Fifty-four cobalt-chrome alloy femoral heads were retrieved at revision surgery or at post-mortem; 23 came from uncemented, eight from hybrid and 23 from cemented hip arthroplasties. The uncemented and hybrid implants had porous coating, metal backing and modular femoral heads; with one exception none of the cemented implants had any of these. Twenty-five of the 31 heads from uncemented and hybrid arthroplasties, and 11 of the 23 heads from cemented arthroplasties showed surface damage involving more than 25% of the surface. Scanning electron microscopy revealed multidirectional fine scratches 1 micron to 10 microns in depth and width which appeared to have been made by fine, hard particles. There was a higher rate of such damage in the uncemented and hybrid arthroplasties than in the cemented implants, suggesting that the abrasive particles were mainly released from the metal, rather than from the cement or polyethylene components of the implants.

Six porous-coated, uncemented femoral components were revised at a mean of 34.5 months for persistent thigh pain. At operation the stems were rigidly stable, difficult to extract, and showed good bony ingrowth. The four men and two women, with an average age of 59 years, all had thigh pain starting within the first year, progressive over time and unresponsive to conservative measures. These cases show that rigid fixation with good bony ingrowth does not guarantee the clinical success of a porous-coated uncemented femoral stem.

The radiographic and histological features of radiolucent areas at the cement-bone interface were correlated in 15 specimens retrieved at post-mortem from patients who had undergone cemented total hip arthroplasty, two weeks to 15 years prior to death. All but one of the components were securely fixed, as demonstrated by direct measurements of micromotion. Extensive radiolucencies were present in all but one case. In 11 of the 14 specimens with radiolucencies, histological examination showed that the radiolucent areas represented regions of osteoporosis and bone remodelling. The remodelling changes were characterised by osteoporosis, cancellisation and thinning of the endosteal cortex, and osteopenia of the trabecular bone. In two specimens the appearance of radiolucency was found to be due to fibrous tissue at the cement-bone interface and in one specimen there was a mixed picture of osteolysis and fibrosis. The study demonstrates that radiolucent lines can occur with well-fixed components and that they may commonly represent osteoporosis rather than the presence of a fibrous membrane at the cement-bone interface.

We studied 16 femora retrieved at post-mortem from symptomless patients who had a satisfactory cemented total hip arthroplasty from two weeks to 17 years earlier, with the aim of delineating the initial mechanisms involved in loosening. Only one specimen showed radiographic evidence of loosening; the other 15 were stable to mechanical testing at 17.0 Nm of torque. In all 16 specimens, the cement-bone interface was intact with little fibrous tissue formation. By contrast, separation at the cement-prosthesis interface and fractures in the cement mantle were frequent. The most common early feature was debonding of the cement from the metal, seen at the proximal and distal ends of the prosthesis. Specimens which had been in place for longer also showed circumferential fractures in the cement, near the cement-metal interface, and radial fractures extending from this interface into the cement and sometimes to the bony interface. The most extensive cement fractures appeared to have started at or near sharp corners in the metal, or where the cement mantle was thin or incomplete. Fractures were also related to voids in the cement. The time relationship in this series suggested that long-term failure of the fixation of cemented femoral components was primarily mechanical, starting with debonding at the interface between the cement and the prosthesis, and continuing as slowly developing fractures in the cement mantle.

We evaluated the initial stability of cemented and uncemented femoral components within the femoral canals of cadaver femurs during simulated single limb stance and stair climbing. Both types were very stable in simulated single limb stance (maximum micromotion of 42 microns for cemented and 30 microns for uncemented components). However, in simulated stair climbing, the cemented components were much more stable than the uncemented components (76 microns as against 280 microns). There was also greater variation in the stability of uncemented components in simulated stair climbing, with two of the seven components moving 200 microns or more. Future implant designs should aim to improve the initial stability of cementless femoral components under torsional loads; this should improve the chances of bony ingrowth.

We have reviewed 25 cases of focal femoral osteolysis in radiographically stable, cemented femoral implants. In three hips retrieved at post-mortem from two patients, we have been able to make a detailed biomechanical and histological analysis. The interval between arthroplasty and the appearance of focal osteolysis on clinical radiographs ranged from 40 to 168 months, and in over 70% of the cases this did not appear until after five or more years. Few had significant pain and there was no relation to age, sex or original diagnosis. The most common site for osteolysis were Gruen zones 2 and 3 on the anteroposterior radiograph and zones 5 and 6 on the lateral radiograph. In 15 cases (60%), the area of osteolysis corresponded to either a defect in the cement mantle or an area of very thin cement. The rate of progression of these lesions was variable, but to date only one has progressed to gross loosening of the femoral component. The back-scatter scanning electron microscopic examination of serial sections and biomechanical testing of the post-mortem specimens demonstrated focal cement fracture around implants that were otherwise rigidly fixed. In eight cases from which tissue was available, histology showed a histiocytic reaction with evidence of particulate polymethylmethacrylate. We consider that this local fragmentation was the stimulus for local osteolysis in an otherwise stable cemented femoral component.

We evaluated 38 hip reconstructions in 36 patients at a mean follow-up of 5.9 years (range 4 to 9.1) after femoral head allografts had been used to augment severely deficient acetabular bone stock. The patients were all relatively young and had many previous operations. Their pre-operative Harris hip rating averaged 46 points (range 18 to 73). All the allografts united and there were no infections. However, 12 acetabular components (32%) became loose; six of these had needed revision using the healed allograft, and two hips had required resection arthroplasty. The 30 surviving hips had a mean Harris hip score of 82 points. Some radiographic evidence of graft resorption was seen in 23 hips, though this was mild in 17. The extent of cover provided by the allograft and the severity of graft resorption both correlated with acetabular loosening. Although structural allografts had allowed successful hip reconstructions in many of these patients with major bone loss, the failure rate had increased from zero at four years to 32% at six years; clearly they provide only a short-term solution.

We have tested the porosity and fatigue life of five commonly used bone cements: Simplex P, LVC, Zimmer regular, CMW and Palacos R. Tests were conducted with and without centrifugation and with the monomer at room temperature and, except for LVC, at 0 degrees C. We found that the fatigue life of different specimens varied by a factor of nearly 100. It did not depend on porosity alone, but was more influenced by the basic composition of the cement. Simplex P when mixed with monomer at 0 degrees C and centrifuged for 60 seconds had the highest fatigue life and was still sufficiently liquid to use easily.