From thSchool of M

SubmitThe Co

found at dReprint

Corporate© 20120883-5doi:10.1

The Journal of Arthroplasty Vol. 27 No. 4 2012

Micromotion of Cementless Tibial BaseplatesUnder Physiological Loading Conditions

Safia Bhimji, MS,* and R. Michael Meneghini, MDy

Abstract: Initial implant stability is crucial to cementless knee arthroplasty success. The objectiveof this study was to develop a physiological relevant methodology that incorporates torsion, shear,and compression forces to evaluate two tibial component designs that feature either a keel orcylindrical porous metal pegs. The data were compared with a simplified compression loadingscenario. Results show a loading profile that combines compressive, shear, and torsional loadsresults in significantly larger motions than occur when loading in compression only. Whencomparing between a keeled and a pegged device, the new method shows significant differences intibial component subsidence/liftoff at the anterior and posterior locations, which were lacking inthe simplified test model. To accurately assess implant stability, studies should use physiologicalrelevant loading. Keywords: micromotion, stability, cementless baseplates, total knee arthroplasty,pegs, keels.© 2012 Elsevier Inc. All rights reserved.

Cementless fixation is undergoing renewed interest intotal knee arthroplasty due, in part, to the evolution ofhighly porous metals and the use of total kneearthroplasty in younger patient populations. However,the optimal design remains elusive. To obtain adequatebone ingrowth and fixation, cementless tibial baseplatesmust exhibit a sufficient amount of initial stability afterimplantation. Previous biomechanical studies haveaimed at quantifying initial stability using simplifiedloading scenarios, yet a consistent and accepted loadingscenario or experimental setup has remained undeter-mined. Some have applied low-cyclic compressive loadsto either the medial or lateral condyles, while subsidenceand liftoff are measured at those respective aspects of thebaseplate [1-4]. Others have applied the compressiveload to the anterolateral or posteromedial compartmentsand measured subsidence and liftoff around the periph-ery of the baseplate [5-8]. A few investigators haveincorporated anterior shear loads or torsional loads inaddition to the compressive load while measuring shear

e *Stryker Orthopaedics, Mahwah, New Jersey; and yIndianaedicine, Indianapolis, Indiana.ted September 16, 2010; accepted June 12, 2011.nflict of Interest statement associated with this article can beoi:10.1016/j.arth.2011.06.010.requests: Safia Bhimji, MS, Stryker Orthopaedics, 325Dr., Mahwah, NJ 07430.Elsevier Inc. All rights reserved.

403/2704-0025$36.00/0016/j.arth.2011.06.010

648

motions at the medial and lateral edges [1,7,9-12].Furthermore, the number of loading cycles reported inthe various studies have varied from a few hundred[1,2,5-7,9,10,13,14] to 300 000 [4].A limitation across all these studies is that the loading

was applied directly to the baseplate, as opposed tosimulating the physiological loading that occurs from afemoral component articulating on an insert. In addi-tion, the load profiles were likely oversimplified and didnot accurately reflect the profiles of daily activities. Theobjectives of this study are to develop a method tomeasure the micromotion of cementless tibial compo-nents during stair descent loading and to use thismethodology to evaluate implants with 2 types offixation features: A full keel currently in clinical use,and two generic 0.5-in diameter cylindrical pegs. Resultsare compared with those collected under a typicalsimplified loading protocol incorporating isolated com-pression loads.

Materials and MethodsTo minimize the interspecimen variability associated

with cadaveric specimens, sawbones mechanical testingmaterial (Pacific Research, Vashon, Wash) was used asthe host material for specimen testing in this study. Toreplicate the proximal tibia, the composite polyurethanefoam constructs were manufactured into a shapematching that of a 9-mm-depth resection plane of amedium-sized tibial sawbones specimen. Two densitiesof foam were incorporated into the polyurethanespecimens, which included an inner core of a 12.5-pcf

Fig. 2. Test setup.

Micromotion of Cementless Tibial Baseplates � Bhimji and Meneghini 649

(pounds per cubic foot) cellular foam to simulatecancellous bone and an outer 2.5-mm ring of higherdensity 40-pcf closed-cell foam to simulate the proximaltibial cortical rim. The 12.5-pcf foam was chosen basedon feedback from a panel of orthopedic surgeonsindicating that it looked and felt most like cancellousbone. Also, the compressive modulus of the foam fallswithin the range of moduli reported for the natural tibiaby Goldstein et al [15]. This foam is softer than thehigher density closed-cell version typically found in atibial sawbones, which the surgeons felt was too stiff,requiring excessive force to impact a baseplate ascompared with natural bone. The 40-pcf foam waschosen for the cortical rim because it was the highestdensity foam readily available from the manufacturer.The thickness of the rim was also chosen based onsurgeon feedback and is thinner than that found in acomposite tibial sawbone. The foam densities andconfigurations chosen for the current study are thoughtto offer a better representation of the natural tibia andyield more clinically relevant micromotion levels.The tibial components were positioned such that each

baseplate had cortical bone support under the anteriorrim and lateral posterior condyle and had cancellousbone support under the posterior medial condyle(Fig. 1). The sawbones specimens designated for thefull keel baseplates were prepped by milling out theprofile of the tibial punch tool into the cancellous core.This yielded a complete press-fit of the lower half of thekeel into the foam and a minimal press-fit of the upperhalf (∼0.3 mm), which matches the amount of press-fitcurrently used clinically. Tibial specimens for thecylindrical pegged baseplates were prepped by drillingtwo 0.5-in diameter holes into the cancellous boneanalog, yielding a line-to-line fit of the pegs into thefoam. This was chosen to represent a worst-case press-fitof the peg devices currently in use.

Fig. 1. Alignment of baseplate on foam constructs.

Four tibial components of each type were prepared fortesting by attaching 0.375-in diameter spheres to themedial, lateral, anterior, and posterior notch aspects oftheir rims. The anterior, medial, and lateral spheres weremounted via 1-in length dowel pins, and the posteriorsphere via a 1.5-in length dowel pin. Each tibialbaseplate was then inserted into its respective foamspecimen via the established surgical protocol. Based onpower analyses performed on preliminary data, asample size of 4 was chosen to achieve an 80% powerlevel for a micromotion effect size of 300 μm at thesphere locations.Six LVDTs (linear variable differential transformer)

were mounted to each foam construct 1 cm below thetop surface and arranged so that the plungers restedagainst the spheres attached to the tibial implant (Fig. 2).One LVDT was oriented superoinferiorly against theanterior sphere, and another LVDT against the posteriorsphere, to measure subsidence or liftoff of the baseplate.Two LVDTs were oriented against the medial sphere,and two against the lateral sphere, with one LVDT

Fig. 3. Stair-descent loading profile.

650 The Journal of Arthroplasty Vol. 27 No. 4 April 2012

oriented superoinferiorly and one oriented anteropos-teriorly (A/P) on each. These were used to measuresubsidence/liftoff and anterior/posterior motion of thebaseplate in the respective directions.

Fig. 4. Total micromot

The components were mounted to the anterior/posterior actuator of a servohydraulic test machine viaan x-y table, and a 0.375-in load applicator was mountedto the axial actuator (Fig. 2). A simplified loading profilesimilar to that used in previous studies was initiallyapplied to each baseplate [1-3]. The profile consisted of acompressive load applied to the posterior third of thelateral condyle. The load was cycled from 115 to 1150 Nat 2 Hz for 100 cycles. Loading was then switched to themedial condyle, and the test was repeated. Peak-to-peakmotion (ie, recoverable motion) from each LVDTwas calculated at every 10 cycles for each sample andthen averaged over the 100 cycles of the test. The datawere then averaged across samples for each loadcondition (medial loading vs lateral loading) of eachbaseplate design.A femoral component was then mounted to the

axial actuator of the test machine using a fixture thatallowed varus/valgus rotation to be unrestricted andthe flexion angle to be locked. The flexion angle wasfixed at 72°, which represents an average angle atwhich the peak A/P shear load occurs at a minimalcompressive load, increasing the potential for rockingmotion between the baseplate and simulated bone.This angle is also deep enough to ensure cam/postengagement on both designs tested. Posterior-stabi-lized polyethylene inserts of 16-mm thickness wereused, because they represent an average thicknesstypically used clinically. A loading profile representinga stair descent activity was applied to the baseplate viathe femoral component articulating on the polyethyl-ene insert. The profile was based on data published byBenson et al [16] and incorporated a peak compres-sive load of 2520 N. This magnitude is in line withcompressive load values reported by D'Lima et al [17].Internal and external rotation was applied throughangle control based on the angles generated when rununder the Benson et al[16] torque profile. The torque

ion—keeled device.

Fig. 5. Total micromotion—pegged device.

Micromotion of Cementless Tibial Baseplates � Bhimji and Meneghini 651

profile was approximately ±2 Nm, which yieldedrotational angles of approximately 6°. This is repre-sentative of the rotation seen in vivo at the kneeduring stair activities [18]. The final loading profilesfor the compressive load, anteroposterior load, andinternal and external rotation as a function of the stairdescent cycle are shown in Fig. 3.Loading was applied at a rate of 0.25 Hz for 10 000

cycles. This represents 6 to 8 weeks of a stair descentactivity [19], which is the approximate length of time tothe initiation of bone ingrowth [20,21]. Motions at eachof the six LVDTs were monitored throughout the test at arate of 25 Hz. The peak-to-peak motion from each LVDTwas calculated at every 1000 cycles for each sampletested. Preliminary testing showed that after the first1000 cycles, these magnitudes remained fairly constantthroughout the remainder of the test. Therefore, the

Fig. 6. Total micromotion dur

peak-to-peak data were averaged across the last 9000cycles and then averaged across samples of each design.Statistical analyses were used to compare results

between loading conditions within each device andbetween devices within each loading condition. For theformer comparison, a single-factor analysis of variancewas used, with α = .05 for each LVDT location. For thelatter comparison, an unpaired t test was used, with α =.05 for each LVDT location.

ResultsFig. 4 displays a comparison of the average peak-to-peak

motion recorded at each LVDT for the keeled device ineach of the three testing scenarios. Fig. 5 shows the samedata for the cylindrical pegged device. Results showedstatistically significant differences between motionsobtained under the single-condyle loading conditions as

ing medial condyle loading.

Fig. 7. Total micromotion during lateral condyle loading.

652 The Journal of Arthroplasty Vol. 27 No. 4 April 2012

comparedwith the stair descent loading condition. For thekeeled device, significant differences were seen at five ofthe six locations, with the largest occurring in thesuperoinferior direction at the posterior site, where themotions were 5 times greater under stair descent loading(∼490μm) than single-condyle loading (∼82μm). For thepegged device, significant differences were seen at alllocations, with the largest again occurring in the super-oinferior direction at the posterior site, where themotionswere as much as 9 times greater under stair descentloading (∼730μm) than single-condyle loading (∼80μm).Figs. 6-8 compare motions between devices under

medial condyle loading, lateral condyle loading, andstair descent loading, respectively. Results showed thatall motions at the medial and lateral sites are similar

Fig. 8. Total micromotion du

between the devices, regardless of which loadingprotocol was used. On the contrary, superoinferiormotions at the anterior sites were similar when testedunder the single-condyle loading conditions, but signif-icantly different when tested under stair descent loading.During stair descent, the pegged tibial implant demon-strated less stability through much larger motions thanthe keeled device (512 μm vs 132 μm, P = .003). Thesame trends were seen at the posterior site as well, withthe single-condyle loading conditions yielding similarresults between devices and the stair descent loadingyielding mean differences in motion of 728 μm for thepegged device and 487 μm for the keeled device. Thisdifference did not reach statistical significance with thenumbers available (P = .09).

ring stair descent loading.

Micromotion of Cementless Tibial Baseplates � Bhimji and Meneghini 653

DiscussionWith cementless knee fixation making a comeback

and enjoying renewed interest, reliable and clinicallyrelevant testing methodology to assist with implantdesign is critical, but this methodology is not yetconclusively established in the literature. The majorityof previous studies on initial stability of cementless tibialcomponents have used simplified conditions that typi-cally only load the medial and/or lateral condyles incompression [1-4]. A few studies added either ante-roposterior or torsional loads to the compressive loads[5-12]. Load magnitudes have varied from 1100 to2400 N in compression, 100 to 1000 N in anteroposteriorshear, and 10 to 40 Nm in torsion. Loading was typicallyapplied directly to the baseplate using a load applicator,as opposed to a femoral component that simulates invivo conditions.This study showed that when loading a baseplate via a

femoral component articulating on an insert, using aphysiologically relevant profile that encompasses com-pressive, shear, and torsional loads together, significantlylarger motions can be expected to occur. This can beexplained by the magnitudes of the loads applied, theirphasing, and the method of application. The magnitudesof the shear and torsional loads are similar to those usedin the literature, whereas the compressive loads arehigher, with the peak load exceeding 2500 N. Loadingthe baseplate through an insert increases rockingmoments reacted at the bone interface. Using a thickposterior-stabilized insert and testing at a flexion anglethat engages the femoral component cam with the postmaximizes these moments, creating a “worse casescenario” that enhances implant testing. Regardingphasing, the loading profile chosen represents a relative-ly high-load activity that applies the maximum shearforces to the tibial component at a low compressive load(at ∼60% gait cycle). This maximizes the potential forrocking motion of the tibial construct as the femoralcomponent imparts its translational force. The shear andtorsional loads also involve reverse loading, which canaid in loosening as the tibial construct is torqued back andforth in both the sagittal and transverse planes. Previousstudies typically phased their loads together and did notuse reverse loading, leading to lower motions.Our study showed that the 2-pegged tibial components

demonstrated greater micromotion than did the keeledimplants, suggesting that they are less resistant to thesephysiologic loads. Intuitively, such a result is plausiblebecause a 2-pegged device is more susceptible to rockingmotions than a keeled device, given the large surfacearea provided by a keel to resist that moment-inducedmotion. The long-term stability of cementless keeleddevices has been well documented in the literature[22-25], and the device tested here has demonstratedyears of clinical success, but the performance of peggeddevices remains to be determined. This finding is not

seen, however, when comparing the motions obtainedfor each device under the single-condyle loadingconditions. Instead, the motions are shown to be similarbetween the two, particularly in the superoinferiordirection at the anterior and posterior locations, wherelarge differences would be expected due to rocking.Because the single-condyle loading conditions do notsimulate the A/P translations of a femoral component onthe insert, along with the A/P forces created when thecam engages the post, the moments that cause rockingare not imparted to the baseplate, and these motions areundetected during this simplified testing method.There are limitations to this study. Polyurethane foam

constructs were used to simulate the tibia instead ofcadaveric specimen. Although they are not humanbone, the foam constructs minimize variability that isinherent in cadaveric specimen, allowing differencesbetween test methods and devices to be more preciselymeasured. Also, the densities of foam chosen to simulatecortical and cancellous bone were not validated, butinstead, chosen based on the recommendations oforthopedic surgeons who perform cementless kneearthroplasties. Cadaveric studies are currently in pro-gress using the same testing protocols in an effort tovalidate the foam constructs.A line-to-line fit was used on the pegged device to

represent a worst-case difference in micromotion ascompared with the keel device, which the single-condyleloading conditionwas not able to detect. Imposing a press-fit on the pegs may lead to different micromotion results,but that effect was not addressed here.Lastly, the micromotion levels measured in this study

are amplified due to the offset of the measurementlocations from the baseplates, making comparisons toclinical levels of motion difficult. However, the keeleddevice tested has demonstrated long-term clinicalsuccess, so the motions measured using this protocolserve as a baseline for acceptable micromotion levelsin vivo.In summary, the main objective of this study was to

evaluate a new methodology for micromotion testing,rather than to rigorously test for differences betweendesigns. Results have demonstrated that physiologicallyrelevant loading conditions can generate larger amountsof micromotion and better detect differences betweendevices than simplified loading scenarios. To accuratelyassess the potential for bone ingrowth, future studiesshould focus on these relevant loading conditions andassess the baseplates as part of a complete knee system.

AcknowledgmentsThe authors would like to thank Gregg Schmidig and

Laura Yanoso for their contribution and supportthroughout this study.

654 The Journal of Arthroplasty Vol. 27 No. 4 April 2012

References1. Kraemer WJ, Harrington IJ, Hearn TC. Micromotion

secondary to axial, torsional, and shear loads in twomodels of cementless tibial components. J Arthroplasty1995;10227.

2. Lee RW, Volz RG, Sheridan DC. The role of fixation andbone quality on the mechanical stability of tibial kneecomponents. Clin Orthop Relat Res 1991;273:177.

3. Peters CL, Craig MA, Mohr RA, et al. Tibial componentfixation with cement: full- versus surface-cementationtechniques. Clin Orthop Relat Res 2003;409:158.

4. Volz RG, Nisbet JK, Lee RW, et al. The mechanical stabilityof various noncemented tibial components. Clin OrthopRelat Res 1988;226:38.

5. Bert JM, McShane M. Is it necessary to cement the tibialstem in cemented total knee arthroplasty? Clin OrthopRelat Res 1998;356:73.

6. Lee TQ, Barnett SL, Kim WC. Effects of screw types incementless fixation of tibial tray implants: stability andstrength assessment. Clin Biomech (Bristol, Avon) 1999;14:258.

7. Miura H, Whiteside LA, Easley JC, et al. Effects of screwsand a sleeve on initial fixation in uncemented total kneetibial components. Clin Orthop Relat Res 1990;259:160.

8. Miyoshi S, Takahashi T, Ohtani M, et al. Analysis of theshape of the tibial tray in total knee arthroplasty using athree dimension finite element model. Clin Biomech(Bristol, Avon) 2002;17:521.

9. Matsuda S, Tanner MG, White SE, et al. Evaluation oftibial component fixation in specimens retrieved atautopsy. Clin Orthop Relat Res 1999;363:249.

10. Sala M, Taylor M, Tanner KE. Torsional stability of primarytotal knee replacement tibial prostheses: a biomechanicalstudy in cadaveric bone. J Arthroplasty 1999;14:610.

11. Walker PS, Greene D, Reilly D, et al. Fixation of tibialcomponents of knee prostheses. J Bone Joint Surg Am1981;63:258.

12. Walker PS, Hsu HP, Zimmerman RA. A comparativestudy of uncemented tibial components. J Arthroplasty1990;5:245.

13. Andriacchi TP, Andersson GB, Fermier RW, et al. A studyof lower-limb mechanics during stair-climbing. J BoneJoint Surg Am 1980;62:749.

14. Dempsey AJ, Finlay JB, Bourne RB, et al. Stability andanchorage considerations for cementless tibial compo-nents. J Arthroplasty 1989;4:223.

15. Goldstein SA, Wilson DL, Sonstegard DA, et al. Themechanical properties of human tibial trabecular bone as afunction of metaphyseal location. J Biomech 1983;16:965.

16. Benson LC, DesJardins JD, Harman MK, et al. Effect ofstair descent loading on ultra-high molecular weightpolyethylene wear in a force-controlled knee simulator.Proc Inst Mech Eng [H ] 2001;216:409.

17. D'Lima DD, Patil S, Steklov N, et al. Tibial forces measuredin vivo after total knee arthroplasty. J Arthroplasty 2006;21:255.

18. Costigan PA, Deluzio KJ, Wyss UP. Knee and hip kineticsduring normal stair climbing. Gait Posture 2002;16:31.

19. Styles CM, Evans SL, Gregson PJ. Development of fatiguelifetime predictive test methods for hip implants: part I.Test methodology. Biomaterials 1998;19:1057.

20. Bobyn JD, Pilliar RM, Cameron HU, et al. The optimumpore size for the fixation of porous-surfaced metalimplants by the ingrowth of bone. Clin Orthop Relat Res1980;150:263.

21. Jasty M, Bragdon C, Burke D, et al. In vivo skeletalresponses to porous-surfaced implants subjected to smallinduced motions. J Bone Joint Surg Am 1997;79:707.

22. Buechel Sr FF, Buechel Jr FF, Pappas MJ, et al. Twenty-year evaluation of the New Jersey LCS Rotating PlatformKnee Replacement. J Knee Surg 2002;15:84.

23. Epinette JA, Manley MT. Hydroxyapatite-coated totalknee replacement: clinical experience at 10 to 15 years.J Bone Joint Surg Br 2007;89:34.

24. Hofmann AA, Evanich JD, Ferguson RP, et al. Ten- to14-year clinical followup of the cementless natural kneesystem. Clin Orthop Relat Res 2001;388:85.

25. Ritter MA, Meneghini RM. Twenty-year survivorship ofcementless anatomic graduated component total kneearthroplasty. J Arthroplasty 2010;25:507.