|Home | About | Journals | Submit | Contact Us | Français|
The medical records and radiographs of 99 patients treated for a periprosthetic femur fracture after total hip arthroplasty over a 17-year period at a single institution were prospectively reviewed. Fractures were classified according to the Vancouver system and stratified as to treatment method. Sixty-six patients had complete records available and a minimum of 12 months follow-up. Overall, 86% of the patients achieved fracture union. The success rate of cemented revision in the B2 and B3 groups was 84%, whereas cement-less revision was 86% successful. The complication rate of surgical treatment was 29%. Fracture union with a stable implant was possible in the majority of cases. Our results support the use of the Vancouver classification as a treatment algorithm.
Periprosthetic fractures of the femur after total hip arthroplasty (THA) represent a difficult treatment challenge. The results of management of periprosthetic fractures have varied greatly due to factors such as bone quality, fracture pattern and method of treatment including nonoperative measures, reduction-fixation, or revision surgery. These treatments have historically been fraught with a high incidence of complications, treatment failures, and compromised outcomes [1–13]. Despite these problems, surgical treatment has become the standard in treating the majority of periprosthetic femur fracture .
The true incidence of periprosthetic fractures is uncertain, with estimates ranging from 0.1 to 2.1% [6, 14, 15]. In data from the Mayo Clinic Joint Registry, the incidence of periprosthetic fracture after primary THA was 1.1%, and it was 4.0% after revision THA [14, 16]. In one series, periprosthetic fracture after THA was the second leading cause of revision, after aseptic loosening . With the increasing prevalence of THA and revision THA in the USA, along with an aging population, the absolute number of periprosthetic fractures can be expected to increase as well.
Currently, the treatment of periprosthetic fractures has included nonoperative measures, open reduction internal fixation (ORIF) as well as revision. Due to the number of confounding variables regarding treatment, Duncan and Masri developed a system of classification of periprosthetic fractures according to location, implant stability, and degree of bone loss (Vancouver Classification ) and advocated an algorithm towards treatment. The purpose of this investigation was to examine our experience with the treatment of periprosthetic fractures of the hip over a 17-year period and to corroborate the recommendations of the Vancouver group.
A retrospective review of all periprosthetic femur fractures after THA treated at our institution from 1984 to 2001 was performed. All surgical procedures were performed at a single institution, but these were performed by a total of 12 different surgeons. The medical records of all patients were reviewed for such factors as patient age, sex, type of treatment, operative findings, number and type of complications associated with the fracture and its treatment, and condition of the patient before and after surgery. Radiographs were reviewed to ascertain fracture type, type of prosthesis (cemented or noncemented), the presence of preexisting osteopenia, evidence of healing, and condition of the prosthesis at latest follow-up.
Fractures were classified according to the Vancouver classification of Duncan and Masri [17, 18]. This classification system, described in previous publications, is now widely accepted in the classification and treatment of periprosthetic fractures of the femur. Type A fractures are fractures of the greater or lesser trochanters. Type B fractures involve the femoral diaphysis and/or metaphysis around the femoral stem and are subdivided into three types. In type B1 fractures, the stem is stable, B2 fractures are associated with a loose stem, and B3 fractures are associated with a loose stem and significant bone loss. Type C fractures are well distal to the tip of the femoral stem.
The treatment was classified as (1) nonoperative; (2) reduction and fixation employing interfragmentary or cerclage fixation, plate fixation (including flat plates, plate-clamp devices, and plates with fixed-angle devices), or plate fixation with use of cortical strut grafts; or (3) revision surgery using either uncemented in-growth stem or cemented stem revision. In cases of femoral stem revision, cables were commonly used for fixation, but supplemental plates were not used. The use and type of bone graft material was also recorded.
The stability of the femoral prosthesis was judged from the radiographs at final follow-up. Criteria for judging the stability of cement-less stems were those described by Engh et al.  and include the presence of spots welds, lack of radiolucent lines, absence of pedestal formation, absence of calcar remodeling, and no evidence of migration. For cemented femoral components, the criteria of Harris et al.  were utilized. Stems that had migrated on serial radiographs were considered as definitely loose, stems with a continuous radiolucent zone at the bone–cement interface on either anteroposterior or lateral radiographs were considered as probably loose, and stems with a radiolucent zone occupying between 50 and 100% of the bone–cement interface were considered as possibly loose.
Patients who had not been recently evaluated were asked to make an office visit for routine clinical evaluation and radiographs. Patients who declined to return for follow-up did so for a variety of reasons, which commonly included remote geographic location, advanced age, and difficulty with travel. Patients who could not come to the hospital for follow-up were interviewed over the telephone. The study had the approval of the Institutional Review Board. Patients who had less than 12 months of follow-up and could not be contacted or had died were excluded.
In this study, the main outcome measure was either fracture union or implant survival. Treatment within each fracture type was stratified and chi-square analysis was performed to determine the level of statistical significance in regards to treatment outcome at the 95% confidence level.
There were 99 periprosthetic femur fractures after THA treated during this time period. These consisted of 32 males and 67 females with an average age of 68.1 years. Demographic data are listed in Table 1.
Twenty-five patients had died with less than 12 months of follow-up. These by and large tended to be elderly patients, frequently with multiple medical comorbidities. An additional eight patients with less than 12 months of follow-up could not be contacted. All of these eight patients were two or more years postsurgery at the time of this review and were lost to follow-up despite all attempts to locate and contact them. These 33 patients were therefore excluded from the outcome analysis. The study group for analysis of treatment outcomes and complications therefore comprised 66 patients with a minimum of 12 months of follow-up. There were 21 males and 45 females. The average age was 64.0 years, with a range of 17 to 96 years. The right hip was involved in 39 patients and the left hip in 27 patients. The diagnoses leading to hip arthroplasty were osteoarthritis in 36 patients, rheumatoid arthritis (RA) or juvenile rheumatoid arthritis (JRA) in 16 patients, posttraumatic arthrosis in eight patients, osteonecrosis in five patients, and developmental dysplasia of the hip in one patient. The periprosthetic fracture involved a primary THA in 33 cases, a revision THA in 29 cases, and a primary hemiarthroplasty in four cases. The Vancouver classification of this group of fractures was 2 type A, 25 type B1, 20 type B2, 9 type B3, and 10 type C.
The patients had an average of 2.2 previous surgeries on the involved hip prior to the periprosthetic fracture. The average time period between the original hip replacement procedure and the periprosthetic fracture was 109 months. The period of follow-up from the time of the periprosthetic fracture ranged from 12 to 180 months, with an average of 62 months.
There were two patients with type A fractures of the greater trochanter. Both of these patients underwent operative treatment consisting of ORIF utilizing a plate/clamp device with implant retention. The fracture united in both cases (100% union). One patient sustained a dislocation 8 months after the fracture surgery, which was treated with closed reduction. The second patient underwent elective hardware removal at 1 year postoperatively for lateral trochanteric pain but had no other complications.
The type B1 subgroup consisted of 25 patients with 7 uncemented and 18 cemented stems. Data for this group are listed in Table 2. Of these, four were treated nonoperatively and 20 were treated with ORIF: three with cerclage wiring, nine ORIF with a plate device, eight ORIF with a plate device plus strut allograft, and one underwent revision with an uncemented femoral stem. At latest follow-up, 18 patients (72%) achieved union of their fracture after the initial treatment, three patients (12%) required one or more further procedures to achieve union, and four patients (16%) had an established nonunion. Three patients who sustained a second periprosthetic fracture of the same femur had union or early healing at latest follow-up. Overall, 22 (88%) patients had a stable implant at final follow-up and three patients had implants that were considered to be loose.
There were four treatment failures: two in the nonoperative group (two of four), one with ORIF with plate alone (one of nine), and one ORIF with plate and allograft strut (one of eight). Two of the four patients initially treated nonoperatively subsequently failed to heal the fracture and were considered treatment failures. Both of these patients had an established nonunion at latest follow-up despite subsequent revision of the femoral component. At final follow-up, 18 of 20 patients treated with ORIF achieved union (90%). Of the nine patients treated with ORIF and a plate device, six achieved fracture union after the primary treatment, two required reoperation for delayed union, and one fracture was a nonunion. Two of these six patients sustained a second periprosthetic fracture between 1 and 3 years later, both of which were healing at latest follow-up after repeat surgical treatment. Of eight patients who had ORIF with a plate device plus a strut allograft, seven had primary fracture union and one went on to an established nonunion. One patient with a B1 fracture around an S-ROM (Depuy Orthopaedics, Warsaw, IN, USA) stem underwent revision to a longer S-ROM stem with retention of the ingrown sleeve component, as well as stabilization with a strut allograft and cables. This patient healed successfully without complications and had a stable component at latest follow-up. Union rates by treatment type were as follows: nonoperative treatment 2/4 (50%) and ORIF treatment 18/20 (90%).
There were nine complications associated with treatment in five patients. These included one nonfatal pulmonary embolism (PE), three soft-tissue infections, one deep joint-space infection, two hematomas requiring surgical drainage, and two patients with one or more subsequent dislocations. All three soft tissue infections were treated successfully with surgical irrigation and debridement plus antibiotics, and all three subsequently had successful union of their fractures. The deep joint space infection was associated with nonunion of the fracture and was treated with long-term antibiotic suppression due to a compromised medical condition and limited global function.
This subgroup consisted of 20 patients with 2 uncemented and 18 cemented stems. Two patients underwent ORIF of the fracture with a plate, 13 underwent treatment with cemented revision of the femoral stem, and 5 underwent uncemented revision of the femoral stem. Data are listed in Table 3. Overall, 17 patients achieved union after the primary treatment (85%), two patients’ fractures united after a second procedure, and one patient had an established nonunion at latest follow-up, for a 95% success rate. Three patients in this group sustained a second periprosthetic fracture and required further surgical treatment. All three of these patients had demonstrated union of the second fracture at latest follow-up. Seventeen patients (85%) had a stable implant at latest follow-up. Three patients had a loose femoral stem, one of which was associated with an established nonunion, one of which was awaiting revision, and one in which the patient had died.
Both of the patients treated by ORIF with a plate and morcelized bone graft (100% union) subsequently healed despite the presence of a loose femoral stem. One of these patients had no further surgery and still had a loose implant at latest follow-up. The other sustained a second periprosthetic fracture 4 years later, which was treated with a cemented femoral revision. This patient had a healed fracture and a stable implant at latest follow-up. Of the 13 patients who had a cemented femoral revision, 11 healed without another procedure (85%), one healed after placement of additional internal fracture fixation, and one had an established nonunion despite having undergone a second revision procedure to an uncemented femoral component with placement of a plate and allograft struts. Of the five patients treated with an uncemented femoral revision, four achieved primary fracture union (80%) and one required a repeat revision of the femoral component before healing. There was no difference in union rates between the ORIF group, the cemented revision group, or the uncemented revision groups (p=0.53).
There were a total of seven complications associated with treatment, which occurred in six patients. They were one postoperative cardiac arrhythmia, one nonfatal PE, one superficial wound infection requiring irrigation and debridement, one septic joint requiring two-stage implant exchange, two dislocations treated nonoperatively, and one postoperative hematoma requiring surgical drainage.
There were nine patients with type B3 fractures around one uncemented and eight cemented stems. Data are listed in Table 4. Six patients were treated with cemented femoral revision, two patients with uncemented femoral revision, and one patient with ORIF. Seven patients achieved fracture union after the primary treatment (78%), one patient’s fracture united after further surgery, and one patient’s fracture never healed despite multiple surgeries. All patients in this group were considered to have stable implants at latest follow-up.
Five of six patients who were treated with a cemented femoral revision (three with strut allografting, two without) went on to fracture union (83%). One patient with RA who was initially treated with a cemented femoral revision had multiple subsequent surgeries for infection, refracture, and nonunion was ultimately converted to a total femur replacement. Two patients treated with uncemented revision subsequently achieved union without further fracture treatment. One woman with JRA and severe bone loss was initially treated by ORIF with a plate, cables, and morcelized allograft. She developed a deep space infection and delayed union of the fracture. After hardware removal and treatment of the infection, she was successfully treated with an uncemented femoral revision. The results demonstrated higher union rates in the revision group vs. the ORIF group (p=0.047)
Only three patients in this group did not have a postoperative complication associated with the fracture treatment. Of the other six patients, four had one or more dislocations, and three had deep infections requiring implant removal and further revision procedures.
Data for type C fractures are listed in Table 5. There were ten patients in this subgroup with two uncemented and eight cemented stems. Two patients were treated initially with nonoperative treatment, six were treated by ORIF with a plate device, and two were treated by ORIF with a plate plus strut allograft. Of the two patients that were initially treated with traction followed by a cast or brace, one went on to heal without complications (50%) but the other patient developed a nonunion, which was subsequently treated with ORIF plus bone grafting. This surgery was complicated by hematoma formation and recurrent infection, and the patient ultimately required a transfemoral amputation. Eight patients underwent ORIF with or without strut grafting. Of these, four achieved primary fracture union (50%) and two required repeat ORIF but were united at latest follow-up. The other two patients went on to a nonunion. One patient who required multiple surgical procedures for recurrent joint sepsis, osteomyelitis, and nonunion was finally treated with a total femur replacement. The other patient had an established nonunion despite a second ORIF and a subsequent revision to a long-stem S-ROM. Eight out of the ten (80%) had a stable femoral implant at the latest follow-up. Complications associated with treatment in this group consisted of the two patients with recurrent infections mentioned above. There was no statistical difference between groups regarding union whether nonoperative or ORIF was the treatment of choice (p=0.49).
Although the overall rate of periprosthetic femur fractures associated with THA is unknown, it appears to be increasing [21, 22]. Estimates in the literature range from 0.1 to 2.1% for postoperative fractures and from 0.3 to 5.4% for intraoperative fractures . Postoperative fracture rates are generally higher for revision THA compared to primary THA, whereas intraoperative rates are generally higher for uncemented THA compared to cemented THA.
The treatment of periprosthetic femur fractures after THA has historically been associated with a high rate of treatment failures, complications, and unsatisfactory outcomes. Difficulty arises in comparison of various results in the literature due to differences in length of patient follow-up, patient demographics, types of implants used, the number of revision arthroplasties, the types of operative techniques employed, and variable outcome measures utilized. To date, no prospective randomized trial studying treatment interventions has been performed. It has been the work of Duncan and Masri  that has clearly aided in the classification and recommendation for treatment of these complex fractures. According to these authors, treatment is guided by the classification system, with type A fractures generally being treated nonoperatively or with cerclage fixation. Type B1 fractures can be treated by ORIF with plate devices and or the use of cortical strut grafts. Type B2 fractures usually require revision to a long-stem femoral component, with or without additional extramedullary fracture fixation. Type B3 fractures, in addition to revision of the femoral stem, generally require the use of significant allograft in an attempt to reestablish bone stock. Type C fractures can usually be treated by ORIF with standard plate systems or fixed angle plate devices. While these recommendations occurred after the onset of our study, our results appear to be consistent with what these authors have found.
The results of this series are similar to those from previous reports and reflect the difficulty in treating periprosthetic fractures of the femur after THA. In this study, 74% of patients achieved fracture union after the primary treatment, whereas another 12% united the fracture after undergoing further surgical treatment, for a final fracture union rate of 86%. Complications relating to surgical treatment occurred in 19 patients, for a rate of 29% confirming the complicated nature of this type of surgery including several patients with multiple complications. There are several limitations inherent to this type of review including different biases towards treatment over the study time period and the limited number of study patients within some groups. However, clearly some trends are worth noting.
While not statistically significant, among the type B1 fractures, those treated with ORIF and the use of a plate plus a cortical strut graft had a higher healing rate (seven of eight, 88%) than those treated with a plate device alone (six of nine, 67%), although the numbers are small in this subgroup. There has been support in the recent literature for use of onlay allograft cortical struts [4, 23], as well as animal studies investigating the use of osteogenic proteins to enhance strut allograft healing .
In the B2 group the success rate with cemented revision (11 of 13, 85%) was equivalent to that of uncemented revision (four of five, 80%). Not surprisingly, our type B3 fractures were associated with the highest complication rates (66%) in our series, as might be expected given the difficulty of surgical reconstruction of many of these fractures. Cemented femoral revision was surprisingly successful in this subgroup, with five of six patients (83%) healing and having a stable femoral component at latest follow-up. Of note, while both patients in whom ORIF was performed gained union, the implants were still both loose at follow-up, making the patients potentially at risk for fracture. Currently, we would recommend revision of the implant.
In the C group of patients, it was interesting to note that union was difficult to obtain regardless of treatment (50%). Clearly, patient factors such as osteopenia and the effect of stress shielding may play a role in getting these fractures to unite. Consideration of adjuvant bone grafting as well as potential use of bone enhancing agents should be given at the time of surgery.
A clear trend was that patients with RA or JRA fared much worse than the overall group. In addition to having compromised immune systems due to the effects of various disease-modifying agents, these patients also frequently had poor bone stock. Of 16 patients, only eight successfully healed after initial treatment. Six fractures ultimately went on to become nonunions, with two patients being converted to total femur replacement due to recurrent infection, nonhealing, and bone loss. Total femur replacement following periprosthetic fractures in RA patients has also been previously reported in the literature . Three patients sustained a refracture and 6 of the 16 (38%) had one or more infections. These high complication rates highlight the risk to RA and JRA patients whose underlying disease can compromise treatment outcomes.
Recently, authors at the Mayo clinic reported the results of revision arthroplasty for 118 hips with Vancouver type-B fractures . The femoral implants used in this series included cemented stems, proximally porous coated uncemented stems, extensively coated stems, and allograft-prosthetic composites. Overall, Kaplan–Meier analysis showed a 5-year survival rate of 90% and a 10-year rate of 79.2%. They concluded that the best results were seen when an uncemented and extensively porous coated stem was used. There have also been recent reports demonstrating good results with the use of Wagner-type fluted tapered stems for the treatment of Vancouver type-B fractures [12, 13]. The patient numbers were small in these two studies, however.
Beals and Tower  reported the results of 93 periprosthetic fractures treated at numerous facilities by multiple surgeons. Overall, outcomes were considered excellent in 32% of cases, good in 16%, and poor in 52%. Most fractures except trochanteric fractures had a poor result with nonoperative treatment. They also found that the results of ingrowth revision were superior to those for cemented revision for all fracture types. The complication rates were 41% relating to the fracture and 33% relating to the arthroplasty.
Sledge and Abiri  reported on seven patients with Vancouver B2 fractures treated in a standardized manner with removal of the implant, cerclage wiring of the fracture with allograft struts, and implantation of a long, uncemented stem. A modular, proximally loading implant was used in three cases, and a curved, fully porous-coated stem was used in four cases. At 33 months average follow-up, they reported no failures, no refractures, and an average Harris hip score of 83. All patients except one returned to their preoperative ambulatory status. They did not comment on complications.
Enthusiasm for operative treatment of complex periprosthetic fracture gained popularity based upon the work of Mont and Maar , whose meta-analysis of 26 published reports looked at periprosthetic fractures in 487 patients. While comparisons are difficult due to a lack of uniformity in fracture classification and reporting of outcomes, they found the most satisfactory outcomes for type 2, 3, and 4 fractures (around the tip of the stem) with cerclage fixation or long-stem revision. Screw/plate fixation and nonoperative treatment with traction generally lead to unsatisfactory outcomes. Highly comminuted type 5 fractures appeared to be best treated with long-stem revision.
Our results demonstrate good success rates in treating the majority of periprosthetic femur fractures. However, surgeons should be alerted to the high rate of complications in this group. When presented with a periprosthetic fracture, we feel that the Vancouver classification system provides excellent stratification and uniform terminology. In addition, our data support the treatment algorithm recommended by the Vancouver group. While one of the limitations of this study includes small numbers of subjects, especially in the nonoperative treatment groups, we believe that depth of analysis of the surgical group provides ample support for our conclusions. Future studies employing prospective randomized trials should be conducted in an attempt to confirm classification systems and outcome measures. However, based upon our data, we recommend the use of the Vancouver classification system and treatment algorithm for these difficult fractures.
No institutional or private financial support was available for this study. The views expressed in this article are those of the authors and do not reflect the official policy or position of the Department of the Navy, the Department of Defense, or the United States Government.