Background: Two-stage exchange arthroplasty after a previous, failed 2-stage exchange procedure is fraught with difficulties, and there are no clear guidelines for treatment or prognosis given the heterogeneous group of patients in whom this procedure has been performed. The Musculoskeletal Infection Society (MSIS) staging system was developed in an attempt to stratify patients according to infection type, host status, and local soft-tissue status. The purpose of this study was to report the results of 2-stage exchange arthroplasty following a previous, failed 2-stage exchange protocol for periprosthetic knee infection as well as to identify risk factors for failure.
Methods: We retrospectively identified 45 patients who had undergone 2 or more 2-stage exchange arthroplasties for periprosthetic knee infection from 2000 to 2013. Patients were stratified according to the MSIS system, and risk factors for failure were analyzed. The minimum follow-up was 2 years (mean, 6 years; range, 24 to 132 months).
Results: At the time of follow-up, twenty-two (49%) of the patients had undergone another revision due to infection and 28 (62%) had undergone another revision for any reason. The infection recurred in 6 (75%) of 8 substantially immunocompromised hosts (MSIS type C) and in 3 (30%) of 10 uncompromised hosts (type A) following the second 2-stage exchange arthroplasty (p = 0.06). The infection recurred in 4 (80%) of 5 patients with compromise of the extremity (MSIS type 3) and 3 (33%) of 9 patients with an uncompromised extremity (type 1) (p = 0.27). Both extremely compromised hosts with an extremely compromised extremity (type C3) had recurrence of the infection whereas 3 (30%) of the 10 uncompromised patients with no or less compromise of the extremity (type A1 or A2) did. Five patients in the failure group underwent a third 2-stage exchange arthroplasty following reinfection, and 3 of them were infection-free at the time of the latest follow-up.
Conclusions: Uncompromised hosts (MSIS type A) with an acceptable wound (MSIS type 1 or 2) had a 70% rate of success (7 of 10) after a repeat 2-stage exchange arthroplasty, whereas type-B2 hosts had a 50% success rate (10 of 20). The repeat 2-stage exchange procedure failed in both type-C3 hosts; thus, alternative salvage procedures should be considered for such patients.
Level of Evidence: Therapeutic Level IV. See Instructions for Authors for a complete description of levels of evidence.
The prevalence of persistent or recurrent infection following 2-stage exchange arthroplasty following failure of a total knee arthroplasty has been reported to be 10% to 25%1-5. However, the results of a subsequent 2-stage exchange arthroplasty have not been well investigated6-8. While 2 or more 2-stage exchange arthroplasties have the potential benefits of increased function and pain relief compared with the outcomes of resection arthroplasty, arthrodesis, or amputation, such procedures are technically demanding. Moreover, patients in whom a 2-stage exchange arthroplasty has failed are a heterogeneous cohort with respect to comorbidities, conditions of local soft tissues, and types of organisms. Thus, a paucity of information is currently available in the literature6-8.
McPherson et al.9 previously described a Musculoskeletal Infection Society (MSIS) staging system to stratify patients on the basis of infection type, host status, and extremity status (Tables I and II). A staging system focused on individual host and local soft-tissue status is helpful in the assessment of patients with chronic periprosthetic joint infection, and may help to guide treatment algorithms.
While 2 previous studies of a second 2-stage exchange arthroplasty for periprosthetic knee infection have shown promising results in terms of infection eradication, they were limited by small patient numbers and short follow-up6,7. Identified risk factors for failure after 2-stage exchange arthroplasty include polymicrobial infection, obesity, chronic lymphedema, and drug-resistant organisms10-12.
The purpose of this study was to determine the rate of infection-free implant survival, risk factors for failure, and complications, based on the patients’ grades according to the MSIS staging system9, associated with a 2-stage exchange arthroplasty following a previous, failed 2-stage exchange procedure for periprosthetic knee infection.
Materials and Methods
We retrospectively identified 45 patients who had undergone 2 or more 2-stage exchange arthroplasties for periprosthetic knee infection at a single academic tertiary-care medical center from 2000 to 2013. Twenty-three (51%) of these patients had had their index total knee arthroplasty and initial 2-stage exchange arthroplasty performed at an outside institution, 14 patients (31%) had had their initial total knee arthroplasty performed elsewhere and their initial 2-stage exchange arthroplasty performed at our institution, and the remaining 8 patients (18%) had had both procedures performed at our institution. Institutional review board approval was obtained prior to initiation of the study.
All patients who had undergone a 2-stage exchange arthroplasty following failure of a previous 2-stage exchange arthroplasty for periprosthetic knee infection were included. The minimum follow-up was 24 months, with a mean of 74 months and a maximum of 132 months. No patient was lost to follow-up. We retrospectively reviewed these patients’ medical records to classify them according to the MSIS staging system9 and to examine their infection-free implant survival rate (primary outcome), potential risk factors for failure (secondary outcome), and complications (secondary outcome). All chart review was performed by an author who was not one of the operating surgeons.
Patients were followed at 3 months, 1 year, 2 years, and 5 years following reimplantation and every 5 years thereafter. Success following reimplantation was defined as retention of the prosthesis without subsequent procedures for infection, regardless of antibiotic suppression status.
The mean age at the time of the second 2-stage exchange arthroplasty was 62 years (range, 47 to 83 years). According to the MSIS staging system9, all infections were classified as late chronic infections (type III). Ten (22%) of the patients were classified as medically healthy (type-A hosts); 27 (60%), as medically compromised (type-B hosts); and 8 (18%), as very medically ill (type-C hosts). The local condition of the extremity was classified as uncompromised (type 1) in 9 patients (20%), compromised (type 2) in 31 (69%), and very compromised (type 3) in 5 (11%). All reinfections were diagnosed in accordance with the criteria for periprosthetic infection defined by the MSIS13.
Surgical Technique for Repeat 2-Stage Exchange Arthroplasty
A static spacer, inserted at the time of the repeat resection arthroplasty, was used in all patients between the 2 surgical procedures of the second 2-stage exchange protocol. Reimplantation was performed once the patient had completed at least a 6-month course of intravenous (IV) antibiotics (targeted against an isolated microorganism if known), there was a downward trend of inflammatory markers, and the clinical course suggested resolution of infection14-16. All patients underwent an institutional standard protocol of antimicrobial precautions including the IV antibiotics in accordance with Surgical Care Improvement Project (SCIP) guidelines, a preoperative chlorhexidine bath, and treatment in a laminar airflow operating room. The mean time to the reimplantation after the resection arthroplasty was 15 weeks (range, 6 to 132 weeks). The mean erythrocyte sedimentation rate was 19 mm/hr (range, 0 to 79 mm/hr; normal: 0 to 22 mm/hr) and the mean C-reactive protein level was 5 mg/L (range, 3 to 17.5 mg/L; normal: <8.0 mg/L) immediately before the reimplantation.
At the time of the repeat reimplantation, metaphyseal fixation utilizing sleeves or cones in addition to stems were used in 26 patients (58%). Cemented stems were used in 41 patients (91%). While antibiotics were used at the discretion of the operating surgeon, the preference was to utilize 1 g of vancomycin and 1.2 g of gentamicin per 40-g batch of bone cement. The reimplantation consisted of a rotating hinge total knee arthroplasty in 21 patients, a varus-valgus constrained total knee arthroplasty in 23, and a posterior-stabilized total knee arthroplasty in 1. The implant selection was based on the particular need for constraint in the individual patient and on surgeon preference.
Descriptive statistics are reported as the number (percentage) or mean (range) as appropriate. Chi-square and Fisher exact tests were used to compare binary variables. All calculations assumed 2-tailed tests. The α level was set at 0.05 for significance.
Infection-Free Implant Survival
At the time of follow-up, at a mean of 74 months (range, 24 to 132 months) after the repeat 2-stage exchange arthroplasty, 23 (51%) of the 45 patients had not undergone another revision for infection; 14 (61%) of these 23 patients were on chronic suppressive antibiotic therapy at the time of follow-up and 6 (26%) underwent revision for mechanical failure, at an average of 72 months following the second 2-stage exchange arthroplasty. One of the 6 patients was on chronic antibiotic suppression at the time of the revision. Overall, 62% (28) of the 45 patients underwent revision for any reason. Of the patients who had infection-free implant survival, 9 (39%) had an extension lag of at least 10° at the time of final follow-up and 16 (70%) walked with gait aids. The mean range of motion following the second 2-stage exchange arthroplasty was 5° to 102°.
The infection recurred in 22 patients (49%). All underwent surgical intervention, with 5 undergoing a third 2-stage exchange arthroplasty and the remaining 17 undergoing irrigation and debridement, resection arthroplasty, or amputation. All 5 patients who underwent a third 2-stage exchange arthroplasty received a rotating hinge construct at that time, and 3 of the 5 had not undergone a fourth revision for infection at the time of final follow-up. Nine (20%) of the 45 patients in this study eventually had a resection arthroplasty or an above-the-knee amputation.
The microorganism leading to the infection treated by the second 2-stage exchange arthroplasty was identified in 34 patients (76%), and in 17 of them the microorganism differed from the one that had caused the initial periprosthetic knee infection. Eight of these 17 patients had infection-free implant survival after the second reimplantation. The most commonly isolated microorganism was coagulase-negative staphylococcus (14 knees; 31%) (Tables III and IV).
Twenty-five patients (14 in the success group and 11 in the failure group) were on suppressive antibiotic therapy at the time of final follow-up. The infection rate did not differ significantly between the 25 patients who were on chronic suppression (44%) and the 20 who were not (55%) (odds ratio [OR] = 0.64; 95% confidence interval [CI] = 0.2 to 2.09; p = 0.46). When this variable was analyzed in isolation for type-B2 hosts (the most common cohort), the reinfection rate was lower, but not significantly so, for those treated with suppressive antibiotics (38% [3/8] compared with 58% [7/12]; p = 0.83).
Host status was found to be a risk factor for failure (p = 0.084 for type A versus type C). We found a relationship between infection recurrence after the second 2-stage reimplantation and worsening grades for the host and local condition of the extremity (p = 0.06). A revision for reinfection was performed in 30% (3) of the 10 uncompromised hosts (type A), 48% (13) of the 27 medically compromised hosts (type B), and 75% (6) of the 8 very medically ill patients (type C) following the second 2-stage exchange arthroplasty. Resection or transfemoral amputation was performed in 4 of the 8 type-C hosts compared with 1 of the 10 type-A hosts. The infection recurred in both of the patients with the worst host and extremity grades (type C3). The reinfection rate was 50% (10 of 20) in the most common group of patients according to host and extremity grade (type B2) (Table V).
Of the 23 patients who had infection-free implant survival, 13 were immunocompromised (type-B or C) hosts and had a compromised (type-2 or 3) lower extremity. Four type-B hosts (15%) eventually had a resection or transfemoral amputation.
Referral to our hospital from an outside institution was not found to be a significant risk factor for failure (p = 0.69).
Complications occurred in 14 patients (31%). The most common was a new disruption of the extensor mechanism (3 cases), intraoperative fracture at the time of resection or reimplantation (3 knees), postoperative fracture (2 knees), aseptic loosening of the implant (2 knees), stem-condyle bolt failure in a varus-valgus constraining knee implant (3 knees), and instability (2 knees). Only 11 patients (24%) did not have a complication or a reoperation for infection, intraoperative complications, or postoperative sequela.
Reinfection after 2-stage exchange arthroplasty is a difficult clinical scenario with limited data on appropriate treatment algorithms. The results of subsequent reimplantation procedures in this setting remain unknown. We sought to define the results of a second 2-stage exchange arthroplasty for reinfection as well as the risk factors for failure based on host and extremity grades derived with the MSIS staging system.
In our study, the overall reinfection rate after a second 2-stage exchange arthroplasty for periprosthetic knee infection was 49% (22 of 45). Previous literature has shown higher success rates in terms of infection eradication following a second 2-stage exchange procedure6,7. However, the patients in those studies were type-A hosts. For example, Azzam et al.6 reported that 14 of 18 self-stated healthy (type-A) patients had successful eradication of infection after undergoing a second 2-stage reimplantation for the treatment of periprosthetic knee infection. Similarly, Backe et al.7 reported a 100% rate of infection eradication at short-term follow-up of patients who were not immunocompromised at the time of reimplantation. Our study differed from these previous reports in that our sample size was larger and we included patients who were compromised hosts (type B or C) and/or had a compromised extremity (type 2 or 3) when they underwent reimplantation. Only 9% of the patients in our study were classified as an uncompromised host with an uncompromised or a not substantially compromised extremity, and 78% were classified as compromised hosts (type B or C). The 30% failure rate (3 of 10) in our medically healthy patients (type-A hosts) was similar to the rates in the previous reports. The reinfection rate substantially increased with declining host and extremity status, with both patients who were considered a compromised host and to have a compromised extremity (type C3) having a reinfection.
To our knowledge, this is the only study in which the results after a second 2-stage exchange arthroplasty were studied in patients stratified according to host status and condition of the extremity. We showed that host factors and overall wound status play a substantial role in outcomes following a second 2-stage exchange arthroplasty. Many of the patients in this study were classified as a compromised host or as having a compromised extremity, or both, according to the MSIS system as described by McPherson et al.9. Reinfection rates were strongly correlated with worsening host and extremity grades. There was also a strong association between worsening host grade and eventual resection arthroplasty or amputation in this cohort. Of the 8 patients classified as very medically ill (type-C hosts), 4 eventually had a resection or amputation at the time of final follow-up whereas only 1 of the 10 medically healthy (type-A) hosts had such a procedure.
Treatment algorithms have evolved over the years. In a previous study of 24 patients treated for reinfection after a failed 2-stage reimplantation, 2 patients received a subsequent reimplantation, 11 underwent arthrodesis, 6 underwent debridement, 4 were treated with antibiotic suppression, and 1 underwent amputation8. Surgeons have become more likely to attempt a subsequent reimplantation procedure in the setting of reinfection with a goal of improving the functional outcome. The overall complication rate in our study was high (31%), indicating that these procedures are fraught with intraoperative and postoperative problems. Likewise, the revision rate for any reason in our study was 62%.
This study has limitations. Patients selected for a subsequent 2-stage exchange arthroplasty are thought to be the best surgical candidates to undergo this treatment regimen. Those who are deemed too compromised to undergo another 2-stage exchange arthroplasty because of host or limb factors are treated with debridement with component retention, amputation, resection, or arthrodesis. Thus, the patient population that we studied consisted of the best possible candidates for an attempt at infection eradication and another reimplantation. The results of a second 2-stage exchange procedure in a less healthy population, or patients with a more compromised limb, are unknown, and our data cannot be extrapolated to all patients.
Functional scores were not obtained for all patients at the time of follow-up; thus, we could not provide the functional outcomes of the procedure. There was also the potential for bias in this study as the reviewer was not blinded to the outcome while identifying patient characteristics to risk-stratify patients.
Treatment for reinfection following 2-stage exchange arthroplasty for periprosthetic knee infection remains challenging. Our results show that enthusiasm and expectations for infection eradication following a second 2-stage exchange arthroplasty should be tempered for patients with a compromised host and extremity status. Selection of the appropriate treatment for a patient with reinfection following an initial 2-stage exchange arthroplasty involves consideration of host status, extremity status, implant status, bone stock, and microorganism profile. Other reconstructive or salvage options should be considered for the most medically compromised patients rather than making another futile attempt at reimplantation.
Investigation performed at the Mayo Clinic, Rochester, Minnesota
Disclosure: The authors indicated that there was no outside source of funding for this study. The Disclosure of Potential Conflicts of Interest forms are provided with the online version of the article.
- Copyright © 2017 by The Journal of Bone and Joint Surgery, Incorporated