A Review of Periprosthetic Femoral Fractures Associated With Total Hip Arthroplasty

Abstract

Periprosthetic fractures of the femur in association with total hip arthroplasty are increasingly common and often difficult to treat. Patients with periprosthetic fractures are typically elderly and frail and have osteoporosis. No clear consensus exists regarding the optimal management strategy because there is limited high-quality research. The Vancouver classification facilitates treatment decisions. In the presence of a stable prosthesis (type-B1 and -C fractures), most authors recommend surgical stabilization of the fracture with plates, strut grafts, or a combination thereof. In up to 20% of apparent Vancouver type-B1 fractures, the femoral stem is loose, which may explain the high failure rates associated with open reduction and internal fixation. Some authors recommend routine opening and dislocation of the hip to perform an intraoperative stem stability test to rule out a loose component. Advances in plating techniques and technology are improving the outcomes for these fractures. For fractures around a loose femoral prosthesis (types B2 and 3), revision using an extensively porous-coated uncemented long stem, with or without additional fracture fixation, appears to offer the most reliable outcome. Cement-in-cement revision using a long-stem prosthesis is feasible in elderly patients with a well-fixed cement mantle. It is essential to treat the osteoporosis to help fracture healing and to prevent further fractures. We provide an overview of the causes, classification, and management of periprosthetic femoral fractures around a total hip arthroplasty based on the current best available evidence.

Introduction

Periprosthetic femoral fracture in association with total hip arthroplasty (THA) was first reported in 1954. ¹ Since then, the incidence has steadily increased as the indications for THA have broadened and the life expectancy of the population has increased. ^{2 –5} The current overall incidence of periprosthetic femur fracture is approximately 4.1%, with higher rates for uncemented and revision THA. ² Late periprosthetic fractures account for approximately 6% of revision cases ⁶ and are the third most common reason, after aseptic loosening and infection, for revision surgery. ⁷ Such fractures are difficult to treat. Historically, nonoperative treatment such as traction has yielded poor results. ⁸ Operative management has the advantages of early mobilization and reduced hospital stay. It also affords a reduction in systemic and local complications such as malunion and nonunion. ⁹ In this article, we review the classification, causes, and management of periprosthetic femoral fractures around a THA, based on the best available evidence.

Diagnosis

Diagnosis is typically made with a clinical history of pain and injury. It is very important to ask about the function of the joint before the injury. A history of pain or dysfunction may indicate prosthetic loosening or infection. Other signs of infection should be evaluated.

Conventional radiographs are the workhorse for diagnosis of periprosthetic fractures. High-quality radiographs are essential to look for radiolucent lines around the prosthesis or cement. A comparison to older radiographs is optimal to give the best chance of detecting loosening. In some cases, computed tomography may allow further visualization of fracture lines and provide evidence of prosthetic loosening including radiolucent lines around the prosthesis or cement mantle or osteolysis.

If there is loosening, a white blood cell count, erythrocyte sedimentation rate, and C-reactive protein analysis should be obtained to evaluate for possible joint infection. Normal inflammatory markers help to rule out infection, but these markers may be elevated from the fracture itself rather than from an infection. One study of 204 patients with periprosthetic hip fractures recorded an elevated white cell count, erythrocyte sedimentation rate, and C-reactive protein in 16.2%, 33.3%, and 50.5% of patients, respectively. ¹⁰ The true infection rate in the group was 11.6%, and the positive predictive value for these markers was poor (18%, 21%, and 29%, respectively). ¹⁰ If infection is highly suspected, a joint aspiration for cell count, differential, and culture should be obtained, even if doing so delays surgery. If the joint is explored, intraoperative cultures and pathology specimens should be checked to help rule out periprosthetic infection.

Classification

Femoral periprosthetic fractures may occur intraoperatively or postoperatively. The widely used Vancouver classification, developed by Duncan and Masri, ¹¹ provides a practical assessment of postoperative femoral fractures according to the level of the fracture and the presence of a well-fixed or loose component. This system has proved to be reliable and valid, showing good correlation between radiographic evaluation and intraoperative findings, even when used by nonexperts. ^12,13

The Vancouver classification for postoperative fractures divides the femur into anatomical zones (Figure 1). Type-A fractures occur within the proximal metaphysis and are subclassified as A_G or A_L for fractures around the greater and lesser trochanters, respectively. Type-B fractures occur at or just below the distal tip of the femoral prosthesis and are subclassified as follows: type-B1 fractures occur around a well-fixed component; in type-B2 fractures, the component is loose but has sufficient bone stock for straightforward revision surgery; and in type-B3 fractures, the component is loose, with substantial osteolysis and bone loss. Type-C fractures occur well distal to the stem, so implant stability is not an issue.

Figure 1.

Vancouver classification of postoperative periprosthetic femur fractures.

The Vancouver classification of periprosthetic femoral fractures has been modified to include intraoperative fractures and perforations: Type-A fractures are confined to the proximal metaphysis, type-B fractures involve the proximal diaphysis, and type-C fractures extend beyond the longest revision stem and may include the distal femoral metaphysis. ¹⁴ Each type is subdivided into simple perforations (subtype 1) and undisplaced (subtype 2) or displaced (subtype 3) fractures. If stress risers are present intraoperatively, bypassing them with a stem that has sufficient length to extend 2 to 3 cortical diameters distal to the defect is considered effective in preventing subsequent fracture ^{4,5,15 –17} For medial calcar fractures that occur on insertion of uncemented stems, cerclage wires placed just proximal and distal to the lesser trochanter are usually adequate to prevent fracture propagation. ¹⁸

Postoperative periprosthetic fractures tend to occur years after primary or revision surgery. In 1049 periprosthetic femoral fractures from the Swedish National Hip Arthroplasty Register, the mean time to fracture after surgery was reported as 7.4 years and 3.9 years in primary and revision THA patients, respectively, and 70% of cases occurred in the presence of a loose component. ⁶ Using the Vancouver classification, the Swedish Hip Register reported that more than 80% of periprosthetic fractures are type-B fractures. ⁶ For primary stems, approximately 24.5% are stable (type-B1 fractures), and 75.5% are associated with a loose stem (types B2 and B3). ⁶ For periprosthetic fractures that occur after revision THA, approximately 51% of stems are type-B1 fractures, and 49% are classified as loose (type B2 or B3). Type-B3 fractures that occur in the presence of massive bone loss account for only 4% of periprosthetic fractures. ⁶

Predisposing Factors/Mechanisms of Injury

Several factors are known to predispose to the development of a periprosthetic fracture. These factors include female gender, the presence of rheumatoid arthritis, the presence of large osteolytic lesions (especially in high-stress anatomical regions ^3,15 and in younger patients with high activity levels ³ ), and advanced age, possibly as a result of osteoporosis. ^19,20

Most postoperative periprosthetic femur fractures result from low-energy falls from sitting or standing heights. ^6,21 A series of 71 cases of postoperative periprosthetic fractures recorded minor trauma in 87% of cases, spontaneous fractures in 9%, and major trauma in 4%. ²¹ Chakravarthy et al ²² reported 12 cases of periprosthetic femur fractures with well-fixed components, and in all cases the mechanism of injury was a low-energy fall. However, in the presence of severe osteolysis, more than 50% of patients will develop spontaneous fractures without prefracture symptoms. ^4,5

Surgical Technique

Surgical technique in particular influences the risk of fracture. ²⁰ For intraoperative fractures, the risk of fracture is directly related to implant type. Results from the Mayo Clinic joint registry indicated an intraoperative periprosthetic femoral fracture incidence of 0.3% in 20 859 patients undergoing primary cemented THA and 5.4% in those undergoing uncemented THA. ² The higher incidence for uncemented THA is likely to be related to the increased force necessary to obtain the stem-press fit required for initial component stability. Revision THA is associated with higher rates of intraoperative femoral fracture for cemented and uncemented components, approximately 3.6% and 20.9%, respectively. ² Technical factors that potentially increase the risk of postoperative fracture include the presence of cortical defects from the creation of a cortical window, inadvertent cortical penetration during reaming, ²³ and empty screw holes.

Metabolic Bone Disease Associated With Periprosthetic Fracture

Osteoporosis is considered to be an independent risk factor for late periprosthetic fracture. ^19,24,25 A study of 16 periprosthetic fractures in 454 arthroplasty patients found that preoperative osteoporosis was a significant predictor of fracture risk. ²⁶ Periprosthetic fractures are also more common in patients who have sustained hip fractures before arthroplasty. ^19,20

Vitamin D deficiency

Vitamin D deficiency may play a role in the development of osteoporosis, subsequently increasing fracture risk. ²⁷ However, it is also evident that vitamin D deficiency in elderly individuals is associated with inferior physical activity levels, gait speed, and balance. ²⁸ Vitamin D supplementation in elderly patients at risk for fracture has been shown to reduce the risk of falling possibly by improving muscle function and reducing body sway. ^29,30 A randomized controlled study of hip fracture patients showed that vitamin D supplementation orally or by injection improved patient balance, increased bone mineral density, and reduced the risk of falls. ³¹ Although research is needed to define the incidence of vitamin D deficiency in periprosthetic femoral fracture patients, a substantial proportion of patients do have osteoporosis, so it could be expected that the incidence of vitamin D deficiency is high. ^19,24,26 We consider it reasonable to assess all patients admitted with a periprosthetic fracture for vitamin D deficiency and commence supplementation as indicated. The exact definition of vitamin D deficiency is unclear, but it is recommended that vitamin D levels above 20 ng/50 mL are optimal for bone health and that a dose of 800 IU of vitamin D3 plus a total calcium intake of 1200 mg/d can reduce fractures. ³² The Institute of Medicine of the National Academies now recommends such treatment for most elderly patients, and it is likely that much higher supplemental doses of vitamin D2 will be advocated in future. ³³ Greater levels of vitamin D deficiency may require repletion at doses of vitamin D2 up to 50 000 IU/d. For patients admitted with a periprosthetic fracture, a metabolic bone consultation with a geriatrics physician would seem sensible to help optimize fracture care.

Few studies exist regarding the efficacy of vitamin D with regard to osteoporotic fracture healing. The results of some animal studies are promising. ^34,35 In an osteoporotic rat model, 1,25-dihydroxy vitamin D3 resulted in greater ultimate load and energy absorption compared with controls. ³⁵ Fracture callus was also better remodeled, leading the authors to conclude that 1,25-dihydroxy vitamin D3 may have a role in the treatment of osteoporotic fractures in humans. ³⁵ Some controversy exists, however, because 1 animal study has shown that a vitamin D3 analogue inhibited callus remodeling by inhibiting osteoclasts, although this process had no apparent effect on the natural healing process. ³⁶ Despite the high prevalence vitamin D deficiency and the strong correlation with increased fracture risk, ²⁷ additional research in human participants is required to assess the role of vitamin D supplementation during fracture healing.

Bisphosphonates

Bisphosphonates are antiresorptive agents and are now the most commonly used antiosteoporotic agent. ³⁷ Although this characteristic may provide a benefit in the battle against osteoporosis, it potentially slows down bone healing, although clinical data are sparse. According to 1 rat model study, alendronate increased bone strength through the production of greater callus volume and increased enzymatic cross-links but did delay the remodeling of woven bone to lamellar bone. ³⁸ One cohort study of 19 731 patients suggested that the risk of humeral fracture nonunion was double when bisphosphonates were used postfracture, but interestingly, the risk of nonunion in patients already on bisphosphonates before fracture was the same as those patients who did not receive this treatment. ³⁹ Suggestions that bisphosphonate use for osteoporosis actually poses an increased risk of developing atypical subtrochanteric fractures are unproven. ⁴⁰

The timing of bisphosphonate therapy remains uncertain. In a hip fracture population, 1 randomized controlled trial showed that the optimal time to give intravenous zoledronic acid was 2 to 12 weeks after hip fracture repair. ⁴¹ Compared with a placebo, there was a reduction in the subsequent fracture risk at 12 months and also in all-cause mortality at 16 months in patients treated within 90 days of fracture. ⁴¹ There is 1 concern regarding the use of intravenous administration: this mode of therapy is relatively expensive and has a potential disadvantage compared with oral administration in the acute fracture situation because a substantial portion of the bisphosphonate may be sequestered at the fracture site rather than being dispersed throughout the skeleton. To avoid unwanted wastage, it may be sensible to withhold intravenous bisphosphonates for 2 to 6 weeks after fracture, although this protocol is not evidence based.

Parathyroid hormone

Parathyroid hormone (PTH) increases serum calcium levels by enhancing gastrointestinal absorption, increasing renal calcium and phosphate reabsorption, and liberating calcium from the skeleton. ⁴² Although prolonged exposure to PTH increases osteoclast activity, short bursts of exposure stimulate osteoblasts and result in increased bone formation. ⁴³ Administration of PTH for fracture repair is at present considered an exciting prospect for the optimization of fracture repair and is already approved for the treatment of osteoporosis in the United States. ⁴²

Animal studies ^44,45 have shown that PTH can speed up fracture healing in simulated conditions of aging and estrogen withdrawal, which is relevant to elderly individuals who sustain periprosthetic fractures. Considering the evidence from animal studies, PTH may have a role in adjunctive fracture treatment after surgical fixation. In 1 study ⁴⁶ of 65 osteoporotic female patients with pubic rami fractures, 21 patients were assigned to receive 100 ug of PTH 1-84. All patients received 1000 mg of calcium and 800 IU of vitamin D. The results showed that the patients who received PTH had significantly faster fracture healing, at a mean of 7.8 weeks compared with 12.6 weeks for the control group. Another study of 102 postmenopausal female patients with nonoperatively managed distal radius fractures suggested that 20 ug of PTH 1-34 resulted in significantly accelerated bone healing compared with a control group. ⁴⁷ However, the higher dose of 40 ug of PTH 1-34 had no apparent benefit compared with the control group who had no adjunctive therapy, raising doubt over the validity of the study. ⁴⁷

It seems that based on animal studies ^44,45 and preliminary clinical data in humans, ^46,47 PTH may enhance bone healing in patients with osteoporosis who sustain a periprosthetic fracture. Before such therapy can be recommended, however, additional clinical research must be conducted.

Surgical Management

Vancouver Type-A Fractures

Fractures of the greater or lesser trochanter tend to be associated with osteolysis, but they also can occur intraoperatively in patients with good bone stock. ⁴⁸

The treatment for greater trochanter fractures (type A_G) depends on the degree of displacement involved. Type-A_G fractures that are minimally displaced and considered stable have historically been treated nonoperatively with protected weight bearing or an abduction brace. ^4,5 However, the clinical outcome may be unreliable. ^49,50 Type-A_G fractures that are displaced can be difficult to manage, especially in the presence of osteolysis where stable fixation is difficult to achieve. ^3,51 Surgical methods of fixation used have been similar to that described for trochanteric osteotomy, namely, with wires, screws, cables, or specialized plates. ⁵² Cable grip systems provide a more rigid fixation than cables alone and have lower rates of nonunion and trochanteric migration. ⁵² One small case series reported the successful use of a simple technique with a tibial locking plate fixation, a single cable, and allogenic bone graft for trochanteric fractures associated with marked osteolysis. ⁵⁰ For additional fixation, the cable plate may be extended beyond the prosthesis to obtain bicortical screw fixation in the femur. Cerclage wires in conjunction with morselized bone graft are another viable alternative, with a reported 95% union rate in 1 study of 19 patients who had trochanteric fractures associated with osteolysis secondary to loose acetabular components; the mean time to union was 5 months. ⁵¹ After such fixation, some authors recommend an abduction brace and protected weight bearing for up to 3 months. ^50,51

Fractures of the lesser trochanter (type A_L) are rare and are usually avulsion fractures occurring through osteoporotic bone or areas of osteolysis. ⁴⁸ They must be carefully assessed because a fracture of the medial cortex may destabilize the implant, necessitating surgical intervention. If the implant is considered stable, nonoperative treatment is warranted but should be followed up closely with radiography to monitor for implant loosening. Surgical treatment is commonly performed with cable fixation when there is adequate bone stock. However, in the presence of osteolysis, revision surgery with or without bone graft is required to achieve stem stability.

In summary, various options exist for the fixation of proximal metaphyseal (Vancouver type-A) fractures, although the optimal method depends on the cause of the fracture. For intraoperative fractures, if the implant is stable, a cerclage wire or cable is sufficient. If the implant is unstable, it must be revised. In the presence of periprosthetic osteolysis, associated with stem instability, revision surgery to a long diaphyseal fitting stem is indicated to remove the particle debris generator. ^4,5,51 Additional fracture fixation and supplemental bone graft may also be necessary. ¹⁴ Nonoperative treatment must be reserved for cases in which the component is stable. ^4,5

Vancouver Type-B1 Fractures

Vancouver type-B1 fractures occur at or just distal to the femoral stem and are associated with a well-fixed component. Nonoperative treatment is no longer recommended because patients do not tolerate prolonged immobilization and increased risk of subsequent complications such as death, pulmonary infection, and skin ulceration. ^53,54 The rates of nonunion after nonoperative management are high because of inadequate fracture stability and sometimes because of the presence of cement in the fracture site. ^54,55 Also, the length of stay after nonoperative treatment is unacceptably long, with 1 study reporting an average stay of 91 days. ¹⁵ With modern treatment strategies, nonoperative management is reserved only for patients who would not be able to tolerate surgery.

Although surgical intervention is thought to offer the best outcome for Vancouver type-B1 fractures, controversy exists regarding the optimal management strategy. According to the Swedish Hip Register, open reduction and internal fixation of type-B1 fractures has a 34% failure rate. ⁶ This high failure rate has been attributed to misdiagnosis of a stable stem. In 1 study of 45 radiographic type-B1 fractures, when the hip joint was opened and dislocated, 20% of stems were found to be loose, requiring revision surgery. ⁵⁶ Accurate assessment of stem stability is key to a good outcome, leading some authors to recommend routine intraoperative stem stability tests before fixation. ⁵⁶ This approach, however, requires more exposure of the joint than is necessary for plating of the femur, and it has the added potential for postoperative dislocation.

Besides stem stability, consideration should be given to fracture configuration, fracture comminution, the presence of infection, any obstruction to screw insertion by stem or cement, adequacy of bone stock, the general medical status of the patient, and the anesthetic risk. ⁵⁷ The aims of fixation include accurate fracture reduction, fracture union, and early mobilization of the patient so that hip and knee function are preserved. ³ Adequate proximal fixation should be achieved without compromising the cement mantle. Good fixation in bone is less easily achieved in the presence of osteoporosis, which is common in this group of patients.

Open Reduction Internal Fixation

Notable advances have been made in the implants used for the fixation of periprosthetic femur fractures. Clinical reports often have heterogeneous populations or small numbers of patients and, to our knowledge, randomized controlled trials are not available. ^{54,55,58 –60} It has become apparent that certain methods of internal fixation are unsuitable. The Mennen paraskeletal clamp is associated with early catastrophic failure, ^61,62 and similarly, Partridge bands have poor results because they do not provide adequate fixation and they cause substantial bone resorption. ^15,63 In contrast, the Ogden plate, which uses heavy-duty Parham bands proximally and nonlocking 4.5-mm cortical screws distally, has good clinical results, ⁵⁵ with 1 study reporting rates of 95% union for femoral periprosthetic fractures in 19 patients. ⁵⁵

Minimally invasive plating and locking plates

Minimally invasive plating is feasible and has the potential advantage of preserving the periosteal blood supply with minimal soft tissue stripping, thereby reducing the risk of nonunion or failure. One of the largest clinical series of minimally invasive fracture fixation for Vancouver type-B1 fractures reported the results of 50 patients using dynamic compression plating. ⁶⁰ Fracture union was achieved by 3 months in all 41 patients available for follow-up. One case was complicated by infection. Of the 41 patients, 30 returned to their prefracture function.

Modern internal fixation is frequently achieved with locking plates, which provide relative fracture stability, and potentially preserve the periosteal blood supply to the fractured bone, especially when minimally invasive surgery and indirect fracture reduction techniques are used. Locking plates allow proximal fixation with flat-tipped unicortical screws combined with cerclage wires (Figure 2). ⁶⁰ These plates are curved to match the osteoporotic femur and often are long enough to extend the entire length of the femur. Recent plate designs also allow for variable angle screws to be placed around the prosthesis. Cerclage wires can now be secured to the plate using grommets that prevent slippage of fraying of the wires. Newer surgical tools also permit the placement of cables with less soft tissue damage and improve fracture reduction with limited incisions. These tools provide marked advantages compared with the systems available only 5 years ago. Recently, small locking plate attachments have been developed to achieve screw fixation at the proximal and distal ends of the plate (Synthes, Inc, Paoli, Pennsylvania), which could potentially replace cerclage wires. These plates typically lock into the standard long locking plate and facilitate insertion of up to 4 locking screws that bypass the stem anteriorly and posteriorly for bicortical purchase. Their fixed angle nature is theoretically beneficial in osteoporotic bone, although the effectiveness of these screws in a clinical situation is unknown.

Figure 2.

Locking plate fixation for Vancouver type-C fractures around a well-fixed stem. Proximal fixation is achieved with screws below the prosthesis and unicortical screws angled around the prosthesis. A, The long condylar plate. B, Anteroposterior view of the outrigger device enabling screws to be placed around the stem of the prosthesis. C, Lateral view with 2 screws anteriorly and 2 screws posteriorly to the stem.

Clinical results are limited to case series. Schutz et al ⁶⁴ analyzed 117 cases of Less Invasive Stabilization System plating (Synthes, Inc) using minimally invasive techniques. Of 14 patients with fractures close to a hip prosthesis, only 1 failed. Chakravarthy et al ²² reported the results of 12 patients treated with locking plates for Vancouver types-B1 and -C periprosthetic femur fractures. In the 6 type-B1 fractures, with a well-fixed femoral component, locking compression plates were applied using an open approach. Bone graft was not used, and union was achieved in all cases at a mean of 5 months, with acceptable alignment maintained, despite patients being allowed to bear partial or full weight immediately after surgery.

Despite the good results reported, ^22,60,64 single locking plate fixation may not offer the optimal fixation, and a recent retrospective review of 14 patients with short oblique or transverse type-B1 fractures reported less satisfactory results. ⁵⁹ Five of the 14 fractures in that report ⁵⁹ were treated with a locking plate and allograft, and 9 were treated with a locked plate alone. The construct was a long locked plate with distal bicortical fixation and proximal unicortical fixation, without cerclage wires. All fractures were anatomically reduced before plate application, and minimally invasive approaches were not used in any case. Of the 14 fractures, 8 healed uneventfully at an average of 5.4 months. Of the remaining 6, 3 had plate fracture at an average of 8.7 months, and 3 had plate pullout at an average of 2.8 months. All but 1 of these failures occurred in constructs in which strut allograft had not been used, leading the authors to conclude that locking plates alone are insufficient for the treatment of periprosthetic femur fractures and should be supplemented with cortical strut grafts.

Absolute recommendations for the use of an additional strut are difficult to make. Certainly, some patterns of fracture seem more unstable than others, particularly simple patterns at the base of a well-fixed implant. More unstable fractures may benefit from an allograft strut at 90° to the plate.

Optimal plate/screw configuration

Great debate exists regarding optimal plate configuration. Although minimally invasive plating may preserve the biology at the fracture site, inadequate fracture reduction may lead to early failure and, therefore, accurate fracture reduction is the priority over soft tissue exposure.

Methods including single- and double-plating, single- and double-allograft struts, and combinations thereof are possible. ^{65 –69} Biomechanical studies have attempted to determine the strongest constructs, although validation with clinical studies is rare. ^{65 –70} A cadaveric study showed that dual fixation with a plate and strut graft or strut grafts only was significantly more stable than a plate-only construct when loaded at 1.5 times the body weight. ⁶⁹ In contrast, a similarly conducted cadaveric study reported that plate fixation with cables proximally and screws distally (Ogden system) was significantly stronger than 2 allograft struts with cables. ⁶⁵

The number of screws used can vary, but engaging at least 8 cortices distally and 4 proximally for fractures around the stem (Vancouver type B1), and only 4 cortices proximally and distally for fractures more proximal around the stem, provide adequate fixation according to 1 clinical series using a locking plate. ⁵⁸ Good results using a 14-hole dynamic compression plate, with at least 8 cortices on either side of the fracture site, have been described. ²³ A 7-mm bone cortex or cement mantle anterior or posterior to the prosthesis allows good screw purchase with nonlocking 4.5-mm screws. ²³ Flat-tipped unicortical locking screws are now available that provide proximal fixation and rotational stability without disruption of the cement mantle and can be combined with alternate adjacent cerclage wires. ⁶⁰

Inserting screws close to and far from the fracture site is advocated. ⁷¹ However, stiff fracture constructs with too many screws at the fracture site may lead to nonunion or plate fracture. The working length of the plate, as defined by the distance between the proximal and distal screws at the fracture site (Figure 3), should theoretically be as long as possible to reduce fracture strain and the subsequent associated risk of nonunion and plate fracture. Therefore, some authors advocate leaving 3 or 4 screw holes empty at the fracture site, ⁷² resulting in an increased working length and avoiding local stress concentration. This construct is thought to provide a stable stimulus for indirect bone healing with callus formation.

Figure 3.

Working length (WL) of a plate. Excessive use of screws close to the fracture site should be avoided because it creates excessive plate stiffness and high fracture strain, potentially leading to nonunion and plate fracture. (Modified with permission from Pletka et al, ⁷³ figure 1B.)

Optimal plate length is not known. However, long plates require greater soft tissue stripping and plate contouring than do short plates, and 1 cadaveric study concluded that long locked plates (20 holes) do not offer a significantly better plate survival over short locked plates (12 holes) when cyclically loaded. ⁷³ It is clear, given the debate that exists, that additional clinical studies are required to determine the optimal plate configuration and length.

Vancouver Type-B2 and -B3 Fractures

Approximately 75% ⁶ of periprosthetic fractures around a primary femoral stem are associated with a loose stem. Loose components, for example, occurring secondary to third-body particle wear and subsequent osteolysis, require revision surgery so that the underlying causes that may have contributed to the fracture are addressed. ^4,5,9 For revision THA, the surgeon should choose a stem for which good long-term outcomes for straightforward revision surgery have already been established.

There are several principles that should be adhered to when performing revision surgery for periprosthetic fracture. Bone stock should be preserved as much as possible. ^4,5 Stable fixation of the stem into intact host bone is best achieved with distal fixation. ^4,5 The fracture should be propagated proximally to allow direct access to the prosthesis, followed by removal of the implant and loose bone and cement debris. Unless the implant is grossly loose and easily extractable, an extended trochanteric osteotomy should be performed to allow for access to the proximal femur. If necessary, the acetabular component should be revised at this stage. ⁷⁴ The distal extremity of the fracture should then be identified, and a cerclage wire should be applied just distal to the fracture to prevent further propagation of the fracture when the femur is reamed or the revision prosthesis is inserted. ⁷⁴ Although fracture healing may be possible in the presence of cement at the fracture site, ^75,76 it is generally recommended that cement extrusion into the fracture site during cementation and implant insertion be avoided because cement interposition may prevent bone union. ^4,5,15,23 If possible, the femur may be reconstructed as a tube, and the implant can then be inserted. In some cases, femoral reconstruction as a tube is not possible, and the implant has to be potted distally and the remaining bone cerclaged around the implant. The stem length should be sufficient to achieve adequate distal fixation and should extend approximately 2 to 3 cortical diameters beyond the fracture site. ^4,5,16,77 Distal fixation may be achieved with different stem designs, such as modular or monoblock stems that are cylindrical or conical. Femoral components with distal locking screws are also available, ^4,5,78 although problems with screw breakage and component subsidence have been reported. ⁷⁸

The dislocation rate is relatively high for patients who have revision surgery for periprosthetic femur fractures, especially when multiple previous operations have rendered the abductor mechanism deficient. ⁷⁹ It is vital to restore the correct soft tissue tension and limb length and also achieve the correct femoral and acetabular rotation and version. Every effort should be made to anchor the greater trochanter to prevent trochanteric escape. The acetabular component should be revised as necessary if loosening or polyethylene wear are evident. Using a constrained acetabular liner and large femoral heads may improve stability. ^79,80

For Vancouver type-B3 fractures with a loose stem associated with severe proximal bone deficiency, revision of the stem and restoration of bone stock are prioritized. ⁸⁰ Options include using a distal fixed stem that is cemented or uncemented, an allograft prosthesis composite, or a tumor prosthesis. ^79,81,82 Regardless of the method used, the proximal part of the femur should be retained and wrapped around the prosthesis. A constrained acetabular liner may be helpful if there is intraoperative instability because of soft tissue deficiency. ⁸⁰

Cemented revision THA

Revision of a loose stem has traditionally been performed with a long-stem cemented prosthesis, although complication rates can be high. ⁸³ Springer at al ⁸² reported the results of 42 patients with type-B2 and -B3 fractures treated with a long-stem prosthesis, with a mean follow-up of 68 months. Six patients underwent revision for loosening, nonunion, and dislocation. Two cases were infected, requiring implant removal. In 18% of the remaining implants that survived, radiographic femoral loosening was evident. Therefore, only 60% of patients had a stable implant with fracture union. An alternative technique in elderly patients (for whom prolonged surgical time is undesirable) is cement-in-cement revision, as long as the cement mantle is well fixed and accurate fracture reduction can be achieved. ⁸⁴

Long-stem uncemented THA

The trend toward uncemented revision hip arthroplasty for type-B2 and -B3 fractures is supported by several reports. ^9,74,85,86 The Mayo Clinic reported lower complication rates for fully coated uncemented THAs in comparison with cemented THAs and proximally porous-coated uncemented stems at 10-year follow-up. ⁸² Uncemented stems tend to be used in younger patients to preserve bone stock. ⁹ At least 5 to 10 cm of bone are required for stable distal fixation ^4,5 to prevent subsidence. ⁸⁰ Overreaming is advised to reduce hoop stresses and the risk of fracture when inserting press-fit stems. ⁸⁷ Poorly designed monoblock stems with proximal porous coating are available but should be avoided because they have been reported to be associated with high rates of loosening and subsidence. ^4,5,82

Clinical results using a fully coated uncemented stem are promising. Union rates at medium-term follow-up range from 83% ⁸² to 100%. ^9,82,86 Supplementary bone graft is frequently required to achieve reliable fracture union. In a series reporting the results of 26 type-B2 and -B3 fractures, ⁹ cortical struts, supplemented with morselized bone graft and demineralized bone matrix, were used in 16 cases. Sixteen patients reported slight-to-moderate thigh pain. No dislocations or infections occurred. Subsidence of tapered implants ranged from 5 to 8 mm. All fractures united, with a mean time to union of 4 months. High union rates were also reported by O’Shea et al, ⁸⁶ who treated 10 type-B2 fractures and 14 type-B3 fractures with an uncemented extensively porous-coated implant. Cortical strut grafts were used in 9 cases, and union occurred in 20 of 22 patients. Nonunion was attributed to infection (1 case) and extensive bone loss (1 case).

Modular systems are available that facilitate accurate positioning of the femoral head. ^80,88 However, the dislocation rate for revision surgery using uncemented stems can be high. Mulay et al ⁷⁴ reported a 21% dislocation rate after revision using a tapered distally fixed modular stem in 24 patients with type-B2 and -B3 fractures. An average of 5 mm of implant subsidence occurred in 17 patients, which may have contributed to joint instability. Union occurred in 91% of patients. Despite the high rate of dislocation and implant subsidence, the mean Harris Hip Score at follow-up was 69, with 90% of patients reporting mild pain to pain-free function, suggesting that if bone union is achieved, hip function is relatively good.

Bone Graft

Bone graft potentially improves fracture healing, improves fracture stability, and restores bone stock, and it has been applied to all types of periprosthetic femoral fracture. Some authors recommend routine cancellous bone grafting of all fracture lines at the time of revision. ^4,5 Autogenous bone graft is osteoconductive and osteoinductive, and it provides osteogenic bone cells. ^89,90 Although effective, supply is limited and donor-site morbidity is common. ⁹¹ Allograft is considered the gold standard of bone augmentation and is typically applied in morselized form or as a strut graft. ⁹⁰

Poor bone stock contributes to the risk of nonunion, as does the presence of devitalized bone. ^55,80 Localized bone defects may act as stress risers and, if not addressed with bone graft, may result in a second fracture. ²³ Generous grafting with cancellous bone is perceived by some authors to provide an optimal environment for bone healing, ^4,5 and some recommend regrafting at 3 months if no evidence of bone healing is seen on radiographs. ¹⁵ Bone graft substitute to enhance strut graft incorporation for selected patients may be useful. ⁸⁰

Cortical strut grafts and impaction grafting

Cortical onlay strut grafts provide mechanical stability, accelerate bone healing at the fracture site, and restore bone stock. ^90,92,93 They have been applied to Vancouver type-B fractures even when good bone stock is present and the femoral component is well fixed. ^53,94 Lewallen and Berry ^4,5 recommend supplemental use of cancellous and onlay cortical strut grafts to achieve rapid fracture union for revision THA used to treat periprosthetic femoral fractures. Duwelius et al ¹⁸ also support the use of allograft struts, placed at 90° to the plate. Dual strut grafting and also a plate and strut graft combination have been shown to provide greater stability compared with isolated plating. ^69,70 Cortical onlay graft plates may also show less stress shielding compared with metal plates. ⁹⁵

Cortical strut grafts offer reliable fracture healing. Chandler et al ⁹⁴ used cortical strut allografts fixed with cerclage wires or cables in 19 patients; 16 resulted in union and excellent clinical function at 4.5 months after surgery. Emerson et al ⁹² described a union rate of 96.6% at a mean 8.4 months after surgery. The disadvantage of strut grafts is that they are technically demanding to apply; extensive soft tissue dissection is often required to place the grafts accurately, and the struts usually need to be contoured with a bur to fit the femur. To preserve the femoral blood supply, the linea aspera must be spared from muscle stripping. ⁹⁰ It has been suggested that each graft should measure 1/3 of the cortical diameter, placed perpendicular to each other on the anterior and lateral aspect of the femur. ⁴⁸ Additional plate fixation potentially protects the strut graft while it is incorporated into the bone. ⁹⁰ One multicenter study of 40 patients with type-B1 fractures reported excellent union rates. Of the 40 patients, 19 were treated with a strut graft and 21 were treated with an anterior strut graft and lateral plate ⁹³ ; 39 fractures healed, and there were 4 malunions. ⁹³

Although fracture healing can occur within 6 months, strut grafts take substantially longer to incorporate into the host bone. Distinct periods of strut incorporation have been described, including commencing with “round off period” at 6 months, followed by “scalloping,” and “bridging.” ^90,96 Complete bone bridging typically occurs as late as 12.5 months. Diffuse loss of radiodensity and the appearance of a trabecular bone pattern within the graft is known as “cancellization” and signifies revascularization.

Impaction grafting with morselized cancellous bone graft is also effective when there is severe bone loss. ⁹⁷ A report of 144 fractures treated with long-stem revision using impaction grafting for type-B2 and -B3 fractures showed that fractures treated with impaction grafting were significantly more likely to unite compared with those fractures simply treated with cemented long-stem revision. ⁹⁷

Proximal femoral replacement

Proximal femoral replacement with allograft is rarely required and may not be suitable in elderly individuals, ⁹⁰ but it remains a viable solution to extensive bone loss and severe comminution. ^80,81 Medium-term clinical results are reasonable although complication rates can be high. ^{79,80,82,93,98} Klein et al ⁷⁹ reported the clinical results for proximal femoral replacement for 21 Vancouver type-B3 fractures with a mean 3.2-year follow-up. The mean age of patients was 78.3 years. Complications included dislocation (2 hips, 1 requiring revision); superficial wound infection (2 hips, requiring debridement and intravenous antibiotics); undisplaced periprosthetic fracture (1 patient, after a fall); and failure of acetabular cage reconstruction (1 patient). Nevertheless, walking function was restored to all but 1 patient.

The major disadvantage of allograft is the potential for disease transmission. ⁵³ Fresh allograft also risks rejection by the host immune system. Freeze drying and irradiation reduces the immunogenicity, allowing long-term preservation of the graft, although irradiation may reduce graft strength. ^90,99 Other disadvantages include the extensive time for healing at the graft-host interface (which usually takes place at 12.5 months ⁹⁶ ), lack of graft availability, and variable graft quality. ⁹⁰ The use of allograft also typically requires more soft tissue stripping around the fracture site for application.

In summary, the treatment of Vancouver type-B fractures generates great discussion and debate, but a logical algorithm has been constructed based on the best available current evidence (Figure 4).

Figure 4.

Algorithm for the management of Vancouver type-B periprosthetic femur fractures. ORIF, open reduction and internal fixation; Intra-op, intraoperative; PFR, proximal femoral replacement.

Vancouver Type-C Fractures

Vancouver type-C fractures occur well below the tip of the stem and do not involve the prosthesis. They account for approximately 10% of periprosthetic femoral fractures around a hip prosthesis. ⁷ Surgical techniques include dynamic compression plates, plates and cerclage wires, and more recently, nonlocking plates. Retrograde intramedullary nailing has also been used. ²¹ Plate length when treating these fractures is an important consideration. Using a short plate may reduce soft tissue dissection and operative time, but it creates a potential stress riser through the weak bone between the tip of the femoral prosthesis and the proximal end of the plate. Therefore, overlapping the femoral prosthesis, using unicortical screws or cerclage wires proximally has been suggested. ¹⁰⁰ Given that these plates span the entire length of the diaphysis, proper centering of the plate is required because inadvertent misplacement may result in pullout of the plate. ^100,101

Although the femoral prosthesis is considered stable, surgical fixation in this group of patients can have high complication rates. In 1 study of 71 periprosthetic femoral fractures, there were 21 type-C fractures; 20 of those patients underwent open reduction and internal fixation with standard plates or plates and cerclage wires ²¹ ; 12 (55%) patients required revision surgery. ²¹

Locking plates for Vancouver type-C fractures may be more reliable. The Less Invasive Stabilization System (Synthes, Inc), for example, has been designed specifically for distal femoral fractures, offering multiple screw holes for use in the distal metaphyseal region of the femur. Such locking plates may have a biomechanical advantage over nonlocking plates in osteoporotic bone and may also reduce soft tissue dissection if a percutaneous approach is used. In a series of 12 patients with type-C fractures treated with this system, 11 fractures healed uneventfully. ¹⁰⁰ Plate pullout occurred in 1 patient and was successfully revised to a longer plate.

In very unstable fracture patterns, a medial buttress placed with an allograft strut or a plate may be helpful. It may be particularly helpful when distal fixation is limited because of a total knee replacement distally that has a box- or stem-type design, which allow only unicortical screw fixation distally.

Postoperative Recovery and Complications

In general, postoperative recovery remains unpredictable and the risk of perioperative and postoperative complications after surgery for periprosthetic fracture is high. Lindahl et al ⁶ reported an overall complication rate of 18%. Up to 23% of 1049 patients required reoperation for late complications, and of the 245 patients in this group, the most common reasons for reoperation were nonunion (24%), refracture (24%), aseptic loosening (21%), and recurrent dislocation (16%). For patients with complications, the length of hospital stay is significantly longer and hip function is significantly worse. ²¹

The mortality rate for periprosthetic femoral fracture is high. The Swedish National Hip Arthroplasty database allowed comparison of 63 582 THA patients with 736 patients who sustained periprosthetic femur fractures. ¹⁰² Patients who sustained periprosthetic femur fractures had a higher mortality rate during the early postoperative phase than did patients undergoing THA. ¹⁰² For patients 70 years old or younger, an increased risk of death was observed at 1-year postfracture. For patients aged 70 years old, the risk of death at 1 year was 2.1% for male patients and 1.2% for female patients, and mortality rates increased with age ¹⁰² Other studies have reported 1-year mortality rates ranging from 9.4% ⁶ to 9.8%. ²¹

Treatment of nonunion after periprosthetic fracture is difficult. The implant stability and bone stock should also be double checked. If the implants are stable and bone stock is adequate, fixation should be performed. The surgeon needs to carefully check the method of fixation used and to determine whether a technical flaw can be corrected. An example of this may be too much stiffness with too many screws, the use of absolute fixation in a case where bridge plating should be used. If the implant is unstable, then revision with fixation needs to be performed. If the bone stock is inadequate, then a mega prosthesis may need to be considered, which may require proximal femoral replacement or even total femoral replacement.

Periprosthetic fractures in the presence of infection are challenging to treat. Options include retention of implants and fixation of the fracture to try to obtain healing, or removal of the implants combined with bracing or fixation of the fracture. In the latter case, second-stage reimplantation can then be considered. A single-stage revision with fixation may also be considered. Suppressive therapy can be tried, but results may or may not be successful.

Frequently, weight bearing is restricted in an attempt to prevent refracture or fixation failure, and it is often tailored to the intraoperative findings when bone quality and the quality of fixation can be accurately assessed. Chakravarthy et al ²² allowed total or partial weight bearing, with union in all 6 cases of type-B1 fractures. In a series of type-B2 and -B3 fractures treated with uncemented revision arthroplasty, touch weight bearing was encouraged for 6 weeks, followed by 6 weeks of partial weight bearing. In most cases, full weight bearing can be achieved by 3 months. ⁷⁴ Others recommend nonweight bearing until fracture healing has occurred, ^11,61 although it is often not possible in frail elderly patients, and it risks pressure sores or pulmonary complications while they are confined to bed or chair. Considering that a substantial proportion of patients with periprosthetic fractures have a degree of cognitive impairment that results in noncompliance, weight bearing as tolerated is a sensible approach, making rehabilitation easier for both the patient and the physiotherapist.

Functional recovery can vary. In 1 series of periprosthetic hip fractures, 60% of patients reported chronic pain. ⁶ Better results were recorded in a smaller series of 21 patients treated with proximal femoral replacement for Vancouver type-B3 fractures. ⁷⁹ Walking was restored in 20 patients, and the average Harris Hip Score was 71 points at a mean 3.2-year follow-up. ⁷⁹ Hip function was considered good or excellent in 11 hips, fair in 9, and poor for 1 patient who was unable to walk after failure of acetabular cage reconstruction. ⁷⁹

Prevention

Given that 70% of patients with periprosthetic femur fractures have stem loosening and that complication rates after surgical treatment are high, it has been suggested that the optimal approach for such patients is probably surgical intervention before the patient sustains a fracture. ²⁵ Early component loosening with or without osteolysis is often asymptomatic, which emphasizes the need for routine follow-up for patients with THA. Routine follow-up also appears to be cost effective ¹⁰³ : treatment of an acute periprosthetic fracture costs substantially more than a planned elective revision for loosening.

Conclusion

Periprosthetic femoral fractures associated with THA remain difficult to treat. The reoperation rates are high, ^21,25 and patients frequently report long-term pain ⁶ and unsatisfactory outcomes. Isolated use of open reduction internal fixation with standard plates may be unreliable. Additional cortical strut grafting is sensible because it offers reliable fracture union. ^18,53 The trend to use locking plates requires further research to evaluate their effectiveness. Most fractures occur around a loose stem. Radiologic assessment alone is probably inadequate because misdiagnosis of a stable stem can occur in up to 20% of cases and, therefore, intraoperative stem stability testing is recommended. ⁵⁶ Internal fixation is inadequate when the stem is loose. ⁵⁷ In such situations, revision surgery to a long stem using an extensively porous-coated long stem with or without supplementary open reduction internal fixation or additional bone graft is advocated. ^{9,18,82,83,86}

Given that outcomes are usually poor, the prevention of periprosthetic femoral fractures remains the best strategy. Routine follow-up of primary and revision THAs allows identification of patients with osteolysis who are at risk of fracture and is likely cost effective. ¹⁰³