In September 1985, the senior one of us (S. T. W.) began to perform primary total hip replacements with a porous-coated acetabular component (Harris-Galante; Zimmer, Warsaw, Indiana) inserted without cement and a femoral component (Precoat; Zimmer) inserted with cement in older patients and in patients who had osteoporosis of the proximal end of the femur. One hundred and forty-four patients who had had 157 primary so-called hybrid total hip replacements between September 1985 and June 1989 were included in this study. Twenty-four patients (17 per cent) (twenty-five hips) died before completion of the study. Twelve of these patients (thirteen hips) had been followed for less than one year. The remaining twelve patients (twelve hips) had last been examined at twelve to forty-six months, at which time they had an average Harris hip score8 of 95 points (range, 87 to 100 points). Six patients (4 per cent) (seven hips) could not be located for re-evaluation after eighteen to fifty-five months. The average hip score for these patients was 95 points (range, 88 to 100 points) at the last follow-up examination.
The remaining 114 patients (79 per cent) (125 hips) had an average of seventy-two months (maximum, 108 months) of clinical follow-up, and all but two patients, who had a revision at twenty-nine and forty-seven months, were followed for at least fifty-six months. Eight patients (eight hips) were interviewed over the telephone and were asked to send a new set of radiographs, and the remaining 106 patients were examined personally by the senior one of us. Four of the eight patients who were interviewed over the telephone refused to have new radiographs made, leaving 110 patients (121 hips) who had both clinical and radiographic follow-up at a minimum of fifty-six months or a revision before that time.
Of the 114 patients who had clinical follow-up, seventy-one were women and forty-three were men, and the average age at the time of the index arthroplasty was sixty-seven years (range, forty to eighty-eight years). The average weight was seventy-three kilograms (range, forty-four to 118 kilograms) and the average height was 166 centimeters (range, 148 to 185 centimeters). Eleven patients had a bilateral total hip replacement. The indication for the hip replacement was osteoarthrosis in seventy-nine hips; rheumatoid arthritis in fourteen; failure of a cup arthroplasty, an endoprosthesis, or fracture fixation in eleven; osteonecrosis in eight; acute fracture of the femoral neck in eight; congenital dysplasia in two; and miscellaneous conditions in three.
A posterolateral approach was used in 119 hips and a transtrochanteric approach, in six. One hundred and one of the Harris-Galante acetabular components were of the original design and were fixed to the pelvis with 4.5-millimeter titanium screws; the remaining twenty-four acetabular components were of the new design and were fixed with 6.5-millimeter screws. The diameter of the femoral head was twenty-eight millimeters in 119 hips and twenty-two or thirty-two millimeters in six.
All patients had prophylaxis for deep venous thrombosis with intermittent pneumatic compression alone or in combination with aspirin or low-dose warfarin, and all had routine surveillance for thrombosis with use of venography or duplex ultrasonography, or both, before discharge from the hospital.
Operative Technique
Before 1988, the acetabular implants were inserted after reaming with a reamer of the same size as the component to be inserted (line-to-line reaming). From 1988 on, the acetabular components were press-fit by underreaming by two millimeters. An average of 2.6 screws (range, one to four) were used for fixation of the metallic shell.
The femoral components were inserted with the so-called third-generation technique18. The femoral canal was widened with a femoral rasp that corresponded in size to that of the femoral component to be used, with added space for cement. The largest rasp that could be inserted easily without reaming of the canal was chosen and was nearly always the same size as that selected on the basis of preoperative planning with the use of a template on the preoperative radiographs. The medial aspect of the femoral neck was planed with a reamer placed over a trunnion on the rasp to provide contact of the collar of the prosthesis with the femoral neck. In all patients, a femoral plug was used to occlude the medullary canal of the femur. A cement plug was used in ninety-eight hips (78 per cent), and a bone or plastic plug was used in twenty-seven hips (22 per cent) because the diameter of the distal end of the femoral canal was too small to allow passage of the cement-plug syringe.
After the canal was plugged, loose cancellous bone was removed with a curet and the femur was then prepared for insertion of the cement with use of a Water-Pik. The canal was packed with a surgical sponge and was suctioned in order to dry it just before the retrograde filling of the canal with doughy cement with use of a cement gun. The porosity of the cement was reduced in all cases with either centrifugation or vacuum-mixing. The cement was then pressurized in the canal with use of a plunger-type femoral pressurizer.
Clinical and Radiographic Analysis
The Harris hip score8 was used to grade the function of the 125 hips (114 patients) that had clinical follow-up. An anteroposterior radiograph of the pelvis centered over the symphysis pubis and a frog-leg lateral radiograph of the hip and proximal aspect of the femur were made for all patients within several weeks after the operation. A cross-table lateral radiograph of the hip was available for 118 hips. At the latest follow-up examination, an anteroposterior radiograph of the pelvis was available for 121 hips and a frog-leg lateral radiograph, for 110 hips.
Acetabular Component
The acetabular component was assessed on the anteroposterior radiograph of the pelvis for initial gaps at the metallic shell-bone interface and the angle of abduction from a horizontal line drawn through the teardrops. It was assessed on the true lateral radiograph for the angle of anteversion. On the follow-up radiographs, a Müller template was used to determine migration of the acetabular component in a superior or medial direction according to the method described by Wilson et al. Migration was defined as superior or medial movement of three millimeters or more or tilting of more than 3 degrees. This method was reported to be accurate to within two millimeters19. The late appearance of radiolucent lines at the acetabular component-bone interface was noted.
Femoral Component
The femoral component was assessed for the grade of the initial cement mantle18 and the position within the femur. A grade-A cement mantle completely filled the femoral canal in all areas (a so-called white-out), extended two centimeters distal to the tip of the prosthesis, and was at least two millimeters thick in all areas without focal voids. A grade-B mantle also filled the femoral canal completely, but areas of trabecular bone surrounding the stem were not completely filled with cement. A grade-C1 mantle had voids or bubbles in the cement, and a grade-C2 mantle had focal deficiencies in the cement so that small portions of the stem were in contact with bone. A grade-D mantle had multiple focal deficiencies, no cement distal to the tip of the stem, or radiolucent lines at the cement-prosthesis interface. The cement mantle at the proximal-medial aspect of the femoral component within three centimeters distal to the collar was graded separately, as completely or focally deficient, poor (a complete mantle that was less than two millimeters thick), or good (a complete mantle that was more than two millimeters thick).
Radiolucent lines at the bone-cement and cement-prosthesis interfaces were recorded and localized according to the system of Gruen et al. Radiographic loosening was defined according to the criteria of Harris et al.10. Definite loosening was indicated by subsidence, a new radiolucent line at the cement-prosthesis interface, a radiolucent line around the entire cement mantle, or a fracture of the cement or stem. Possible loosening was indicated by a new radiolucent line occupying 50 to 99 per cent of the cement-bone interface. Subsidence was measured from the collar of the prosthesis to the lesser trochanter. Remodeling or thinning of the inner cortex of the femur was localized to the zones, described by Gruen et al., on the anteroposterior and frog-leg lateral radiographs.
Heterotopic ossification was graded according to the classification of Brooker et al. All areas of femoral endosteal lysis were noted.
Statistical Analysis
Stepwise regression analysis was done to determine significantly independent variables for the development of osteoporotic remodeling of the femur. The level of significance was p < 0.01.
Clinical
The average preoperative Harris hip score for the 114 patients (125 hips) who had clinical follow-up was 47 points (range, 5 to 74 points). The average postoperative score for the 109 patients (120 hips) who did not have revision of the femoral component or a Girdle-stone procedure was 91 points (range, 25 to 100 points). The result was excellent (90 to 100 points) for eighty-seven hips (73 per cent), good (80 to 89 points) for twenty-one (18 per cent), fair (70 to 79 points) for seven (6 per cent), and poor (less than 70 points) for five (4 per cent). Of the twelve patients (twelve hips) who had a fair or poor result, nine had no or only slight pain in the hip; two had mild pain; and one, who had had recurrent dislocations and was not able to walk because of generalized physical disability, had moderate pain (this patient had a hip score of 25 points). Current radiographs were not available for the patient who had moderate pain. The remaining ninety-seven patients (108 hips) had no or only slight pain in the hip. Over-all, 98 per cent (118) of the 121 hips that were not revised for loosening of the femoral component were associated with either no or slight pain.
Seven patients (seven hips; 6 per cent) had a revision operation, a reoperation for disassembly of the component, or a resection arthroplasty. Four femoral components (3 per cent) in four patients were revised for aseptic loosening at an average of fifty-five months (range, twenty-nine to seventy-four months). Pain in the thigh had developed in two of these four patients immediately after a fracture of the femoral component at seventy-one and seventy-four months (Figs. 1-A and 1-B). The other two patients, who had the revision at twenty-nine and forty-seven months, had subsidence of the component with a complete radiolucent line at the bone-cement interface. A fifth patient had a Girdlestone procedure at seventy-four months because of late infection. None of these five patients had loosening of the acetabular component. A sixth patient had a reoperation for reassembly because a modular femoral head had dissociated from the Morse taper lock of the stem during a closed reduction of a late posterior dislocation23. The final patient had replacement of a polyethylene liner that had displaced from the metallic shell during a fall on the knee at fifty-four months.
Thirty-four patients had a postoperative complication, including a wound hematoma in two (both of whom were managed with operative evacuation), a urinary tract infection in two, deep venous thrombosis in twenty-two (sixteen of whom had involvement of the proximal veins and six of whom had involvement of only the veins in the calf), and a non-fatal pulmonary embolus in one. Seven patients had a posterior dislocation.
Radiographic
Acetabular Component
Three hips had gaps between the bone and the acetabular component on the initial postoperative radiographs; none of these gaps were more than one millimeter wide or involved more than one of the zones described by DeLee and Charnley. The cups were positioned in an average of 39 degrees (range, 18 to 55 degrees) of abduction. One hundred and eight components were positioned in anteversion; six, in neutral; and four, in slight retroversion.
At the latest follow-up examination, none of the 121 acetabular components had migrated. No radiolucent lines were seen at the bone-prosthesis interface on any of the anteroposterior radiographs, and only two components had a radiolucent line at the bone-prosthesis interface that involved one or two zones on the frog-leg lateral radiograph (the oblique view of the acetabulum). There were no complete radiolucent lines at the bone-prosthesis interface, and there was no acetabular osteolysis. There were no fractures, and there was no evidence of movement, of any screw.
Femoral Component
The cement mantle about the femoral component was assessed for all 121 hips on the anteroposterior radiograph and for all but eleven hips on the frog-leg lateral radiograph. The column of cement extended distal to the tip of the component in all but one hip, which had had a femoral bone plug. Voids (or air bubbles) were seen within the cement mantle in fifty-six hips (46 per cent). Thirty-eight hips had at least one void within the proximal aspect of the cement mantle, and eighteen had an air bubble at the tip of the stem. Seven hips had both a proximal void and a bubble at the tip of the stem. It was discovered that the bubble was caused by air caught within a hole at the tip of the stem that had been designed by the manufacturer to hold a cement-centralizer. After this discovery, the hole was filled with liquid cement before the stem was inserted. A distal modular cement-centralizer was not used in any of the patients.
The cement mantle was grade A in eighteen hips (15 per cent), grade B in thirty-six (30 per cent), grade C1 in forty-three (36 per cent), and grade C2 in twenty-two (18 per cent)18. The cement mantle was grade D in two hips (2 per cent) because of incomplete seating of one of the components within the mantle and because of the absence of cement at the tip of the stem of the other. The cement mantle at the proximal-medial aspect of the femoral component was deficient in twenty-three hips (19 per cent), poor in seventeen (14 per cent), and good in eighty-one (67 per cent).
The femoral component was in neutral alignment in relation to the femoral canal in ninety-two hips (76 per cent), in less than 10 degrees of valgus in twenty-six (21 per cent), and in less than 5 degrees of varus in three (2 per cent).
At the latest follow-up examination, fourteen hips (12 per cent) demonstrated a punched-out lesion of the bone at the medial aspect of the femoral neck, indicative of focal osteolysis. Eleven hips (9 per cent) had severe heterotopic ossification (class III or IV), twenty-eight (23 per cent) had class-I or class-II heterotopic ossification, and eighty-two (68 per cent) had no heterotopic ossification.
New radiolucent lines at the femoral bone-cement interface were found on the latest anteroposterior radiograph of twenty-three hips (19 per cent) and on the frog-leg lateral radiograph of twenty-seven hips (25 per cent); however, only one hip that had not been revised for loosening of the femoral component had a new radiolucent line that occupied more than 50 per cent of the bone-cement interface on both radiographs. Migration of the femoral component was seen in one other patient, in whom the component had subsided five millimeters. One patient had an undisplaced fracture of the cement mantle at the medial aspect of the tip of the stem but did not report any clinical symptoms related to the hip. The fracture was detected at thirty-nine months, but it did not increase in size and there was no more subsidence of the femoral component after sixty-three months. Therefore, in addition to the four femoral components that were revised, two were definitely loose and one was possibly loose.
Four of the seven patients who had a revision for loosening of the femoral component or who had definite or possible radiographic loosening of that component had a focal deficiency of the cement mantle or a mantle that was poor in the proximal-medial area (three had a grade-C1 or C2 mantle and one had a grade-D mantle), and two had a good mantle in the proximal-medial area (grade A in one and grade C1 in the other). The seventh patient, who had a grade-D cement mantle, had radiographic findings of possible loosening and extensive areas of endosteal lysis, although the hip was asymptomatic at the latest follow-up examination. Thus, six of these seven hips had a grade-C or D cement mantle.
Remodeling of the medullary bone at the proximal aspect of the femur was assessed on the anteroposterior radiographs and occurred to some degree in seventy-three (60 per cent) of the 121 hips. Forty-six hips (38 per cent) had extensive osteoporosis of the endosteal femoral cortex—that is, the development of osteoporosis in at least four of the seven zones of Gruen et al. evident on either the anteroposterior or the frog-leg lateral radiograph. This remodeling process involved the entire proximal aspect of the femur (all seven zones of Gruen et al. on the anteroposterior radiograph) in sixteen patients (sixteen hips; 13 per cent); two of these patients were men and fourteen were women (Figs. 2-A and 2-B. Seventy-five hips had slight or no remodeling of the femur. Stepwise analysis revealed that women (p = 0.008) and patients who had had a longer duration of follow-up (p = 0.002) had more extensive femoral osteoporosis. Other independent variables, including the age and weight of the patient, the size of the stem, the percentage of the canal filled by the femoral stem at the mid-point and at the tip of the stem, and the thickness of the medial femoral cortex at the tip of the stem, could not be shown to have an effect on this remodeling process.
Endosteal lysis of the femoral canal occurred in ten (8 per cent) of the 121 hips, including two that were revised for loosening of the femoral component, and was extensive in two hips that had not been revised at the time of writing. The patient who had definite radiographic loosening as determined by subsidence of the femoral component had a small lytic lesion, but the two patients who had major lytic lesions had no symptoms at the latest follow-up examination, although one had evidence of possible loosening of the component.
Insertion of an acetabular component without cement and a femoral component with cement, or so-called hybrid fixation, has become commonplace in primary total hip replacement, but there have been few intermediate follow-up studies documenting the clinical and radiographic results of this technique6,14,17,18. Our study of a consecutive series of patients in whom the operation was performed in standard fashion by one surgeon was designed to provide such information with few confounding variables.
The clinical results were excellent in our series. Of the 120 hips that were not revised for loosening or late infection, 108 (90 per cent) were not painful, nine (8 per cent) were slightly painful, two (2 per cent) were mildly painful, and one (1 per cent) was moderately painful at the time of the latest follow-up examination. One hundred and eight (90 per cent) of the 120 hips demonstrated a negative Trendelenberg sign.
The radiographic findings for the 121 Harris-Galante acetabular components for which radiographic follow-up data were available were remarkable in that there was no loosening, evidence of acetabular lysis, or problems related to any of the screws. Failure of the liner-locking mechanism in one acetabular component necessitated a reoperation. Only two cups were associated with new radiolucent lines. To our knowledge, no previous study with an average of six years of follow-up has demonstrated such good radiographic results for fixation of the acetabular component. These good results may have been due to the liberal use of screws during insertion of the metallic shell, which provided good fixation with no initial gaps as well as subsequent bone ingrowth into the porous surface of the implant. One of us (S. T. W.) and Murphy reported on the rate of wear of the polyethylene of this component22, and it appears to be comparable with the rates of wear documented for other implants.
In contrast to the excellent results with the acetabular component, the femoral component failed because of loosening (was revised or was radiographically loose) in 5 per cent (six) of the 121 hips that were followed radiographically. We attribute four of these failures to a deficient cement mantle at the proximal-medial aspect of the femoral component. Thirty-three per cent (forty) of the 121 cement mantles in this series were less than two millimeters thick or had focal deficiencies at the proximal-medial aspect (zone 7) on the initial radiographs. We believe that one of the causes of these deficiencies was that the rasp that corresponded to each implant size was not oversized enough proximally to create sufficient space for a two-millimeter-thick cement mantle proximally. It is unlikely that the proximal-medial deficiency of the cement mantle was caused by poor positioning of the stem within the column of cement, as only three of the 121 stems were in slight varus. However, a good mantle of cement and good fixation were achieved at the distal portion of the stem, causing a cantilever bending mode of failure that resulted in loosening of the stem in four patients and fracture of the stem in one. Although a varus position of the stem within the cement mantle was not common, 21 per cent (twenty-six) of the stems were positioned in slight valgus. Only one (4 per cent) of the stems placed in valgus failed, whereas one of the three stems placed in varus fractured.
The 2 per cent rate of fracture of the femoral stem (two of 121 hips) is high and causes some concern because this is a forged chromium-cobalt prosthesis. We are aware of only one other reported case of fracture of a forged chromium-cobalt stem13, a Harris Design-2 implant (Howmedica, Rutherford, New Jersey), which is similar in design to the Precoat stem used in this series. One of the fractures in the present study occurred in a sixty-seven-year-old man who weighed eighty-six kilograms. This patient had focal deficiencies in the cement at the proximal-medial aspect of the cement mantle and excellent distal fixation (a white-out) of a small-sized prosthesis (Figs. 1-A and 1-B). The stem fractured at the junction of the middle and distal thirds. The other fracture occurred in a sixty-five-year-old man who weighed 110 kilograms. The large-sized stem also fractured at the junction of the middle and distal thirds. The proximal-medial aspect of the cement mantle was completely absent as the prosthesis had not been completely introduced within the femur (it was left proud by one centimeter) because the cement plug had been placed too proximal in the femoral canal; the distal fixation was excellent. Both of these fractures, which occurred approximately six years postoperatively, appeared to be caused by fatigue and were not associated with any gross metallurgical defects.
Although the rate of femoral loosening in this series was relatively low, we are concerned that it was higher than that reported by other authors who followed similar implants for longer periods of time. Barrack et al. reported radiographic failure of one (2 per cent) of fifty Harris Design-2 prostheses in patients who were less than fifty years old. No component had been revised at ten to fifteen years (average, twelve years). Mulroy and Harris reported a 3 per cent rate of failure (a 2 per cent rate of revision and a 1 per cent rate of radiographic loosening) of the Harris Design-2 implant in 105 hips at an average of 11.2 years. These two studies involved use of so-called second-generation cementing techniques with a stem that was the same as the Precoat prosthesis with regard to metallurgy, but a monolithic femoral component design and a non-modular acetabular component with cement were employed. It could be speculated that use of an improved (so-called third-generation) cementing technique and precoating of the components improves the results of fixation. However, the rate of failure in our series was approximately twice as high as that in these two studies after half the average duration of follow-up. Modularity of the femoral component, the quality of the proximal-medial aspect of the cement mantle, and polyethylene wear of components inserted without cement may have contributed to the higher rate of failure in our series.
The over-all prevalence of femoral lysis was 8 per cent (ten of 121 hips), although only two of these lytic reactions in patients who had not had a revision involved large portions of the femur. This rate does not include the fourteen hips that had small lytic lesions at the proximal-medial aspect of the femoral neck distal to the collar of the femoral component; it was calculated for the lytic lesions that were seen in areas other than distal to the collar. This rate was much higher than that reported by Goetz et al., who found no lysis an average of six years after forty-one Precoat hip implants had been inserted with so-called hybrid fixation, but it was consistent with that reported by Oishi et al., who found a 6 per cent prevalence an average of seven years after insertion of eighty-nine Precoat hip replacements. Barrack et al. found the rate of lysis to be 12 per cent (six of fifty hips), with all of the lesions being focal, at an average of twelve years, and Mulroy and Harris reported a rate of 7 per cent (seven of 103 hips) at an average of 11.2 years. The rate of osteolysis in these previous studies and in the present investigation of cemented femoral components is dramatically less than that in a previous series reported on by one of us (S. T. W.) and Maloney21 of patients who had a Harris-Galante femoral implant inserted without cement combined with the same Harris-Galante acetabular component used in the present study. The rate of osteolysis in that series21 was 22 per cent (fifteen of sixty-nine hips) after an average of 3.5 years. This rate increased to 53 per cent (thirty-five of sixty-six hips) at five years, and osteolysis was the major reason for the 12 per cent rate of revision of the femoral component (eight of sixty-six hips). It appears that the risk of femoral endosteal osteolysis from particulate polyethylene shed from the Harris-Galante acetabular component is sharply decreased with the use of bone cement for fixation of the femoral component.
Remodeling of the medullary bone of the femur as manifested by the development of osteoporosis of the inner cortex was a common finding and was significantly related to female gender and the duration for which the component had been in place. Engh et al. found that bone-remodeling of the proximal aspect of the femur after hip replacement with insertion of the AML (anatomic medullary locking) prosthesis (Depuy, Warsaw, Indiana) without cement occurred to a greater degree in patients who had more severe osteoporosis at the time of the arthroplasty. It appears that women have a greater risk of adaptive remodeling of bone because they have a higher prevalence of osteoporosis than male patients. We found no association between bone-remodeling and the risk of loosening of the femoral component.
So-called hybrid fixation for primary total hip replacement provided excellent results in this series of patients. We hope that our six-year results will hold true for the long term, and we are guardedly optimistic as very few of the femoral implants had radiographic evidence of impending failure (the formation of radiolucent lines at the bone-cement interface) and only two hips had extensive areas of osteolysis. We encourage the use of this technique in all elderly patients and in young patients for whom femoral fixation without cement is not suitable because of femoral osteoporosis. However, we caution that an adequate mantle of cement should be obtained in the proximal portion of the femur with use of a rasp that allows enough space for cement and possibly with use of a proximal cement-spacer.