The insertion of a stemmed prosthesis into the femur initiates adaptive bone-remodeling that can lead to varying degrees of bone loss4,12,14. Interest in this phenomenon of so-called stress-shielding has led to the performance of numerous biomechanical and finite-element studies in an attempt to identify the factors that lead to bone-remodeling. The bone-remodeling response has been quantified, both in vivo and in specimens obtained at autopsy, with use of such methods as comparison of plain radiographs, video densitometry, and dual-energy x-ray absorptiometry5,6,11-13,15,16. These studies have led to the understanding that bone-remodeling is a ubiquitous phenomenon that occurs with all types of hip arthroplasty implants and is not unique to devices that are inserted without cement13.
It has been theorized that periprosthetic bone-remodeling secondary to stress-shielding may contribute to increased pain or decreased function, fracture of the femur or the femoral component, loss of fixation of the implant, increased prevalence or severity of osteolysis, and difficulty in performing a revision10,12. However, the prevalence of these adverse consequences has not been reported, to our knowledge.
We describe the long-term clinical outcome of total hip arthroplasty performed without cement in patients who had evidence of stress-shielding on the two-year postoperative radiographs.
*One or more 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. No funds were received in support of this study.
†Department of Orthopaedic Surgery, University of California at San Diego, 200 West Arbor Drive, San Diego, California 92103.
‡Anderson Orthopaedic Research Institute, 2445 Army Navy Drive, Arlington, Virginia 22206. Please address requests for reprints to Dr. Engh, Sr. E-mail address for Dr. Culpepper and Dr. Engh: aori@ix.netcom.com.
The present study group was drawn from our larger (parent) series, which consisted of 215 patients who had had a total of 223 consecutive primary total hip arthroplasties with use of the anatomic medullary locking hip system (AML; DePuy, Warsaw, Indiana)7. This system consists of a non-modular femoral component with a porous coating over four-fifths of its surface and a non-modular acetabular cup; the stem is available in seven sizes and the cup is available in eight. Both components consist of a cobalt-chromium substrate with a porous coating of sintered cobalt-chromium beads (mean pore size, 200 micrometers). The stem has a fixed thirty-two-millimeter-diameter femoral head.
The extent of bone-remodeling (stress-shielding) was graded on the two-year postoperative radiographs according to the method of one of us (C. A. E., Sr.) and Bobyn5. Anteroposterior and Lowenstein lateral radiographs were made until both the contrast and the orientation were similar to those on the immediate postoperative radiographs5. The femur was divided into sixteen discrete sites (Fig. 1). The criterion for bone-remodeling at each site was simply whether the bone appeared darker, thinner, or more osteopenic on the two-year radiographs as compared with the immediate postoperative radiographs. Remodeling was considered to be mild when the changes were confined to the four sites around the proximal quarter of the prosthesis (zones 1M, 1L, 1A, and 1P, Fig. 1) and was considered to be pronounced and of potential clinical importance when the changes involved more than these four zones and, specifically, when they were observed distal to the level of the lesser trochanter (Fig. 2).
Of the 199 patients (207 hips) from our larger (parent) series7 for whom two-year radiographs were available, forty-seven (forty-eight hips) had evidence of pronounced femoral bone-remodeling. The initial diagnoses were osteoarthrosis (thirty-six hips; 75 per cent), aseptic necrosis (six hips; 13 per cent), rheumatoid arthritis (four hips; 8 per cent), dysplasia of the hip (one hip; 2 per cent), and femoral fracture (one hip; 2 per cent). Three patients (three hips) died within ten years after the index arthroplasty, leaving forty-four patients (forty-five hips) who were followed clinically for a minimum of ten years. Thirteen patients were male and thirty-one were female; the average age at the time of the arthroplasty was fifty-seven years (range, sixteen to seventy-nine years). The forty-four patients responded to a questionnaire that included twenty-one items related to pain, function, and over-all satisfaction. The results obtained from this questionnaire are the focus of the present study. Data regarding complications and reoperations also were recorded. Thirty-two patients (thirty-three hips) had both a physical examination and a radiographic evaluation at least ten years postoperatively. Ten of the remaining twelve patients had been followed radiographically for at least five years (mean, seven years; range, five to nine years). All of the most recent radiographs were compared with the immediate postoperative and two-year radiographs to determine the status of fixation of the component7 and the presence of osteolysis18.
The femoral component was considered to be stably fixed by either bone or fibrous tissue, or to be unstable8. The component was thought to be fixed by fibrous tissue if it had not migrated but lacked definite signs of bone ingrowth, and it was considered to be unstable if serial radiographs demonstrated a change in its position within the femur—that is, subsidence of two millimeters or more or varus or valgus tilting.
The acetabular component was considered to be unstable if it had tilted 5 degrees or more, had migrated two millimeters or more, was associated with a circumferential radiolucent line, or demonstrated shedding of beads.
Areas of localized loss of trabecular bone or cortical erosion were defined as osteolysis18. Because these cystic lesions had discrete borders, their size was measured by multiplying their length by their width on the anteroposterior radiograph. These lesions also were characterized by their location in the femur or the pelvis.
The most recent clinical data were compared with the preoperative data with use of the Wilcoxon rank-sum test; p < 0.05 was considered to be significant. The results for the larger (parent) series were reported previously7, and they are not reported here. The clinical results for the subgroup of forty-four patients (forty-five hips) who had evidence of bone-remodeling were compared with those for the 167 patients (174 hips) who were followed for at least ten years as well as with the long-term clinical results for sixty-nine patients (seventy-eight hips) who had been managed with a porous-coated anatomic prosthesis (PCA; Howmedica, Rutherford, New Jersey) that had been inserted without cement17.
Patients Who Died before the Ten-Year Evaluation
Three patients (three hips) died within ten years after the index arthroplasty. At the time of the latest follow-up (at three, four, and eight years), none of these patients had pain in the hip and only one used a walking aid. There were no known complications or reoperations. All of the femoral components appeared to have bone ingrowth, and all of the acetabular components appeared to be stable. No osteolytic lesions were observed.
Patients Who Were Followed for a Minimum of Ten Years
Reoperations and Complications
Two of the forty-four patients who were followed for at least ten years had a reoperation, but neither operation involved the femoral component. One patient had a revision of the acetabular component 7.9 years after the index arthroplasty because of loosening. The second patient had an exchange of the polyethylene liner within the acetabular cup ten years postoperatively because of concern regarding wear; we believed that if the wear had continued, the resultant damage to the femoral head or the acetabular shell, or both, would have necessitated a more complex revision procedure. Both of these patients were followed clinically and radiographically for more than two years after the reoperation. At the time of the most recent follow-up, the patient who had had an exchange of the polyethylene liner had evidence of bone ingrowth, reported no pain or functional limitations related to the hip, and could walk an unlimited distance without a walking aid. The patient who had had a revision of the acetabular cup had signs that the new cup was loose at the most recent follow-up evaluation. Although this patient had rheumatoid arthritis with involvement of multiple joints and could only walk in the house with a walker, she did not complain of pain in the hip. Despite the need for another procedure, both of these patients had less pain in the hip than they had had before the index arthroplasty and were satisfied with the outcome of the second operation.
One other patient had a dislocation of the hip, which was treated with closed reduction and did not recur. There were no other known complications in any of the patients.
Clinical Findings
Pain: At the time of the latest follow-up, thirty patients (68 per cent) reported that they had no pain related to the hip, eight (18 per cent) had mild pain that did not limit activities, five (11 per cent) had moderate pain that limited some activities, and one (2 per cent) had severe pain that limited most activities (Table I). We could not explain the activity-limiting pain in two patients: one had isolated pain in the thigh despite a well fixed femoral component and one had pain in the region of the greater trochanter but no history or signs of bursitis. Over-all, forty-two (95 per cent) of the forty-four patients noted that the pain was less severe than it had been preoperatively (p < 0.01); no patient had pain that was worse than it had been before the arthroplasty.
Function: Twenty-two patients (50 per cent) could walk an unlimited distance, nine (20 per cent) could walk four to six blocks, six (14 per cent) could walk two or three blocks, four (9 per cent) could walk without support indoors only, and three (7 per cent) needed support for all walking. The limited mobility of one of the latter three patients was due to severe rheumatoid arthritis with involvement of multiple joints, as already described, and the limited walking ability of the other two patients was due to health problems that were unrelated to the hip. All three patients reported that they were satisfied with the outcome of the arthroplasty. Thirty-three patients (75 per cent) stated that they could ascend and descend stairs normally, and thirty-six (82 per cent) reported no difficulty putting on shoes and socks or stockings. Forty-two (95 per cent) of the forty-four patients still lived independently, either alone or with a spouse, at the time of the latest follow-up.
Over-all satisfaction: Forty (91 per cent) of the forty-four patients noted that their over-all function had improved after the arthroplasty, and thirty-nine patients (89 per cent) reported no or only slight limitations related to the hip. Forty-one patients (93 per cent) were satisfied with the results of the arthroplasty. One of the three patients who was not satisfied had severe rheumatoid arthritis and clinical depression. The second dissatisfied patient, who had moderate pain in the region of the greater trochanter, stated that her health problems were related mostly to the hip; however, she also noted that her over-all function was better than it had been before the operation. The third dissatisfied patient had moderate pain in the thigh but stated that the pain was less severe than it had been before the arthroplasty. All three of these patients had a stable femoral component with bone ingrowth.
Radiographic Findings
The ten patients (ten hips) for whom the latest radiographs were made at a mean of seven years (range, five to nine years) had evidence of bone growth into the femoral component and a stable acetabular component. Osteolysis was not observed in any of these patients.
Of the thirty-two acetabular components that had not been revised and for which a radiograph had been made at a minimum of ten years, thirty (94 per cent) were stable and two (6 per cent) had migrated and were considered to be unstable. One cup was revised because of loosening, as already described. Therefore, the over-all rate of loosening of the acetabular component at a minimum of ten years was 9 per cent (three of thirty-three hips).
There were radiographic signs of bone growth into thirty-two (97 per cent) of the thirty-three femoral components for which radiographs were available at a minimum of ten years. One femoral component appeared to be encapsulated in fibrous tissue. No femoral component had migrated, had tilted, or showed any sign of loosening. The rate of loosening of the femoral component was therefore 0 per cent.
Osteolytic lesions were observed in the pelvis and the femur in two patients (two hips) and in the femur only in two patients (two hips). All of the lesions were smaller than 1.5 square centimeters. The pelvic lesions were found in the ilium, ischium, and pubis in one patient and were isolated to the ilium in the other patient. The femoral lesions were confined to zone 1 of Gruen et al. in one patient, to zone 7 in one, and to zones 1 and 7 in two. No patient had evidence of osteolysis in the distal part of the femur.
An understanding of both the etiology and the consequences of periprosthetic bone-remodeling is important. Many authors have discussed the theory, physical and biological mechanisms, risk factors, identification, and measurement of the bone-remodeling related to stress-shielding, and some have used computer modeling to investigate the phenomenon5,6,11-13,15,16. However, there have been few studies on the clinical outcome for patients in whom this type of bone-remodeling developed following total hip arthroplasty1,6,10,16.
For the present study, we identified forty-eight hips (forty-seven patients) with evidence of stress-shielding that we believed could become a clinically important problem. Forty-five hips (forty-four patients) had a minimum of ten years of clinical follow-up. No femoral component was revised, one acetabular component (2 per cent) was revised because of loosening, and one polyethylene liner (2 per cent) was exchanged because of wear.
The rates of revision of the femoral and acetabular components in the subgroup of hips that had evidence of stress-shielding (0 per cent and 2 per cent [one of forty-five hips]) were lower than those previously reported7 for the larger (parent) series (1 per cent [three] and 5 per cent [ten] of 201 hips for which the outcome was known) after a mean of eleven years. The rates in the present study also were lower than the rates of 5 per cent (four hips) and 8 per cent (six hips) in the report by Xenos et al., who studied seventy-eight hips in a similar group of patients who were followed for 9.5 years after the insertion of a porous-coated anatomic prosthesis.
It has been suggested that bone-remodeling secondary to stress-shielding renders the femur more prone to debris-induced osteolysis10,12. We did not find this to be the case. The over-all prevalence of radiographically detectable acetabular and femoral osteolysis in the present study (12 per cent; four of the thirty-three hips for which ten-year radiographs were available) was lower than that in the larger (parent) series7 (39 per cent; fifty-four of 137 hips for which ten-year radiographs were available) and that in the series of Xenos et al. (45 per cent; thirty-five of seventy-eight hips). In addition, all of the osteolytic lesions were smaller than 1.5 square centimeters. The absence of femoral osteolysis distal to zones 1 and 7 of Gruen et al. suggests that, despite notable bone-remodeling, the porous coating-host bone interface continued to provide an effective barrier to the distal migration of wear debris.
The information obtained from the questionnaires reflected the patients' perceptions of the outcome of the arthroplasty. The findings of decreased pain and improved function as well as the high rate of over-all satisfaction (93 per cent; forty-one of forty-four patients) were similar to those reported in other studies with similar durations of follow-up7,17. We were particularly interested in the prevalence of pain in the thigh, which frequently has been mentioned as a complication of arthroplasty without cement. It has been postulated that, when fixation occurs predominantly in the distal portion of the femoral prosthesis and bone-remodeling occurs in the proximal part of the femur, there is a greater likelihood of so-called end-of-stem pain, or pain in the thigh. Activity-limiting pain in the thigh was associated with 7 per cent (three) of the forty-five hips in the present study, compared with 5 per cent (seven) of the 154 hips in the larger (parent) series7 that were followed for at least ten years but did not have a reoperation and 12 per cent (nine) of the seventy-eight hips in the study by Xenos et al.
The rate of adverse clinical events (revision, reoperation, and osteolysis) in the present series of patients who had bone-remodeling of the proximal part of the femur was lower than that in the studies of patients who had the same implant or a similar implant and who were followed for a similar period of time7,17. It has been argued that bone-remodeling occurs most frequently in older individuals who have osteoporosis16 and who place a lower demand on the implant; this may account for the decreased rates of complications in the present study. We compared the demographic data of the forty-seven patients (forty-eight hips) who had evidence of stress-shielding with those of the 152 patients (159 hips) from the larger (parent) series7 who had two-year radiographs but did not have evidence of stress-shielding. We found no apparent differences between the groups with regard to age, weight, or Charnley classification. However, there was a greater proportion of hips in female patients in the present study (69 per cent; thirty-three of forty-eight hips) than in the larger (parent) series (46 per cent; seventy-three of 159 hips), as well as a greater percentage of hips that received a femoral component with a diameter of more than fifteen millimeters (eighteen [38 per cent] of forty-eight hips, compared with twenty [19 per cent] of 159 hips).
We recognize that the assessment of plain radiographs may be a relatively insensitive method for the identification of bone loss due to stress-shielding. However, the use of plain serial radiographs is the most common and most practical method with which to evaluate the results of total hip arthroplasty. One of us (C. A. E., Sr.) developed a standardized method3 for making radiographs, and we insist that they be made with comparable contrast and orientation. In the present study, multiple exposures occasionally were necessary to achieve this goal. Pritchett, in an in vivo study in which dual-energy x-ray absorptiometry was used to evaluate fifty femora with five different components that had been in place for a minimum of three years, found that a 10 per cent decrease in bone mineral density was apparent on plain radiographs that were made more than one year postoperatively. Thus, it is likely that the semiquantitative method employed in the present study, combined with our experience in making consistently reproducible high-quality radiographs, enabled us to detect most, if not all, instances of clinically notable bone-remodeling.
The semiquantitative method that was used in the present study did not allow us to assess whether the degree of bone loss increased between the second and tenth postoperative years. We previously showed that radiographically apparent bone-remodeling occurs predominantly in the first year and rarely occurs after two years1,4,5. In a study in which dual-energy x-ray absorptiometry was used to analyze seventy-two femora, Kilgus et al. found that bone-remodeling sometimes continued for as long as seven years after the arthroplasty. However, the changes in bone mineral density after two years were less than the 10 per cent change that could be detected on plain radiographs. In a study of specimens obtained at autopsy, Sychterz and one of us (C. A. E., Sr.) found no correlation between the extent of bone-remodeling and the duration that the implant had been in situ.
In summary, we characterized the long-term clinical outcome of total hip arthroplasty without cement in patients who had evidence of stress-shielding (pronounced femoral bone-remodeling). This type of bone-remodeling was found in 23 per cent (forty-eight) of 207 hips that had been followed radiographically for at least two years. In the first ten years postoperatively, we did not encounter any adverse clinical consequences. We are encouraged by these results but recognize that there is still cause for concern as these patients enter a second decade with the hip prosthesis in place. Additionally, prospective studies involving quantitative methods are needed to define the prevalence of femoral bone-remodeling associated with implants inserted both with and without cement.