Sir John Charnley introduced cement fixation of prosthetic femoral implants in total hip arthroplasty in 1961. Since that time there have been many innovations in the technique of cement insertion and in femoral stem design which have helped make insertion of a prosthetic femoral implant with cement a very reliable and reproducible procedure. There are many studies in the literature that document revision rates for cemented femoral stems due to aseptic loosening of 5% or less at follow-up of at least 15 years1-4 and the results have improved as cementing techniques have evolved. In recent years, however, there have been several reports of early failure of cemented femoral stems that have a rough surface finish or precoating.5-9 These stems include the Iowa (Zimmer, Warsaw, Indiana) and Zimmer Precoat¨ (Zimmer, Warsaw, Indiana) stems--implants which feature a coat of polymethylmethacrylate (PMMA) applied in the factory in a process referred to as pre-coating. Failure of some pre-coated stems and some non-smooth stems has been associated with debonding of the implant-cement interface. Subsequent motion between the roughened implant surface and the cement is believed to generate wear debris leading to periprosthetic osteolysis.

Surgeons presenting data on early femoral failures have implicated the pre-coating or roughened surface finish as the primary source of problems. Other factors have been incompletely considered. In our opinion, mechanical properties may be equally important in the early aseptic failure of cemented femoral stems. In particular, certain stem sizes and shapes may be associated with an increased risk for fatigue failure of the proximal cement mantle, particularly in heavier, more active patients. We have set out to test this hypothesis using mechanical testing methods developed in our laboratory.

Figure 1: Debonding of the cement-prosthesis interface has been associated with early failure of pre-coated and rough finished femoral stems.

Problems with Rough-Surfaced Femoral Implants

          Mohler and colleagues7 reported revision because of failure of fixation in 29 of 1941 (1.5%) Iowa femoral stems inserted with cement between two and 10 years follow-up. This series included both precoated and non-precoated stems. Twenty of the 29 stems that failed before 10 years of follow-up were pre-coated with PMMA. The overall revision rate was 2% in pre-coated and 1% in non-precoated femoral stems. All of the stems failed by debonding of the cement-stem interface. The average weight of the patients with failed stems was 190 pounds with some patients weighing as much as 300 pounds. Of the 29 loose stems, 11 had defects in the cement mantle visible on an anteroposterior radiograph. No lateral films were taken.

          Callaghan and co-workers6 documented aseptic loosening requiring revision in 8 of 131 (6%) pre-coated Iowa femoral implants followed for five to eight years. Again, all of the stems failed by debonding with rotation into retroversion. Debonding was followed by progressive osteolysis. The average weight of these patients was 97.6 kg (215 lbs.).

          The same authors then reviewed the results of total hip arthroplasty in patients under 50 years of age and compared the results of polished Charnley femoral stems at an average of 20 years follow-up with the results of grit-blasted and pre-coated Iowa femoral stems at 8 years average follow-up.5 The rates of revision and of radiographic loosening were 5% and 13% respectively for the polished Charnley stems as compared to 18% and 24% for the pre-coated, grit-blasted Iowa stems.

          Woolson and Haber8 noted similar problems with early loosening when using the Zimmer Precoat¨ femoral stem. In their series, four of 121 stems (3.3%) required revision for aseptic loosening at an average of six years follow-up. Another two stems were radiographically loose but had not been revised, giving a total of 5% loose stems at an average of six years. One year later three more of these stems were revised and one more was radiographically loose.9 Thus at just over six years, 8% of the Zimmer Precoat¨ femoral stems were revised or loose. These stems also failed by debonding. The failed stems were all noted to have defects in the proximal cement mantle.

Good Results with Rough Surfaced or Precoated Femoral Implants

          The excellent experience of other surgeons using rough surfaced and/or precoated femoral implants establishes that other factors may be important. Using matte finished Charnley stems in patients under 50 years of age, Wroblewski and Siney(10) had a stem revision rate for aseptic loosening of less than 1% at an average of 14 years follow-up. In a consecutive, prospective series of cemented HD-2 stems (Howmedica, Rutheford, NJ) in patients aged 50 years and younger, Barrack and colleagues11 documented debonding of a single stem (2%) with no revisions for aseptic loosening at an average 12-year follow-up. Harris12, in a summary of several series of Precoat stems noted a revision rate of 0.7% for aseptic loosening with follow-up of five to 8.5 years in 279 hips. These studies when added to his own recent 10 year follow-up of 102 Precoat stems1 yield a combined average revision rate of 1% for aseptic loosening over 10 years for 381 Precoat stems.

           Given the differences in success rates when using pre-coated femoral implants, it seems unlikely that the early failures reported can be attributed to surface finish alone. Loosening and subsequent failure of cemented femoral stems involves many different factors that are closely interrelated. These include patient related factors such as size, age, activity level, and bone quality; technical factors such as the technique of cement insertion and the quality of the cement mantle; and implant features such as geometry, offset, and surface finish.

Mechanical Investigations

           Many of the factors related to early failure of a femoral implant can be characterized in terms of mechanical stability, particularly under torsion. The mechanical behavior of cemented femoral implants has been an area of interest in the Biomechanics Laboratory at Massachusetts General Hospital for many years. Previous studies focusing on implant micromotion and cement mantle strains around femoral implants subjected to various loading conditions have laid the groundwork for current investigations regarding the mechanism of early failure of cemented pre-coated femoral implants.

Background Work           Burke and colleagues13 examined the in situ micromotion of cemented and cementless femoral stems during single leg stance and stair climbing. Models of stair climbing are useful for examining the forces on femoral implants because they produce substantial out-of-plane loading, particularly in torsion. The forces encountered by the proximal femur during stair climbing - including those from the hip abductor and extensor musculature - were modeled in the laboratory on the basis of data from radiographs and computed tomography scans. When compared to the forces encountered in normal gait, stair climbing produced similar axial, but greater torsional micromotion of a cemented femoral prosthesis.           O'Connor and co-workers14 used the same model to examine strain in the cement mantle during gait and stair climbing. They noted that the highest strains occurred near the tip of the prosthesis and in the proximal portion of the cement mantle. The proximal mantle strains were significantly higher in stair climbing than in gait. Some strains were greater than 1000 microstrain, which has been estimated as the fatigue limit of polymethylmethacrylate in vitro.

Figure 2: We have developed a special loading jig to simulate the complex forces encountered by the proximal femur during stair climbing.

Current Investigations

          In light of the conflicting clinical reports regarding the success and failure of fixation of certain cemented femoral stems, it is essential to examine individually the many potential contributors to early debonding and loosening. Studies underway in our laboratory are using the previously established model of stair climbing to measure strain in the proximal cement mantle as a measure of the torsional stability of cemented femoral stems. Body weight, component offset, stem size, and the presence or absence of calcar-collar contact are being considered as potential contributors to early loosening.

           The investigations use Zimmer Centralign¨ (Zimmer, Warsaw, IN) femoral stems cemented into cadaver femora. Strain gage rosettes are mounted in the proximal cement mantle just below the collar and the principle strains in the proximal cement mantle are measured under loading conditions that simulate stair climbing. Strain data are collected at different joint reaction forces in specimens with a variety of stem sizes and neck lengths. Specimens with and without calcar-collar contact are tested.

          Early data suggest that increased offset and body weight raise the peak principle strain in the proximal cement mantle in a linear fashion. The effect of body weight is more than four times greater than that of offset. Calcar-collar contact reduces peak strains in the proximal cement mantle approximately 50% even during stair climbing. Never the less, stair climbing can produce peak strains greater than 1500 microstrain, particularly when smaller stems without calcar-collar contact are tested. It is believed that this level of strain could lead to fatigue failure of the cement.

Conclusions

          It may not be appropriate to condemn pre-coated or rough surfaced femoral implants as the only reason for the early implant failures noted in recent series. Our preliminary laboratory data suggest that patient and implant variations that affect the torsional strains on the proximal cement mantle may be important in the torsional stability of cemented femoral implants and thus in the rate and timing of failure of fixation of cemented femoral stems. In particular, stair climbing can produce peak strains in the proximal cement mantle that may exceed the fatigue limit of polymethylmethacrylate, particularly when smaller stems are used without calcar-collar contact. In this regard, our laboratory data concur with a finite element computer model developed by Chang, Mann and Bartel.15 The importance of surface finish cannot be accurately evaluated until additional clinical and laboratory data on other contributing factors such as stem design, offset, and patient weight have been fully analyzed.

Melvyn Harrington, MD is a Fellow in the Adult Reconstructive Unit, Massachusetts General Hospital

Daniel O. OÕConnor and Andrew J. Lozynsky are researchers in the Orthopaedic Biomechanics Laboratory at Massachusetts General Hospital

Ian Kovach, MD is a former Fellow, Adult Reconstructive Unit, Massachusetts General Hospital

William H. Harris, MD is Chief of the Adult Reconstructive Unit at Massachusetts General Hospital, and Allen Gerry Clinical Professor of Orthopaedic Surgery at Harvard Medical School

Address Correspondence to:
William H. Harris, MD; Massachusetts General Hospital; Jackson 1126; 55 Fruit St.; Boston, MA 02114