Impaction bone grafting has become an accepted technique to treat severe femoral bone loss during complex revision total hip arthroplasty (THA) procedures [14, 23, 25, 34, 37, 40]. One of the essential requisites for this method is containment of the impacted cancellous bone within the medullary cavity. In patients with segmental bone loss, this procedure can be performed with metal mesh on the outside of the femur [39, 40]. We have been using metal meshes to reconstruct femoral deficiencies since 1990. Because of favorable outcomes in the literature [15, 20, 25, 39] as well as in our patients [4] and laboratory tests [18], we extended the indication of these devices to reconstruct complete circumferential proximal cortical bone loss. However, this circumferential bone loss has been considered a contraindication for the impaction bone grafting technique in the femur by one group [39].

We classified femoral deficiencies radiographically before surgery and confirmed the class after removal of components during the procedure using the system reported by Della Valle and Paprosky [9] and the Endoklinik (Engelbrecht and Heinert) classification [12] (Table 1). There were 14 cases presenting a Type IV and one case with a Type IIIB defect using the system of Della Valle and Paprosky. In a type III femoral defect, the metaphysis is damaged severely and nonsupportive. There is some intact cortical bone present distal to the isthmus. In a type IV defect, there is extensive metadiaphyseal damage in conjunction with a widened femoral canal, the isthmus is nonsupportive, and distal fixation cannot be achieved [9]. All the cases presented a Grade 4 Endoklinik deficiency, characterized by gross destruction of the proximal third of the femur with involvement of the middle third, precluding the insertion of even stemmed prosthesis [12]. The average height of the circumferential proximal femoral bone loss was 12 cm (range, 9–15 cm).

Five patients presented having had three previous transtrochanteric approaches, and one patient four previous transtrochanteric approaches. In four patients, severe major trochanteric osteolysis was present and in five the major trochanter was absent due to failed surgeries.

Surgeries were performed by the three authors (MB, FC, FP) following the general principles that have been described for the original impaction grafting technique [15]. Under epidural hypotensive anesthesia, a posterolateral approach was performed in 13 cases with an additional intrapelvic approach to remove an excessively protruded cup in one case and a transtrochanteric approach in two cases. The presence of active infection during one-stage surgeries or second-stage reimplantations was ruled out in all patients using intraoperative frozen section biopsy [33] and bacteriologic and pathologic analysis. All implants, polymethylmethacrylate, granulation tissue, and interface were completely removed and the remaining distal femur was identified. The removed failed femoral stems presented an average length of 14.4 cm (range, 12–26 cm) and they were all cemented except one uncemented implant.

The femoral metal mesh was 0.15 inches thick and had 2-mm² perforations. Fico^® (Fico^®, Buenos Aires, Argentina) metal mesh was implanted in 13 cases, and X-Change metal mesh (Stryker Howmedica, Newbury, UK) was used in two cases. To make reconstruction easier and to measure the length of the needed mesh, a proximal impactor is temporarily inserted as a form around which the meshes are bent. The meshes were fixed with three double cerclage metal circumferential monofilament wires (Ortron 90; diameter, 1 mm; DePuy Int., Leeds, UK, Warsaw, Ind.) (Fig. 2A). Approximately 10 cm of circumferential metal mesh was fixed to the remaining bone.

We obtained bone allografts from frozen femoral heads from our own bank following the protocol of the American Association of Tissue Banks for the harvesting and processing of grafts [13, 29]. We made the cancellous chips 7 to 10 mm with a grinder from the unwashed bone allografts and mixed for 15 minutes with 1 g dry powdered vancomycin (Lilly, Indianapolis, Ind.) per femoral head according to previous investigations [5]. Grafts were impacted according to the technique described by Gie et al. [15] using specific instruments (Primary Impaction Grafting Instruments; DePuy Int., Leeds, UK in 13 cases and Exchange Revision Instruments System; Stryker Howmedica Osteonics, Allendale, NJ in two cases). An additional three double cerclage wires were placed in the proximal mesh to contain the impaction force that was applied uniformly until rotational stability of the impactor was achieved. An average of 3.5 allograft femoral heads were needed per revision, including the acetabulum (range, 3–5). A cancellous bone thickness of 10 mm was intended to be obtained in the proximal part of the reconstructed femur (Fig. 2B).

We retrogradely injected bone cement CMW1 with gentamicin (DePuy, Warsaw, IN) in 13 cases and Simplex with tobramycin cement (Stryker Howmedica Osteonics, Rutherford, NJ) in two. A C-Stem (DePuy) (Ra 0.05 um) triple-taper polished femoral stem was implanted in 10 cases (Fig. 3), a Charnley stem (DePuy) (Ra 0.64 um) was cemented in three cases, and an Exeter stem (Stryker Howmedica Osteonics) (Ra 0.05 um) was used in two cases. The average length of the femoral stems was 222 mm (range, 130–260 mm). A 28-mm femoral head was implanted in 12 cases and a 22.225-mm in three cases. The femoral stems bypassed the circumferential mesh for an average of 5 cm (range, 1–9). The average operative time was 196 minutes (range, 150–240 minutes) (Fig. 3).

In aseptic cases we prescribed antibiotic prophylaxis with cefazolin 1 g every 8 hours for 48 hours. All patients with a previous infection received the same intravenous antibiotic treatment that had been used after the first stage; antibiotics were stopped when the infection was controlled. Routine prophylaxis for thromboembolic disease was continued for the first postoperative month. This consisted of intravenous heparin during surgery, early postoperative mobilization, enoxaparin 0.4 mg in patients with a high clinical risk of thromboembolic disease (i.e., malignancy, particularly if associated with chemotherapy; antiphospholipid syndrome, immobility, or a history of venous thromboembolism; administration of tamoxifen, raloxifene, oral contraceptives, or estrogen; morbid obesity; stroke; atherosclerosis; and an American Society of Anesthesiologists physical status classification of 3 or greater), and aspirin 325 mg in patients with a low clinical risk [38]. We did not routinely prescribe prophylaxis against heterotopic calcification. The mean inhospital transfusion requirements were 3.5 concentrated red blood cell units per case (range, 2–6).

The rehabilitation protocol included early mobilization 48 hours after surgery and ambulation with a walker and toe-touch weightbearing on the operated side for 90 days. After that, we encouraged patients to progressively weight bear as tolerated with the use of a cane for at least 1 month.

Two of us (MB, FC) evaluated patients clinically and radiographically at 15, 45, 90, and 180 days postoperatively and then yearly. We used the Merle D’Aubigné and Postel scoring system [8]. As a result of the presence of metal meshes and cerclage wires, we could not evaluate the incorporation of the graft. Femoral radiolucencies and periprosthetic osteolysis, defined as progressive bone loss in an area larger than 5 mm² were assessed in the distal part of the femoral stems, where metal mesh was not present. Femoral radiographic loosening of the Charnley stems was classified according to Harris and McGann in three categories: definitive, probable, and possible loosening [21]. Subsidence of the femoral stem was determined using the method described by Loudon and Charnley [26], which depends on measuring the distance from a selected (but variable) point in the femoral prosthesis to a fixed point in the bone.

One patient having repeat surgery agreed to have biopsy specimens from six of the seven zones described by Gruen et al. [17] through the fenestrations of the metal mesh using a 4-mm Jamshidi needle. These samples were evaluated by an independent orthopaedic pathologist.

We defined clinical failure as the need for further femoral revision irrespective of the reason. The radiographic definition of failure of the polished tapered stems was progressive migration by more than 5 mm in any direction on standard anteroposterior (AP) and lateral radiographs or mesh fracture. Success was defined as the presence of a functional hip prosthesis at the time of the latest followup. A functional hip was considered when pain was absent or slight during walking (with or without the use of a cane) without radiographic findings that indicated the possibility of a future operative intervention.

There was an improvement in the functional score from a mean of 4.8 to 14.4 points.

Radiographically five major trochanters were stable, there were four cases with a trochanteric pseudarthrosis, and there were six cases with an absent trochanter. In 13 cases, we observed no femoral radiolucencies or periprosthetic osteolysis in the distal part of the femoral stems, where metal mesh was not present. None of these cases is awaiting femoral surgery and there were no radiographic signs of impending failure in any of these femurs at last followup. One patient presented a progressive acetabular migration that started with symptoms after 3 years postoperative and now is waiting for acetabular revision surgery. We observed two fractures of the metal mesh at 31 and 48 months in cases reconstructed with a short stem that did not bypass the mesh (Fig. 5). Both cases were the first two reconstructions and presented a Charnley stem that incompletely bypassed the metal mesh 1 cm in the first failure and 1.5 cm in the second one. The first failure was treated with an alloprosthesis, and fractured two osteosynthesis plates due to lack of the osteotomy consolidation. Three years after the mesh failure, a third surgery consisting of a distal osteotomy with a new alloprosthesis fixed with two plates showed the osteotomy consolidation. The second failure (Case 2) (Fig. 5) was treated with a new circumferential metal mesh, impacted bone allografts, and a long 200-mm cemented stem bypassing the mesh 6 cm (Case 13). One patient presented a single episode of dislocation that was treated nonoperatively with reduction and an abduction device for 6 weeks. No new episodes of dislocation were present at last followup in this patient. Three cases presented recurrent dislocations that needed open reduction. These dislocations were treated with trochanteric advancement and femoral head lengthening in one case, and femoral head lengthening in another case. Three acute deep infections were successfully treated with irrigation, débridement, antibiotics, and component retention.

Limitations of this study include a small patient cohort with a short term followup and the lack of a control group of patients with similar age, weight, and femoral defects. Longer followup will determine whether any of these reconstructions will require more reoperations. We implanted three geometrically different stem designs, two with a polished surface finish and one with a matte finish. One study suggested triple-tapered stems performed similar to double-tapered designs in primary THA [11]. In one study the Charnley stems in combination with impaction grafting achieved 93% survivorship at 54 months followup [37]. Other authors reported similar incidences of subsidence without clinical consequence [34, 39]. In a recently published single-surgeon experience on impaction grafting, no stems subsided more than 2.5 mm. Most of the stems (66%) demonstrated no measurable subsidence. The authors believed the vigorous impaction of bone graft with containment of the metaphysis of the stem with impacted bone graft was responsible for the minimal subsidence [25]. These patients presented different diagnoses, which included septic, aseptic, and traumatic situations. Different outcomes might have been obtained in one of these isolated situations. Although we had the opportunity to take bone biopsies and observe their histological appearance, the microscopic appearance of these sections seems inconclusive after 10 months postoperatively. Additional biopsies and autopsy retrievals of femurs reconstructed with this technique are needed to confirm our findings. However, the presence of active bone marrow may be interpreted as future areas of vascular invasion, remodeling bone graft, and new bone formation.

Reported treatment alternatives for these complex situations include modular fluted stems [1, 30], tumoral prostheses that replace the proximal femur [19, 27, 35, 36], and allograft-prostheses [2, 16, 28]. Revision arthroplasty with use of a modular fluted stem has been described as an option in these situations. However, stress due to the absence of proximal femoral bone support could be sufficiently high to put these implants at risk for fatigue fracture [6]. Furthermore, the long-term outcomes of such reconstructions are not yet available. Because of concerns about possible early failure, the use of modular prosthetic replacements (so-called megaprostheses) is typically reserved for elderly and sedentary patients as implantation can be performed in a more expeditious manner than other, more complex reconstructive procedures [19, 27, 35, 36]. The diaphyseal cement fixation predisposes the bone-cement-prosthesis to high torsional and compressive stresses leading to early loosening [35] (Table 3). Although allograft-prosthesis composite appears to provide better survival than proximal femoral replacement for the treatment of neoplastic conditions [41], complications including unhealed osteotomies, allograft resorption, trochanteric migration and dislocation are frequently described with this treatment method [2, 16, 28] (Table 3).

Although we could not find comparative data in the literature, we consider more than 15-cm segmental defects a contraindication for the use of circumferential metal mesh, impaction grafting, and a cemented stem. In these patients, we currently use an alloprosthesis composite. We also prefer this method in neoplastic conditions compromising the proximal femur [32]. However, it is essential to have a good bone quality in the distal femur in order to avoid unhealed osteotomies if an allograft-prosthesis composite is chosen [31]. Multiple failed previous surgeries, as well as a previous history of local infection (seven of our 15 cases) prompted us to avoid the use of a massive proximal femoral allograft. The fact that cancellous bone can act as an antibiotic carrier [5] and the 20 years’ experience in our institution with the bone impaction grafting technique played a major role in the decision to use this surgical technique.

High survival and functional scores have been reported with the use of impaction bone grafting and metal meshes in segmental defects [14, 25, 39, 40]. We observed circumferential metal meshes, impaction bone grafts, and long cemented triple-tapered femoral stems presented a favorable initial resistance, reaching more than 600 kg to axial forces in bovine femurs [18].

Recent experimental reports recommend the use of open wire mesh instead of strut grafts for segmental femoral defect reconstruction in combination with impaction grafting to allow for optimal revascularization in an area with impaired vascularity [3]. Our histologic observations of biopsies obtained from impacted bone allografts contained by calcar femoralis metal mesh demonstrated morselized bone incorporation under these devices was not affected [4]. Although we observed signs of active hematopoietic bone marrow, osteoid and viable new bone apposition in the biopsied case, further investigations are needed to completely understand the biology of impacted bone allograft under circumferential meshes.

Dislocation was the most frequent complication in our series. Although it was observed in four of our 15 cases, this incidence is comparable to the ones reported with megaprosthesis [36] and allograft-prosthesis composites [7] (Table 3). Multiple failed previous surgeries through transtrochanteric approaches with an absent abductor mechanism and difficult proper leg length could explain this high incidence. This situation prompted us to consider constrained cups at first attempt in the last cases we performed this technique. We are also concerned about the high incidence of acute infections that might also be related to the number of previously failed surgeries and the prolonged operative time. Other authors have recently published up to 420 minutes of operative time with the use of megaprostheses [36]. Additional followup is essential to assess long-term outcomes of these reconstructions.