A prospective study of hip revision surgery using the Exeter long-stem prosthesis: function, subsidence, and complications for 57 patients

Background The long-stem Exeter femoral component is commonly used in revision hip surgery. Subsidence of the femoral stem in primary hip arthroplasty has been studied extensively, but much less is known about its significance in revision surgery. This prospective study examined the relationship between radiological subsidence, Western Ontario and McMaster (WOMAC) osteoarthritis index pain score, patient satisfaction and complication rates for the long-stem Exeter hip prosthesis. Materials and methods Data was prospectively collected for a single-surgeon series of 96 patients undergoing revision surgery with a mean follow-up period of 36 months. Pre- and post-operative clinical evaluation was carried out using the validated WOMAC osteoarthritis index. Radiographic evaluation was carried out on magnification-adjusted digital radiographic images. Results Data from 57 patients were analysed. The mean rate of subsidence recorded was 0.43 mm/year, with a mean total subsidence of 0.79 mm [95% confidence interval (CI) 0.57–1.01] at 36.3 months. There was no correlation between subsidence and post-operative WOMAC score, complication rate or patient satisfaction. There was a statistically significant reduction between pre-operative and post-operative WOMAC scores, with means of 33.5 and 10.7, respectively (P < 0.001), and high patient satisfaction. Conclusion Our subsidence rates for long-stem revision femoral components are lower than the published data but demonstrate the same plateau. Radiographic subsidence does not appear to relate to functional outcome or complication rates in our data.


Introduction
With an increasing number of younger patients undergoing total hip arthroplasty and the finite lifespan of prosthesis, the incidence of revision hip surgery is on the rise. A previous study by Coyte et al. projected a 5.1% annual growth in the number of revision hip arthroplasties [1]. Revision arthroplasty is associated with higher complications rates [2] and a smaller improvement in functional outcome. Common causes for revision surgery include prosthesis loosening (both septic and aseptic), recurrent dislocation, femoral and acetabular osteolysis and periprosthetic fracture of the femur. The method of hip reconstruction is usually individualised to each patient depending on the extent of bone loss, surgeon experience and anticipated patient lifespan. The revision femoral stem is usually longer than the retrieved stem in order to bypass the deficient proximal defect and achieve a tighter diaphyseal fixation. Revision for an infected prosthesis is usually performed in two stages. In aseptic loosening, cementwithin-cement is sometimes employed where removal of the cement mantle proves difficult, thus avoiding the complications of removing acrylic cement (i.e. femoral fracture, reduced bone stock and blood loss). This is still an area of much debate [3,4].
Many options are now available for complex femoral reconstruction, including the use of long stems, uncemented or cemented implants, hydroxyapatite (HA)-coated implants, impaction grafting, massive allografts or proximal femoral replacement type implants. These prosthetic implants in a variety of stem lengths demonstrate varying clinical and radiological success [5][6][7]. Long-stem femoral components have been shown to be favourable in revision hip arthroplasty [8,9]. The Exeter stem, however, enjoys a significantly long survivorship [9] and is a popular choice amongst orthopaedic surgeons in both primary and revision total hip arthroplasty. We report a cohort of 57 patients with a mean follow-up of 36 months following revision total hip arthroplasty using cemented Exeter long-stem implant. This paper focuses on functional outcomes using the Western Ontario and McMaster (WOMAC) osteoarthritis index pain score, complication rates and radiological subsidence.

Materials and method
A prospective cohort study was undertaken of a singlesurgeon series at our institution from 1999 to 2007. Revision cemented hip arthroplasty was performed by the senior author (TL). Ninety-six, Exeter double-tapered long-stem femoral prostheses (Stryker/Howmedia Osteonics Newbury, UK) were inserted into 96 patients. Informed consent was obtained prior to patient inclusion in our study. The study was planned and implemented in accordance with the ethical standards of the Helsinki Declaration as amended in 2000 and was approved by our institution's ethics committee. Our indications for long-stem femoral revision were as follows: stem loosening with distal osteolysis, intraoperative femoral perforation, intra-operative femoral window, periprosthetic fractures, proximal segmental bone-stock deficiencies, lack of proximal femoral cancellous bone and complex reconstruction with re-alignment osteotomy.

Surgical technique
All operations were performed at the Heart of England NHS Trust by the senior author in dedicated orthopaedic theatres. A modified Hardinge approach was used in all patients. The affected components were removed along with the cement, debris and fibrous material. Of the extracted femoral prosthesis, 58 (57.4%) were cemented and 38 were uncemented. The femoral canal was curetted and reamed, and a range of broach sizes was tried. Impaction grafting was used in cases of cancellous bone stock deficiency. Sclerotic bone was removed and replaced with impaction grafting. Distal plugs were used to fill femoral windows. Third-generation cementing technique using pulsed lavage, tamponade and vacuum mixing with pressurization of gentamicin-loaded cement [SIMPLEX (Stryker, Newbury, UK)] was standard for all patients. A total of six cement mixes were routinely used when cementing the femoral component. One patient received PALACOS cement (Zimmer, UK). Two types of acetabular components were used in 96 patients who underwent total revision arthroplasty. Eighty-eight (93.6%) patients received the Elite cup (Depuy, Leeds, UK), and eight (6.4%) patients received the Ogee cup (Depuy). Eighty-eight (91.6%) patients received a 28-mm femoral head, four (4%) received a 32-mm head and four (4%) received a 22.2-mm head. All patients received prophylactic parenteral administration of the antibiotics flucloxacillin 1 g and gentamicin 160 mg pre-operatively, as well as 10 days of post-operative parenteral administration of flucloxacillin 500 mg QDS. Antithrombotic prophylaxis consisted of tedgradcoman (T.E.D.S) stockings and subcutaneous administration of enoxaparin (Clexane, Sanofi-Aventis) injection daily for 5 days, followed by aspirin 75 mg daily for 6 weeks. Post-operative rehabilitation consisted of 6 weeks of non-weight bearing for peri-prosthetic fractures, followed by partial/full weight bearing or 1 week non-weight bearing followed by full weight bearing for the other indications.

Clinical evaluation
Pre-operatively, patients underwent clinical evaluation and were assessed according to the validated WOMAC osteoarthritis score [10]. Endpoints for the study were re-revision, pain, stiffness, function and complications. Data was collected prospectively at 6 and 12 months post-operatively and then yearly. Interview and clinical examination were carried out in the outpatient clinic. The WOMAC osteoarthritis index and a patient satisfaction questionnaire were used at each follow-up appointment. Patients were asked if they would recommend the procedure to a friend or relative in a four-point scale (definitely not, probably not, possibly yes, definitely yes).

Radiological evaluation
All patients were followed up with pre-and post-operative radiographic investigations. All investigations were performed at our hospital and included standardised anteriorposterior and lateral projections of the hip. Two observers evaluated the radiographs. Subsidence was evaluated by comparing the immediate post-operative radiographic images with those obtained at follow-up. All radiographs were digitalised. Subsidence was measured as a vertical drop of the stem, characterised by a radiolucent line at the proximal most aspect of the cement-prosthesis interface. Magnification was adjusted based on the known diameter of the prosthetic femoral head. Our measurements were aided by the use of an on-screen calliper and protractor software (www.iconico.com).

Statistics
Statistical analysis was carried out using the XLSTAT module (Addinsoft, NY, USA) for Microsoft Excel. The paired Students' t test and analysis of variance (ANOVA) for parametric data, set at a 0.05% significance level, was used for data analysis. Linear regression analysis was used to correlate Paproski scores to subsidence and percentage WOMAC score improvements to subsidence.

Follow-up
At review, 13 patients had died. The cause of death on review of patient's notes was not directly related to the revision. Eleven patients were lost to follow-up, and four patients did not have adequate follow-up time (less than 6 months), leaving 68 patients for analysis with a mean follow-up time of 36 (range 6-84) months (Fig. 1).

Complications
There were four cases (4.1%) of superficial infection, which were treated by intra-operative washout. Three patients (3.1%) suffered deep infection, two of whom underwent two-stage re-revision surgery (at 34 and 33 months post-operatively), and the other patient underwent a washout with prosthesis retention. One patient presented with loosening of both components and underwent a re-revision 5 years after the primary operation. Four patients (4.1%) presented with peri-prosthetic fractures; of these, three underwent re-revision femoral stem surgery, and one was treated conservatively. Two patients presented with deep vein thrombosis and were treated with anti-coagulation therapy. One patient suffered recurrent dislocations and underwent an acetabular revision (Table 1).
Of these 57 patients, 25 had a recorded WOMAC score at more than two occasions post-operatively (12,24,36 and/or 48 months post-operatively). We analysed this  subgroup to ascertain whether there was a trend of postoperative WOMAC score with time. There was a trend to improvement in terms of overall WOMAC scores (Fig. 2). In considering the WOMAC pain scores, graphical analysis suggested some worsening in pain at late follow-up (Fig. 3a). However, following statistical analysis using ANOVA, this trend was found to be non-significant (P = 0.777). The same analysis was performed on WOMAC activity scores (Fig. 3b). There was no significant difference in WOMAC activity scores over time (P = 0.751). Fifty-seven patients answered the WOMAC satisfaction questionnaire, of whom only one patient would not recommend revision arthroplasty to a friend. This patient had to undergo a re-revision total hip arthroplasty due to resistant infection. Sixty patients answered ''definitely yes'', six patients answered ''probably yes'', and one patient answered ''probably not'' to the same question.

Radiological findings
Of the 57 patients available for X-ray analysis and complete WOMAC questionnaires, 17 cases showed no subsidence. The mean subsidence for the 57 patients was 0.79 mm (CI 0.57-1.01, range 0-2.81). The mean rate of subsidence was 0.43 mm/year (CI 0.24-0.62), as outlined in Table 2.
The patient with the largest subsidence (2.81 mm) had a postoperative WOMAC score of 6, down from 37 at 21 months post-operatively, and was highly satisfied with the procedure. The subsidence-time graph for all the patients is shown in Fig. 4. Of the 57 patients with subsidence data, 23 showed a subsidence of between 1.0 and 2.81 mm at an average of 39.2 months' follow-up. Only one of these patients, who recorded a subsidence of 2.09 mm, suffered from a complication (superficial wound infection).
We performed a regression analysis to assess the correlation between the percentage change in WOMAC score and subsidence and found no correlation between these two variables ( Fig. 5: coefficient of determination, R 2 = 0.0015). In addition, we performed a regression    analysis by plotting Paproski femoral scores against subsidence and found no correlation between the two variables. (Fig. 6, R 2 = 0.0017).

Discussion
Revision hip arthroplasty is a technically difficult procedure. As younger patients undergo primary hip arthroplasty, the age of the revision hip patient is falling. Needless to say, failure rates in revision hip arthroplasty are higher than in primary hip surgery [11][12][13]. Surgeons are faced with the task of augmenting the reduced bone stock and weakened soft tissues. The options for revision are vast. Augmenting femoral bone loss is the main difficulty in achieving stable fixation. Cemented revision with impacted cancellous allograft is one of the ways to overcome this difficulty [14][15][16]. Previous studies on cemented long-stem revisions have shown high failure rates [15,17,18]; however, this can be attributed to older cementing techniques. Cementless revision techniques are suitable in some cases. However, results are unfavourable [19,20], and patients often complain of thigh pain and must undergo long periods of post-operative immobility [21]. In our study, all patients received the Exeter long stem. At 220-mm long, it has the advantage of bridging the bony deficiencies in the femoral canal. Seven patients underwent re-revision surgery; three for peri-prosthetic fractures, two for deep infection, one for aseptic loosening and one for recurrent dislocation. Periprosthetic fractures are a common complication of revision hip surgery. In our series, this was secondary to falls. Only one patient suffered aseptic loosening. This occurred 5 years after the initial surgery. In comparison, the aseptic loosening rate in other studies varied between 7% and 26% [8-10, 15, 16, 20].
Subsidence is defined as the inferior migration of the femoral stem following total hip arthroplasty. It is a phenomenon that applies to all stems, regardless of surface finish, presence of cement and type of stem and is due to axial forces acting on the prosthesis. Subsidence for primary total hip arthroplasty has been studied extensively. Less is known about its significance in revision hip arthroplasty. Subsidence of a femoral stem is governed by a number of factors. It was initially thought that an unpolished, roughened surface would provide stability and prevent femoral-stem loosening. These stems enjoyed good to excellent results and demonstrated minimal subsidence [22]. On the other hand, polished tapered stems make good use of cement creep and stress relaxation of cement to achieve stability [23]. The most common pattern is rapid initial subsidence followed by secondary stabilisation [24,25]. This characteristic, however, is stem dependent [30]. Excessive subsidence has been shown to predict early failure [26,27]. Cement viscosity was found not to significantly affect subsidence [28], and neither was the crosssectional shape of prosthesis [29]. Subsidence in revision femoral stems is further affected by impaction grafting and the integrity of the cortical envelope [30].
Our subsidence rate averaged 0.79 mm with a mean follow-up of 36.3 months. Seventeen hips exhibited no subsidence, with a mean follow-up of 34.6 months. Overall, these values for subsidence are far lower compared with published results. Eldridge et al. reported nine cases of massive early subsidence following cemented revision total hip arthroplasty [31]. Meding et al. found a mean subsidence of 10.1 mm in 13 of their 28 patients [32]. Other studies have reported similar results [33,34]. Our good results might be attributed to surgical cementing technique. Furthermore, our subsidence figures are comparable with those of primary hip arthroplasty.
There was no correlation between subsidence and WOMAC scores. This is likely due to our low subsidence figures. Surprisingly, bone loss (as graded by Paproski femoral score) did not affect subsidence. The patient who had the least bone stock prior to surgery did not have any subsidence 60 months post-operatively. We attribute this result to good cementing and allograft impaction techniques.  The gold standard for measuring subsidence is by Roentgen stereo-photogrammetric analysis (RSA) [35]. This represents an accurate method of determining threedimensional stem migration and has been used in numerous studies. However, it is time consuming, expensive and requires pre-operative implantation of tantalum beads. Our method may not be as accurate as RSA but allows the surgeon to measure subsidence rapidly and to reproduce the measurements with future X-rays. This will allow the surgeon to differentiate the high-risk group of patients who may require more frequent follow-up from those who have stable hips and may be discharged from follow-up. Further studies are required to evaluate this theory. We confirm that small subsidence (\3.0 mm) is not significant. Our method allows accurate measurement of subsidence that could be clinically significant.
The WOMAC score is a patient-oriented functional questionnaire. It takes into account patient perception of pain and activity levels both pre-and post-surgery. Our results show that after an initial improvement of WOMAC scores post-operatively, some patients show a slight worsening of their WOMAC scores. ANOVA testing revealed no significant difference between the post-operative scores at 12, 24, 36 and 48 months. A similar trend was seen for WOMAC activity score. Once again, there was no significant difference between the score at the various post-operative periods. Prior studies have shown little correlation between subsidence and clinical scoring provided there was no loosening [36]. In our study, patients satisfaction score was excellent. All but one patient would recommend the procedure to a friend.
In conclusion, the Exeter long femoral stem is an excellent option in revision hip arthroplasty. Our subsidence rates for long-stem revision femoral components are lower than the published data but demonstrate the same plateau. Radiographic subsidence does not appear to relate to functional outcome or complication rates in our data, but other studies show subsidence to be a significant indicator of premature prosthesis failure. Longer follow-up will be required to see if this is the case in our prospective cohort. Our method of measuring subsidence is simple, reproducible and can be used in the clinical setting.