Revision total hip replacement using cemented collarless double-taper femoral components

At revision hip replacement the choices for femoral reconstruction include cemented or cementless fixation, and the use of impaction grafting. Using cemented fixation, early series reported high rates of re-revision, which were probably a result of poor stem design, first-generation cementing techniques, the concomitant use of cemented acetabular components and perhaps an over-reliance on a standard-length femoral component.^1–3 Initially, cementless femoral revision was also associated with poor results.^4,5 Successful results have been achieved with modern cementless designs,^6–10 but there are concerns which include loosening in the presence of severe metadiaphyseal deficiency,^7,10–13 intra-operative fracture, stem subsidence, thigh pain, stress shielding and the need for an extensive femoral osteotomy to fit many of the longer stem cementless designs.^14–17 A long period of restricted weight-bearing is usually advised after revision to a cementless femoral component, which may hamper rehabilitation.

There have been reports of good results for cemented femoral revision,^18–20 but there has been no report of a large cohort of femoral revisions using cemented collarless double-taper components without the addition of impaction grafting, despite this design having theoretical advantages in the way it loads cement and bone.^20–23 We have prospectively studied the outcome of such components used at femoral revision. We aimed to determine the survival to re-revision and re-operation and the clinical and radiological results of this design. We specifically examined the extent of pre-operative bone deficiency and the incidence of complications reported with other techniques of femoral revision.

Patients and Methods

Between September 1984 and December 2003, 211 patients underwent 219 revision total hip replacements (THRs) at which the femoral component was revised to a cemented collarless double-taper design (Table I, Figs 1 and 2). Bilateral femoral revisions were performed in seven patients (3%). One patient had two revisions on the same hip. The mean age of the patients at revision surgery was 72 years (30 to 90). The operations were undertaken by four consultant surgeons (DWH and three others who were not authors) or by arthroplasty fellows and residents under their supervision.

The revision was the first in 174 hips (79%) (Table I). The reasons for revision were aseptic loosening in 163 hips (75%), peri-prosthetic fracture in 26 (12%), recurrent dislocation in seven (3%), to match an acetabular revision for loosening in six (3%), chronic pain in three (1%) and implant fracture in three (1%). A total of 11 femoral components (5%) were implanted at the second stage of a two-stage revision, nine for previous infection. During the study period, 132 other types of femoral revision were undertaken, but were excluded from the study because they were either one-stage revisions for infection, involved impaction grafting or a proximal femoral allograft, used cementless femoral components or other designs to match a retained acetabular component.

The posterior approach alone was used in 189 hips (86%) or combined with a transtrochanteric approach in 21 (10%), a trochanteric slide in three (1%), an extended trochanteric slide in three (1%) and a longitudinal femoral split in three (1%). Osteotomies were used as necessary to extract ingrown or fractured femoral stems. These osteotomies were reduced and held with cables before cementing the revision component. No femoral osteotomies were used to correct a pre-operative angular deformity alone.

Revision was undertaken with 137 long-stemmed and 82 standard-stemmed femoral components (Table I). During part of the study, standard stems were used in younger patients because of our reluctance to use components which crossed the isthmus of the femur in patients with a long life expectancy. Standard stems are now rarely used without the addition of impaction grafting.

All femoral components were a collarless double-taper design and included the stainless steel Exeter (Stryker-Howmedica Osteonics, Newbury, United Kingdom) or stainless steel or cobalt-chrome collarless polished tapered (CPT) (Zimmer Ltd, Warsaw, Indiana) components. The early monobloc Exeter continuous-taper steel stems included one long matt stem, nine long matt distally polished, six long polished and one standard matt stem. The modular polished Exeter stems were 75 proximal taper, 17 continuous taper long, and 79 standard stems. The CPT stems included 26 long and two standard-length steel stems and three cobalt-chrome long stems. The long taper stems ranged between 50 mm and 130 mm longer than a standard stem. The length of stem was chosen to extend 5 cm into undamaged bone and at least two femoral diaphyseal diameters below the lowest cortical bone deficiency (Figs 1 and 2).

The cementing technique included the use of a distal plug, lavage, retrograde insertion of hand-mixed cement with a cement gun, a proximal seal and cement pressurisation. If present, any sclerotic neoendocortex was burred or was removed to expose cancellous bone in the metaphysis. Simplex antibiotic cement (Howmedica International, Staines, United Kingdom) was used routinely.

In 145 hips (66%) the acetabular component was revised and in 23 hips (11%) an acetabular component was implanted during conversion of a hemiarthroplasty to a THR. The acetabular component designs used included 42 Exeter (Howmedica International) and two ZCA (Zimmer Ltd) cemented components, and 100 Cluster (Howmedica International, Rutherford, New Jersey) or Vitalock (Howmedica International) and 24 Trilogy (Zimmer Ltd) cementless components. We retained nine cemented and 36 cementless acetabular implants (Table I). Of the 36 cementless components retained, the liner was exchanged in 22. Six hips (3%) were revised to a hemiarthroplasty.

We used 24 long and two standard stems to treat pre-operative peri-prosthetic fractures (Table I). For femoral segmental deficiencies which extended to below the lesser trochanter, we used 25 long and 14 standard stems. Fractures were fixed with cerclage wires or cables and plates, and bone defects were supplemented with bone graft as necessary. Patients bore full weight on the first post-operative day unless there was a peri-prosthetic fracture, a femoral shaft osteotomy, reattachment of the greater trochanter or a graft of the acetabulum providing structural support.

Femoral bone deficiencies were graded radiologically according to Aribindi et al¹⁷ and the EndoKlinik.²⁴

There were 102 of 129 (79%) femora revised with long stems; 58 of 80 (73%) femora revised with standard stems were graded as Paprosky grade IIIA or worse, indicating considerable metaphyseal bone loss or damage (Table I).

Radiological analysis was performed on hips without rerevision and with five or more years of follow-up. Of 139 hips in this category with a mean follow-up of 8 years (5 to 18), 28 long stems and 19 standard stems could not be radiologically reviewed because of the patient’s general health, or for reasons of distance or access, leaving 48 hips with long stems and 44 hips with standard stems available.

Radiographs were assessed using manual techniques previously described.^25–28 Definite radiological loosening of the stem was defined as migration within the bone by more than 5 mm,²⁹ or stem fracture.³⁰ The criteria of Harris, McCarthy and O’Neil³¹ were used to distinguish a probably loose, possibly loose or stable stem. Osteolysis was defined as a non-linear demarcated radiolucent lesion more than 3 mm in diameter.³² Major stress shielding was defined as a major reduction in bone density or cortical thinning in the absence of radiolucent lines or osteolytic lesions.³³ Postoperative femoral fractures were classified as peri-prosthetic or non-peri-prosthetic.³⁴

Patients were assessed pre-operatively and at regular intervals post-operatively. The mean follow-up of the 129 surviving patients was 6 years (2 to 18). Outcomes collected included the Harris pain score and the Harris hip score.³⁵ Apart from the physical examination of the hip, all outcomes were independently reported by the patient.^35,36 Patients unable to attend clinic were reviewed by either postal questionnaire or telephone interview.

Data were analysed using GraphPad Prism software (version 4.01, GraphPad, San Diego, California). Survival analyses were undertaken using the Kaplan-Meier method. Survival curves were compared using the log-rank test. Fisher’s exact test was used to compare incidences. A p-value of less than 5% was set to indicate statistical significance.

Results

Of the 211 patients (219 revisions), 82 (86 revisions, 39%) had died leaving 129 patients (133 hips). The number of hips at risk for re-revision is shown in Table II. After excluding nine revisions at the latest review of these surviving hips, 81 were reviewed by clinical examination, seven by postal questionnaire and 36 by telephone interview.

There were eight intra-operative femoral shaft fractures in femora implanted with long stems (6%) and six in femora implanted with standard stems (7%; Fischer’s exact test, p = 0.77) (Table III). There were 11 intra-operative femoral cortical perforations in femora implanted with a long stem (8%) and three in femora implanted with a standard stem (4%; Fischer’s exact test, p = 0.26). There were one or more post-operative hip dislocations reported for 19 hips (14%) with long stems and for 11 hips (13%) with standard stems (Fischer’s exact test, p = 1.00). There were no stem fractures.

The median pre-operative Harris hip score was 41 (7 to 94). This improved to 82 (23 to 100) at two years, to 76 (30 to 100) at a mean of seven years (6 to 8), and to 90 (78 to 97) at a mean of 13 years (12 to 14) follow-up. There was a trend towards better hip scores for patients with standard stems at longer follow-up, consistent with their younger age and higher levels of activity. The median Harris pain score at each review period after three months was 40 to 44 indicating no pain or slight or occasional pain. The number of patients with severe or worse thigh pain was, for long and standard stems respectively, at one year 4% and 4%, at two years 4% and 3%, at three to five years 9% and 9% and at six years or more 1% and 0%.

We reviewed the radiographs of 48 long stems and 44 standard stems at a mean follow-up of eight years (5 to 18) and nine years (5 to 13), respectively. One long stem was definitely loose and one standard stem probably loose. Both had been used in patients with Paprosky grade¹⁶ IIIA femoral deficiencies.

In four hips with long stems (3%) post-operative avulsion of the greater trochanter occurred, but none required surgery. The mean total subsidence of the 91 stable stems was 2 mm (0 to 4). Of this total subsidence, a mean of 1 mm (0 to 3) occurred within the cement mantle, as expected with this design. Femoral osteolysis was present in two hips, both with radiologically stable standard stems; in one, osteolysis was present in Gruen zone²⁸ 1 and in the other, in Gruen zone 7. No femur had major stress shielding seen on radiographs taken at five years or more post-operatively.

The re-operations are listed in Table III. There were nine re-revisions (4%) involving removal of the femoral component and cement. One polished long stem and four polished standard stems were revised for aseptic loosening.

When a major re-revision, involving removal of the femoral component and cement for aseptic loosening was used as the end-point, the nine-year rate of survival was 98.0% (95% confidence interval (CI) 94 to 100) for long stems and 92.8% (95% CI 85 to 100) for standard stems (Fig. 3). The difference in survival curves for long and standard stems was not statistically significant (log-rank test, p = 0.28). Worst-case analysis did not affect the survival estimates because no hip was lost to follow-up. Survival at nine years to any re-operation involving the femoral component, excluding infection, was 93.9% (95% CI 87 to 100) for long stems and 91.4% (95% CI 83 to 100) for standard stems (log-rank test, p = 0.45). Survival at nine years to any re-operation involving the hip, including acetabular revision and infection, was 85.5% (95% CI 78 to 93) for long stems and 86.0% (95% CI 76 to 96) for standard stems (log-rank test, p = 0.60).

Discussion

In this study a cemented long collarless double-taper femoral component performed well when used at revision, and had a 98% nine-year survival for aseptic loosening. There were no further revisions for aseptic loosening at between nine and 18 years. Importantly, the status of all hips was known, and therefore the reported survival to re-operation is a worst-case analysis. There are a number of explanations for the low incidence of loosening and the good results using this stem, including its design and the relatively older age of the patients. The design included a collarless double taper, a rectangular proximal cross-section enhancing rotational stability, a polished surface in the majority of components and a relatively thin distal stem which optimises cement mantle thickness. The taper-slip principle of limited subsidence of a femoral component within the cement mantle has proved successful at primary THR in optimising fixation and load transfer,²³ and has probably contributed to the good results at revision. The low incidence of osteolysis is probably a result of the polished surface of the stem, which minimises the generation of cement debris. The mean age of our patients was 74 years at the index revision with a long stem and, given that young age at revision is known to be a risk factor for failure,^18,37 the older age of our patients may have contributed to the good results.

Stem length had some influence on the results. In the medium term, the survival of long and standard stems was similar. However, with further follow-up the long stems had a lower incidence of loosening. This difference did not reach statistical significance because of the low incidence of re-revision in both groups and the smaller numbers reviewed in the long term, although the trend is consistent with the findings of others.¹⁸

The results of the cemented, long collarless taper femoral components were better than many cementless designs in terms of survival to aseptic loosening, and especially so if a worst-case analysis, or radiological subsidence was considered. Although the Solution (DePuy, Warsaw, Indiana) and AML (DePuy) porous-coated femoral components have a reported survival at 14 years of 96% to loosening or radiologically unstable stems,⁶ 11 of 188 patients (6%) were lost to follow-up and excluded from survival analysis, and subsidence occurred in 16%. Another cementless approach, the use of a modular S-ROM (DePuy) fluted stem, has a reported five-year survival of only 88% to either radiological loosening or revision for loosening, and six of 75 hips (8%) were lost to follow-up.¹³ The use of distal fixation with a rough-blasted fluted femoral stem is reported to have good results, but in a report of 129 hips one femoral component was revised early for subsidence and 20% had subsided more than 10 mm.³⁸ Others have reported an incidence of subsidence of 20%.³⁹ A high incidence of radiological failure is reported with other cementless designs.⁴

Our medium-term results are as good as for many other cementless femoral components despite the femoral deficiencies frequently being severe. For example, in the study reporting the best results of porous-coated femoral components,⁶ only 11% of the femora were Paprosky grade¹⁶ IIIB, and none was grade IV, compared with 34% of femora in our study which were Paprosky grade IIIB¹⁶ or IV. Good results are reported for the Furlong hydroxyapatite stem (JRI Limited, London, United Kingdom),^7,8 but 36% of femora were Paprosky grade¹⁶ III or IV, compared with 73% of femora in our study.

The clinical results of cemented revision of the femoral component were good. Our incidence of significant thigh pain was low, affecting a mean of 5% of hips (3% to 9%) up to five years and 0.5% of hips thereafter. This was self-reported by patients and given the prevalence of other disabilities, including lumbar spine problems, it is likely that some of the thigh pain was not hip related.⁴⁰ In contrast, cementless fully porous-coated stem designs have been associated with significant thigh pain, in up to 31% of hips with AML femoral components;¹⁴ in 9%, this pain was of sufficient severity to require medication or to limit activity.¹¹ An advantage of a cemented long stem for the older patient is the guaranteed pain relief. Immediate fixation also allows full weight-bearing post-operatively, thereby enhancing rehabilitation. After cementless revisions many patients are able to bear only partial weight for eight weeks or longer.^6,7,10 Dislocation was reported in 14% of our cemented revisions, which appears high, but with extensively-reviewed patients there is a reporting bias for dislocation⁴¹ and so this, combined with the long duration of our study, is relevant. Furthermore, many patients were elderly, a known risk factor for dislocation.

Severe stress shielding was not seen in our study and osteolysis was rare. Although it is difficult to compare stress shielding between cemented and cementless femoral components, it is worth noting that some degree of stress shielding has been reported in 39% of S-ROM femoral components¹³ and 90% of femora with fully porous-coated stems,¹⁴ in which 6% were graded as severe.^6,14

An advantage of using cemented, long tapered stems at revision is the avoidance of a transfemoral osteotomy, and the only infrequent need for an extended trochanteric osteotomy. Significantly bowed femora can usually be handled by combining a relatively narrow taper stem with cement as a form of customised device to fit the deformity. In contrast, straight large-diameter cementless femoral components frequently require a corrective femoral osteotomy in order to fit the deformity of the femur. An osteotomy creates a major segmental deficiency, which weakens the femur⁴² and may excessively load the stem, leading to fracture.

Another important advantage of the cemented tapered femoral components seen in our study was that they are potentially modular within the cement mantle. This allows a cement-within-cement exchange in order to treat complications, including recurrent dislocation, and allows for anteversion and leg length to be adjusted as necessary. This simple solution is in contrast to the major problem of treating a recurrently dislocating long, ingrown stem in elderly osteopenic bone.

At long-stem cemented revision, our incidence of intra-operative femoral shaft fracture was 6% and of trochanteric fracture 2%, comparable to, or better than, with other techniques. For example, the incidence of intra-operative fracture has been reported as 9% with fully porous-coated femoral components,⁶ 27% with modular proximal cementless components,¹³ 29% and 18% with taper grit-blasted cementless components,^38,39 and 12%, 5% and 6% with hydroxyapatite-coated designs.^7–9 In addition, many of the cementless revisions would have required a femoral osteotomy.

In conclusion, our study has found successful long-term results for cemented collarless double-taper long femoral components used at revision hip replacement, including those in which there is significant femoral bone loss. An advantage of the technique is rigid fixation, thereby allowing full weight-bearing immediately and without the risk of early subsidence. The relatively thin stem combined with cement avoids the need for an osteotomy when treating most angular deformities of the femur. Major stress shielding and osteolysis were not problems. Importantly, these cemented femoral components are relatively inexpensive and, because only a small inventory is necessary, can be held available for unplanned revisions. Based on the predictably good results, cemented femoral revision with long collarless polished taper stems can be recommended for the majority of older patients.

The authors acknowledge the contribution of the consultant surgeons, Mr S. Graves, Mr A. Mintz and Mr R. Clarnette. The authors thank Mr P. Ward, Ms K. Costi and Ms S. Golding for their assistance with the conduct of this study. The authors also thank all arthroplasty fellows and residents involved in the cases. This research was financially supported by the Australian Orthopaedic Association, the Royal Adelaide Hospital, the Flinders Medical Centre, the Adelaide Bone and Joint Research Foundation, Bristol Myers Squibb-Zimmer and Zimmer Ltd.

Although none of the authors has received or will receive benefits for personal or professional use from a commercial party related directly or indirectly to the subject of this article, benefits have been or will be received but will be directed solely to a research fund, foundation, educational institution, or other nonprofit organisation with which one or more of the authors are associated.