It is unclear whether there is a limit to the amount of distal bone required to support fixation of supracondylar periprosthetic femoral fractures. This retrospective multicentre study evaluated lateral locked plating of periprosthetic supracondylar femoral fractures and compared the results according to extension of the fracture distal with the proximal border of the femoral prosthetic component.
Between 1999 and 2008, 89 patients underwent lateral locked plating of a supracondylar periprosthetic femoral fracture, of whom 61 patients with a mean age of 72 years (42 to 96) comprising 53 women, were available after a minimum follow-up of six months or until fracture healing. Patients were grouped into those with fractures located proximally (28) and those with fractures that extended distal to the proximal border of the femoral component (33).
Delayed healing and nonunion occurred respectively in five (18%) and three (11%) of more proximal fractures, and in two (6%) and five (15%) of the fractures with distal extension (p = 0.23 for delayed healing; p = 0.72 for nonunion, Fisher’s exact test). Four construct failures (14%) occurred in more proximal fractures, and three (9%) in fractures with distal extension (p = 0.51). Of the two deep infections that occurred in each group, one resolved after surgical debridement and antibiotics, and one progressed to a nonunion.
Extreme distal periprosthetic supracondylar fractures of the femur are not a contra-indication to lateral locked plating. These fractures can be managed with internal fixation, with predictable results, similar to those seen in more proximal fractures.
Supracondylar fractures of the femur after total knee replacement (TKR) are an uncommon but highly challenging injury. Despite occurring in only 0.2% to 0.54% of TKRs,1,2 their incidence is expected to increase as TKRs are more frequently performed and patients become older and their activity level increases. According to the Center of Disease Control, in 2006 Medicare patients underwent over 50% more TKRs than in 2000,3 which represents an increase of over 100 000 replacements.
In the presence of a stable implant, periprosthetic supracondylar femoral fractures are preferably managed with reduction and internal fixation, as anatomical alignment and stable fixation of the distal articular component allows for early movement of the knee and patient mobilisation.4 Reliable fixation can nevertheless be difficult to achieve, owing to interference of the prosthesis with standard implants and the presence of poor bone stock.5
Over recent years, locked plating has led to good clinical results for the management of supracondylar femoral fractures.6–10 Advantages include angular stability resisting varus collapse, and improved fixation in osteoporotic bone.6,11 Furthermore, plates can be used as an aid to the reduction and allow more critical assessment of the accuracy of intra-operative reduction of the fracture,10 as opposed to intramedullary nails, which have been associated with considerable rates of malalignment.12 Although available bone stock has been identified as a factor to consider when stabilising supracondylar fractures around TKRs, so far no study has critically examined this variable. The purpose of this study was to evaluate the use of lateral locked plating in a cohort of patients with a periprosthetic femoral fracture adjacent to a TKR, and to compare results based on the distal extent of the fracture.
Our null hypothesis was that extreme distal fractures extending into the region of the femoral component are amenable to reduction and internal fixation, with results similar to those of more proximal fractures with greater available distal bone stock.
Patients and Methods
After approval by the institutional review board at each of the three participating Level I trauma centres (Barnes-Jewish Hospital, St Louis, Missouri; Harborview Medical Center, Seattle, Washington; Harris Methodist Hospital, Fort Worth, Texas), patients with a supracondylar femoral fracture above an existing TKR admitted between 1999 and 2008 were identified from the prospective institutional orthopaedic trauma databases. Inclusion criteria were non-pathological Lewis and Rorabeck type II fractures (stable implant, displaced fracture),13 open reduction and internal fixation using a lateral locked plate, and clinical and radiological follow-up of at least six months or until fracture healing. Operating notes and medical records were reviewed for patient characteristics (age, gender, body mass index (BMI), diabetes mellitus, and smoking), fracture characteristics (side and mechanism of injury, open fracture type according to Gustilo and Anderson14 and Gustilo et al15), implant details, healing and complications (infection, construct failure, additional procedures).
Anteroposterior (AP) and lateral radiographs of the femur and knee were reviewed from the time of admission, post-operatively, and at final follow-up. Fractures were classified according to the Orthopaedic Trauma Association (OTA)16 with all included fractures being OTA 33, and the method of Su, DeWal and Di Cesare.17 The latter classification establishes three periprosthetic fracture types according to the most distal extent of the fracture relative to the proximal border of the femoral component. Type I fractures are located proximal to the component, type II extend from the proximal aspect of the anterior prosthetic flange proximally, and type III extend distally beyond the proximal border of the femoral component.17 As distal bone stock is most limited in type III fractures, we grouped fractures into those without distal fracture extension beyond the proximal border of the femoral component (Su types I and II; Fig. 1⇓), and those with distal extension (extreme distal fractures, Su type III; Fig. 2⇓). Results were compared between groups without (group A) and with distal extension (group B).
Out of a total of 441 patients with 457 supracondylar fractures of the femur admitted for lateral locked plating between 1999 and 2008, 83 patients with 85 fractures occurring around an existing stable TKR were identified. Of these, 22 patients with 24 fractures were excluded. Five patients had died before completing six months of follow-up, and 14 were lost to follow-up. Of the latter, one patient each died at 11, 18 and 79 months after surgery. Finally, three patients were excluded because radiological follow-up was not available. Telephone interviews more than one year after surgery revealed that, of these patients, one was asymptomatic, one complained of pain, and one complained of painless deformity at the fracture site. Therefore, a total of 61 patients with unilateral fractures were included in the study. The fracture was located proximal to the femoral component in 28 patients (46%; group A) and extended distally in 33 (54%; group B). The mean follow-up was 12 months (2 to 49) in group A and nine months (3 to 35) in group B (Table I⇓).
A fall was the most frequent mechanism of injury in both groups. All fractures related to a fall were closed, and all those resulting from a motor vehicle collision were open (Table I⇑). All patients were treated with a distal femoral locking plate, either the Peri-Loc Distal Lateral Femur locking plate (Smith and Nephew, Memphis, Tennessee), the Locking Condylar Plate (Synthes, Paoli, Pennsylvania) or the Less Invasive Stabilization System (Synthes). In group A, no supplementary implants were used, whereas two fractures in group B required additional proximal cable fixation owing to the presence of long femoral stemmed components of previous revision hip replacements. Additionally, one highly comminuted OTA 33A3 fracture required fixation of an unstable intercalary coronal fragment with a small fragment plate. Finally, two patients in group B received a calcium phosphate/calcium sulphate graft for segmental defect filling at the time of fracture fixation (ProDense, Wright Medical Technology, Arlington, Tennessee). The indication was a comminuted fracture in a patient aged 96 years, and an open type IIIA fracture in another. In the latter, the osteobiological cement was mixed with vancomycin for prophylaxis against infection (Table II⇓).
Sagittal and coronal alignment was measured to the nearest degree using a standard goniometer for printed radiographs and the calibrated digital measuring tools of the proprietary software for DICOM image viewing at each participating centre. Coronal alignment was measured as the angle between the anatomical axis of the femoral shaft and a line tangential to the distal femoral articular surface. In the sagittal plane, the angle between the anatomical axis of the femur and the anterior flange of the femoral component was measured. Extension and flexion of the distal fragment were recorded as negative and positive values, respectively.
Delayed healing was defined using the upper limit proposed by Hammacher, van Meeteren and van der Werken18 as a fracture requiring more than six months to unite without requiring an additional surgical procedure. Nonunion was defined as a painful fracture which after six months of follow-up did not show progressive healing on three consecutive radiographs taken at one-month intervals,19 and requiring an additional procedure to obtain healing. Construct failure was defined as any hardware breakage or implant-bone dissociation leading to loss of reduction visible on any radiographic view. Isolated screw breakage or loosening without loss of alignment was not considered construct failure. Location of the construct failure was classified as occurring either proximally (loss of fixation to the femoral shaft), at the plate’s working length (plate breakage at segment spanning the fracture site) or distally (failure of the fixation of the distal fracture component).
Healing and construct rates of failure were defined as the primary outcome; secondary outcomes included complications and fracture alignment.
Categorical variables were analysed using chi-squared or Fisher’s exact tests. Continuous data were analysed for normality using the one-sample Kolmogorov-Smirnov test. Normally distributed variables were compared using Student’s t-tests. Non-parametric variables were compared using the Mann-Whitney U test. Statistical significance was set at p < 0.05, and confidence intervals (CI) at 95%. Statistical analysis was performed with SPSS version 16.0 for Windows (SPSS Inc., Chicago, Illinois).
Patient and fracture characteristics.
There were no differences in age, gender, BMI, incidence of diabetes mellitus and smoking between the two groups. Similarly, there were no differences in fracture characteristics between groups, including side, mechanism, open fractures, and OTA fracture pattern (Table I⇑).
Plate length ranged from six to 18 holes in both groups. In group A, given the more proximal nature of the fractures, plates were on average one hole longer, but similar distal fixation was achieved in both groups (Table II⇑).
No significant differences could be found for healing after the index procedure between groups (Table III⇓). Septic nonunion occurred in one case in each group, and two aseptic nonunions occurred in group A and four in group B. The septic nonunion in group A required three additional procedures, including debridement and irrigation, hardware removal, and ultimately a transfemoral amputation. The septic nonunion in group B required five additional procedures for healing to occur, including serial debridement and irrigation, hardware removal, and staged revision with grafting. All aseptic nonunions healed after revision surgery with hardware exchange and grafting, except for one patient in group B who was lost to follow-up after 1.4 years without having undergone revision surgery. One peri-implant fracture that occurred in an aseptic nonunion from group B was successfully treated with revision surgery of the nonunion. A total of five of eight patients with nonunion had a history of diabetes mellitus, represented by two of three nonunions in group A and three of five nonunions in group B. Similarly, obesity was present in two of three nonunions in group A and four of five non-unions in group B.
Similar coronal and sagittal angulations obtained after initial fixation, as well as change in sagittal alignment at final follow-up, were observed in both groups (Table III⇑).
The complications that arose are summarised in Table III⇑. In each group, one of two deep infections healed after debridement, irrigation and antibiotic treatment, and the other developed into a septic nonunion. Proximal construct failures occurred in the presence of three and five proximal screws in group A and four proximal screws in group B, respectively. Distal construct failures occurred in the presence of five screws in group A and five and seven screws in group B, respectively.
With the advent of modern implants and surgical techniques, operative treatment has become the choice for displaced supracondylar fractures of the femur around a TKR with a stable femoral component,4,20,21 as high rates of malunion and nonunion have been reported after non-operative treatment.2 Several surgical options have evolved, including hybrid external fixation, intramedullary nailing, and conventional and locked plate fixation.22–27 The complication rate for conventional, non-locked plating is high. Using conventional plating, Figgie et al5 had a 50% nonunion rate in ten supracondylar periprosthetic femoral fractures. Furthermore, augmentation with cement is often required in order to obtain adequate fixation.28 Adequate clinical results have been reported with both retrograde intramedullary nailing and locked plating.6–10,29–35 As comminution is frequently present at the fracture site, angular stability is crucial to resist varus deformity.36,37 Although biomechanical studies have shown intramedullary nailing to achieve increased resistance to varus stress compared with locked plating,38 this difference has not been shown to be clinically significant.32,39,40 Furthermore, intramedullary nailing is limited by the availability of distal bone stock, as well as the size and position of the femoral component notch.41 Therefore, enlargement of the notch with a diamond burr is often required, raising concern about possible third-body wear.42 Locked plating, on the other hand, allows for distal fixation regardless of the design of the femoral component and the presence of intramedullary implants. Additionally, multiple angularly stable fixation points around existing obstacles and in close proximity to the distal and anterior femoral cuts can be obtained.10,36
Wick et al40 evaluated 18 patients who underwent operative treatment of periprosthetic supracondylar femoral fractures using either locked plating or retrograde intramedullary nailing. The mean age of their patients was 80 years in nine treated with a nail, and 77 years in patients treated with locked plating. Duration of surgery and length of stay were not different. One infection occurred in the plating group and one malalignment in the nail group. Each of these required revision surgery. Even though the results were similar, the authors concluded that locked plating may enable fixation of more distal fractures.40 In our institutions, lateral locked plating has been used as the treatment of choice for this type of fracture, as it does not depend on the design of the TKR or on distal bone stock in order to achieve adequate fixation.
Despite the advantages of locked plating for fixation of more distal fractures compared with retrograde intramedullary nailing, bone stock of the distal segment continues to be considered a key limiting factor to obtain adequate fixation.32 As a consequence, Su et al17 proposed a classification based on fracture location in order to guide treatment options. According to their classification, type I fractures are amenable to either retrograde or antegrade intramedullary nailing, and type II fractures require either retrograde intramedullary nailing or fixed-angle plating. Type III fractures, on the other hand, may, according to the authors, require fixation with a fixed-angle device or revision arthroplasty when distal bone stock is poor.17 Our results suggest that extreme distal periprosthetic supracondylar fractures of the femur can be adequately managed with open reduction and internal fixation using locked plate constructs, with results that are comparable with those of the treatment of more proximal fractures.
Locked plate fixation of periprosthetic supracondylar femoral fractures has shown good clinical results.6–10 Raab and Davis43 had excellent results in nine patients managed with the Condylar Locking Plate, all of whom healed with adequate alignment. Erhardt et al44 reported similar results in the management of 24 periprosthetic supracondylar femoral fractures using a variable-angle locking plate. All patients healed, with only one malunion of 12° of varus, and 82% of patients had a high level of satisfaction. Anakwe et al,7 reported a 100% healing rate in 28 patients undergoing locked plating of periprosthetic distal femoral fractures. Two patients required revision surgery for plate removal, one for pain and one because of deep infection. A wide range of complication rates has nevertheless been reported in several other studies. Herrera et al32 performed a systematic review of the literature on outcome of surgery for periprosthetic supracondylar femoral fractures in 415 cases published over a 25-year period. An overall non-union rate of 9%, failure of fixation of 4%, infection of 3% and revision surgery of 13% were reported using different methods of fixation. Studies using locked plating reported a mean rate of nonunion of 5.3% (1.8% to 14%), of fixation failure of 3.5% (0.9% to 12%), of deep infection of 5.3% (1.8% to 14%) and of further procedures of 8.8% (3.8% to 19%).32 Additionally, Large et al39 described five malunions and no nonunions in 29 patients with a mean age of 75 years undergoing locked plate fixation with either LISS or Condylar Locking Compression Plates. Fulkerson et al45 reported two nonunions, one delayed union, one deep infection and one implant failure in 18 patients undergoing locked plating of supracondylar periprosthetic femoral fractures. Additional procedures included one revision of reduction and fixation, one removal of hardware and arthrodesis, one nonunion repair and two revision TKRs.45
Ricci et al10 had a primary healing rate of 86% in a study of 24 periprosthetic supracondylar femoral fractures managed with lateral locked plating. All three nonunions were present in diabetic, obese patients. These findings were similar in the current study, where, of eight patients with non-union, five were diabetic and six obese. Despite our sample being too small to establish diabetes mellitus and obesity as risk factors for nonunion, these conditions certainly pose an additional challenge to the management of these injuries.
Even though not evaluated in our outcome, mortality rates of up to 18% in the first year after surgery have been reported.46 Similarly, only 50% of patients return to their pre-injury place of living.7,46 Furthermore, complications may include transfemoral amputation for the management of recalcitrant chronic osteomyelitis. Similar to our results, Bezwada et al28 described one patient who required transfemoral amputation after deep infection following retrograde nailing in a series of 30 periprosthetic supracondylar femoral fractures.
The rate of union for periprosthetic supracondylar femoral fractures managed with locked plating in our study is similar to that in the literature. Additionally, extremely distal fractures, where the fracture extended to the prosthesis itself, showed similar results to those of more proximal fractures, supporting the idea that fractures around stable implants can be adequately fixed with locked plates, despite a limited amount of distal bone stock. It would be desirable to establish the minimum number of distal screws required to prevent failure of distal fixation. The relatively low rate of implant failure in our study did not allow us to calculate this figure. However, it has become our practice to obtain fixation with at least five screws proximally and distally.
Despite being the largest study on periprosthetic fractures of the distal femur published to date, from a statistical perspective our study sample is too small to allow conclusive statements regarding the equality of the outcome among fracture groups. The lack of power and the possible presence of type 2 error may be of special relevance for the prevalence of diabetes mellitus, which was almost twice as high in group A as in group B. In contrast, group B fractures included more severe open fractures than group A. Similarly, our treatment outcomes are subject to the presence of type 2 error. According to post hoc power analysis for a β error of 20%, a sample of 92 fractures per group would have been required for the measured nonunion rates to be statistically equal. However, a study on 184 periprosthetic distal femur fractures would seem difficult to achieve practically.
Of the 22 patients excluded from our study, almost one-third (eight patients) had died, six in the first year after operation and two thereafter. Although this loss to follow-up is within the expected range for similar studies dealing with an elderly trauma population, results from the present study have to be interpreted bearing in mind that dissimilar results in the missing sample could theoretically influence our findings.
In the presence of a stable femoral component, our results suggest that extreme distal periprosthetic supracondylar fracture is not a contra-indication to lateral locked plating. These fractures can be managed with internal fixation with predictable results, similar to those seen with more proximal fractures. This study nevertheless reflects the outcomes of specialised trauma surgeons managing highly complex fractures in level one trauma centres. Although the outcomes are encouraging, distal femoral fractures around a TKR represent highly demanding injuries that may lead to considerable complications if inadequately treated.
This study was supported by an educational and research grant from Smith and Nephew (Memphis, Tennessee), AO North America (Paoli, Pennsylvania), Synthes (Paoli, Pennsylvania) and the Foundation of Orthopaedic Trauma.
No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article.
- Received June 26, 2009.
- Accepted December 8, 2009.
- © 2010 British Editorial Society of Bone and Joint Surgery