Introduction

The number of primary and revision TKAs performed annually continues to increase with estimates of upward of 1.3 million primary TKAs and 125,000 revision TKAs to be performed per year by 2020 [13]. As indications for joint arthroplasty have expanded, there has been a disproportionate increase in the number of TKAs performed in younger patients [14]. These trends are expected to contribute to the rising number of revision arthroplasties performed each year. Berend et al. [3] described the concept of trading arthritic disease for “prosthetic disease,” which represents all possible failure modes of a prosthetic implant during the life of the patient. Multiply revised TKAs for infection, fracture, or instability are not uncommon and represent a complex reconstructive challenge. With revision after revision, these patients can develop accompanying bone loss, poor bone quality, soft tissue insufficiency, instability, and infection, all of which further complicate efforts to achieve a successful outcome.

Endoprostheses (also known as megaprostheses or tumor prostheses) have been used with success in the orthopaedic oncology setting for decades in the setting of limb salvage after large bony resections. Their use in the nontumor setting is less common, accounting for only approximately 7% of all endoprosthetic reconstructions [16]. An endoprosthesis is an effective alternative to structural allograft reconstruction for revision TKA in the setting of severe femoral bone loss. The relative ease of insertion and early mobility it affords are attractive benefits, especially for older and frailer individuals [5, 9, 10].

Numerous case series have been published for the use of endoprostheses in the revision arthroplasty setting [2, 7, 10–12, 15, 17, 19–21]. These series describe the outcomes related to individual types of reconstructions, ie, distal femoral or diaphyseal femoral. However, no study to date to the authors’ knowledge has directly compared outcomes between these two types of reconstruction of different endoprosthetic length and bone replacement. It is unknown whether the amount of bone loss and size of the endoprosthesis are correlated to complication risk, infection, reoperation, and ambulatory status.

We therefore asked: (1) Are longer endoprosthetic reconstructions associated with higher risk of infection? (2) Does increasing endoprosthetic length increase the risk of reoperation and decrease implant survivorship? (3) Is increasing length of an endoprosthesis associated with poorer ambulatory status?

Patients and Methods

An institutional review board-approved retrospective case series was performed consisting of all nononcologic femoral endoprosthetic reconstructions for revision TKA with distal femoral bone loss performed at a single tertiary care institution from 1995 to 2013. Cases were categorized as distal femoral replacement or diaphyseal femoral replacement. Patient demographics and outcome variables including number of prior surgeries, reoperations, infections, implant survival, and ambulatory status were collected by electronic medical record review.

Patients were initially screened for inclusion by keyword search of all operative notes of revision total joint arthroplasty cases (keywords: hinge, endoprosthesis, diaphyseal, diaphysis, distal femoral replacement, distal femoral arthroplasty, megaprosthesis, tumor prosthesis). A retrospective chart and radiograph review were then performed to identify all patients who had undergone endoprosthetic reconstruction for revision of a failed TKA. Cases were separated in groups by length of endoprosthetic reconstruction as follows: (1) distal femoral reconstructions, which included extension from the knee to the supracondylar metaphyseal-diaphyseal junction; or (2) diaphyseal femoral reconstructions, which included the supracondylar metaphyseal-diaphyseal junction to the proximal femoral diaphysis (Fig. 1). It is important to note that the reconstruction had to include the entirety of the metaphysis at a minimum to be a distal femoral reconstruction and then any extension further up the diaphysis constituted a diaphyseal reconstruction. Hinged revisions using resurfacing components with distal augments and/or metaphyseal cones did not therefore meet inclusion criteria.

Fig. 1A–B
figure 1

Examples of the two lengths of endoprosthetic reconstruction studied for revision TKA with distal femoral bone loss are shown. A distal reconstruction extends from the knee to the supracondylar metaphyseal-diaphyseal junction (A) versus a diaphyseal reconstruction, which included cases from the supracondylar metaphyseal-diaphyseal junction to the proximal femoral diaphysis (B).

Full size image

Initial keyword search resulted in 573 potential cases. After chart and radiograph review, 32 cases in 32 patients of endoprosthetic reconstruction for nononcologic revision TKA were identified (17 distal femoral reconstruction, 15 diaphyseal femoral reconstruction). Five patients from each group did not satisfy the minimum length of followup of 2 years leaving a total of 22 reconstructions for analysis (12 distal, 10 diaphyseal). Of the 10 patients without long-term followup, one patient died in the immediate perioperative period (diaphyseal), two had died before 2-year followup (both diaphyseal), and seven patients (five distal, two diaphyseal) were lost before 2 years. Attempts to connect with the seven lost patients by available contact information was unsuccessful. Of the patients included for analysis, the mean length of followup was 4 years for distal reconstructions (range, 2–8 years) and 6 years for diaphyseal reconstructions (range, 2–16 years). Ten of the 12 distal and six of the 10 diaphyseal reconstructions were implanted after January 2006 and 11 of the 12 distal and eight of the 10 diaphyseal reconstructions had been evaluated within the last 5 years.

With the numbers available, patients in each group were not different in terms of age, sex, Charlson Comorbidity Index, mean number of prior procedures, history of soft tissue rotational flaps, history of periprosthetic joint infection (PJI), primary operative diagnosis at the time of revision reconstruction, or tibial component type (Table 1). A large proportion of patients across all groups had a history of PJI treated with a two-stage exchange at some point before (not necessarily immediately prior) the implantation of the endoprosthesis (distal = seven of 12 versus diaphyseal = four of 10; p = 0.670). The organism history for distal reconstruction were methicillin-resistant Staphylococcus aureus × 2, methicillin-sensitive S aureus, Enterococcus faecalis × 2, Propionibacterium acnes, and Streptococcus. For the diaphyseal reconstruction group, prior organisms were methicillin-resistant S aureus, Staphylococcus lugdunensis, Gram-positive rods, and Candida albicans. Additionally, no difference was seen in the proportion of endoprostheses implanted for segmental bone loss immediately after a two-stage exchange for active PJI (distal = four of 12 versus diaphyseal = three of 10; p = 0.867).

Table 1 Patient characteristics
Full size table

Total femoral replacements were not included in this study because the majority were related to failed THA and felt to represent a different clinical entity. Patients were excluded for any oncologic diagnoses given the inherent differences between patients undergoing revision arthroplasty and the oncologic population.

All patients selected had undergone reconstruction with an endoprosthesis for revision TKA in the nononcologic setting. Two fellowship-trained, Knee Society member, joint arthroplasty surgeons performed all cases included in analysis except for three (one distal and two diaphyseal reconstructions). For reference, from 2006 through 2013, the same two surgeons performed 1013 revision total knee procedures (average > 60 cases each per year). It is our institution’s practice to preserve as much viable bone as possible in all cases and use distal or diaphyseal segmental replacement only as a final reconstructive alternative to amputation. Despite practicing at a tertiary care center with frequent cases involving large amounts of bone loss, the number of endoprosthetic reconstructions performed accounted for only 2.4% of all revision total knees (24 cases including those lost to followup over the same time period of 2006–2013).

All infections were treated with a standard two-stage exchange protocol. At the initial explantation, all devitalized and compromised soft tissues and bone were thoroughly débrided followed by placement of a static antibiotic-impregnated cement spacer. A minimum of 6 weeks of targeted intravenous antibiotic was administered followed by a minimum 2-week antibiotic holiday. Then serum erythrocyte sedimentation rate and C-reactive protein were drawn as well as aspiration and culture of joint fluid. Reimplantation was performed only with normalization of the serum inflammatory markers and sterile aspiration. Patients were given 24 hours of perioperative antibiotic prophylaxis with cefazolin, vancomycin, or cefazolin and vancomycin. Chronic long-term oral antibiotic therapy was not used routinely after revision with either distal or diaphyseal reconstruction. Postoperative infection was defined by the Musculoskeletal Infection Society criteria for PJI. A subset of patients who developed postoperative infections after their distal or diaphyseal replacement were treated with débridement, polyethylene liner exchange, and retention of components. These patients received intravenous antibiotics for 6 weeks followed by use of oral antibiotic therapy under the direction of an infectious disease consultant.

Once cases had been identified, retrospective chart and radiograph review was performed by study authors other than operating surgeons to identify variables of interest including age, sex, length of followup, number of procedures before endoprosthetic implantation, history of knee PJI before endoprosthetic reconstruction, surgical diagnosis at the time of endoprosthetic reconstruction, reoperation risk, endoprosthetic PJI risk, prevalence of chronic suppressive antibiotic use postoperatively, implant survival (defined as time to explant OR revision of endoprosthetic components), infection-free, revision-free implant survival (defined as time to diagnosis and débridement of infection OR explant OR revision of endoprosthetic components), and ambulatory status at most recent followup. Ambulation status was separated into able to walk (regardless of aid required) or full-time wheelchair user. Those who were able to walk were further divided into those without any aids and those that required aids (cane, crutches, or walker). Patient administrative records were reviewed to calculate Charlson Comorbidity Index without age adjustment using 16 comorbidities identified through International Classification of Diseases, 9th Revision, Clinical Modification coding [18].

Statistical Analysis

Categorical variables were compared using Fisher’s exact test or chi square when applicable. Continuous variables were compared using t-test or analysis of variance F-test when comparing more than three groups. Statistical significance was defined as p < 0.05. Kaplan-Meier survival curves were calculated with 95% CI included. Log-rank testing was used to determine significant differences in survival between groups. All statistical analysis was performed using R Version 3.0.2 (R-Project, R Foundation for Statistical Computing, Vienna, Austria).

Results

Patients with diaphyseal femoral replacements were more likely to develop deep infections after their reconstruction than were patients treated with distal femoral replacements (distal = three of 12 versus diaphyseal = nine of 10; p = 0.004; Table 2). Infection recurred after endoprosthetic implantation in patients with a history of PJI in three of seven distal and four of four diaphyseal reconstructions (p = 0.190). New infections occurred in previously aseptic limbs in zero of five distal and five of six diaphyseal reconstructions (p = 0.015). With the numbers available, the incidence of implant retention with débridement and long-term suppressive antibiotic therapy for limb salvage after postoperative infection were not different between the distal and diaphyseal reconstruction groups (distal = two of 12 versus diaphyseal = six of 10; p = 0.074). Nearly all organisms responsible for postoperative infection in the diaphyseal group were coagulase-negative Staphylococcus species (seven of nine infections; other organisms were Streptococcus and Escherichia coli) compared with the distal femoral reconstruction group in which there was one E. faecalis infection and two culture-negative infections that met other Musculoskeletal Infection Society PJI criteria.

Table 2 Results
Full size table

Implant survival (Kaplan-Meier survivorship free from explant or revision of components) of diaphyseal replacing reconstructions was worse than distal femoral replacing reconstructions at 2 years (distal = 100%, 95% CI, 100%–100% versus diaphyseal = 40%, 95% CI, 19%–86%; p = 0.001) and 5 years (distal = 90%, 95% CI, 75%–100% versus diaphyseal = 30%, 95% CI, 12%–73%; p = 0.001) (Fig. 2). Infection-free, revision-free survival (no infection, component revision or explant) was also worse for diaphyseal replacing reconstructions than for distal femoral replacements at 2 years (distal = 70%, 95% CI, 48%–100% versus diaphyseal = 20%, 95% CI, 6%–69%; p = 0.037) and 5 years (distal = 70%, 95% CI, 48%–100% versus diaphyseal = 10%, 95% CI, 2%–64%; p = 0.012; Fig. 3). The vast majority of reoperations were associated with infections (distal = three of four versus diaphyseal = eight of nine; p = 1.000). Only one patient from each group underwent a noninfectious reoperation (one distal reconstruction for patellar component revision, one diaphyseal reconstruction for a periprosthetic fracture).

Fig. 2
figure 2

Kaplan-Meier survival curve for implant survival (free from explant or component revision) comparing distal (DIS) and diaphyseal (DIA) femoral endoprosthetic reconstructions are shown. Bars represent the 95% CI.

Full size image
Fig. 3
figure 3

Kaplan-Meier survival curve for infection-free, revision-free survival comparing distal (DIS) and diaphyseal (DIA) femoral endoprosthetic reconstructions are shown. Bars represent the 95% CI.

Full size image

Compared with distal femoral endoprostheses, with the numbers available, diaphyseal endoprostheses showed no difference in the likelihood a patient would regain the ability to walk (distal = eight of 12 versus diaphyseal = seven of 10; p = 1.000) (Table 2). One patient from each group was able to ambulate without any assistive devices. The remainder of the ambulatory patients required a cane or walker for assistance. Above-knee amputations were ultimately performed on zero of 12 distal femoral replacements and four of 10 diaphyseal femoral replacements (p = 0.029). All patients who underwent amputation used wheelchairs to ambulate except one patient who used a walker after an above-knee amputation for infection.

Discussion

The use of endoprosthetics as a salvage option for complicated revision arthroplasty may become more commonplace as the number of revisions performed continues to increase. Prior series have typically analyzed a single implant type [2, 7, 10–12, 15, 17, 19–21]. To our knowledge, direct comparisons between different endoprosthesis subtypes for the treatment of distal femoral bone loss have not previously been performed. The purpose of this study was to evaluate if the length of the endoprosthesis implanted had an effect on outcomes with midterm followup. We found that prostheses that extended past the metaphyseal-diaphyseal junction of the distal femur were associated with higher risks of infection, reoperation, and implant revision.

The biggest limitation of endoprosthesis studies is the small number of patients given the rarity of surgical indications and use. Our institution’s practice is to only use endoprosthetic reconstruction as a last resort for the most severe bone loss cases and so these cases represent only 2.4% of revision total knees. When combined with the retrospective nature of this work, detailed statistics such as multivariate analysis would be greatly underpowered. Ten patients (five from each group) were not available for 2-year followup with three of five of the diaphyseal patients confirmed dead before 2 years and the rest unable to be contacted. Additionally, of those with sufficient followup duration, one of 12 distal and two of 10 diaphyseal reconstructions had not been evaluated in the past 5 years. Given this substantial portion of the patients lost to followup, our already sobering results likely represent a best-case scenario because those lost to followup are often faring worse. However, followup loss was equal between the two groups and a known problem in this older, frail population [2]. Other limitations include that our study spans nearly two decades with two main primary surgeons with potential confounding from implant design, surgical technique, and postoperative management. The three patients from other surgeons also increase this potential confounding. The wide study period may have affected the diaphyseal group more given four of 10 were performed before 2006 compared with only two of 12 of the distal reconstructions. Additionally, the study population is quite heterogeneous with many patients arriving at a salvage operation for a variety of reasons.

In our series, half of patients (11 of 22) had a history of PJI treated with a two-stage exchange at some stage before their endoprosthetic implantation. Seven of 22 had the two-stage exchange immediately before endoprosthetic implantation. For comparison, previously reported infection history preimplantation has ranged from 16% to 20% for distal and 4% to 16% for diaphyseal endoprostheses [10, 11, 20]. Previously reported infection risk after endoprosthetic reconstruction of various lengths for revision arthroplasty in the nononcologic setting have varied from 13% to 35% for distal through total femoral replacements [1, 2, 6, 10–12, 20]. Our high proportion of patients with a PJI history compared with previously reported cohorts likely contributed to the high proportion of patients who developed postoperative infection overall in this series; however, we feel the bigger cases through larger exposures in patients with multiple comorbidities and compromised soft tissues likely all contribute to the higher prevalence in the diaphyseal group. Interestingly, nearly all reinfections occurred with different organisms and the majority of postoperative infections involved Staphylococcus epidermidis, a biofilm-producing organism. The standard practice for these replacements at our institution had previously been to use 24 hours of antibiotic postoperative prophylaxis. Interestingly, Berend et al. [3] reported on 59 total femoral replacements for nononcologic revision and had 14% infections at 5 years followup after using prolonged antibiotics in all patients despite a preoperative infection history in 24% of patients. Similar to their group, given the morbidity of infection in these implants, it is now our practice to place all patients with prostheses that extend into the diaphysis on empiric chronic antibiotic therapy postoperatively. Investigation of this prolonged antibiotic course as well as collaborative efforts with oncology colleagues on interventions like local antibiotic delivery or silver-coated prostheses warrants future attention [8, 22].

Compared with distal femoral replacement, diaphyseal femoral replacement was associated with decreased implant survival (explant or component revision) at 5 years postoperatively (distal = 90%, 95% CI, 75%–100% versus diaphyseal = 30%, 95% CI, 12%–73%; p = 0.001). However, in our small series, this already sobering survival rate may still falsely estimate true “success,” because there was a disparity in both groups between implant survival and infection-free, implant survival (cases without infection AND without revision/explant–Kaplan-Meier 5 years [distal = 70%, 95% CI, 48%–100% versus diaphyseal = 10%, 95% CI, 2%–64%; p = 0.012). The difference comes from patients treated with multiple débridements and chronic antibiotic suppression given the morbidity associated with revision, staged exchange, or amputation. Although the implant “survives,” it is difficult to argue these cases as success. Our revision-free survival at 5 years for distal reconstructions is in line with previously reported rates for endoprosthetics in the nontumor setting; however, revision risk for the diaphyseal reconstruction group was substantially higher. Reoperation or major complication prevalence has been reported as 13% to 59% for distal and 27% to 31% for diaphyseal reconstructions previously when reported separately [2, 3, 7, 10–12, 15, 19, 20]. However, as noted, our cohort’s infection proportion before implantation was higher and all reoperations except for two were infection-related. Survivorship for noninfectious indications in revisions total joints have been noted to be better historically [4, 6].

Despite the high morbidity associated with endoprosthetic reconstruction for complex revision arthroplasty, the likelihood a patient would be able to walk at final followup was high if they had not undergone an amputation (14 of 17). Amputations only occurred in the diaphyseal group in our small series. When used as a salvage option, endoprosthetic reconstruction can offer an immediate mobility benefit over amputation, especially in the older, frail individuals in whom bone loss is often a problem and immobility can be fatal. Our ambulatory proportion at final followup is similar to those reported for other endoprosthetic reconstructions for revision arthroplasty such as Höll et al. [10] who reported 19 of 20 distal and diaphyseal endoprosthetic reconstructions were ambulatory, although 16 required a walker and 10 were homebound. Similarly, Amantullah et al. [1] reported 14 of 20 patients ambulatory at an average 73 months after total femoral replacement.

Endoprosthetic distal and diaphyseal femoral reconstruction can be used as a final salvage alternative to amputation for the treatment of failed TKA with segmental distal femoral bone loss. In our small limited series even with substantial loss to followup and likely best-case estimates of success, reconstructions with extension proximal to the supracondylar metaphyseal-diaphyseal junction result in higher infection and reoperation risk. This finding may be related to the greater area of bone and soft tissue devascularization and larger dead space associated with more proximal segmental replacement. Limb salvage in infected reconstructions remains possible, however, with infection source control and chronic antibiotic suppression. We now routinely use chronic empiric antibiotics for all endoprosthetic reconstructions extending into the diaphysis or higher in an effort to minimize the morbidity of infection in these salvage situations.