Joint Line Elevation in Revision TKA Leads to Increased PatellofemoralContact Forces

Christian Konig, Alexey Sharenkov, Georg Matziolis, William R. Taylor, Carsten Perka, Georg N. Duda, Markus O. Heller

Julius Wolff Institut and Center for Musculoskeletal Surgery, Charite—Universitatsmedizin Berlin, Augustenburger Platz 1, D-13353 Berlin, Germany

Received 15 January 2009; accepted 11 June 2009

Published online 27 July 2009 in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jor.20952

ABSTRACT: One difficulty in revision total knee arthroplasty (TKA) is the management of distal femoral bone defects in which a joint lineelevation (JLE) is likely to occur. Although JLE has been associated with inferior clinical results, the effect that an elevated joint line has onknee contact forces has not been investigated. To understand the clinical observations and elaborate the potential risk associated with a JLE,we performed a virtual TKA on the musculoskeletal models of four subjects. Tibio- and patellofemoral joint contact forces (JCF) werecalculated for walking and stair climbing, varying the location of the joint line. An elevation of the joint line primarily affected thepatellofemoral joint with JCF increases of as much as 60% of the patient’s body weight (BW) at 10-mm JLE and 90% BW at 15-mm JLE, whilethe largest increase in tibiofemoral JCF was only 14% BW. This data demonstrates the importance of restoring the joint line, as it plays acritical role for the magnitudes of the JCFs, particularly for the patellofemoral joint. JLE caused by managing distal femoral defects withdownsizing and proximalizing the femoral component could increase the patellofemoral contact forces, and may be a contributing factor topostoperative complications such as pain, polyethylene wear, and limited function. � 2009 Orthopaedic Research Society. Published by

Wiley Periodicals, Inc. J Orthop Res 28:1–5, 2010

Keywords: knee; TKA; joint line; biomechanics; joint contact forces

The influence of an elevated joint line on the post-operative outcome of total knee arthroplasty (TKA) isdiscussed controversially. Although several studieshave found no correlation between joint line elevationand clinical outcome,1–3 others have linked an elevatedjoint line to inferior clinical and functional results.4–9 Inparticular, it was found that joint line elevation can beassociated with patellofemoral problems,4 lower kneescores,5–7 limited knee flexion capability,8 and mid-flexion joint instability9—all factors that can lead torevision surgery. Even though restoration of the jointline in primary and revision TKA seems to be possiblewithin an accuracy of 5 mm,10–12 recent studies haveshown that despite the possible clinical complicationsassociated with a joint line elevation, the joint lineremains elevated by more than 5 mm in over 36% ofrevised TKAs.7,13 Besides difficulties in estimating thephysiological location of the joint line13 and an overlycautious tibial resection,13,14 a common reason for anelevated joint line is the way in which distal femoralbone loss is managed in revision situations.1,7,13 Whilethe joint line can be restored using femoral aug-ments,1,13 this procedure produces additional implantcosts for the augments, as well as for the associatedstem. Addressing distal and posterior femoral bone losswith a proximalization and undersizing of the femoralcomponent is therefore common.1,15 However, thenecessity to fill the resulting larger flexion or extensiongaps with a thicker insert leads to an elevated jointline.7,13

Until now, the effect of an either intended orunintended joint line elevation on knee biomechanics,and in particular on the loading conditions at the joint,has received little interest. Increased joint loading isthought to have an influence on the development of

osteoarthritis,16 and has been related to pain17–20 as wellas to the wear rate of polyethylene inserts21 after TKA.Altered contact forces that are induced by changing thejoint line position in TKA are therefore likely to have aneffect on the postoperative outcome. In the broadspectrum of human motion, these altered contact forceswould have the largest effect during activities which arephysically challenging and frequently performed. Stairclimbing is such an activity that already exposes the kneeto considerable contact forces.22–24 Any further increaseto the contact forces may lead to overloading conditionsin the joint, possibly contributing to the increasedincidence of post-TKA patellofemoral pain during thisactivity.25 In addition, prevalent daily activities such asnormal walking26 may be subject to the risk of perma-nent overloading, which can increase polyethylene wearand contribute to aseptic loosening of the implants.

Understanding the biomechanical situation afterjoint line elevation in walking and stair climbingactivities is an important step towards comprehendingthe role that restoration of the joint line has on theclinical outcome, and for determining whether surgicalprocedures that cause a joint line elevation should beavoided. The aim of this study was therefore to analyzethe effects that an elevation of the joint line has onthe tibio- and patellofemoral joint contact forces duringstair climbing and normal walking.

METHODSMusculoskeletal ModelTo calculate joint contact forces, we used a previously validatedmusculoskeletal model of the human lower limb.27–29 Loca-tions of the joint contact forces and muscle attachments weredetermined by scaling a computer model of the human lowerlimb to match each subject’s anatomy. Muscles were modeledas straight lines spanning between the origin and insertion,and bony wrapping points were introduced where necessary torepresent a more realistic curved path of the muscle. Thephysiological cross-sectional area of each muscle was obtained

JOURNAL OF ORTHOPAEDIC RESEARCH JANUARY 2010 1

Correspondence to: Georg N. Duda (T: þ49 30 450 559048; F: þ4930 450 559969; E-mail: georg.duda@charite.de)

� 2009 Orthopaedic Research Society. Published by Wiley Periodicals, Inc.

from the literature30,31 and scaled to the patients’ bodyweights.29 Intersegmental resultant forces were calculatedusing an inverse dynamics approach, using measured groundreaction forces and gait data from four subjects. Since trackingof the patella was not possible during gait analysis, the motionof the patella was derived from an in vitro simulation of acomplete flexion–extension cycle32 of an intact knee joint andintegrated into the musculoskeletal model. To calculate thejoint contact forces, quasi-static optimization was performedwith the goal of minimizing the sum of the square of the musclestresses.22 The sum of all muscle forces acting on a joint,together with the intersegmental resultant forces from theinverse dynamics analysis, defined the joint contact forces.Specifically at the patella, the patellofemoral joint contactforces were the sum of the forces of the patellar ligament andthe muscle forces that were calculated for the vastus medialis,vastus lateralis, vastus intermedius, and the rectus femoris.Instrumented femoral prostheses in the same four subjectsgave access to in vivo hip contact forces during normal walkingand stair climbing activities, allowing the validation of thecalculated joint contact forces at the hip.27–29

A TKA with an ultra-congruent, fixed-bearing, and cruciate-sacrificing implant design (Columbus UC, Aesculap AG,Tuttlingen, Germany) was virtually performed on the muscu-loskeletal models of the same four subjects to simulate thepostoperative anatomy and allow surgical variation to beinvestigated. CAD models of the tibial and femoral componentswere orientated in the sagittal, coronal, and axial planesaccording to standard surgical procedures (Fig. 1) using avisualization and volume modeling software (AMIRA, MercuryComputer Systems Inc., Chelmsford, MA). The correct sizingand positioning of the components was supervised by anexperienced orthopedic surgeon (G. M.).

Kinematic AdaptationIn the original study, a clinical gait analysis was performed tocapture the lower limb kinematics for walking and stairclimbing activities.29 To adapt this kinematic data to a post-TKA situation, a generic kinematic model of tibiofemoralmotion based on the geometry of the ultra-congruent prosthe-sis’ articulating surfaces was developed in the sagittal plane.

Tibiofemoral kinematics were modeled using three flexionangle-dependent rotation axes. The locations of these axes were

determined by creating sagittal cross-sections of the femoralcomponent in the center of each condyle. Three arcs were fittedin each of the medial and lateral articulating contours (NX3,Siemens PLM Software, Cologne, Germany) (Fig. 2). The axisconnecting the corresponding medial and lateral centers oftheses arcs defined the flexion axes. The first axis defined thecomponent rotation from hyperextension up to 178knee flexion,at which point the radius of the second took over from 178 to 528,and likewise the third from 528 to maximum flexion. Todetermine the complete tibiofemoral kinematics during walk-ing and stair climbing, this kinematic model was then drivenusing the knee flexion angle determined from the gait analysisdata. For both activities, adaptations in tibiofemoral kine-matics were reflected in the hip joint, whereas the ankle jointwas not affected.

The virtual TKA aimed to provide an optimal anterior–posterior position of the femoral component and to avoid varus/valgus or internal/external rotation errors. Under theseconditions, the trochlear groove of the femoral component witha sulcus angle of 78 was considered to closely approximate thenatural trochlea. The patella kinematics of the intact knee wastherefore considered to provide a good first approximation of thepostoperative situation.

Variation of the Joint LineLoss of bone stock in the distal femur, a too-small tibialresection, or the use of a too-thick inlay, all situations resultingin joint line elevation, were simulated by modifying thecranial-caudal position of the femoral and tibial prosthesiscomponent and the size of the inlay. The anatomical femoraland tibial axes were used as a reference for respective com-ponent translations. A joint line distalization was alsosimulated by distalizing both the femoral and tibial compo-nents. Leg length was not altered in the process.

The joint line was distalized 5 mm from its anatomicallocation, as well as 10 and 15 mm elevated (Fig. 3). The length of

Figure 1. Joint contact forces after total knee arthroplasty werecalculated using a validated musculoskeletal model for instances ofstair climbing (displayed) and normal walking.

Figure 2. Cross-sectional view of a femoral component (Colum-bus UC, Aesculap AG, Tuttlingen, Germany). Centers of circles thatwere fitted in the articulating surfaces on the medial and lateralcondyle defined the flexion axes of the femoral component.

2 KONIG ET AL.

JOURNAL OF ORTHOPAEDIC RESEARCH JANUARY 2010

the ligamentum patellae remained unchanged, thus creating apseudo patella baja in the case of an elevated joint line.14

Joint Contact ForcesPatellofemoral and tibiofemoral joint contact forces werecalculated for repetitive trials of normal walking and stairclimbing for all four subjects, and all joint line variations,using inverse dynamics and optimization algorithms incorpo-rated in the musculoskeletal model.29,33 Using the condition ofan anatomically reconstructed joint line as a reference for thesituations with an altered joint line, the time point at whichthe peak contact force occurred was identified within eachactivity cycle and the maximal deviations in contact force weredetermined for all subjects for all altered joint line levels. AStudent’s t-test was performed to verify the reproducibility ofthe calculated contact force deviations within the repetitivetrials of each individual subject. Significance was assumed atp< 0.05.

RESULTSElevating the joint line increased the contact forces inboth the tibio- and patellofemoral joints in almost allcases (Figs. 4 and 5). The patellofemoral joint was moreaffected than the tibiofemoral joint, and stair climbingcaused a larger increase in contact force than walking inboth joints.

A joint line elevation of 10 mm caused an increase inpatellofemoral joint contact force of 60% of the subject’sbody weight (BW) during stair climbing, and 30% BW

during normal walking. At the tibiofemoral joint, only aslight increase of less than 10% BW was observed in bothactivities. A further elevation of the joint line from 10 to15 mm only minimally affected the tibiofemoral joint ineither activity, with an additional increase in contactforce of less than 6% BW. In the patellofemoral joint,however, this additional joint line elevation furtherincreased the contact forces by about 30% BW duringstair climbing, resulting in a total increase of 90% BWcompared to the anatomically reconstructed joint line.During normal walking, the additional joint line eleva-tion increased the patellofemoral contact forces by about10% BW, resulting in a total increase of 40% BW.

Distalizing the joint line by 5 mm reduced the contactforces in both the tibio- and patellofemoral joints. Thetibiofemoral contact forces during normal walking andstair climbing, however, were only minimally affected asthey were reduced by less than 6% BW. A larger effectwas seen in the patellofemoral joint, where a 5-mm jointline distalization resulted in a reduction of contact forcesof 11% BW during normal walking and up to 30% BWduring the stair climbing activity.

At the same joint line levels, there were consistentfindings for contact forces in all repetitive trials of theindividual subjects and between the individual subjects(p< 0.05).

DISCUSSIONRecent studies have shown that in more than 36% of allrevision TKAs, the joint line remains elevated by morethan 5 mm.7,13 Although clinical observations haveindicated that an elevation of the joint line is associatedwith inferior clinical results,4–7,13 the effects of joint lineelevation on knee biomechanics remains relativelyunknown. However, this knowledge is essential forunderstanding the observed inferior clinical results. Forthe first time, to our knowledge, we have shown thatelevating the joint line can lead to an increase in jointcontact forces in both the tibio- and patellofemoral jointsduring walking and stair climbing, activities of daily

Figure 3. The joint line level was simulated between a 5-mmdistalization (left), the anatomical reconstruction (center), and up to15-mm elevation (right).

Figure 4. Exemplary patellofemoral joint contact force patternduring a cycle of normal walking calculated for one patient (thicksolid line). The contact forces were recalculated after simulating ajoint line elevation of 10 and 15 mm (dashed lines), and a joint linedistalization of 5 mm (thin solid line).

Figure 5. The average changes and the respective standarddeviations in patellofemoral and tibiofemoral contact forces of foursubjects, caused by a distalization (green bars) or elevation (yellowand red bars) of the joint line, are shown in multiples of body weight(BW) compared to the contact forces in the situation of ananatomically reconstructed joint line.

JOINT LINE RECONSTRUCTION IN TKA 3

JOURNAL OF ORTHOPAEDIC RESEARCH JANUARY 2010

living that are prevalent26 and known to expose theknee to considerable internal loading.22–24

The patellofemoral joint was particularly affectedduring stair climbing activities, in which the contactforces increased by more than half of a patient’s bodyweight (60% BW) at 10 mm of joint line elevation, adistance that can commonly occur in revision surgery.5

An even more pronounced increase in contact forces wasobserved when the joint line was further elevated to15 mm: During stair climbing, our calculations showedpatellofemoral joint contact forces of up to 90% BW morethan the contact forces that occur in an anatomicallyreconstructed knee. Thus, a patient of 85 kg couldexperience an increase in patellofemoral contact forcesof up to 750 N. Increased forces at the patellofemoral jointwere also observed during walking, although the contactforce increase was always less than during stair climb-ing. However, due to the prevalence of walking,26 eventhe 30% BW increase calculated at 10-mm joint lineelevation further contributes towards a considerableand constant overloading situation in the load-bearingstructures. Overloading of the patellofemoral joint canincrease the risk of patellar degeneration, and can alsoput a patellar replacement at risk of early failure due toan increased wear rate of the polyethylene insert.21

Overloading conditions have also been associatedwith pain in several studies.17–20 While a number ofother factors such as instability or patellar overstuffingcan also contribute to anterior knee pain, the results ofour analyses indicate that the clinically observedcomplications in patients with an elevated joint line5–9

may well be linked to the increase in patellofemoralcontact forces demonstrated in this study. Specifically,Porteous and co-workers7 reported that patients with ajoint line elevated by more than 5 mm had a lower BristolKnee Score, a score that has a high sensitivity to painlevels. A study by Figgie and co-workers6 showed thatjoint line elevation correlated with lower functional kneescores, patellofemoral pain, and the need for revision.Avoiding an elevation of the joint line, and thus avoidingan overloading of the load-bearing structures in theknee, may therefore be crucial for further improving thepostoperative outcome in revision TKA.

Even though direct measurements of in vivo loadingconditions in the knee have been performed previ-ously,34–38 a study of anatomical variation such as theelevation or distalization of the joint line in an individualis, however, not possible in vivo. The approach used inthis study utilizing validated musculoskeletal models isan established procedure to gain information about theloading conditions in structures that are hardly acces-sible for in vivo measurements and also to investigate theeffects of anatomical variations.39,40

Several studies have investigated tibiofemoral kine-matics after TKA.41–44 However, no general kinematicpatterns have been identified so far. Dennis et al.43,44

showed that postoperative kinematics depend upon theimplant type used and on the surgeon performing thesurgery. To allow standardization and comparative

results, a geometric approach was chosen for this studyto derive a generic kinematic pattern for the ultra-congruent, fixed-bearing prosthesis design. While thismodeling approach of describing tibiofemoral kinematicsin the sagittal plane reflects the guiding characteristicsof this specific prosthesis design, a certain variation ofthis generic kinematic pattern is likely to occur in vivo.44

The sensitivity of the patellofemoral joint contactforces to the joint line location, as shown in this study,underlines the importance of carefully considering jointline reconstruction, especially in revision TKA. Theresults from this study indicate that an elevated jointline, creating a pseudo patella baja,14 can expose thepatient’s patellofemoral joint to unphysiologically highcontact forces and thus expose the patient to a high risk ofdeveloping clinical and functional problems. Orthopedicsurgeons should be aware of the procedures in TKA thatmay lead to an elevated joint line.1,6,15 Our data suggeststhat distal femoral defects should be addressed withaugmentation,1 rather than proximalizing and under-sizing the femoral component using a thicker tibialinsert. Since errors in joint line reconstruction have asignificant impact on joint contact forces, the correctlevel of the tibial resection13,14 should be verified usingrobust methods to identify the location of the naturaljoint line.11,45 Especially in revision situations, naviga-tion systems could be of great benefit when assisting thesurgeon to restore the natural joint line.

In conclusion, the contact forces in the knee, partic-ularly in the patellofemoral joint, are sensitive to jointline reconstruction. Elevating the joint line in revisionTKA can increase the joint contact forces and shouldtherefore be avoided, where possible, in order to mini-mize the risk of biomechanically induced postoperativecomplications.

ACKNOWLEDGMENTSThis study was supported by a grant from the GermanResearch Foundation (DFG SFB 760) and by Aesculap AG,Tuttlingen, Germany.

REFERENCES1. Bellemans J. 2004. Restoring the joint line in revision TKA:

does it matter? Knee 1:3–5.2. Ritter MA, Montgomery TJ, Zhou H, et al. 1999. The clinical

significance of proximal tibial resection level in total kneearthroplasty. Clin Orthop Relat Res 360:174–181.

3. Selvarajah E, Hooper G. 2008. Restoration of the joint line intotal knee arthroplasty. J Arthroplasty (in press).

4. Laskin RS. 1998. Management of the patella during revisiontotal knee replacement arthroplasty. Orthop Clin North Am2:355–360.

5. Partington PF, Sawhney J, Rorabeck CH, et al. 1999. Jointline restoration after revision total knee arthroplasty. ClinOrthop Relat Res 367:165–171.

6. Figgie HE III, Goldberg VM, Heiple KG, et al. 1986. Theinfluence of tibial-patellofemoral location on function of theknee in patients with the posterior stabilized condylarknee prosthesis. J Bone Joint Surg [Am] 7:1035–1040.

7. Porteous AJ, Hassaballa MA, Newman JH. 2008. Does thejoint line matter in revision total knee replacement? J BoneJoint Surg [Br] 7:879–884.

4 KONIG ET AL.

JOURNAL OF ORTHOPAEDIC RESEARCH JANUARY 2010

8. Chiu KY, Ng TP, Tang WM, et al. 2002. Knee flexion aftertotal knee arthroplasty. J Orthop Surg (Hong Kong) 2:194–202.

9. Martin JW, Whiteside LA. 1990. The influence of joint lineposition on knee stability after condylar knee arthroplasty.Clin Orthop Relat Res 259:146–156.

10. Wyss TF, Schuster AJ, Munger P, et al. 2006. Does total kneejoint replacement with the soft tissue balancing surgicaltechnique maintain the natural joint line? Arch OrthopTrauma Surg 7:480–486.

11. Rouvillain JL, Pascal-Mousselard H, Favuto M, et al. 2008.The level of the joint line after cruciate-retaining total kneereplacement: a new coordinate system. Knee 1:31–35.

12. Mahoney OM, Kinsey TL. 2006. Modular femoral offset stemsfacilitate joint line restoration in revision knee arthroplasty.Clin Orthop Relat Res 446:93–98.

13. Laskin RS. 2002. Joint line position restoration duringrevision total knee replacement. Clin Orthop Relat Res 404:169–171.

14. Grelsamer RP. 2002. Patella baja after total knee arthro-plasty: is it really patella baja? J Arthroplasty 1:66–69.

15. Scuderi GR, Insall JN. 1992. Total knee arthroplasty. Currentclinical perspectives. Clin Orthop Relat Res 276:26–32.

16. Wu JZ, Herzog W, Epstein M. 2000. Joint contact mechanicsin the early stages of osteoarthritis. Med Eng Phys 1:1–12.

17. Dye SF. 2005. The pathophysiology of patellofemoral pain: atissue homeostasis perspective. Clin Orthop Relat Res 436:100–110.

18. Grana WA, Kriegshauser LA. 1985. Scientific basis ofextensor mechanism disorders. Clin Sports Med 2:247–257.

19. Heino Brechter J, Powers CM. 2002. Patellofemoral stressduring walking in persons with and without patellofemoralpain. Med Sci Sports Exerc 10:1582–1593.

20. Dye SF, Vaupel GL, Dye CC. 1998. Conscious neurosensorymapping of the internal structures of the human kneewithout intraarticular anesthesia. Am J Sports Med 6:773–777.

21. Estupinan JA, Bartel DL, Wright TM. 1998. Residual stressesin ultra-high molecular weight polyethylene loaded cyclicallyby a rigid moving indenter in nonconforming geometries.J Orthop Res 1:80–88.

22. Taylor WR, Heller MO, Bergmann G, et al. 2004. Tibio-femoral loading during human gait and stair climbing.J Orthop Res 3:625–632.

23. Costigan PA, Deluzio KJ, Wyss UP. 2002. Knee and hipkinetics during normal stair climbing. Gait Posture 1:31–37.

24. D’Lima DD, Patil S, Steklov N, et al. 2005. The ChitranjanRanawat Award: In vivo knee forces after total kneearthroplasty. Clin Orthop Relat Res 440:45–49.

25. Mayman D, Bourne RB, Rorabeck CH, et al. 2003. Resurfacingversus not resurfacing the patella in total knee arthroplasty:8- to 10-year results. J Arthroplasty 5:541–545.

26. Morlock M, Schneider E, Bluhm A, et al. 2001. Duration andfrequency of every day activities in total hip patients.J Biomech 7:873–881.

27. Bergmann G, Deuretzbacher G, Heller M, et al. 2001. Hipcontact forces and gait patterns from routine activities.J Biomech 7:859–871.

28. Bergmann G, Graichen F, Siraky J, et al. 1988. Multichannelstrain gauge telemetry for orthopaedic implants. J Biomech2:169–176.

29. Heller MO, Bergmann G, Deuretzbacher G, et al. 2001.Musculo-skeletal loading conditions at the hip during walkingand stair climbing. J Biomech 7:883–893.

30. Brand RA, Pedersen DR, Friederich JA. 1986. The sensitivityof muscle force predictions to changes in physiologic cross-sectional area. J Biomech 8:589–596.

31. Duda GN, Brand D, Freitag S, et al. 1996. Variability offemoral muscle attachments. J Biomech 9:1185–1190.

32. Durselen L, Claes L, Kiefer H. 1995. The influence of muscleforces and external loads on cruciate ligament strain. Am JSports Med 1:129–136.

33. Heller MO, Matziolis G, Konig C, et al. 2007. Musculoskeletalbiomechanics of the knee joint: principles of preoperativeplanning for osteotomy and joint replacement. Orthopade7:628–634.

34. Wallace AL, Harris ML, Walsh WR, et al. 1998. Intraoperativeassessment of tibiofemoral contact stresses in total kneearthroplasty. J Arthroplasty 8:923–927.

35. D’Lima DD, Townsend CP, Arms SW, et al. 2005. Animplantable telemetry device to measure intra-articular tibialforces. J Biomech 2:299–304.

36. Zhao D, Banks SA, D’Lima DD, et al. 2007. In vivo medial andlateral tibial loads during dynamic and high flexion activities.J Orthop Res 5:593–602.

37. Heinlein B, Graichen F, Bender A, et al. 2007. Design,calibration and pre-clinical testing of an instrumented tibialtray. J Biomech 40(Suppl 1):4–10.

38. Wasielewski RC, Galat DD, Komistek RD. 2005. Correlationof compartment pressure data from an intraoperative sensingdevice with postoperative fluoroscopic kinematic results inTKA patients. J Biomech 2:333–339.

39. Heller MO, Taylor WR, Perka C, et al. 2003. The influence ofalignment on the musculo-skeletal loading conditions at theknee. Langenbecks Arch Surg 5:291–297.

40. Heller MO, Bergmann G, Deuretzbacher G, et al. 2001.Influence of femoral anteversion on proximal femoral loading:measurement and simulation in four patients. Clin Biomech(Bristol, Avon) 8:644–649.

41. Zihlmann MS, Gerber H, Stacoff A, et al. 2006. Three-dimensional kinematics and kinetics of total knee arthro-plasty during level walking using single plane video-fluoro-scopy and force plates: a pilot study. Gait Posture 4:475–481.

42. Nilsson KG, Karrholm J, Ekelund L. 1990. Knee motion intotal knee arthroplasty. A roentgen stereophotogrammetricanalysis of the kinematics of the Tricon-M knee prosthesis.Clin Orthop Relat Res 256:147–161.

43. Dennis DA, Komistek RD, Mahfouz MR, et al. 2003. Multi-center determination of in vivo kinematics after total kneearthroplasty. Clin Orthop 416:37–57.

44. Dennis DA, Komistek RD, Mahfouz MR, et al. 2004. Amulticenter analysis of axial femorotibial rotation after totalknee arthroplasty. Clin Orthop Relat Res 428:180–189.

45. Mason M, Belisle A, Bonutti P, et al. 2006. An accurate andreproducible method for locating the joint line during arevision total knee arthroplasty. J Arthroplasty 8:1147–1153.

JOINT LINE RECONSTRUCTION IN TKA 5

JOURNAL OF ORTHOPAEDIC RESEARCH JANUARY 2010