Rigidity of initial fixation with uncemented tibial knee implants

Rigidity of Initial Fixation With U n c e m e n t e d Tibial Knee Implants

P h i l i p J. B r a n s o n , M D , * J o h n W . S t e e g e , MSME,-IR i c h a r d L. W i x s o n , MD,:[: J a c k L e w i s , P h D , § a n d S. D a v i d S t u l b e r g , MD:[:

Abstract: This study quantifies the in vitro motion occurring between bone and cemented and noncemented tibial components. Liquid metal strain gauges were used to measure the motion between the tibial component and bone at four locations in eight cadaver tibia at near-point cyclic loads ranging from 10 to 2,000 N. Two types of motion were observed: inducible displacement, which is reversible, followed the oscillating load and occurred in both cemented and uncemented tests, and liftoff or separation of the component and bone, which occurred only for the noncemented cases and remained even after removal of the load. For both motion types, noncemented tests exhibited significant (P < .05) and dramatic increased interface motion compared to the cemented cases for all load types. The results suggest that the magnitudes of implant-bone interface separation at loads in the low physiologic range for noncemented implants can be sufficiently large to inhibit bony ingrowth into a prosthesis with an average pore size of 300 ixm. K e y words: porous ingrowth, implant fixation, noncemented implants, cemented implants

reported by Collier et al. (6) and Haddad (10). These have shown fibrous tissue rather than bony ingrowth underneath the tibial surface. Rosenqvist et al. (20) have reported progressive radiolucencies beneath the tibial c o m p o n e n t of PCA T M components (Porous Coated Anatomic Knee Prosthesis, Howmedica, Inc.). These have been accompanied by some implant migration and loosening of some of the c h r o m e - c o b a l t beads. Ryd et al. (2 I), using s t e r e o p h o t o g r a m m e t r y , demonstrated migration of the noncemented knees from 0.2-3.5 m m over a period of 1 - 2 years. By comparison, cemented knees, with and without metal backing, demonstrated less migration. Stulberg et al. (24), with uncemented Microloc T M (Microloc Total Knee System, Johnson & J o h n s o n Orthopaedic Division, Braintree, MA) knee arthroplasties, using an accurate, repeatable fluoroscopic

Biological fLxation of total knee implants, through bone ingrowth into porous surfaces, has been developed to provide long-term stable fixation. While there are n o w reports that indicate short-term successful clinical results with noncemented bone ingrowth total knee systems (9, 13, 14, 19, 23), bone ingrowth into the tibial porous surface m a y be unreliable and variable. Retrieval studies involving several different types of total knee designs have been

* From Princeton, West Virginia. f From the Rehabilitation Engineering Program, Northwestern University, Chicago, Illinois. From the Department of Orthopaedic Surgery, Northwestern University, Chicago, Illinois. § From the Department of Orthopaedic Surgery, University of Minnesota, Minneapolis, Minnesota.

Reprint requests: Richard L. Wix~on, MD, Suite 404, 233 East Erie Street, Chicago, IL 60611.

21

22 The Journal of Arthroplasty Vol. 4 No. 1 March 1989 method, demonstrated radiolucencies of variable size beneath 39 of 40 noncemented tibial trays. Assumptions involved in the design of porous implants were developed empirically on the basis of a large number of animal studies that demonstrated bony ingrowth (8). Factors that were considered important for reliable bony ingrowth into a porous surface included the use of an inert material (22), proper pore size (15), total initial implant bone contact (2), and relatively rigid initial fixation (5). Although bone ingrowth has been demonstrated under some conditions of dynamic loading (5), studies by Ducheyne et al. (7) with a canine porous knee and Pilliar (17) with a canine intramedullary stem demonstrated fibrous rather than bony fixation under conditions of dynamic loading. Development of a fibrous attachment has been attributed to gross motion or micromovement at the bone-prosthesis interface. Because of a high incidence of radiolucencies noted beneath the tibial component in the Microloc knee system, we attempted to quantify the motion occurring between bone and the Microloc tibial ingrowth design at the time of initial fixation to understand better the role of relative motion in ingrowth failure.

Materials and Methods Eight tibial cadaver specimens were obtained at the time of autopsy and stored at -20°C until tested. The tibial specimens came from men between the ages of 23 and 60 years without evidence of arthritis. Following complete stripping of the soft tissues, the specimens were potted in PMMA 12 cm below the joint line. The potted specimen was then held in a machinist clamp such that the articular surface was perpendicular to the long axis of the tibia in a frontal plane and sloped 5° posteriorly in the sagittal plane. Using an industrial-quality milling machine, the proximal articular surface was milled flat beginning with 0.5-ram increments. Final finishing was done in 20-p.m increments at 1 mm below the lowest level of the cartilage-bone junction. This technique allowed the creation of the flattest and most even surface possible with a precision that could not be obtained with conventional total knee guides and hand-held power saws. The surface was checked for flatness using a precision machinist's square equipped with a dial gauge. The prothesis design utilized in the study, Microloc, is a titanium oval-shaped metal plate with three short, porous-coated fixation pegs extending 7 mm below the prosthesis. Various thicknesses of

polyethylene are attached to the prosthesis. Two pegs 7 mm in diameter are located beneath the center of each tibia1 plateau, and an oval peg (7 x 10 ram) is located anteriorly. The entire surface (including the pegs) is coated with three layers of titanium beads, resulting in a mean pore size of 300 p.m. Figure 1 shows the bottom surface of the prosthesis design. The drill size used to prepare the peg fixation holes on the tibial surface was 1 mm undersized and produced a 12% interference fitbetween the pegs and bone. To determine the amount of tibial coverage by the implant, the surface area of each tibial specimen was measured by tracing the proximal surface on the digitizing tablet of a Zeiss Videoplan. Tibial coverage was expressed as the ratio of the surface area of the undersurface of the prosthesis to the surface area of the milled proximal tibia. The prosthesis used for each tibia was determined by templating the four available widths (65, 71, 77, or 83 mm) and selecting the largest Component that could be accommodated without significant overhang. The amount of tibial coverage ranged from 77% to 94% (average, 85%). Following creation of the peg fixation holes on the surface at the tibia, the metal baseplate of the selected tibial component was axially impacted onto the tibial surface. This was done using a 2-1b. mallet and tibial impactor in the same manner as the surgical technique. Mechanically calibrated 1-cm-long liquid metal strain gauges (LMSG, Parks Medical Electronics, Beaverton, OR), whose design has been characterized thoroughly by Brown et al. (3), were attached to the metal baseplate and the proximal 5 mm of the tibia at anterolateral, anteromedial, posterolateral, and posteromedial locations (Fig. 2). The gauge was cemented to notches in the titanium plate and screwed to the bone with a 1.5-ram machine screw. The specimens were then mounted in a MTS machine and loaded perpendicular to the tibial titanium baseplate. The output from the four gauges was amplified, filtered, and monitored on a four-

Fig. 1. Underside of the Microloc baseplate with peg location.

Rigidity of Initial Fixation *

23

Branson et al.

with an injection nozzle, attempting to achieve approximately 4 m m of penetration over the entire surface of tibia. The prosthesis was then placed onto the tibia and manually impacted until the cement had set. The liquid metal strain gauges were then reattached to the tibial baseplate and mounting screws and recalibrated, and the entire loading sequence was repeated. The tibias were radiographed at the completion of testing, showing 4 - 5 m m of cement penetration in all specimens.

Results

Fig. 2. Liquid metal strain gauges attached to the prosthesis baseplate and proximal tibia. channel stripchart recorder. Due to the inherent low resistance of approximately 1 ohm for the LSMGs, each gauge was configured in series with a 120-ohm resistor as one arm of a conventional Wheatstone bridge circuit. Near-point loads were applied axially at a rate of one cycle every 4 seconds. The loads were initially applied centrally in increments from 10 to 2,000 N. A series of 10 cycles for each load level was performed with the displacement for each gauge recorded separately. The interface motion was also visually monitored with a 10 x microscope. Following achievement of the maximal load, the loading increments were reversed down to 10 N to compare the reproducibility of the displacement measurements. Following the central loading, the same loading sequence was performed at posteromedial, posterolateral, and anterior locations on the edge of the prosthesis. In t h e eccentric loading situations, to avoid destruction of the proximal tibial bony bed or damage to the gauges, further incremental loading was terminated if separation exceeded 2 - 3 ram, thus not all specimens were taken to the 2,000-N level. Following completion of the loading tests, the implant was removed from the proximal tibia. The top surface of the cut proximal tibia was inspected for depressions or areas of bone deformation. The peg holes were then enlarged to 1 mm diameter greater than the peg diameter and the proximal tibia prepared for cement fixation of the same implant. The cancellous surface was cleaned with pulsatile lavage, dried, and injected with Simp!ex-P, 2 minutes after mixing at room temperature, using a cement gun

All eight specimens were loaded centrally from 10 to 2,000 N without visible evidence of tilting or liftoff. The displacement varied from specimen to specimen, with the maximum displacement in either anteriorly or posteromedially. The mean maximum motion induced with central loading, recorded at any of the four gauges, varied from 6 to 290 p.m over the loading range. With the same loading regime, the cemented implant mean maximum motion varied from 0 to 100 p.m. Using l0 x magnification, the uncemented implant appeared to be pushed down into the cancellous bone, where there was no visible interface motion with the cemented system. Table 1 presents the displacement seen at 2,000 N at each location for all eight specimens. With uncemented fixation, the displacement began immediately and increased with the applied load. When the prosthesis was cemented, the amount of measured displacement did not substantially increase until 1,000 N of load. This can be seen in Figure 3, which illustrates the mean of the max-

3OO

200

u

~

Cemented

~oo

10

50

...... be',wee~ m~l~c~s

ns

n$

100

500

I tl

, , Load ( t ~ o n $ )

i

!

.04

,03

.C2

.COS

1000

i ! .,i . i .C07

,

.001

! J

ii

I

i

i

i

.000

.C00 000

2000

I

.000

Fig. 3. Mean at the maximum inducible motion deter-

mined by any of the four gauges at each load increment for all specimens for control loading.

The Journal of Arthroplasty Vol. 4 No. 1 March 1989

24

T a b l e 1. Bone-Implant Interface Displacement (p.m) with Central Loading at 2,000 N Anterior-Lateral

Anterior-Medial

Posterior-Medial

Posterior-Lateral

Specimen

Noncemented

Cemented

Noncemented

Cemented

Noncemented

Cemented

Noncemented

Cemented

1 2 3 4 5 6 7 8

137 177 * 84 177 97 371 514

22 0 59 97 71 71 84 371

204 71 * 46 137 259 177 177

0 22 22 22 22 97 22 48

59 191 * 124 97 259 22 287

0 0 22 46 22 22 22 315

46 232 * 273 71 97 97 204

0 0 22 0 22 71 0 22

Mean Plevel

222

99

153

55

148

61

146

.02

.01

.05

17 .01

*Missing specimens.

imum inducible motion determined by any of the four gauges at each load increment for all specimens. Eccentric or edge-loading with the uncemented implants produced dramatically more motion with tilting and liftoff of the implant. The spearation became visible to the naked eye at 200-300 p.m. For some specimens, posteromedial loading produced the greatest separation at the lowest loads. Separation of 2 0 0 - 5 0 0 p.m occurred at 2 0 - 2 0 0 N. The harder bone in the medial plateau of the tibia appeared to act as a fulcrum for liftoff measured at the anterolateral strain gauge. Posterolateral loading produced medial separation at higher loads from 150 to 1,500 N. Anterior loading produced liftoff posteriorly at 4 0 - 2 0 0 N and was accompanied by displacement into the anterior cancellous bone of 100250 p.m. Figure 4 shows the mean for all noncemented specimens of the maximum liftoffs as recorded by either of the two posterior gauges for increasing anterior loads. Liftoff of the anterolateral gauge due to posteromedial loading and liftoff of the

800 -

), 600 A i

v

~

400

-

200

-

o~ lo

Non-Cemented Liftoff due tO Eccentric Loading 41 A~I.,~, toad ~ PO$1IrI~.~LII [Old I1~ Pos~lnof.La~eril Load

5

/ /

Cemented Displacement Beneath Eccentric Loads {~ A~:mr,oFtoad 0 Posleno~Med~aJ Los¢ 1~ Pos~ermf-Lllm,al Load

~ 5o

too

Load

S00

1000

2000

(Newtons)

Fig. 4. The mean for all noncemented specimens of the maximum liftoffs and cemented inducible displacement recorded for increasing anterior, posteromedial, and posterolateral loads.

anteromedial gauge due to posterolateral loading is also shown. Loading of the cemented implant in any of the eccentric modes resulted in no visible or measured separation or liftoff. Figure 4 also shows the mean for all cemented specimens of the maximum measured interface displacements for the same loading situations.

Discussion In this study of the tibial Microloc component, the cemented fixation, which provides the greatest degree of initial rigid fixation, served as a basis for comparison to the relative rigidity of the noncemented fixation. With normal postoperative activities after surgery, some degree of loading of the interface will occur. Since the preparation of the tibia was done with a milling machine and the prosthesis implanted under laboratory conditions, the quality of fixation should be superior to what could be obtained in surgery and represents a best-case situation. The posterior slope of 5° is parallel to the true joint surface and allowed preservation of stronger subchondral bone anteriorly than a 0 ° cut, which would have removed more bone. These results demonstrate significantly greater initial micromotion between the Microloc implant and bone, for uncemented fixation compared with cemented fixation. The displacement recorded is a combination of deformation of the tibia, compaction of the implant into the subchondral cancellous bone, and interface separation. The gauges were attached within 5 nun of the tibial surface to minimize measurement of fibial deformation. Maximal motion induced at 2,000 N in cemented applications was 100 p,m, as compared with 290 y.m in the noncemented applications. Observation with 10 x magnification in the nonce-

Rigidity of Initial Fixation •

mented application demonstrated visible motion occurring at the edge of the prosthesis between the metal and the bone. There was no visible interface motion at 10 x magnification with the cemented application. Since the cement was injected into the proximal cancellous bone, the measured motion probably is a reflection of deformation in the b o n e cement composite and the underlying cancellous bone. The deformation was not associated with gross trabecular disruption or depression. The difference, therefore, between the m a x i m u m motion shown with cement and the noncemented prosthesis would largely represent true interface motion. Interface motion has been shown to prevent ingrowth into porous surfaces. Cameron et al. (5) demonstrated that gross motion at a canine osteotomy site fixed with a porous-coated staple precluded bone ingrowth. Ducheyne (7) found in a dynamically loaded hinged canine total knee with m e a n pore sizes of 87 and 110 p.m that bone ingrowth did not occur. Pilliar et al. (17) found, in a canine intramedullary model with an implant pore size of 5 0 400 ~m, that with dynamic loading, fibrous rather than b o n y ingrowth occurred. He calculated that with this pore size, "the m a x i m u m m o v e m e n t possible between bone and implant was 100 microns if the implant was loaded in compression alone, or 150 microns if the implant was cyclically loaded between tension and compression." By comparison, Pilliar (18) also determined that bone ingrowth would occur in an endodontic implant with a pore size of 5 0 - 2 0 0 ~ m with m a x i m u m tooth m o v e m e n t restricted to 28 p.m. For noncemented components in this study with central compressive loading, displacements of 1 0 0 500 p.m occurred at loads of 1 5 0 - 5 0 0 N. Larger degrees of motion with lower loads were found in cases of eccentric loading. This a m o u n t of motion, on the basis of Pilliar's conclusions from his animal studies, m a y be sufficient to prevent bone ingrowth. Motion of the implant relative to the bone could effectively reduce the pore size available for ingrowth. Motions observed in the noncemented applications, which are near the pore size of the implant, may effectively close he pores to ingrowth. Micromotion produced by compression-no load, as in this study, could have a different effect than similar amounts of motion produced by shear forces. The time required for vascular ingrowth and bone formation to occur in humans is u n k n o w n but is thought to be 6 - 8 weeks, as compared to 4 - 6 weeks in dogs (22). During this postoperative period, low loads would be expected to occur on the implant. Therefore, the b o n e - m e t a l interface might be subject to micromotion at the levels seen in this study. Har-

Branson et al.

25

rington (l 1) has shown in both normally aligned knees and those with varus or valgus deformity that the center of pressure varies from the lateral to the medial p l a t e a u during different phases of the gait cycle. This would indicate that during gait, regardless of the knee alignment, some degree of eccentric loading normally occurs. For the Microloc prosthesis used in this study, with eccentric loading, the situation would be expected to be worse, with significant amounts of liftoff occurring at relatively low loads. The interference fit of the pegs used in this prosthetic design was inadequate in resisting the liftoff and tilting observed, in comparison to the fixation achieved by cement.

Conclusions In comparison to the rigidity achieved at the time of initial fixation with bone cement, significant (P < .05) micromotion was observed with both compressive central loading and eccentric loading of a Microloc porous-coated noncemented tibial knee prosthesis on cadaver tibias. The motion observed in the u n c e m e n t e d application may be sufficient to explain the lack of b o n y fixation observed in m a n y n o n c e m e n t e d implants.

References 1. Bobyn JD, Pilliar RM, Cameron HU, Weatherly GC: The optimum pore size for the fixation of porous surfaced metal implants by the ingrowth of bone. Clin Onhop 150:263, 1980 2. Bobyn JD, Pilliar RM, Cameron HU, Weatherly GC: Osteogenic phenomena across endosteal bone implant spaces with porous surfaced intramedullary implants. Acta Orthop Scand 52:145, 1981 3. Brown TD, Sigal L, Njus GO et al: Dynamic performance characteristics of the liquid metal strain gauge. J Biomechanics 19:165, 1986 4. Cameron HU, Pilliar RM, Macnab I: The rate of bone ingrowth into porous metal. J Biomed Mater Res 10:295, 1976 5. Cameron HU, Pilliar RM, Macnab I: The effect of movement on the bonding of porous metal to bone. J Bi0med Mater Res 7:301, 1973 6. Collier JP, Mayor MB, Townley CO et al: Histology of retrieved porous-coated knee prosthesis. Presented at the 53rd AAOS Meeting, New Orleans, 1986 7. Ducheyne P, DeMeester P, Aemoudt E: Influence of a functional dynamic loading on bone ingrowth into surface pores of orthopedic implants. J Biomed Res 11:811, 1977

26 The Journal of Arthroplasty Vol. 4 No. 1 March 1989 8. Galante J, Rostoker W, Lueck R, Ray R: Sintered fiber metal composites as a basis for attachment of implants to bone. J Bone Joint Surg 53A:101, 1971 9. Galante JO, Sumner DR, Turner 3",Barden R: Fixation in total knee arthroplasty. In: Proceedings of the Knee Society, 1985-1986. Aspen, Rockviile0 MD, 1986 10. Haddad ILl, Cook SD, Thomas KA et al: Histologic and microradiographic analysis of noncemented retrieved PCA knee components. Presented at the 53rd AAOS Meeting, New Orleans, 1986 11. Harrington IJ: Static and dynamic loading patterns in knee joints with deformities. J Bone Joint Surg 65A:247, 1983 12. Hulbert SF, Cooke FW, Klawitter JJ el al: Attachment of prostheses to the musculoskeletal system by tissue ingrowth and mechanical interlocking. J Biomed Mater Res Symp 4:1, 1973 13. Hungerford DS, Kenna RV: Preliminary experience with a total knee prosthesis with porous coating without cement. Clin Orthop 176:95, 1983 14. Hungerford DS, Krackow KA, Kenna RV: Two to five year experience with cementless porous-coated total knee prosthesis. In: Proceedings of the Knee Society, 1985-1986. Aspen, Rockville, MD, 1986 15. Klawitter JJ, Hulbert SF: Application of porous ceramics for the attachment of load bearing internal orthopedic applications. J Biomed Mater Res Syrup 2:161, 1971 16. Miller J, Burke DL, Stachiewicz JW, Kelebay LC: The fixation of major load bearing metal prosthesis to bone: an experimental study comparing smooth to porous surfaces in a weight bearing Model. Presented

17.

18.

19.

20.

21. 22.

23.

24.

at the 1 lth International Conference on Medical Biological Engineering, Ottawa, Canada, 1976 Pilliar RM, Cameron HU, Welsh RP, Binnington AG: Radiographic and morphologic studies of load bearing porous surfaced structured implants. Clin Orthop 156:249, 1981 Pilliar RM, Lee JM, Manatopoulos C: Observation on the effect of movement on bone ingrowth into poroussurfaced implants. Clin Orthop 208:108, 1986 Rand JA, Bryan RS, Chao EYS, Ilstrup DM: A comparison of cemented versus cementless porous-coated anatomic total knee arthroplasty, p. 195. In: Proceedings of the Knee Society, 1985-1986. Aspen, Rockville, MD, 1987 Rosenqvist R, Bylander B, Knutson K et al: Loosening of the porous coating of bicompartmental prostheses in patients with rheumatoid arthritis. J Bone Joint Surg 68A:538, 1986 Ryd L: Micrumotion in knee arthroplasty. Acta Ortho Scand 57:Suppl, 1986 Spector M: Bone ingrowth into porous polymers, p. 55. In Williams DF (ed): Biocompatibility of orthopedic implants. Vol. 2. CRC Press, Boca Raton, FL, 1982. Stulberg SD, Stulberg BN: The biologic response to uncemented total knee replacements. In: Proceedings of the Knee Society, 1985-1986. Aspen, Rockville, MD, 1986 Stulberg SD, Stulberg BN, Hendrix R et al: The relationship of interface radiographs and the quality of fLxation for porous coated uncemented tibial components. Onhop Trans 11:534, 1987