- Email: [email protected]
The effect of joint loading on acetabular wear measurement in total hip arthroplasty
The Journal of Arthroplasty Vol. 15 No. 4 2000
The Effect of Joint Loading on Acetabular Wear Measurement in Total Hip Arthroplasty John M. Martell, MD,* Seth S. Leopold, MD,† and Xiling Liu, MS‡
Abstract: All radiographic calculations of acetabular wear assume concentric reduction of the prosthetic articulation. To date, no studies have shown that the femoral head is fully reduced on standard radiographs, and we have seen cases on early postoperative radiographs in which this assumption is not met. Using our computerized radiographic technique, 78 paired anteroposterior pelvic radiographs in 46 patients at a mean of 14 months after surgery (range, 1–92 months) were evaluated with and without joint loading. Displacement with loading was analyzed against time since surgery, a surrogate for acetabular wear. Regression analysis found a statistically significant increase in femoral head displacement after loading with longer duration of follow-up, but the rate of this increase was small (0.027 mm/y). This difference affects calculated wear values by ⬍15%. We conclude that in a low-wear cohort, joint loading does not affect radiographic calculations of acetabular polyethylene wear in a clinically important way. Key words: joint loading, acetabular wear, polyethylene wear, total hip arthroplasty.
tem examination and analysis of revised cups do not provide sufficient numbers for an accurate determination of polyethylene wear rates in the general population. Several techniques have been developed to determine polyethylene wear using clinical radiographs. Manual methods [12–14] as well as methods using image analysis [15–18] have been applied to standard 2-dimensional radiographs. Another group has described a manual method of analysis using 3-dimensional reconstruction from anteroposterior (AP) and lateral films [19,20]. We have developed a computerized radiographic technique for the determination of acetabular polyethylene wear based on digital AP pelvic radiographs. Our technique involves computerized image analysis of digitized radiographs and has a mean accuracy of 0.01 mm when tested against phantom radiographs of polyethylene liners milled to exactly 2.0 mm of wear and a mean accuracy of 0.08 mm as validated by autopsy retrievals [17]. The technique has intraobserver and interobserver repeatability coefficients of 0.004 mm and 0.06 mm [17].
Polyethylene wear debris is an important cause of prosthetic component loosening in total hip arthroplasties (THAs) [1–3]. Wear particles incite a robust immune response that involves multinucleated giant cells, macrophages, histiocytes, and inflammatory mediators that are known to cause bone resorption [4–8]. The results of this biologic response include osteolysis, prosthesis failure, and structural defects in the femur and pelvis that are difficult to reconstruct [9–11]. In view of the grave consequences of polyethylene wear debris, it is critical that acetabular component wear be quantified accurately in situ. PostmorFrom the *Section of Orthopaedic Surgery and Rehabilitation Medicine, University of Chicago Hospitals, Chicago, IL; †Orthopaedic Surgery Service, William Beaumont Army Medical Center, El Paso, TX; and the ‡Department of Health Studies, University of Chicago, Chicago, Illinois. Submitted July 17, 1998; accepted October 26, 1999. No benefits or funds were received in support of this study. Reprint requests: John M. Martell, MD, 5841 South Maryland Avenue, MC 3079, Chicago, IL 60637. Copyright r 2000 by Churchill Livingstonet doi:10.1054/arth.2000.4336 0883-5403/00/1504-0017$10.00/0
512
Joint Loading and Wear Measurements in THA ●
Martell et al.
513
Fig. 1. (Left) Enlargement of an anteroposterior pelvic radiograph taken in the recovery room immediately after total hip arthroplasty. The inferolateral subluxation of the femoral head within the acetabular component is 1.5 mm. (Right) Enlargement of an arteroposterior pelvic radiograph of the same patient taken at 13 months’ follow-up. Concentric reduction of the femoral head within the acetabular component is apparent.
All previous research on in vivo radiographic techniques for measuring acetabular polyethylene wear after THA [13,15,18,19,21], including our own method [17], assumes that the femoral head is concentrically and fully reduced within the acetabular liner on plain AP views of the components. Preliminary work using our method of computerized image analysis called this assumption into question. We observed inferior and inferolateral femoral head subluxation of 1.5 mm on radiographs of patients taken in the recovery room and during the early postoperative period after THA (Fig. 1). Serial radiographs on those patients taken in the weeks after surgery showed the earlier femoral head subluxation to have resolved (Fig. 2). We attributed the early subluxation to inhibition of hip musculature by pain and residual anesthetic effects. This phenomenon has not been described in the context of acetabular wear analysis and was noticed only as a result of the increased accuracy and precision offered by the computerized image analysis method used. Telemetric studies have similarly shown that hip joint reaction forces in the early postoperative period can be variable and unpredictable [22,23]. To calculate acetabular polyethylene wear accurately, the femoral head must be completely seated within the polyethylene bearing on the radiographic views used for wear calculations. Any subluxation of the prosthetic femoral head within the acetabular component necessarily affects wear calculations, which are based on initial and final femoral head positions. The observation that significant subluxation of the prosthetic articulation can occur highlights the potential importance of joint loading on the position
of the femoral head during pelvic radiographs. To our knowledge, the need for hip loading during radiographic evaluation of acetabular wear has not previously been evaluated. The objective of this study is to determine whether load bearing significantly alters the position of the prosthetic femoral head within the acetabular component on AP pelvic radiographs. We also investigate whether the effect of load bearing varies with time since surgery. We do not attempt to quantify absolute wear because this study tests one of the important assumptions underlying the calculation of acetabular wear.
Fig. 2. Graph of femoral head position relative to the acetabular center over time for the patient whose radiograph is shown in Fig. 1. The early postoperative inferolateral subluxation resolves, and the femoral head is well reduced at 13 months after surgery.
514 The Journal of Arthroplasty Vol. 15 No. 4 June 2000
Materials and Methods Patients and Prosthetic Components A total of 46 patients with 51 primary THAs formed the cohort for this study. The patients were examined at an average of 14 months since the THA (range, 1–92 months) and represented a consecutive series of follow-up visits to 1 surgeon. Only patients more than 4 weeks’ postsurgery were included. No patient had symptoms referable to the prosthetic hip at the time of the visit. None had clinical or radiographic signs of prosthetic loosening or infection. The prostheses included a radiopaque femoral head and a cementless metal-backed acetabular cup. The femoral head size was 28 mm in all hips. Radiographic Technique An adjustable marker was placed 6 inches above the anterior tibia at a level 2 cm proximal to the transmalleolar axis to ensure a reproducible leg raise for each examination. Each patient first underwent a standard AP pelvis radiograph. Without changing patient position, the patient was then asked to perform a 6-inch straight-leg raise on the operative side while the AP pelvis film was repeated. Telemetric data indicate that a straight-leg raise generates a joint-reaction force equal to 1.8 ⫻ body weight [24]. This is a greater force than is transmitted across the hip in double-limb stance and is nearly as great as the load seen by the joint in the ipsilateral single-leg stance phase of gait. All footwear was removed before the radiographs. Radiographs were digitized with a Konica LD 4500 (Tokyo, Japan) laser scanner using a pixel size of 0.171 mm ⫻ 0.171 mm and a matrix size of 2,000 ⫻ 2,430. The gray scale used by the scanner was 10 bits. Once digitized, the 10-bit image was converted to an 8-bit image using histogram equalization. This transformation procedure minimizes the loss of contrast information in the image during gray level reduction. The image was then converted into a tag image file format (TIFF) for later analysis. Analysis of Paired Radiographs The method used for computerized image analysis based on vector displacements of the femoral heads as well as validation of the technique is outlined elsewhere [17]. Following is a brief description of the technique as it applies to femoral head position with joint loading. The underlying approach of this method is to find reproducibly the acetabular and femoral head centers, using a com-
puterized method of image analysis and digital edge detection. Changes in the location of the femoral head center with respect to the acetabular center are tracked with and without joint loading. The radiographs are analyzed to determine the direction and amount of displacement the femoral head undergoes during joint loading. Our vector analysis technique defined the center of the acetabular metal shell as a (0,0) reference point for a 2-dimensional coordinate grid. The center of the femoral head is identified, and its position on the coordinate grid is calculated for the paired radiographs. Magnification correction is based on the known femoral head diameter. Femoral head position on the standard AP pelvis radiograph is compared with the matched straight-leg raise radiograph by plotting the femoral head position relative to the known acetabular center first without, then with, joint loading. A vector subtraction of each pair of points is then performed, and the actual displacements are graphically depicted. As a vector quantity, this result contains a magnitude and a direction of displacement of the femoral head with respect to the unloaded prosthesis. The absolute magnitude of femoral head displacement with joint loading, calculated as described previously for each radiograph pair, is also plotted against time since surgery. Time since surgery was taken to be an indirect measure of acetabular wear; we chose not to use a calculated value for acetabular wear because any calculated value depends on the assumptions being tested in this study. Assessment of Joint-Loading Protocol Repeatability Four patients in this series had repeated studies within 1 to 5 months of each other (mean 1.9 months) for clinical reasons. These film pairs were analyzed to assess the repeatability of the jointloading technique. Statistical Methods The effect of joint loading on femoral head position was evaluated in several ways. The null hypothesis that joint loading has no effect on femoral head position was assessed for all radiograph pairs in the study using a chi-square test. The possible resultant displacement directions were divided graphically into quadrants with respect to the initial head position: superomedial, inferomedial, inferolateral, and superolateral. Observed displacements were compared with the expected result of no effect with joint loading, that is, each quadrant containing 25% of the points. The magnitude of displacement with
Joint Loading and Wear Measurements in THA ●
joint loading is summarized by calculating the mean, standard deviation, and 95% confidence interval for the mean. The 46 patients in this study represent 51 hips, 20 of which have repeated measurements (2–4 radiographs). To estimate the effect of joint loading over time since surgery, regression analysis was performed using the general estimating equation model with 95% confidence intervals for the slope provided. The effects of patient age, height, weight, inclination angle, and radiographic anteversion on the magnitude of femoral head displacement with joint loading were also estimated by multiple regression analysis. The effects of surgical approach (anterior ⫽ 13%, posterolateral ⫽ 84%, transtrochanteric ⫽ 3%), surgical procedure (primary ⫽ 73%, revision ⫽ 27%), femoral neck length (short ⫽ 38%, medium ⫽ 21%, long ⫽ 23%, extra-long ⫽ 18%), and acetabular liner type (0° ⫽ 87%, 5° ⫽ 5%, 10° ⫽ 8%) on femoral head displacement were determined using analysis of variance employing a general linear model.
Results Femoral Head Displacement Femoral head displacement vectors observed with joint loading are shown in Fig. 3. This graph depicts the vector magnitude and direction of displacement in the 4 possible quadrants: superomedial, superolateral, inferolateral, and inferomedial. The mean femoral head displacement with active straight-leg raise was 0.226 mm (range, 0.001–
Martell et al.
515
Fig. 4. Regression analysis of the effect of time since surgery on displacement of the femoral head with joint loading. 95% confidence intervals (CI) for the regression line are shown. (Slope ⫽ 0.027 mm/y increase in displacement; P ⬍ .002.) Solid line, regression; broken line, 95% CI.
0.802 mm; SD, 0.153 mm) for the 78 paired radiographic evaluations (Fig. 3). As seen in Fig. 3, the direction of displacement appeared random, with vector directions evenly divided among all 4 quadrants with joint loading. Using the chi-square test, this distribution of vector directions was not significant relative to the null hypothesis of no directional preference as a result of joint loading (P ⫽ .98). Examining the subcohorts of patients who were ⬎1 (n ⫽ 31) and ⬎2 years (n ⫽ 20) since surgery also failed to yield a directional preference of femoral head position change with joint loading (P ⫽ .75 and P ⫽ .28). Repeatability of the Joint-Loading Protocol In the 4 patients with repeated studies within 5 months of each other, the average standard deviation for the head displacement and displacement direction between studies was 0.088 mm and 18.5°. This excellent agreement suggests that variations observed in displacement with joint loading are not secondary to errors in the radiographic analysis. Femoral Head Displacement Versus Time Since Surgery
Fig. 3. Graph depicting the displacement of the femoral head within the acetabulum after the application of joint loading. The direction of femoral head displacement with joint loading does not follow a consistent direction but appears randomly divided among the 4 quadrants (chisquare, P ⫽ .98).
Multiple regression analysis by the general estimating equation model showed a small but statistically significant effect of time on femoral head displacement with joint loading (Fig. 4). The rate of change of displacement with time was found to be 0.027 mm/y. Because of the high precision of the
516 The Journal of Arthroplasty Vol. 15 No. 4 June 2000 image analysis technique used and the relatively large sample size, this clinically small rate of change was found to be statistically significant (P ⫽ .002). The 95% confidence interval for this slope ranges from 0.008 mm/y to 0.040 mm/y. Although this rate of change in the joint-loading effect over time is statistically significant, even the largest effect possible within the 95% confidence interval is not clinically important (0.040 mm/y) with traditional polyethylene articulations. Regression analysis also suggested a small but significant effect of acetabular inclination angle on the magnitude of femoral head displacement (P ⫽ .004). As the acetabular component became more vertical (inclination angle increased), the effect of joint loading on femoral head displacement decreased slightly (slope ⫽ ⫺0.054 mm/°). There is no evidence to show that the additional variables of height, weight, age, and acetabular anteversion had a significant impact on the slope of the regression line. Analysis of variance showed no significant effect of the surgical approach, surgical procedure, femoral neck type, or acetabular liner type on femoral head displacement with joint loading.
Discussion The present study addresses the question of whether AP pelvic radiographs taken without joint loading are equivalent to weight-bearing views, for purposes of calculating acetabular polyethylene wear. There are several potential limitations on interpreting the data in this study. First, the length of follow-up allowed regression analysis out to 92 months (7.6 years), but the mean time since surgery was only 14 months. This cohort is presumed to be a low-wear group. It is conceivable that as follow-up duration and acetabular polyethylene wear increase, the effect of joint loading on femoral head position might also increase. Second, the technique of image analysis used in this study is based on AP pelvic radiographs only, making it a 2-dimensional technique. As such, any motion of the prosthetic femoral head with joint loading that occurs in the AP direction is not detected. Although this is a potential limitation, most of the hip’s joint reaction force as well as the expected displacement is directed superomedially or superolaterally with joint loading [14,25]. Three-dimensional analysis techniques may improve the overall sensitivity of acetabular polyethylene wear measurement, but to this point, all published techniques, including those using 3-dimensional analyses [19], rely on supine, non–weight-bearing radiographs.
A final potential limitation of the study design was the method of achieving joint loading for the radiographs. Consistent patient positioning and use of an adjustable marker to ensure a constant leg raise regardless of limb girth helped ensure reliability of joint loading. Our analysis of 4 patients in whom repeated examinations had been obtained within 5 months of each other showed excellent repeatability for the magnitude of head displacement as well as displacement direction. We cannot know precisely, however, the forces transmitted across each patient’s hip with the straight-leg raise protocol. The absolute load varied from patient to patient depending on the length and weight of the limb involved. Several conclusions can be drawn from these data. In the cohort of patients studied, applying a joint-reaction force caused the femoral head to reposition within the acetabulum by a small amount (0.226 mm). This position shift did not occur consistently in line with the joint reaction force, and the femoral head did not move reproducibly in any uniform direction (Fig. 3). These results suggest that although there is some play in the prosthetic articulation, joint loading does not cause a more concentric reduction. Some investigators have found that low machining tolerances of acetabular polyethylene bearings may allow for a gap between the bearing and the metal shell [1,26]. These gaps may contribute to femoral head displacement with joint loading. Errors in the measurement technique may also contribute to the apparent motion of the femoral head; however, with an intraobserver repeatability of 0.004 mm [17], these should not have a significant effect. Differences in prosthetic position and design may explain the variability observed in the direction of femoral head shift. Certain combinations of prosthetic position and component design may cause impingement with the straight-leg raise maneuver, forcing femoral head displacement in a direction other than the joint reaction force. Our observations that hip prostheses may be subluxed inferiorly or inferolaterally on radiographs taken in the 1st postoperative month (Figs. 1 and 2) suggest that these films should not be used as baseline examinations for the study of acetabular wear. Based on these observations and on the telemetry data of other investigators [22–24] indicating normalization of hip joint forces by 4 weeks postoperatively, we recommend that radiographs for the purpose of calculating acetabular wear at later follow-up be taken at 6 weeks after surgery. We question the accuracy of studies that use radiographs taken in the 1st postoperative month as a
Joint Loading and Wear Measurements in THA ●
baseline for calculation of polyethylene wear. Regression analyses found that the amount of motion with joint loading increased with time since surgery. This result was statistically significant (Fig. 4), but the absolute value of the effect was small. The change in femoral head position with time would introduce an error of only 7% to 10% in the radiographic measurement of linear wear rates reported by others [19,21,27,28]. The fact that our subcohorts with longer duration of follow-up did not have femoral head migration in a uniform direction further suggests that radiographs taken with joint loading do not produce a more concentrically reduced prosthesis. Statistically, the percentage of variability in displacement with loading explained by the time since surgery is low (R 2 ⫽ 7%). This low percentage indicates that other factors in addition to time since surgery may have an effect on femoral head displacement with joint loading. In an attempt to quantify these, we considered patient height, weight, age, acetabular anteversion, and inclination angle. Of these, only acetabular inclination angle had an additional effect on femoral head displacement with joint loading. Combining the time since surgery with the acetabular inclination angle improved the regression equation and accounted for R 2 ⫽ 11.4% of variability in displacement with loading. The decision to perform routine joint-loading radiographs results in increased time, cost, and patient exposure to ionizing radiation. We believe that clinically significant—and not merely statistically significant—differences between joint-loaded and nonloaded films need to be shown before assuming the costs, time, and risk of additional radiographs. The results of this study do not justify additional joint-loading radiographs because our regression analysis predicts a difference in the loaded versus nonloaded femoral head position of only 0.27 mm at 10 years’ follow-up. With longer followup, the femoral head does not move in a consistent direction with joint loading. Our finding of a statistically significant difference in femoral head position with joint loading in a low-wear group raises the question of joint-loading effects in a high-wear group. We continue to follow the patients in this series and are evaluating a group with longer follow-up to determine if the amount of displacement with joint loading increases more dramatically with time and whether the direction of displacement develops a more consistent pattern as more acetabular polyethylene wear occurs.
Martell et al.
517
Acknowledgment The authors thank Theodore Karrison, PhD, for his help with statistical analysis and interpretation.
References 1. Rimnac CM, Wilson PD, Fuchs MD, et al: Acetabular cup wear in total hip arthroplasty. Orthop Clin North Am 19:631, 1988 2. Ritter MA, Keating EM, Faris PM, et al: Metal-backed acetabular cups in total hip arthroplasty. J Bone Joint Surg Am 72:672, 1990 3. Tanser M, Maloney WJ, Jasty M, et al: The progression of femoral cortical osteolysis in association with total hip arthroplasty without cement. J Bone Joint Surg Am 74:404, 1992 4. Appel AM, Sowder WG, Silverhus SW, et al: Prosthesisassociated pseudomembrane-induced bone resorption. Br J Rheumatol 29:32, 1990 5. Howie DW, Cornish BL, Vernon-Roberts B, et al: A rat model of resorption of bone at the cement-bone interface and the presence of polylethylene wear particles. J Bone Joint Surg Am 70:257, 1988 6. Howie DW, Cornish BL, Vernon-Roberts B: Resurfacing hip arthroplasty classification of loosening and the role of prosthesis wear in particles. Clin Orthop 255:144, 1990 7. Kim KJ, Chiba J, Rubash HE: In vivo and in vitro analysis of membranes from hip prostheses inserted without cement. J Bone Joint Surg Am 76:172, 1994 8. Murray DW, Rushton N: Macrophages stimulate bone resorption when they phagocytose particles. J Bone Joint Surg Br 72:988, 1990 9. Amstutz HC, Campbell P, Kossovsky N, et al: Mechanism and clinical significance of wear debris-induced osteolysis. Clin Orthop 276:7, 1992 10. Johanson NA, Callaghan JJ, Salvati EA: Fourteen year follow up study of a patient with massive calcar absorption—a case report. Clin Orthop 213:189, 1987 11. Mirra JM, Marder RA, Amstutz HC: The pathology of failed total joint arthroplasty. Clin Orthop 117:175, 1982 12. Bankston AB, Faris PM, Keating EM, et al: Polyethylene wear in total hip arthroplasty in patient-matched groups: a comparison of stainless steel, cobalt chrome, and titanium-bearing surfaces. J Arthroplasty 8:315, 1993 13. Livermore J, Ilstrup D, Morrey B: The effect of femoral head size on wear of the polyethylene acetabular component. J Bone Joint Surg Am 72:518, 1990 14. Wroblewski BM: Direction and rate of socket wear in Charnley low-friction arthroplasty. J Bone Joint Surg Br 67:757, 1985
518 The Journal of Arthroplasty Vol. 15 No. 4 June 2000 15. Hardinge K, Porter ML, Jones PR, et al: Measurement of hip prostheses using image analysis: the maxima hip technique. J Bone Joint Surg Br 73:724, 1991 16. Jones PR, Taylor CJ, Hukins DW, et al: Prosthetic hip failure—retrospective radiograph image analysis of the acetabular cup. J Biomed Eng 11:253, 1989 17. Martell JM, Berdia S: Determination of polyethylene wear in total hip replacements with use of digital radiographs. J Bone Joint Surg Am 79:1635, 1997 18. Shaver SM, Brown TD, Hillis SL, et al: Digital edge detection measurement of polyethylene wear after total hip arthroplasty. J Bone Joint Surg Am 79:690, 1997 19. Devane PA, Bourne RB, Rorabeck CH, et al: Measurement of polyethylene wear in metal-backed acetabular cups—I. three dimensional technique. Clin Orthop 319:303, 1995 20. Devane PA, Robinson EJ, Bourne RB, et al: Measurement of polyethylene wear in acetabular components inserted with and without cement: a randomized trial. J Bone Joint Surg Am 79:682, 1997 21. Kabo JM, Gebhard JS, Loren G, et al: In vivo wear of polyethylene acetabular components. J Bone Joint Surg Br 75:254, 1993
22. Rydell NW: Forces acting on the femoral-head prosthesis: a study on strain gauge supplied prostheses in living persons. Acta Orthop Scand 37:(Suppl)88:1, 1966 23. English TA, Kilvington M: In vivo records of hip loads using a femoral implant with telemetric output (a preliminary report). J Biomed Eng 1:111, 1979 24. Davy DT, Kotzar GM, Brown RH, et al: Telemetric force measurements across the hip after total arthroplasty. J Bone Joint Surg Am 70:45, 1988 25. Wroblewski BM: Wear and loosening of the socket in the Charnley low-friction arthroplasty. Orthop Clin North Am 211:30, 1986 26. Bartel DL, Wright TM, Edwards DL: The effect of metal backing on stresses in polyethylene acetabular components. In: The Hip: proceedings of the Hip Society. CV Mosby, St Louis, 1983 27. Hernandez JR, Keating EM, Faris PM, et al: Polyethylene wear in uncemented acetabular components. J Bone Joint Surg Br 76:263, 1994 28. Zichner LP, Willert HG: Comparison of aluminapolyethylene and metal-polyethylene in clinical trials. Clin Orthop 282:86, 1992
The Effect of Joint Loading on Acetabular Wear Measurement in Total Hip Arthroplasty John M. Martell, MD,* Seth S. Leopold, MD,† and Xiling Liu, MS‡
Abstract: All radiographic calculations of acetabular wear assume concentric reduction of the prosthetic articulation. To date, no studies have shown that the femoral head is fully reduced on standard radiographs, and we have seen cases on early postoperative radiographs in which this assumption is not met. Using our computerized radiographic technique, 78 paired anteroposterior pelvic radiographs in 46 patients at a mean of 14 months after surgery (range, 1–92 months) were evaluated with and without joint loading. Displacement with loading was analyzed against time since surgery, a surrogate for acetabular wear. Regression analysis found a statistically significant increase in femoral head displacement after loading with longer duration of follow-up, but the rate of this increase was small (0.027 mm/y). This difference affects calculated wear values by ⬍15%. We conclude that in a low-wear cohort, joint loading does not affect radiographic calculations of acetabular polyethylene wear in a clinically important way. Key words: joint loading, acetabular wear, polyethylene wear, total hip arthroplasty.
tem examination and analysis of revised cups do not provide sufficient numbers for an accurate determination of polyethylene wear rates in the general population. Several techniques have been developed to determine polyethylene wear using clinical radiographs. Manual methods [12–14] as well as methods using image analysis [15–18] have been applied to standard 2-dimensional radiographs. Another group has described a manual method of analysis using 3-dimensional reconstruction from anteroposterior (AP) and lateral films [19,20]. We have developed a computerized radiographic technique for the determination of acetabular polyethylene wear based on digital AP pelvic radiographs. Our technique involves computerized image analysis of digitized radiographs and has a mean accuracy of 0.01 mm when tested against phantom radiographs of polyethylene liners milled to exactly 2.0 mm of wear and a mean accuracy of 0.08 mm as validated by autopsy retrievals [17]. The technique has intraobserver and interobserver repeatability coefficients of 0.004 mm and 0.06 mm [17].
Polyethylene wear debris is an important cause of prosthetic component loosening in total hip arthroplasties (THAs) [1–3]. Wear particles incite a robust immune response that involves multinucleated giant cells, macrophages, histiocytes, and inflammatory mediators that are known to cause bone resorption [4–8]. The results of this biologic response include osteolysis, prosthesis failure, and structural defects in the femur and pelvis that are difficult to reconstruct [9–11]. In view of the grave consequences of polyethylene wear debris, it is critical that acetabular component wear be quantified accurately in situ. PostmorFrom the *Section of Orthopaedic Surgery and Rehabilitation Medicine, University of Chicago Hospitals, Chicago, IL; †Orthopaedic Surgery Service, William Beaumont Army Medical Center, El Paso, TX; and the ‡Department of Health Studies, University of Chicago, Chicago, Illinois. Submitted July 17, 1998; accepted October 26, 1999. No benefits or funds were received in support of this study. Reprint requests: John M. Martell, MD, 5841 South Maryland Avenue, MC 3079, Chicago, IL 60637. Copyright r 2000 by Churchill Livingstonet doi:10.1054/arth.2000.4336 0883-5403/00/1504-0017$10.00/0
512
Joint Loading and Wear Measurements in THA ●
Martell et al.
513
Fig. 1. (Left) Enlargement of an anteroposterior pelvic radiograph taken in the recovery room immediately after total hip arthroplasty. The inferolateral subluxation of the femoral head within the acetabular component is 1.5 mm. (Right) Enlargement of an arteroposterior pelvic radiograph of the same patient taken at 13 months’ follow-up. Concentric reduction of the femoral head within the acetabular component is apparent.
All previous research on in vivo radiographic techniques for measuring acetabular polyethylene wear after THA [13,15,18,19,21], including our own method [17], assumes that the femoral head is concentrically and fully reduced within the acetabular liner on plain AP views of the components. Preliminary work using our method of computerized image analysis called this assumption into question. We observed inferior and inferolateral femoral head subluxation of 1.5 mm on radiographs of patients taken in the recovery room and during the early postoperative period after THA (Fig. 1). Serial radiographs on those patients taken in the weeks after surgery showed the earlier femoral head subluxation to have resolved (Fig. 2). We attributed the early subluxation to inhibition of hip musculature by pain and residual anesthetic effects. This phenomenon has not been described in the context of acetabular wear analysis and was noticed only as a result of the increased accuracy and precision offered by the computerized image analysis method used. Telemetric studies have similarly shown that hip joint reaction forces in the early postoperative period can be variable and unpredictable [22,23]. To calculate acetabular polyethylene wear accurately, the femoral head must be completely seated within the polyethylene bearing on the radiographic views used for wear calculations. Any subluxation of the prosthetic femoral head within the acetabular component necessarily affects wear calculations, which are based on initial and final femoral head positions. The observation that significant subluxation of the prosthetic articulation can occur highlights the potential importance of joint loading on the position
of the femoral head during pelvic radiographs. To our knowledge, the need for hip loading during radiographic evaluation of acetabular wear has not previously been evaluated. The objective of this study is to determine whether load bearing significantly alters the position of the prosthetic femoral head within the acetabular component on AP pelvic radiographs. We also investigate whether the effect of load bearing varies with time since surgery. We do not attempt to quantify absolute wear because this study tests one of the important assumptions underlying the calculation of acetabular wear.
Fig. 2. Graph of femoral head position relative to the acetabular center over time for the patient whose radiograph is shown in Fig. 1. The early postoperative inferolateral subluxation resolves, and the femoral head is well reduced at 13 months after surgery.
514 The Journal of Arthroplasty Vol. 15 No. 4 June 2000
Materials and Methods Patients and Prosthetic Components A total of 46 patients with 51 primary THAs formed the cohort for this study. The patients were examined at an average of 14 months since the THA (range, 1–92 months) and represented a consecutive series of follow-up visits to 1 surgeon. Only patients more than 4 weeks’ postsurgery were included. No patient had symptoms referable to the prosthetic hip at the time of the visit. None had clinical or radiographic signs of prosthetic loosening or infection. The prostheses included a radiopaque femoral head and a cementless metal-backed acetabular cup. The femoral head size was 28 mm in all hips. Radiographic Technique An adjustable marker was placed 6 inches above the anterior tibia at a level 2 cm proximal to the transmalleolar axis to ensure a reproducible leg raise for each examination. Each patient first underwent a standard AP pelvis radiograph. Without changing patient position, the patient was then asked to perform a 6-inch straight-leg raise on the operative side while the AP pelvis film was repeated. Telemetric data indicate that a straight-leg raise generates a joint-reaction force equal to 1.8 ⫻ body weight [24]. This is a greater force than is transmitted across the hip in double-limb stance and is nearly as great as the load seen by the joint in the ipsilateral single-leg stance phase of gait. All footwear was removed before the radiographs. Radiographs were digitized with a Konica LD 4500 (Tokyo, Japan) laser scanner using a pixel size of 0.171 mm ⫻ 0.171 mm and a matrix size of 2,000 ⫻ 2,430. The gray scale used by the scanner was 10 bits. Once digitized, the 10-bit image was converted to an 8-bit image using histogram equalization. This transformation procedure minimizes the loss of contrast information in the image during gray level reduction. The image was then converted into a tag image file format (TIFF) for later analysis. Analysis of Paired Radiographs The method used for computerized image analysis based on vector displacements of the femoral heads as well as validation of the technique is outlined elsewhere [17]. Following is a brief description of the technique as it applies to femoral head position with joint loading. The underlying approach of this method is to find reproducibly the acetabular and femoral head centers, using a com-
puterized method of image analysis and digital edge detection. Changes in the location of the femoral head center with respect to the acetabular center are tracked with and without joint loading. The radiographs are analyzed to determine the direction and amount of displacement the femoral head undergoes during joint loading. Our vector analysis technique defined the center of the acetabular metal shell as a (0,0) reference point for a 2-dimensional coordinate grid. The center of the femoral head is identified, and its position on the coordinate grid is calculated for the paired radiographs. Magnification correction is based on the known femoral head diameter. Femoral head position on the standard AP pelvis radiograph is compared with the matched straight-leg raise radiograph by plotting the femoral head position relative to the known acetabular center first without, then with, joint loading. A vector subtraction of each pair of points is then performed, and the actual displacements are graphically depicted. As a vector quantity, this result contains a magnitude and a direction of displacement of the femoral head with respect to the unloaded prosthesis. The absolute magnitude of femoral head displacement with joint loading, calculated as described previously for each radiograph pair, is also plotted against time since surgery. Time since surgery was taken to be an indirect measure of acetabular wear; we chose not to use a calculated value for acetabular wear because any calculated value depends on the assumptions being tested in this study. Assessment of Joint-Loading Protocol Repeatability Four patients in this series had repeated studies within 1 to 5 months of each other (mean 1.9 months) for clinical reasons. These film pairs were analyzed to assess the repeatability of the jointloading technique. Statistical Methods The effect of joint loading on femoral head position was evaluated in several ways. The null hypothesis that joint loading has no effect on femoral head position was assessed for all radiograph pairs in the study using a chi-square test. The possible resultant displacement directions were divided graphically into quadrants with respect to the initial head position: superomedial, inferomedial, inferolateral, and superolateral. Observed displacements were compared with the expected result of no effect with joint loading, that is, each quadrant containing 25% of the points. The magnitude of displacement with
Joint Loading and Wear Measurements in THA ●
joint loading is summarized by calculating the mean, standard deviation, and 95% confidence interval for the mean. The 46 patients in this study represent 51 hips, 20 of which have repeated measurements (2–4 radiographs). To estimate the effect of joint loading over time since surgery, regression analysis was performed using the general estimating equation model with 95% confidence intervals for the slope provided. The effects of patient age, height, weight, inclination angle, and radiographic anteversion on the magnitude of femoral head displacement with joint loading were also estimated by multiple regression analysis. The effects of surgical approach (anterior ⫽ 13%, posterolateral ⫽ 84%, transtrochanteric ⫽ 3%), surgical procedure (primary ⫽ 73%, revision ⫽ 27%), femoral neck length (short ⫽ 38%, medium ⫽ 21%, long ⫽ 23%, extra-long ⫽ 18%), and acetabular liner type (0° ⫽ 87%, 5° ⫽ 5%, 10° ⫽ 8%) on femoral head displacement were determined using analysis of variance employing a general linear model.
Results Femoral Head Displacement Femoral head displacement vectors observed with joint loading are shown in Fig. 3. This graph depicts the vector magnitude and direction of displacement in the 4 possible quadrants: superomedial, superolateral, inferolateral, and inferomedial. The mean femoral head displacement with active straight-leg raise was 0.226 mm (range, 0.001–
Martell et al.
515
Fig. 4. Regression analysis of the effect of time since surgery on displacement of the femoral head with joint loading. 95% confidence intervals (CI) for the regression line are shown. (Slope ⫽ 0.027 mm/y increase in displacement; P ⬍ .002.) Solid line, regression; broken line, 95% CI.
0.802 mm; SD, 0.153 mm) for the 78 paired radiographic evaluations (Fig. 3). As seen in Fig. 3, the direction of displacement appeared random, with vector directions evenly divided among all 4 quadrants with joint loading. Using the chi-square test, this distribution of vector directions was not significant relative to the null hypothesis of no directional preference as a result of joint loading (P ⫽ .98). Examining the subcohorts of patients who were ⬎1 (n ⫽ 31) and ⬎2 years (n ⫽ 20) since surgery also failed to yield a directional preference of femoral head position change with joint loading (P ⫽ .75 and P ⫽ .28). Repeatability of the Joint-Loading Protocol In the 4 patients with repeated studies within 5 months of each other, the average standard deviation for the head displacement and displacement direction between studies was 0.088 mm and 18.5°. This excellent agreement suggests that variations observed in displacement with joint loading are not secondary to errors in the radiographic analysis. Femoral Head Displacement Versus Time Since Surgery
Fig. 3. Graph depicting the displacement of the femoral head within the acetabulum after the application of joint loading. The direction of femoral head displacement with joint loading does not follow a consistent direction but appears randomly divided among the 4 quadrants (chisquare, P ⫽ .98).
Multiple regression analysis by the general estimating equation model showed a small but statistically significant effect of time on femoral head displacement with joint loading (Fig. 4). The rate of change of displacement with time was found to be 0.027 mm/y. Because of the high precision of the
516 The Journal of Arthroplasty Vol. 15 No. 4 June 2000 image analysis technique used and the relatively large sample size, this clinically small rate of change was found to be statistically significant (P ⫽ .002). The 95% confidence interval for this slope ranges from 0.008 mm/y to 0.040 mm/y. Although this rate of change in the joint-loading effect over time is statistically significant, even the largest effect possible within the 95% confidence interval is not clinically important (0.040 mm/y) with traditional polyethylene articulations. Regression analysis also suggested a small but significant effect of acetabular inclination angle on the magnitude of femoral head displacement (P ⫽ .004). As the acetabular component became more vertical (inclination angle increased), the effect of joint loading on femoral head displacement decreased slightly (slope ⫽ ⫺0.054 mm/°). There is no evidence to show that the additional variables of height, weight, age, and acetabular anteversion had a significant impact on the slope of the regression line. Analysis of variance showed no significant effect of the surgical approach, surgical procedure, femoral neck type, or acetabular liner type on femoral head displacement with joint loading.
Discussion The present study addresses the question of whether AP pelvic radiographs taken without joint loading are equivalent to weight-bearing views, for purposes of calculating acetabular polyethylene wear. There are several potential limitations on interpreting the data in this study. First, the length of follow-up allowed regression analysis out to 92 months (7.6 years), but the mean time since surgery was only 14 months. This cohort is presumed to be a low-wear group. It is conceivable that as follow-up duration and acetabular polyethylene wear increase, the effect of joint loading on femoral head position might also increase. Second, the technique of image analysis used in this study is based on AP pelvic radiographs only, making it a 2-dimensional technique. As such, any motion of the prosthetic femoral head with joint loading that occurs in the AP direction is not detected. Although this is a potential limitation, most of the hip’s joint reaction force as well as the expected displacement is directed superomedially or superolaterally with joint loading [14,25]. Three-dimensional analysis techniques may improve the overall sensitivity of acetabular polyethylene wear measurement, but to this point, all published techniques, including those using 3-dimensional analyses [19], rely on supine, non–weight-bearing radiographs.
A final potential limitation of the study design was the method of achieving joint loading for the radiographs. Consistent patient positioning and use of an adjustable marker to ensure a constant leg raise regardless of limb girth helped ensure reliability of joint loading. Our analysis of 4 patients in whom repeated examinations had been obtained within 5 months of each other showed excellent repeatability for the magnitude of head displacement as well as displacement direction. We cannot know precisely, however, the forces transmitted across each patient’s hip with the straight-leg raise protocol. The absolute load varied from patient to patient depending on the length and weight of the limb involved. Several conclusions can be drawn from these data. In the cohort of patients studied, applying a joint-reaction force caused the femoral head to reposition within the acetabulum by a small amount (0.226 mm). This position shift did not occur consistently in line with the joint reaction force, and the femoral head did not move reproducibly in any uniform direction (Fig. 3). These results suggest that although there is some play in the prosthetic articulation, joint loading does not cause a more concentric reduction. Some investigators have found that low machining tolerances of acetabular polyethylene bearings may allow for a gap between the bearing and the metal shell [1,26]. These gaps may contribute to femoral head displacement with joint loading. Errors in the measurement technique may also contribute to the apparent motion of the femoral head; however, with an intraobserver repeatability of 0.004 mm [17], these should not have a significant effect. Differences in prosthetic position and design may explain the variability observed in the direction of femoral head shift. Certain combinations of prosthetic position and component design may cause impingement with the straight-leg raise maneuver, forcing femoral head displacement in a direction other than the joint reaction force. Our observations that hip prostheses may be subluxed inferiorly or inferolaterally on radiographs taken in the 1st postoperative month (Figs. 1 and 2) suggest that these films should not be used as baseline examinations for the study of acetabular wear. Based on these observations and on the telemetry data of other investigators [22–24] indicating normalization of hip joint forces by 4 weeks postoperatively, we recommend that radiographs for the purpose of calculating acetabular wear at later follow-up be taken at 6 weeks after surgery. We question the accuracy of studies that use radiographs taken in the 1st postoperative month as a
Joint Loading and Wear Measurements in THA ●
baseline for calculation of polyethylene wear. Regression analyses found that the amount of motion with joint loading increased with time since surgery. This result was statistically significant (Fig. 4), but the absolute value of the effect was small. The change in femoral head position with time would introduce an error of only 7% to 10% in the radiographic measurement of linear wear rates reported by others [19,21,27,28]. The fact that our subcohorts with longer duration of follow-up did not have femoral head migration in a uniform direction further suggests that radiographs taken with joint loading do not produce a more concentrically reduced prosthesis. Statistically, the percentage of variability in displacement with loading explained by the time since surgery is low (R 2 ⫽ 7%). This low percentage indicates that other factors in addition to time since surgery may have an effect on femoral head displacement with joint loading. In an attempt to quantify these, we considered patient height, weight, age, acetabular anteversion, and inclination angle. Of these, only acetabular inclination angle had an additional effect on femoral head displacement with joint loading. Combining the time since surgery with the acetabular inclination angle improved the regression equation and accounted for R 2 ⫽ 11.4% of variability in displacement with loading. The decision to perform routine joint-loading radiographs results in increased time, cost, and patient exposure to ionizing radiation. We believe that clinically significant—and not merely statistically significant—differences between joint-loaded and nonloaded films need to be shown before assuming the costs, time, and risk of additional radiographs. The results of this study do not justify additional joint-loading radiographs because our regression analysis predicts a difference in the loaded versus nonloaded femoral head position of only 0.27 mm at 10 years’ follow-up. With longer followup, the femoral head does not move in a consistent direction with joint loading. Our finding of a statistically significant difference in femoral head position with joint loading in a low-wear group raises the question of joint-loading effects in a high-wear group. We continue to follow the patients in this series and are evaluating a group with longer follow-up to determine if the amount of displacement with joint loading increases more dramatically with time and whether the direction of displacement develops a more consistent pattern as more acetabular polyethylene wear occurs.
Martell et al.
517
Acknowledgment The authors thank Theodore Karrison, PhD, for his help with statistical analysis and interpretation.
References 1. Rimnac CM, Wilson PD, Fuchs MD, et al: Acetabular cup wear in total hip arthroplasty. Orthop Clin North Am 19:631, 1988 2. Ritter MA, Keating EM, Faris PM, et al: Metal-backed acetabular cups in total hip arthroplasty. J Bone Joint Surg Am 72:672, 1990 3. Tanser M, Maloney WJ, Jasty M, et al: The progression of femoral cortical osteolysis in association with total hip arthroplasty without cement. J Bone Joint Surg Am 74:404, 1992 4. Appel AM, Sowder WG, Silverhus SW, et al: Prosthesisassociated pseudomembrane-induced bone resorption. Br J Rheumatol 29:32, 1990 5. Howie DW, Cornish BL, Vernon-Roberts B, et al: A rat model of resorption of bone at the cement-bone interface and the presence of polylethylene wear particles. J Bone Joint Surg Am 70:257, 1988 6. Howie DW, Cornish BL, Vernon-Roberts B: Resurfacing hip arthroplasty classification of loosening and the role of prosthesis wear in particles. Clin Orthop 255:144, 1990 7. Kim KJ, Chiba J, Rubash HE: In vivo and in vitro analysis of membranes from hip prostheses inserted without cement. J Bone Joint Surg Am 76:172, 1994 8. Murray DW, Rushton N: Macrophages stimulate bone resorption when they phagocytose particles. J Bone Joint Surg Br 72:988, 1990 9. Amstutz HC, Campbell P, Kossovsky N, et al: Mechanism and clinical significance of wear debris-induced osteolysis. Clin Orthop 276:7, 1992 10. Johanson NA, Callaghan JJ, Salvati EA: Fourteen year follow up study of a patient with massive calcar absorption—a case report. Clin Orthop 213:189, 1987 11. Mirra JM, Marder RA, Amstutz HC: The pathology of failed total joint arthroplasty. Clin Orthop 117:175, 1982 12. Bankston AB, Faris PM, Keating EM, et al: Polyethylene wear in total hip arthroplasty in patient-matched groups: a comparison of stainless steel, cobalt chrome, and titanium-bearing surfaces. J Arthroplasty 8:315, 1993 13. Livermore J, Ilstrup D, Morrey B: The effect of femoral head size on wear of the polyethylene acetabular component. J Bone Joint Surg Am 72:518, 1990 14. Wroblewski BM: Direction and rate of socket wear in Charnley low-friction arthroplasty. J Bone Joint Surg Br 67:757, 1985
518 The Journal of Arthroplasty Vol. 15 No. 4 June 2000 15. Hardinge K, Porter ML, Jones PR, et al: Measurement of hip prostheses using image analysis: the maxima hip technique. J Bone Joint Surg Br 73:724, 1991 16. Jones PR, Taylor CJ, Hukins DW, et al: Prosthetic hip failure—retrospective radiograph image analysis of the acetabular cup. J Biomed Eng 11:253, 1989 17. Martell JM, Berdia S: Determination of polyethylene wear in total hip replacements with use of digital radiographs. J Bone Joint Surg Am 79:1635, 1997 18. Shaver SM, Brown TD, Hillis SL, et al: Digital edge detection measurement of polyethylene wear after total hip arthroplasty. J Bone Joint Surg Am 79:690, 1997 19. Devane PA, Bourne RB, Rorabeck CH, et al: Measurement of polyethylene wear in metal-backed acetabular cups—I. three dimensional technique. Clin Orthop 319:303, 1995 20. Devane PA, Robinson EJ, Bourne RB, et al: Measurement of polyethylene wear in acetabular components inserted with and without cement: a randomized trial. J Bone Joint Surg Am 79:682, 1997 21. Kabo JM, Gebhard JS, Loren G, et al: In vivo wear of polyethylene acetabular components. J Bone Joint Surg Br 75:254, 1993
22. Rydell NW: Forces acting on the femoral-head prosthesis: a study on strain gauge supplied prostheses in living persons. Acta Orthop Scand 37:(Suppl)88:1, 1966 23. English TA, Kilvington M: In vivo records of hip loads using a femoral implant with telemetric output (a preliminary report). J Biomed Eng 1:111, 1979 24. Davy DT, Kotzar GM, Brown RH, et al: Telemetric force measurements across the hip after total arthroplasty. J Bone Joint Surg Am 70:45, 1988 25. Wroblewski BM: Wear and loosening of the socket in the Charnley low-friction arthroplasty. Orthop Clin North Am 211:30, 1986 26. Bartel DL, Wright TM, Edwards DL: The effect of metal backing on stresses in polyethylene acetabular components. In: The Hip: proceedings of the Hip Society. CV Mosby, St Louis, 1983 27. Hernandez JR, Keating EM, Faris PM, et al: Polyethylene wear in uncemented acetabular components. J Bone Joint Surg Br 76:263, 1994 28. Zichner LP, Willert HG: Comparison of aluminapolyethylene and metal-polyethylene in clinical trials. Clin Orthop 282:86, 1992