between the rate of penetration and the direction of wear. Despite the theoretical advantage of penetration in the superolateral direction, i.e., along the margin of.
Observations on the direction of wear in Charnley sockets retrieved at revision R. M. Hall, P. Siney, B. M. Wroblewski, A. Unsworth From the University of Durham and Wrightington Hospital, Wigan, England
he direction of wear in the acetabular socket has implications for the amount of debris that is generated during movement, for the magnitude of eccentric loading and for the incidence of impingement of the neck. We observed the direction of penetration with respect to a global co-ordinate system in 84 acetabular components retrieved at reoperation. The mean direction of wear relative to the open face of the sockets was found to be 37° with a range from 0° to 87°. For those values determined using the inclination of the socket on the prerevision radiograph, the mean direction of penetration in the coronal plane had a lateral, rather than a medial, component. The mean angle was 84° (SD 17°) with respect to the horizontal. The angle of penetration was found to correlate significantly with the depth, in that the lateral component became larger as the wear progressed. There was also a significant correlation between the rate of penetration and the direction of wear. Despite the theoretical advantage of penetration in the superolateral direction, i.e., along the margin of the socket, in reducing the probability of impingement of the neck, no significant correlation was seen between the angle of penetration and the period of use in vivo. This may suggest that impingement of the femoral neck on the rim of the socket may not be the dominant factor in loosening of the socket but can still be important in a few cases.
T
J Bone Joint Surg [Br] 1998;80-B:1067-72. Received 31 January 1997; Accepted after revision 28 July 1998
R. M. Hall, PhD, Lecturer St James's University Hospital, Academic Department of Orthopaedic Surgery, Level 5, Clinical Sciences Building, Beckett Street, Leeds LS9 7TF, UK. P. Siney, BA, Research Fellow B. M. Wroblewski, FRCS, Consultant Orthopaedic Surgeon, Professor of Orthopaedic Biomechanics Wrightington Hospital for Joint Disease, Centre for Hip Surgery, Appley Bridge, Wigan, Lancashire WN6 9EP, UK. A. Unsworth, FEng, Professor Centre for Biomechanical Engineering, School of Engineering, University of Durham, Durham DH1 3LE, UK. Correspondence should be sent to Dr R. M. Hall ©1998 British Editorial Society of Bone and Joint Surgery 0301-620X/98/67647 $2.00 VOL. 80-B, NO. 6, NOVEMBER 1998
Wear of the acetabular socket remains the dominant issue in hip arthroplasty and has been linked to the premature failure of these components through the resultant loosen1,2 ing. The current most-favoured theory is that an immune response to wear debris is the principal cause of loosen3,4 ing, but mechanically-based aetiologies cannot be ruled 5 out. In both theories, the direction of penetration is an important factor. In the immunological theory, the volume of debris generated depends by a factor of two on the 6,7 direction of wear for a given depth of penetration. In the mechanically-based theory, the effects of eccentric loading are increased when the angle between the line of action of the joint reaction force (JRF) and the direction of wear 8 becomes larger. In addition, from simple geometrical arguments it can be shown that impingement of the femoral neck against the rim of the socket is less likely to occur if the penetration is along the superolateral edge, rather than the axis, of the cup. Wear of the acetabular socket is dependent both on the sliding distance of the head as it articulates across the socket and the normal load carried by the interacting 9 surfaces. Studies of friction in new and explanted artificial hips indicate that they operate in a mixed lubrication 10,11 regime, in that there is a significant amount of contact between the surfaces of the head and socket. The sliding distance depends on both the type and duration of the activities undertaken by the patient, but in terms of the walking cycle the major movement is flexion-extension coupled with smaller amounts of internal-external rotation 12 and abduction-adduction. The load applied to the joint in vivo is approximately three times the body-weight and the mean direction of this load vector has been estimated to be in the superomedial direction with an angle of 12° to the 12 13 vertical. Elson and Charnley reported that the line of action of the wear vector was 8° to the vertical, again with a medial component. In a study in which only flexion14 extension was considered McLeish and Skorecki found that a JRF with a medial direction of 12° with respect to the vertical would produce a wear direction of 7° relative to the same axis. The JRF and wear direction need not be collinear. An additional important result was that the rate of wear, if the force were collinear with the axis of rotation, would be half that observed if the load were perpendicular to the axis. The direction of wear could be altered if the 1067
1068
R. M. HALL,
P. SINEY,
B. M. WROBLEWSKI,
Material and Methods
Table I. Details of the patients in both groups
Mean implant period in years (SD) Mean age at primary surgery in years (SD) Mean weight (N); (SD) Loose sockets (%)
A. UNSWORTH
Group A (n = 84)
Group B (n = 24)
11.2 (4.8) 51.4 (13.9) 710 (150) 92
13.2 (3.7) 51.4 (11.4) 690 (130) 97
Fig. 1 Photograph of an explanted Charnley socket with the marks indicating its anatomical orientation at the time of revision.
axis of rotation was tilted in the coronal plane. This model took no account of the creep component in any penetration 15 of the socket by the femoral head. Wroblewski reported that a significant proportion of a small series of explanted sockets showed superolateral wear. Similar observations 8 were made by Hall et al who also noted that the tendency for superolateral wear increased as the depth of penetration 16 became greater. Murray and O'Connor undertook a theoretical analysis of the results of Hall et al noting that penetration was a linear combination of wear and creep and that the two variables did not have to act along the same line. It was assumed that the direction of penetration due to wear alone was perpendicular to the rotational axis regardless of the position of the JRF. Comparison with the model 14 of McLeish and Skorecki indicates that for a linear model this assumption may not be justified. Our aim was to identify, from a large series, the mean angle of penetration and its association with the depth of penetration, the period of implantation, the rate at which the head bores into the socket and the volumetric wear. We attempted to explain these observations in terms of the current knowledge of human gait and the understanding of the process of wear. We also discuss the importance of the direction of penetration in terms of impingement of the neck and load eccentricity.
We removed 84 acetabular components at revision and marked them so that the orientation of the socket about its axis at the time of revision could be determined (Fig. 1). Each of the removed prostheses had the accompanying radiograph before revision available for inspection. In addition, for 29 of the components the radiograph after the primary operation was available. We divided the components into two groups: in group A both primary and revision radiographs were available, and in group B only the radiographs at revision. The parameters which were derived by measurements taken from the postprimary radiographs were assigned the subscript 'p' while those determined at revision were denoted as 'r'. It was important to have these groups so that the effect on the direction of wear of any migration of the socket, i.e., change in inclination, could be assessed. Table I gives the details of the patients. We used 17 the method of Wroblewski to determine whether the socket was fixed or loose at the time of revision. 7 Wear measurement. We used the shadowgraph technique to determine both the angle of penetration () and the amount of linear wear (P). The former is the angle between the line joining the two centres and the rim of the socket (Fig. 2). A cast of the socket bore was prepared using a high-definition silicone-based moulding agent (RTV 2039; Ambersil, UK). The replica was subsequently projected, at a magnification of 10, on to the screen of the shadowgraph apparatus. Profiles of the socket bore in both the coronal and wear planes were recorded. A template was used to locate the initial and final positions of the centre of the femoral head. The depth of penetration was equivalent to the distance between the centres at these two positions (Fig. 2). Penetration is composed of both creep and wear although, in general, the creep part is small and of the order
Fig. 2 Definition of the angles of penetration with respect to both the socket and the body axes, and angle of tilt. THE JOURNAL OF BONE AND JOINT SURGERY
OBSERVATIONS ON THE DIRECTION OF WEAR IN CHARNLEY SOCKETS RETRIEVED AT REVISION
1069
Table II. The mean values (SD) of the parameters determined in both groups
Mean penetration angle with respect to the socket () (degrees) Tilt of the socket from the postprimary radiographs (pB) (degrees) Tilt of the socket from prerevision radiographs (rA and rB) (degrees) Direction of penetration derived from primary radiographs (pB) (degrees) Direction of penetration derived from revision radiographs (rA and rB) (degrees) Depth of penetration (P) (mm) Rate of penetration (P/t) (mm/yr) 3 Volumetric wear (V) (mm ) 3 Volumetric wear rate (V/t) (mm /yr)
Group A (n = 84)
Group B (n = 29)
38 (16) 46 (10) 84 (17) 2.2 (1.6) 0.21 (0.14) 610 (430) 56 (39)
39 (18) 46 (6) 46 (10) 85 (15) 85 (18) 2.3 (1.3) 0.19 (0.12) 660 (370) 52 (31)
Table III. The correlation between the direction of penetration and other explanatory parameters Direction of penetration
Implant period (yr) Penetration depth (P) (mm) Penetration rate (P/t) (mm/yr) 3 Volumetric wear (V) (mm ) 3 Volumetric wear rate (V/t) (mm /yr)
18
of 0.1mm. The wear volume (V) was calculated using a 7 modified version of the formula previously presented by 5 Kabo et al. Using the radiographs the tilt of the socket () with respect to a horizontal line joining teardrops was determined with a goniometer. The direction of penetration, in the coronal plane, with respect to the horizontal () was then calculated as the sum of and . Further, theoretical predictions were made of the direction of penetration as a function of depth. Murray and 16 O'Connor have previously indicated that the penetration for creep alone is approximately 23° from the vertical in a medial direction. For wear alone the direction would be superolateral at an angle of 14° to the vertical. It was suggested that the direction of wear would be determined by the addition of the two vectors describing the penetration due to creep and wear. The exact rate of penetrative creep, however, is difficult to predict because its initial component is very rapid and the amount of creep in newly exposed material is difficult to quantify. Further, there may be effects of interaction between the creep and wear elements of the penetration as both alter and are influenced by the contact regime. In this assessment, we used a simplified model in which the creep component was taken to be constant, at a value of 0.1 mm, which appears to be the 18 maximum value attained in vivo. Empirical evidence suggests that this value is attained after a few hundred 19 thousand cycles of use. The direction of this constant displacement vector is taken as 23° from the vertical in a medial direction. It must be emphasised that this simplification, in which there is the vector addition of the initial creep displacement and the accumulative penetration due to wear, does not take into account the continuing creep of the local ultra-high-molecular-weight polyethylene (UHMWPE) as the penetration occurs. VOL. 80-B, NO. 6, NOVEMBER 1998
Group A (n = 84)
p value
Group B (n = 29)
p value
0.010 -0.253 -0.338 -0.154 -0.267
0.921 0.021 0.002 0.162 0.014
0.123 -0.456 -0.458 -0.379 -0.387
0.525 0.012 0.012 0.043 0.039
Statistical analysis. We used the STATA 4.0 statistical 20 package. Association between variables was undertaken using Pearson's correlation coefficient, r. The difference between two cohorts was determined using Student's t-test. We analysed the inequalities in variances using the F20 statistic.
Results Table II gives the mean values of the direction of penetration, the tilt of the socket, the magnitude of the linear and volumetric wear and their respective rates. For group B, there was no significant difference in the mean value of derived from the postprimary and prerevision radiographs (t = 0.10, p = 0.92) in spite of a measurable change in tilt of 11 of the sockets, which was reflected in a significant increase in the variance observed. This result indicated that there was no preferred direction of rotational migration, in the coronal plane, once the sockets had become loose. The mean value of the direction of penetration with respect to the body, , was found to be significantly different from the vertical for the larger group A (t = 3.34, p = 0.001) and had a lateral, rather than a medial, component. No significant difference between the direction of penetration and the vertical was observed for the smaller cohort (t = 1.63, p = 0.11), although again the lateral component was detected. Figure 3 illustrates the association between and the penetration depth, P, for groups A and B. The correlation coefficients for the direction of wear against other parameters are shown in Table III. As the depth of penetration at the time of revision increases the direction of wear appears to become more lateral. A reasonably strong correlation was observed between linear wear and the direction
R. M. HALL,
P. SINEY,
B. M. WROBLEWSKI,
A. UNSWORTH
Penetration angle (rA) (degrees)
1070
Depth of penetration in the coronal plane (P) (mm) Figure 3 – The relationship between depth of penetration and penetration angle for a) group A (data from prerevision radiographs) and b) group B (data from postprimary radiographs).
Penetration angle (pB) (degrees)
Fig. 3a
Depth of penetration in the coronal plane (P) (mm) Fig. 3b
of wear especially for those components in which data were calculated from the postprimary radiographs. However, the correlations tended to be weaker for volumetric data. Figure 3 also shows the comparison between the observed data and the simplified version of the model proposed by Mur16 ray and O'Connor. A number of points lie outside the rather flexible bounds that the model allows and, from a purely qualitative standpoint, it is difficult to ascertain its suitability due to the variability in the data. Of probably greater importance is the very high degree of association between the rate of penetration and the direction of wear to 16 which Murray and O'Connor give greater emphasis (Table III), although, again, a quantitative assessment is difficult.
Discussion In the light of our results, a number of important issues need to 14 be explained. Clearly, the analysis by McLeish and Skorecki, 13 while explaining the results of Elson and Charnley, cannot account for the predominantly superolateral wear observed in 16 our series. Murray and O'Connor, however, have indicated that the mean axis of rotation was tilted by 14° so that it would enter the superomedial quadrant. Using this value, the model 14 proposed by McLeish and Skorecki would predict the penetration to be in the superior direction and therefore more in keeping with our results. The scatter in the data probably reflects the variance in the direction of the JRF that has been 21 observed during gait analyses. THE JOURNAL OF BONE AND JOINT SURGERY
OBSERVATIONS ON THE DIRECTION OF WEAR IN CHARNLEY SOCKETS RETRIEVED AT REVISION
With regard to the correlation of the depth of penetration and direction, two explanations are possible. First, these are retrieved specimens and as such only the direction at the end of the useful life of the implant can be observed. It may well be possible that each of the sockets has a different direction of wear at the beginning of use in vivo and maintains this constant angle of penetration until the point of revision. While the penetration is greater for those sockets with superolateral wear, the correlation between the direction of penetration and the volume of wear is only marginally significant at best. Therefore these observations do not depart from the currently favoured theories which suggest that loosening is related to the volume of wear 14 debris produced. The analysis of McLeish and Skorecki would not conflict with this interpretation of the results since no change in direction for each individual socket is envisaged. The second explanation is that the direction of wear does change during use in vivo and is related to the depth of penetration. The model proposed by Murray and O'Con16 nor appears to have reasonable agreement with our empirical data. As the head penetrates the socket the biomechanics of the joint may also change. In particular, the gluteus muscles may not be able to provide the same tension as they did before the penetration. In such a circumstance, the movement of the socket in the coronal plane over the head of the femoral component will be less pronounced leading to more wear in the lateral margin of the socket. Surgical factors, such as trochanteric nonunion or impairment of function of the abductors, may also play a role in modifying the gait of the individual and therefore the direction of wear observed in the socket. The issue of whether the direction of wear changes as the wear progresses lends itself to resolution with the use of roentgen stereophotogrammetric analysis in which the penetration of the socket by the head can be accurately monitored in vivo and over an extended period. While difficulties pertain to improve the analytical solutions formulated by McLeish 14 and Skorecki, finite-element models such as those devel22 oped by Maxian et al may assist in accurately predicting directions of wear from gait data. There was an increase in variance of the tilt of the socket in the coronal plane for those measurements taken from the prerevision, rather than the postprimary, radiographs. Observations on 11 of the 29 sockets, for which the initial postoperative radiographs were available, had shown migration. If the same percentage of all the sockets in group B is loose then this may affect the results pertaining to our series due to an increased variance. Migration did not appear to affect the mean value of the angle of penetration as the movement of the loose socket resulted in a reduction in in some and an increase in others. The effects of superolateral wear on the performance of the fixation also needs to be assessed. Once the socket has become worn, the direction of the JRF will no longer pass through the centre of rotation, although it does pass through VOL. 80-B, NO. 6, NOVEMBER 1998
1071
the centre of the femoral head. This will produce a torque on the socket in which the magnitude is dependent on the direction and size of the JRF, the depth and direction of penetration and the disparity between the centre of the back of the socket and the current centre of the femoral head. Since it is highly probable that the strength of fixation at the cement-bone interface does not have a fatigue endurance limit, i.e., a stress below which the interface will not fail, the cyclic nature of the torque produced by eccentric loading may contribute to the loosening process. The analysis is complicated, however, by any eccentricity in the cement mantle. A simple geometrical argument would indicate that the femoral component would become more constrained when the direction of wear with respect to the outer rim of the socket, , is towards the pole rather than along the rim. The position of the socket will also have a role in determining the level of impingement but the amount of encapsulation is the principal factor here. Those sockets with superolateral wear should have less risk of impingement and therefore have a greater life in vivo. In our study those with superolateral wear lasted no longer than those with superomedial wear. This observation, together with the fact that there is only a marginal correlation between the volume of wear at the time of retrieval and the direction of penetration, lends support to those theories which advocate that the volume of wear debris is one of the critical parameters in loosening. Impingement has a major role in only a few sockets. An alternative hypothesis would be that the advantage of superolateral wear in terms of impingement is offset by another effect such as increased bone resorption behind the lateral margin of the socket. We wish to acknowledge the Arthritis and Rheumatism Council for Research (ARC) who are the sole financial supporters of this research (Grant numbers U0010 and U0505). We also thank Mr R. Fraser who allowed us access to the shadowgraph equipment at the Gear Metrology Unit, University of Newcastle upon Tyne, and Mrs P. Fleming of the Centre for Hip Surgery, Wrightington Hospital, for help in acquiring the clinical data. No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article.
References 1. Schmalzried TP, Kwong LM, Jasty M, et al. The mechanism of loosening of cemented acetabular components in total hip arthroplasty: analysis of specimens retrieved at autopsy. Clin Orthop 1992; 274:60-78. 2. Wroblewski BM. Cementless versus cemented total hip-arthroplasty: a scientific controversy. Orthopedic Clinics of North America 1993; 24:591-7. 3. Harris WH. Osteolysis and particle disease in hip replacement: a review. Acta Orthop Scand 1994;65:113-23. 4. Willert H-G, Buchhorn GH. Particle disease due to wear of ultrahigh molecular weight polyethylene. In: Morrey BF, ed. Biological, material and mechanical considerations of joint replacement. 1st ed. New York: Raven Press Ltd, 1993:87-102. 5. Wroblewski BM. Direction and rate of socket wear in Charnley lowfriction arthroplasty. J Bone Joint Surg [Br] 1985;67-B:757-61. 6. Kabo JM, Gebhard JS, Loren G, Amstutz H. In vivo wear of polyethylene acetabular components. J Bone Joint Surg [Br] 1993; 75-B:254-8.
1072
R. M. HALL,
P. SINEY,
B. M. WROBLEWSKI,
7. Hall RM, Unsworth A, Craig PS, et al. Measurement of wear in retrieved acetabular sockets. Proct Inst Mech Engh 1995;209: 233-42. 8. Hall RM, Unsworth A, Wroblewski BM, Hardaker C. Long term changes in the frictional behaviour and wear in implanted Charnley prostheses. J Bone Joint Surg [Br] 1993;67-B:Supp III:267. 9. Williams JA, ed. Engineering tribology. First ed. Oxford: Oxford University Press, 1994. 10. Hall RM, Unsworth A, Wroblewski BM, Burgess IC. Frictional characterization of explanted Charnley hip prostheses. Wear 1994;175: 159-66. 11. Hall RM, Unsworth A, Wroblewski BM, Siney P, Powell NJ. The friction of explanted hip prostheses. Br J Rheumatol 1997;36:20-6. 12. Paul JP. Forces transmitted by joints in the human body. Proc Inst Mech Engrs 1967;181(3J):8-13. 13. Elson RA, Charnley J. The direction of the resultant force in total prosthetic replacement of the hip joint. Med and Biol Eng 1968;6: 19-27. 14. McLeish RD, Skorecki J. Analysis of wear in a spherical joint. Proc Instn Mech Engrs 1968-69;183:389-94.
A. UNSWORTH
15. Wroblewski BM. Wear and loosening of the socket in the Charnley low-friction arthroplasty. Orthop Clin North Am 1988;19:627-30. 16. Murray DW, O’Connor JJ. Superolateral acetabular wear of the acetabulum. J Bone Joint Surg [Br] 1998;80-B:197-200. 17. Wroblewski BM. Revision surgery in total hip arthroplasty: surgical technique and results. Clin Orthop 1982;170:56-61. 18. Hall RM, Unsworth A, Siney P, Wroblewski BM. Wear in retrieved Charnley acetabular sockets. Proc Instn Mech Engrs 1996;210: 197-207. 19. Elloy M. Simulator testing of joint prostheses: the need for realistic simulator testing. In: Proc SERC/MechE annual expert meeting: failure of joint prostheses. Bury St Edmunds: Mech Engrs Pub Ltd, 1993:79-82. 20. StataCorp. Stata statistical software. Release 4.0, College Station, TX: Stata Corporation. 21. Brown TRM, Nicol AC, Paul JP. Comparison of loads transmitted by Charnley and CAD Muller total hip arthroplasties. Proc Instn Mech Engrs 1984;210:63-8. 22. Maxian TA, Brown TD, Pedersen DR, Callaghan JJ. A slidingdistance-coupled finite element formulation for polyethylene wear in total hip arthroplasty. J Biomech 1995;29:687-92.
THE JOURNAL OF BONE AND JOINT SURGERY