism of prosthesis infection, gram positive cocci, were tried as a single therapy or in combination, including cefazolin, ciprofloxacin, fusidate Na, and rifampicin until the definite causative microorganism, A. xylosoxidans was identified. The eradication of infection was impossible at an early stage because of the lack of information about causative organism and antibiotic susceptibility test. However, the prosthetic joint infection could not be cured despite the appropriate antibiotic therapy, imipenem. In chronic infections of a prosthetic joint, biofilms are likely to develop and implant retention is hardly successful because the resistance of biofilm to antibiotics which is even recommended in vitro sensitivity test. A complete cure usually requires both implant removal and appropriate antibiotics treatment.8) In this case, revision of the total knee arthroplasty was successful after removal of the joint prosthesis and imipenem treatment after initial failure of imipenem treatment alone.

This case demonstrates that A. xylosoxidans is a pathogen capable of causing prosthetic joint infection even in immunocompetent patients and implies that to cure the infection of late stage, removal of implant, thorough debridement & appropriate antibiotic treatment is essential for the success of revision surgery.