101

Design forms of total knee replacement

P S Walker* and S Sathasivam

Department of Biomedical Engineering, University College London and Royal National Orthopaedic Hospital Trust,
Stanmore, Middlesex, UK

Abstract: The starting point of this article is a general design criterion applicable to all types of
total knee replacement. This criterion is then expanded upon to provide more specifics of the required
kinematics, and the forces which the total knee must sustain. A characteristic which differentiates
total knees is the amount of constraint which is required, and whether the constraint is translational
or rotational. The different forms of total knee replacement are described in terms of these constraints,
starting with the least constrained unicompartments to the almost fully constrained fixed and rotating
hinges. Much attention is given to the range of designs in between these two extreme types, because
they constitute by far the largest in usage. This category includes condylar replacements where the
cruciate ligaments are preserved or resected, posterior cruciate substituting designs and mobile bearing
knees. A new term, ‘guided motion knees’, is applied to the growing number of designs which control
the kinematics by the use of intercondylar cams or specially shaped and even additional bearing
surfaces. The final section deals with the selection of an appropriate design of total knee for specific
indications based on the design characteristics.

Keywords: total knee replacement, unicompartmental, condylar replacement knees, mobile bearing
knees, guided motion knees, rotating hinges

1 DESIGN CRITERION FOR TOTAL KNEE that patient, given the status of the muscles and soft
REPLACEMENT tissues around the knee, as well as the overall condition
of the patient.
Although there are numerous individual designs of total
knee replacement ( TKR) available today, there are only
a limited number of design forms or types. This article 1.1 Knee motion and kinematics
is restricted to the design and functional aspects of total
knees, without reference to aspects such as wear and A number of investigators have measured the motion of
fixation. A design criterion can be proposed which cadaveric knee specimens under defined conditions with
relates to the mechanical function of the knee with the generally similar findings. In the sagittal plane, as the
TKR implanted: after implantation, the forces in the femur is flexed on a fixed tibia, the contact point of the
remaining structures and the kinematics of the joint are femur on the tibia displaces posteriorly, controlled by
restored to normal. the geometry of the four-bar linkage consisting of the
This criterion can be applied to all design forms and cruciate ligaments and the joint surfaces [1]. The
is effectively a measure of the performance of the design. mechanical implications of this motion are that the lever
The criterion is clearly applicable to those knees with arm of the posterior muscles is increased towards exten-
limited pathology and where the patient is potentially sion, thus providing an effective brake to hyperextension
capable of an active lifestyle. In such cases, normal func- whereas, in flexion, the lever arm of the quadriceps is
tion for that age of patient may be achievable. On the increased for efficiency of activities such as stair climbing
other hand, if the knee already has loss of ligaments and and rising from a seated position. In addition, the pos-
bone in an elderly patient with a low activity level, the terior displacement of the femur in high flexion allows
word ‘normal’ must be referenced to what is possible for clearance of the posterior structures, increasing the range
of flexion.
The MS was received on 27 May 1999 and was accepted after revision To describe the motion in rigid-body terms, reference
for publication on 31 August 1999. axes need to be defined in the femur and tibia. For the
* Corresponding author: Department of Biomedical Engineering, Royal
National Orthopaedic Hospital, Brockley Hall, Stanmore, Middlesex femur, some researchers have used a transverse line pass-
HA7 4LP, UK. ing through the centres of the posterior lateral and
H02599 © IMechE 2000 Proc Instn Mech Engrs Vol 214 Part H

Downloaded from pih.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on March 12, 2015

102 P S WALKER AND S SATHASIVAM

medial condyles [2]. Using this axis system, the six The motion determined under passive or controlled
degrees of freedom of motion were determined with the conditions as described above can be regarded as a ‘mid-
knee in a modified Oxford knee testing rig [3], where a path motion’. This motion path will deviate under three
moment was applied to the femur, and the quadriceps different conditions:
was continuously shortened and lengthened to produce
flexion–extension. It was found that, as the knee flexed, 1. An anterior–posterior (A–P) force or an internal–
the predominant motions were posterior displacement external torque is applied. When such forces are
and internal rotation of the femur on the tibia. Other applied, displacements and rotations are produced
researchers [4, 5] have proposed a transverse femoral called ‘laxity’. The concept of an ‘envelope of passive
axis through the epicondylar line and have used an opti- motion’ has been advanced to describe the boundaries
mization technique to show that, with reasonable accu- within which the laxity of the knee can lie, for defined
racy, the passive motion could be described by a femoral forces and torques [8]. The laxity of the knee is least
rotation about this axis, coupled with a rotation of the in extension, increases up to about 20° of flexion,
tibia about an axis parallel to and medial to its long axis remains reasonably constant throughout most of the
(Fig. 1). Feikes et al. [6 ] also expressed the motion par- motion and then reduces slightly towards full flexion.
ameters as a function of flexion angle, calling it the The model of Churchill et al. [5] (Fig. 1) has been
‘unique path of passive motion’. More recently, Williams expanded upon to represent laxity about the mid-
et al. [7] have used magnetic resonance imaging data path motion.
from both cadaveric specimens and human subjects, to 2. If an A–P force is applied and then a torque is sub-
explain the above motion pattern in terms of the sequently applied, the rotational laxity is less than in
geometry of the knee, in particular that of the lateral the absence of the A–P force. Similarly, a torque fol-
and medial femoral condyles in the sagittal plane. An lowed by an A–P force will result in a reduced A–P
important implication of reproducing normal motion in displacement. This is termed ‘coupled motion’ and is
a TKR is in obtaining the correct length patterns of the due to the pre-tightening of soft tissues [9].
cruciate and collateral ligaments because these structures 3. If a compressive force is applied along the axis of the
will determine the laxity and stability during function. tibia, all the laxities are reduced, A–P more so than

Fig. 1 A mechanical representation of knee motion: flexion–extension about the epicondylar axis and
internal–external rotation about a medial axis in the tibia. Note that the posterior displacement of
the centre of the femur, and internal rotation of the tibia, are coupled with the flexion angle. (The
model is based on the work of Churchill et al. [5].) The model has been extended to include A–P
and rotational laxity, restrained by soft tissues (see the small figure on the right)
Proc Instn Mech Engrs Vol 214 Part H H02599 © IMechE 2000

Downloaded from pih.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on March 12, 2015

 DESIGN FORMS OF TOTAL KNEE REPLACEMENT 103

rotation. The mechanism is that the dishing of the with the input of gait analysis and force-plate data,
joint surfaces, especially on the medial side, requires together with geometrical measurements of the limb
that the two joint surfaces distract under A–P or [13, 14]. The highest forces were obtained for descending
rotation, requiring energy input, including that due stairs or a slope and then ascending, and the lowest for
to increased tensions in the soft tissues. level walking. The more vigorous the activity, the higher
are the forces, as shown for active subjects walking
At this time most of the data available on the kinemat-
downhill where forces of 8 body-weight (BW ) were
ics of the natural knee in activity has been obtained by
obtained [15]. In one study, a telemeterized distal fem-
gait analysis techniques. Such techniques produce accu-
oral replacement was used to measure the forces directly,
rate data on flexion–extension, but the other motion
the first such measurements of its kind in the knee [16 ].
data are less accurate due to the relative motions between
In walking activities, where the flexion angles in stance
the skin markers and the bones themselves [10].
are about 20°, the patello-femoral forces are less than
However, in one study, pins were fixed into the bones
2 BW but, in higher flexion, forces as high as 7 BW have
themselves and the motion was measured during walking
been calculated. From several studies, the shear forces
[11]. The magnitudes of the A–P displacements and the
were determined to be higher in a direction that would
rotations were similar to those measured in the cadaveric
tense the posterior cruciate ligament (PCL). In the
experiments noted above, but the relations between these
telemetry study [16 ], the forces that would tense the
motions and the flexion angle were clearly influenced by
the external forces and the muscle forces. More recently, anterior cruciate ligament (ACL) in walking were found
fluoroscopic techniques have been used to determine to be small ( Fig. 2). There are significant axial torques
motions in squatting and climbing a step [12], but data occurring during walking and other activities. The tor-
for walking and other activities will become available in ques act at the foot as a consequence of the twisting of
the near future. the body as it swings over the planted foot. The direction
of the torque is internal, such that the lateral tibial pla-
teau tends to move anteriorly. In walking, Taylor et al.
1.2 Forces in the knee [16 ] measured the torque at around 8 Nm.
Knowledge of the force magnitudes is important for
The forces acting in the knee during activity were calcu- a number of reasons. In designing TKRs, the radii of
lated as long ago as the late 1960s using a knee model curvature of the bearing surfaces should be chosen to

Fig. 2 Typical data for the forces at the knee joint in level walking, determined from telemetry of a distal
femoral replacement [16 ]. PAT-LIG, force in the patella ligament; PAT-FEM, force between the
patella and the anterior femur. The thick lines refer to the scale on the left. The thin lines refer to
the scale on the right
H02599 © IMechE 2000 Proc Instn Mech Engrs Vol 214 Part H

Downloaded from pih.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on March 12, 2015

104 P S WALKER AND S SATHASIVAM

produce acceptable values of the contact areas and An important geometrical parameter is the posterior-
stresses, taking account of the forces which occur in the distal transition angle (PDTA). If this is 20° or more,
more strenuous activities and not only in level walking. the large distal radius will contact the tibia during the
The varus or valgus moments redistribute the stresses entire stance phase of gait, reducing the contact stresses
over the tibial surface and in extreme cases can result in on the plastic. However, there are two possible disadvan-
failure of trabecular bone or cause an intramedullary tages of this scheme. Firstly, the distal-anterior trochlea
stem to penetrate a cortex progressively due to bone can be too prominent, which can result in gait abnor-
resorption. The shear forces and torques are of particular malities due to the excessive forces required in the quad-
interest to TKR because these forces are carried by the riceps and patellar ligament [19]. Secondly, there can be
remaining cruciate ligaments, the condylar surfaces, a kinematic abnormality such that, in moving from
plastic stabilizer posts or linkages such as hinges. In extension to mid-flexion, the origin of the femur will
addition, these forces act across the implant–bone displace anteriorly, which is opposite to the behaviour
interface and can affect fixation. In order to predict the in the natural knee. This is because, in the presence of
long-term durability of TKRs, realistic force values are axial compressive forces and relatively low shear forces,
needed for input to knee-simulating machines. the femur locates at the bottom of the tibial dish. This
phenomenon may be seen in fluoroscopy studies of deep
knee bends or step ascending and descending, where the
1.3 Geometry of the bearing surfaces femur is found to slide anteriorly with flexion and pos-
teriorly with extension [12, 20].
To describe the geometry and function of condylar Constraint is the resistance to a particular degree of
TKRs, certain definitions are useful. A parametric freedom, such as A–P, internal–external rotation or
description of the geometry of conventional condylar varus–valgus, when there is a compressive force acting
replacements is useful for modelling the knee, for pre- across the joint. This is measured in newtons of applied
dicting kinematics and ligament-length patterns and for force per millimetre of A–P displacement, or newton-
calculating the stresses on the bearing surfaces ( Fig. 3) millimetres of applied torque per degree of rotation. If
[17]. In the sagittal view, most symmetric TKR designs the tibial surfaces are flat, the constraint is nominally
take an average between the lateral and medial profiles zero, except for friction. For dished tibial surfaces, the
of the natural knee [18], although some designs preserve curves are non-linear with an increase in constraint with
a lateral–medial difference to obtain a differential roll- displacement from neutral. In a hinge the varus–valgus
back in early flexion. If the posterior-superior radius and A–P constraint are infinite.
(RPSF ) is reduced, this is sometimes claimed to increase Laxity is the inverse of constraint and is measured in
the range of flexion, but in reality there is only a small millimetres per newton or degrees per newton-metre. For
linear difference between the profiles whether a constant flat or shallow tibial surfaces, the laxity is higher than
or decreasing radius is used. Likewise, the linear differ- for more dished surfaces. For the latter, the laxity
ences in the profile between using several arcs rather reduces with distance or rotation from neutral. Laxity
than two or three for the sagittal femoral profile is small. can also mean the total displacement or rotation away

Fig. 3 Definition of the geometry of condylar replacements in the frontal and sagittal planes (the frontal
plane is shown in the two drawings on the left; the sagittal plane shown in the drawing on the right):
ROF, outer radius, femoral; RIF, inner radius, femoral; ROT, outer radius, tibial; RIT, inner radius,
tibial; BS, bearing spacing; RDF, radius at the distal part of the femur; RPF, radius in the posterior
part of the femur; RPSF, radius at the posterior-superior part of the femur; PDTA, posterior-distal
transition angle, where the large distal-anterior radius RDF meets the smaller distal-posterior radius
RPF; RPT, posterior radius, tibial; RAT, anterior radius, tibial. The two frontal geometries shown
have very different mechanical functions. (Redrawn from Sathasivam and Walker [17])
Proc Instn Mech Engrs Vol 214 Part H H02599 © IMechE 2000

Downloaded from pih.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on March 12, 2015

 DESIGN FORMS OF TOTAL KNEE REPLACEMENT 105

from the neutral position on the application of a force simplified geometric analogue of the natural femur
or a torque. Constraint and laxity can be measured in ( Fig. 5) [22]. The general concept was introduced earlier
simple test equipment and are useful measures of the by Seedhom et al. [23], who moulded a femoral shell to
mechanical function of the knee in situ. replicate the exact femoral surface anatomy and then
Conformity is a geometrical measure of the closeness derived the shape of the tibial surface by a wax replica
of fit of the contacting regions of the femoral and tibial of the inside of the joint with the menisci removed and
bearing surfaces. An appropriate measure of conformity where the femur was put through a range of flexion. In
is relative radius of curvature. If the femoral convex Walker’s [22] method, equations were written to describe
radius is RF and the tibial concave radius is RT, the the laxity and stability characteristics of the natural knee,
contact can be considered as a femoral radius (RR) on and then the femoral component was displaced and
a flat tibial surface, where 1/RR=1/RF−1/RT. Initially rotated, simultaneously moving the component upwards
it might be thought that constraint and conformity are according to the equations. The downward locus of the
synonymous. However, consider two knees, one with multitude of points on the femoral surface accumulated
almost matching shallow profiles in the frontal view, and at all the flexion angles, displacements and rotations
the other almost matching ‘dished’ profiles. They are described the tibial surface. It was assumed that the PCL
both equally conforming but, in rotation, the shallow would induce the posterior translation with flexion. This
surfaces are less constrained. This is illustrated in Fig. 3. method of generating TKR surfaces was applied to the
design of the Kinemax TKR [24].
A factor which makes it difficult to replicate exactly
1.4 Replicating normal constraint the natural constraint is the friction between the metal
and plastic surfaces. In the natural knee, the friction is
The ideal is that the constraint characteristics of the so small as to have a minimal effect on the kinematics.
natural knee are replicated after insertion of the TKR, In a TKR, however, the coefficient of friction is in the
consistent with the design criterion stated above. One range 0.05–0.1. The latter value will apply when small
solution to replicating both the constraint and the laxity amounts of debris become embedded in the plastic sur-
characteristics of the natural knee was proposed in the face [25] and is a phenomenon which has been measured
geometry of the original total condylar design [21] in hip joints also, attributed to small scratches on the
(Fig. 4). For a geometry with two partially conforming metal surface [26 ]. The importance of friction can be
radii in both frontal and sagittal planes, when there is appreciated in that, when a compressive force of 2000 N
an axial compressive force acting, application of an A–P is acting, the shear force required to overcome friction
shear force or a torque will result in a displacement or is up to 200 N, which is at least half the maximum shear
a rotation, with an equilibrium position being reached forces occurring in activity. The effect is a reduction in
so as to produce reasonably normal laxity values. laxity, an increase in ‘stability’, reduced forces on the
However, in the absence of the cruciate ligaments, there cruciates and periods of ‘stick’ when the direction of the
was no mechanism for inducing the displacements and shear force is reversed. This latter phenomenon was mea-
rotations described above and shown in Fig. 1. sured in a simulating machine, both under the appli-
Another approach to obtaining normal constraint in cation of shear forces and torques and in a simulated
condylar TKRs was to generate the tibial surface from walking cycle [27].
a computer model of the femoral surface that was a It is important to take into consideration the effect of

Fig. 4 The double-dish geometry of the original total condylar design where the surfaces are partially con-
forming in the frontal and sagittal planes, to provide a combination of laxity and stability. The
cruciate ligaments are not required for this type of geometry [21]
H02599 © IMechE 2000 Proc Instn Mech Engrs Vol 214 Part H

Downloaded from pih.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on March 12, 2015

106 P S WALKER AND S SATHASIVAM

Fig. 5 Generation of the tibial surfaces of a TKR (different views of the same surface are shown) by defining
a femoral surface and then moving this in multiple positions in three dimensions according to equa-
tions describing the laxity and stability of the natural knee in both A–P and rotation [22]

muscle and gait patterns on the motion and laxity behav- anatomical structures as possible, thus minimizing the
iour of the natural (and TKR) knee in vivo. Such con- shear forces on the plastic surface and on the fixation
siderations can place into perspective the studies of laxity and maintaining the proprioceptive effects of the liga-
and stability carried out on cadaveric knees, or on the ments. Preservation of both cruciates was used in early
living knee in laboratory conditions. In addition, if a unicompartmental designs (Marmor and Unicondylar),
change occurs in the knee such as a ruptured ligament, where one or both sides of the joint were replaced,
a resected ligament as part of a surgical procedure, or and in designs with connected components ( Townley,
the presence of a TKR, muscle and gait patterns may Duocondylar, Duopatella, Geomedic and Cloutier).
well adapt to the altered mechanics. Examples of adap- Gait analysis of such unicompartmentals has shown
tation have been described for ACL-deficient knees and almost normal kinematic patterns [29], supporting the
for stair climbing with different TKR designs [28, 29]. principle of this scheme. The ideal indications for pre-
serving both cruciates are in the younger and more active
patients where the bone geometry and the surrounding
2 FIXED-BEARING CONDYLAR ligaments are not severely compromised. However, cer-
REPLACEMENTS tain factors have limited the practice of preserving both
cruciates:
This type of design represents the major clinical usage
today. Typical modern-day condylar replacements are 1. Many designs have used close-to-flat plastic surfaces,
shown in Fig. 6. which have a high sensitivity to geometric placement,
tightness and slope of the tibial components in the
sagittal plane, resulting in an incidence of instability.
2.1 Preservation of both cruciates 2. The plastic components have often been thin,
resulting in excessive wear and deformation.
The closest scheme for restoring normal constraint is to However, in the Marmor design, which used net-
retain the cruciate ligaments and to resurface the femoral shape-moulded polyethylene, the destructive delami-
condyles in combination with shallow plastic tibial sur- nation type of wear was not reported [30].
faces. This scheme is based on preserving as many of the 3. The surface area and the fixation methods of many
Proc Instn Mech Engrs Vol 214 Part H H02599 © IMechE 2000

Downloaded from pih.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on March 12, 2015

 DESIGN FORMS OF TOTAL KNEE REPLACEMENT 107

Fig. 6 Typical condylar replacement TKRs: unicondylar, posterior cruciate-retaining and posterior cruciate-
substituting designs

of the unicompartmental components were inad- 1. Posterior displacement of the femur was considered
equate, leading to a much higher rate of loosening necessary for increasing the lever arm of the quadri-
than in total condylar types of TKR. ceps in flexion, and the PCL was shown to perform
4. The surgical procedure was difficult, partly due to the that role in the natural knee.
limited exposure when both cruciates were preserved, 2. PCL preservation in combination with shallow tibial
and to the lack of adequate bony landmarks for surfaces (and in some designs a posterior slope) was
instrument alignment. thought to increase the range of flexion.
3. One-piece tibial components which covered the major
In recent years, efforts have been made to overcome
area of the upper tibia (not possible with ACL+PCL
these disadvantages, such as by improved surface
preservation) produced durable fixation.
designs, more accurate instrumentation and techniques,
4. The surgical technique was easier than for ACL
more wear-resistant polyethylenes, better fixation and +PCL preservation.
more rigorous selection of patients. The possibilities of
using small incisions with more rapid recovery has also Laboratory studies using cadaveric knees demonstrated
led to a resurgence of interest in unicompartmental the rationale for the above [33]. This study compared the
replacement [31, 32]. femoral–tibial contact points in the sagittal plane for flat
and dished tibial surfaces, and for PCL preservation versus
PCL resection. Preserving the PCL prevented an excess-
ively anterior contact point and the shear force was shared
2.2 Preservation of the posterior cruciate between the ligament and the curved plastic surfaces. In
the absence of the PCL, the contact points were too
Today, almost two-thirds of the knees used world-wide anterior for both flat and curved surfaces. However, the
are of the PCL-preserving type, although there is a grad- proportion of the A–P shear force carried by the PCL was
ual trend towards the posterior-cruciate-substituting later shown to be very dependent on the radius of the tibial
designs. There are several reasons why PCL preservation surfaces in the sagittal plane (RAT and RPT, in Fig. 3).
became the most popular approach: In laboratory studies where compressive and shear forces
H02599 © IMechE 2000 Proc Instn Mech Engrs Vol 214 Part H

Downloaded from pih.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on March 12, 2015

108 P S WALKER AND S SATHASIVAM

were applied, of magnitudes typical of those in activity, 3. For high constraint, the PCL is not required (it may
even a moderate sagittal radius of about 60–70 mm limited even produce a ‘kinematic mismatch’) but the laxity
the A–P displacement to only a few millimetres [25]. The values are much lower than for a natural knee.
tests were extended to TKR patients by applying shear In recent years, fluoroscopic techniques have been intro-
forces when the patients were standing on the knee. The duced whereby accurate data of the relative motion
A–P displacements were inversely proportional to the between the femur and the tibia can be determined in load-
weight of the patient and, in a PCL-preserving design, bearing activities [12, 20, 36 ]. Interesting differences have
the PCL was calculated to be carrying about one-third of been demonstrated between designs which preserve or sub-
the shear force. In a design without cruciates but with stitute for the ACL and the PCL. When normal subjects
more dished tibial surfaces, the displacements were small performed deep knee bends, there was a posterior trans-
and were limited almost entirely by the bearing surfaces lation of the femur relative to the tibia. For ACL-deficient
rather than by the soft tissues. knees, however, the motion patterns were inconsistent and
In condylar replacements in general, it is important to less continuous. This was due to the possibility of contact
determine the relative contributions of the bearing surfaces points that were more posterior than normal in the absence
and the soft tissues in carrying the various forces and of the ACL. A similar behaviour was noted for a PCL-
moments. This was studied using a specially designed knee retaining design with low sagittal constraint. The abnor-
simulating machine where patterns of compressive force, mal motion is due to the large sagittal tibial radius and is
shear force, torque and flexion–extension were pro- not due to the overall concept of a PCL-preserving design.
grammed in, and the resulting A–P displacements and The data from this study can be explained more by the
internal–external rotations measured [34]. One feature of shape of the bearing surfaces rather than by the role of the
the simulator was that springs of appropriate stiffness were PCL or the cam, based on the above data (Fig. 7) of Luger
used to simulate the soft tissue restraint. Full-strength et al. [35]. In the PCL-preserving designs, the tibial sur-
springs were used if a cruciate ligament (PCL) was present; faces were shallow, allowing A–P sliding to occur
one-third-strength springs were used for a resected cruciate depending on the resultant of the external and muscle
ligament to represent the secondary restraints. For a knee forces. In the PCL-substituting designs on the other hand,
with low sagittal constraint and PCL preservation (sagittal the dishing of the tibial surface would restrict A–P sliding,
tibial radius of 80 mm or more), the displacement–force resulting in less erratic motion.
( laxity) curves were compared with those of the normal
intact knee under different loads [35]. This is represented
by knee 1 in Fig. 7. 2.3 Posterior stabilized designs
Comparing the laxity curves between the cadaveric
knee and condylar knee 1 shows the following: Returning to the type of design where both cruciates are
resected, the deficiencies of the original total condylar
1. The magnitudes of the A–P laxity are similar. design were addressed in two ways. Firstly, designs were
2. The curves all exhibit ‘strain stiffening’. introduced where the bottom of the dish was more pos-
3. There is a shift in the neutral position after 10° of terior and the anterior upsweep was higher. This design
flexion for the TKR. was used both with and without the PCL and showed
4. For the TKR, when the direction of the shear force excellent clinical results in both configurations. Secondly,
changes, there is a period of ‘stick’ due to friction. an intercondylar cam was added to control the anterior
5. For both knees, the A–P laxity steadily reduces as the position of the femur on the tibia ( Fig. 8). One of the
compressive force is increased. first posterior stabilized designs was the kinematic stabil-
izer where the cam operated throughout the flexion
Consider the bar graphs on the right of Fig. 7. For range to produce a continuous roll-back [37]. In the
the low-constraint knee, the PCL reduced the A–P and Insall–Burstein posterior stabilized design, the inter-
rotational laxities by an average of 30 per cent. Without condylar cam engaged at about 70° flexion and produced
the PCL, the laxities are higher than ideal at more than posterior roll-back up to full flexion. Clinical results have
30° flexion, especially in rotation. For the high-constraint indicated that this feature has increased the range of
knee (sagittal tibial radius of 60 mm or less) the PCL flexion by 10–15° compared with the original total
made little difference to the laxities. The laxities were condylar design and that gait analysis was comparable
much smaller than normal, especially in rotation at less with PCL-retaining designs [38].
than 30° flexion.
The conclusions from this study are as follows:
3 ALTERNATIVE CONDYLAR DESIGNS
1. Condylar TKRs of low constraint require at least the
PCL to limit the laxity. 3.1 Guided motion knees
2. With moderate constraint (sagittal tibial radius,
60–80 mm), the laxity values would be satisfactory A guided motion knee is defined as one which provides
both with and without the PCL. some control or guidance to A–P translation and/or
Proc Instn Mech Engrs Vol 214 Part H H02599 © IMechE 2000

Downloaded from pih.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on March 12, 2015

H02599 © IMechE 2000

DESIGN FORMS OF TOTAL KNEE REPLACEMENT

Proc Instn Mech Engrs Vol 214 Part H

Fig. 7 Laxity comparisons between a typical cadaveric knee, and condylar TKRs measured in a knee simu-
lator: knee 1, low sagittal constraints; knee 2, medial-high sagittal constraint. An axial compressive
force was first applied and then a cyclic A–P force or a cyclic internal–external torque. Laxity curves
at different compressive forces and flexion angles are shown on the left, and total A–P and rotational
laxities at 2000 N compressive force (3 BW ) are shown on the right. (From Luger et al. [35])

109
Downloaded from pih.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on March 12, 2015
110 P S WALKER AND S SATHASIVAM

Fig. 8 The first condylar designs using an intercondylar cam to produce roll-back of the contact point with
flexion, namely the kinematic (now Kinemax) stabilizer and the Insall–Burstein posterior stabilized
design

internal–external rotation, during flexion–extension, by (e) to guide the contact points posteriorly with flexion
the interaction of femoral and tibial cams, bearing sur- (especially on the lateral side), and anteriorly with
faces or other means. The overall goal of a guided extension, and
motion knee is to achieve the specified A–P translation (f ) to produce internal tibial rotation progressively with
and internal–external rotation patterns by the use of flexion and external rotation with extension.
guide surfaces located, for example, in the intercondylar
It should be noted that these statements are consistent
region or by using the femoro-tibial bearing surfaces
with the mid-path motion described earlier in this article.
themselves to guide the motion. The motion control can
The rationale is to provide optimal muscle lever arms,
be discrete or, preferably, guide the knee in the desired
soft tissue tensions, and overall knee and leg alignment,
neutral motion path, with regions of laxity about this
in order to restore natural function as closely as possible.
neutral path.
By using special computer programs, the above and
As described in Section 2, versions of guided motion
other types of intercondylar guide surface with required
knees which have been in use for many years include the
patterns of A–P motion control can be synthesized
kinematic (or Kinemax) stabilizer and the Insall–
( Fig. 9). One type is a ‘saddle’ which can provide control
Burstein posterior stabilized design ( Fig. 8). The aims
of A–P motion throughout a large fraction of the flexion
of these particular designs were as follows:
range [39] (Fig. 10). By varying the parameters of the
(a) to increase the quadriceps lever arm at high FGS, the A–P control can be biased to the mid-range,
flexion angles, with slight loss of control at each extreme, or shifted to
(b) to increase the range of flexion by preventing pos- cover early or late flexion. It can be visualized that the
terior impingement of bone and soft tissues and posterior surface of the anterior part of the tibial saddle
(c) to prevent posterior tibial subluxation in flexion can be used to produce roll-back of the femoral–tibial
under the action of the hamstrings. contact point throughout the flexion range, in a similar
way to the kinematic stabilizer, and other designs in cur-
However, there are other arrangements of intercon-
rent use. In this arrangement, roll-forward as the knee
dylar cams or guide surfaces that can achieve control of
is extended would need to rely on the anterior surface
A–P translation and offer possible advantages. In this
of the posterior part of the saddle, or dished tibial bear-
context, the following criteria can be added to those
ing surfaces into which the femoral condyles would tend
above:
to locate. Another possible scheme is a saddle that pro-
(d) to control the location of the femoral–tibial contact vides motion control in the first half of the flexion range
points in the A–P direction and internal–external combined with a posterior stabilized feature providing
rotation throughout the flexion range but to allow posterior displacement in late flexion.
some laxity about these contact points, Ingenious designs where convex tibial ‘saddles’ were
Proc Instn Mech Engrs Vol 214 Part H H02599 © IMechE 2000

Downloaded from pih.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on March 12, 2015

 DESIGN FORMS OF TOTAL KNEE REPLACEMENT 111

used have been presented in the past. In one design [40]

the saddle moved the contact from anterior to posterior
during early flexion, the second part of flexion using a
ball-in-socket device. In another design [41], the saddle
shape was such as to induce pure rolling of the guided
surfaces during flexion–extension. In order to allow
internal–external rotation in all the above configur-
ations, the plastic component could be moved on a rotat-
ing platform.
Particular designs which can be classified as ‘guided
motion knees’ and which have been introduced up to
this time, include the following:

1. The saddle design shown in Fig. 10 has a saddle that

controls A–P displacement throughout the flexion–
extension range. In this design (the Genius knee,
Astro Medical ) the plastic component consists of a
Fig. 9 Synthesis of intercondylar guide surfaces. The femoral rotating platform that allows additional freedom of
intercondylar guide surface is defined by three para- A–P and unlimited internal–external rotation.
metric circular arcs constrained to join tangentially. In 2. A conforming medial compartment in the form of
this way, a wide range of femoral guide surfaces spherical surfaces, and a lateral compartment with
(FGSs) can be defined. By flexing the femoral compo-
low conformity ( Fig. 11) (medial pivot knee, Wright
nent with a defined contact roll-back pattern, the
Medical, Memphis) is a close geometrical and
matching tibial guide surface (TGS ) is generated [39]
mechanical analogue to the natural joint surfaces if
the menisci are taken into consideration. However, in
the absence of one or both cruciates, there may be
no means to control the A–P contact point location
on the lateral side, although an advantage of the

Fig. 10 A single circle for the FGS generates a saddle for the TGS. By varying the orientation angle of the
circle, motion control in both the anterior and the posterior directions can be achieved in the first
half of the flexion range, in the middle range or in the second half of the flexion range. The saddle
design can be used in fixed-bearing or mobile-bearing configurations
H02599 © IMechE 2000 Proc Instn Mech Engrs Vol 214 Part H

Downloaded from pih.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on March 12, 2015

112 P S WALKER AND S SATHASIVAM

Fig. 11 The medial pivot knee ( Wright Medical ). The highly

conforming medial side provides a large contact area
and a pivot for internal–external rotation. The low
constraint of the lateral side allows for A–P trans-
lation, as in the natural knee

design is that the joint is fully conforming medially

where the highest joint forces are normally trans- Fig. 12 The TRAC PS (Biomet). In early flexion, the inner
mitted. femoral–tibial bearing surfaces provide an anterior
3. The TRAC PS design [42] ( Fig. 12) is such that, from contact location; in the remainder of the flexion range,
0 to 8° flexion, the femoro-tibial contact is on an inner the outer bearing surfaces provide more posterior
pair of bearing surfaces with the bottom of the dish contact. Transfer between the pairs of bearing sur-
anterior of centre and, from 8° to maximum flexion, faces is achieved using an intercondylar cam
the contact is on an outer pair of bearing surfaces
with the bottom of the dish posterior of centre.
Intercondylar guide surfaces (with some similarity to spherical surfaces. In allowing freedom of internal–
the saddle configuration in Fig. 10) produce the pos- external rotation and a posterior axis of rotation, the
terior femoral displacement to transfer from the first goal of this design is to facilitate a high range of
to the second pair of bearing surfaces. This design flexion.
achieves the ideal of anterior contact towards exten- Guided motion knees represent an interesting classifi-
sion and posterior contact in high flexion, as well as cation of TKR. In the absence of one or both of the
obtaining a large contact area for each pair of bearing cruciate ligaments, they offer a means of reproducing at
surfaces. One disadvantage of the design is that, for least some of the characteristics of the natural knee, and
the inner pair of bearing surfaces, the varus–valgus possibly even similar laxity behaviour.
stability is reduced. Rotational freedom is provided
by the plastic being pivoted in a ‘rotating platform’
configuration. 3.2 Mobile-bearing or meniscal knees
4. The Kyocera Bi-Surface design ( Fig. 13) uses conven-
tional, partially conforming bearing surfaces for the Designs where a moving plastic bearing is interposed
approximate range 0–90° flexion. At 90°, the femoral between the femoral condyle and the tibial plate, called
intracondylar convex spherical surface engages the meniscal knees or mobile-bearing knees, were introduced
spherical concave surface at the posterior of the tibial in the mid-1970s. These designs imitate the natural knee
plastic. The main lateral and medial bearing surfaces in that the plastic components resemble the menisci, the
disengage and further flexion takes place on the plastic accommodates A–P translations (as well as
Proc Instn Mech Engrs Vol 214 Part H H02599 © IMechE 2000

Downloaded from pih.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on March 12, 2015

 DESIGN FORMS OF TOTAL KNEE REPLACEMENT 113

rotations), and a large contact area is maintained

throughout all or much of the flexion range. The purpose
of such designs is to minimize the wear and deformation
of the plastic and to allow for natural kinematics. There
are a number of mechanical schemes that can be specified
for a mobile-bearing knee (Fig. 14):

1. Internal–external rotation only is allowed. This has

the advantage of allowing the knee to locate at
a preferred rotational orientation, and to adjust to
a new position without resistance (except friction)
during activity.
2. Internal–external rotation about a medial axis is
allowed. This produces more anatomical motion that
occurs where the centres of internal–external rotation
in the horizontal plane of the upper tibia are located
on the medial side.
3. Internal–external rotation and A–P translation are
allowed. This allows the knee to locate at a preferred
rotational and translational orientation during func-
tion. It can also allow natural knee kinematics where
the location of the medial axis of rotation varies due
to knee laxity, although this would ideally require
that both cruciates are present at the correct tensions.
4. Internal–external rotation is allowed and A–P trans-
lation is guided by a femoro-tibial cam, such as a
posterior stabilizer or ‘saddle’. This is potentially
preferable to a configuration with rotation only
because it produces posterior translation in high
Fig. 13 The Bi-Surface ( Kyocera). After 90° flexion, the inter- flexion, but it requires partially conforming femoro-
condylar femoral spherical surface articulates in a tibial bearing surfaces.
dish in the plastic surface. This combination of a pos-
5. This is as above, but where the tibial cam is fixed to
terior contact and low constraint to rotation is
the tibial plate. Because the plastic meniscus (or two
designed to maximize the range of flexion
menisci) slides in an A–P manner, the femoral–plastic
surfaces can be fully conforming. A design of this
type has so far not emerged.

Fig. 14 Motion schemes for meniscal bearing or mobile-bearing knees (see text for detailed explanations)
H02599 © IMechE 2000 Proc Instn Mech Engrs Vol 214 Part H

Downloaded from pih.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on March 12, 2015

114 P S WALKER AND S SATHASIVAM

In all the above, rotation and A–P translation can be In the New Jersey meniscal knee (LCS), introduced
unlimited or limited by stops. Clinical experience has at about the same time as the Oxford, a one-piece metal
shown few problems with unlimited rotation for uncom- tibial component was provided with tracks for sliding of
plicated primary cases. However, in the absence of one two plastic menisci. In terms of freedom of motion, this
or both cruciates, limits to A–P translation seem was similar to the Oxford. When the LCS was used with-
advisable. out the ACL, the menisci sometimes displaced too far
Referring now to specific designs, the Oxford knee posteriorly, causing problems including fracture of the
(Fig. 15) was intended to be used as a unicompartmental menisci themselves [45]. In terms of replicating natural
replacement with preservation of both of the cruciates knee mechanics, both the Oxford and the New Jersey
so that the A–P pattern of motion could be achieved by were less constrained than the natural knee on the
a four-bar linkage mechanism. The lateral and medial medial side.
bearings are independent and there is sufficient clearance A new version of the New Jersey knee was later intro-
to allow for internal–external rotation. By providing duced, namely the rotating platform. The femoro-tibial
only for the femoro-tibial surfaces, the sagittal radii contact was designed to be fully conforming up to about
could be made constant, thus producing complete area 20° of flexion, covering walking, but at higher flexion
contact throughout the range of motion. Studies of pass- there was partial conformity up to full flexion. It is
ive motion in patients demonstrated that the menisci difficult to avoid this compromise in a femoral compo-
translated posteriorly with flexion, the lateral more than nent design that includes a patellar flange and where the
the medial [43]. However, such passive studies do not sagittal profile is similar to anatomical. This design has
replicate functional conditions where variable shear and the advantage of simplicity and the long-term clinical
torque force will be present. The wear depth in retrievals results have been excellent with few mechanical prob-
has been much less than for total hips [44]. lems. In terms of reproducing natural mechanics, while
there is freedom of internal–external rotation, there
is no provision for A–P translation and, when, for
example, the lateral condyle moves posteriorly, the
medial condyle moves anteriorly which is unphysiologi-
cal. The consequence of this mechanism may be less than
optimal muscle lever arms and a possible loss of some
flexion. Recently, a new version of the LCS has been
introduced, the A–P glide, where freedom of A–P trans-
lation is superimposed on the rotation.
The Polyzoides Rotaglide [46 ] was probably the first
to include both rotation and A–P translation in a one-
piece plastic design. This was achieved by locating the
plastic on a smooth metal plate, with stops acting in
slots and recesses to limit the motion. Limitation of
motion was considered necessary because the design
required the resection of both of the cruciates. Another
feature of this design was complete conformity through-
out the entire flexion range. This was achieved by a nega-
tive PDTA ( Fig. 3). The compromise was a patellar
flange which was recessed more than anatomical in the
distal-anterior region.
The MBK design solved the problem of obtaining
complete femoro-tibial conformity throughout flexion,
while maintaining an anatomically shaped patellar flange
and sagittal profile. This was achieved by separating the
lateral and medial, and patello-femoral bearing surfaces.
This resulted in notches at the sides of the anterior fem-
oral component, where the posterior-distal radius is car-
ried forwards. There is sufficient medial–lateral width to
provide for all three bearing surfaces.
Fig. 15 Different mobile bearing designs. Each design can be Although it might be considered that mobile bearing
classified mechanically according to Fig. 14: top left, knees are less constrained in certain degrees of freedom
Oxford (Biomet); top right, LCS rotating platform than most fixed-bearing designs, there are two factors
(DePuy); bottom left, MBK (Zimmer); bottom right, which modify this view. Firstly, in a fixed-bearing knee,
Polyzoides Rotaglide (Corin) relative femoral-tibial displacements can take place by
Proc Instn Mech Engrs Vol 214 Part H H02599 © IMechE 2000

Downloaded from pih.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on March 12, 2015

 DESIGN FORMS OF TOTAL KNEE REPLACEMENT 115

rolling, which is not the case for a mobile-bearing knee. 1. Stability is provided in all degrees of freedom, varus–
Secondly, friction between the plastic mobile bearing and valgus and hyperextension being particularly import-
the tibial plate can have a significant effect in a mobile ant, although there can be some laxity (e.g. rota-
bearing design, as well as in a fixed-bearing design [27]. tional ) in one or more degrees of freedom.
The result can be periods of stick and slip with reduced 2. A linkage of some type, such as a hinge, provides
overall displacements and rotations in a mobile bearing stability and prevents subluxation or dislocation.
design [47]. 3. Intramedullary stems are required to provide
adequate fixation.

4 DESIGNS PROVIDING VARUS–VALGUS The most conservative type of linked TKR is the inter-
STABILITY condylar hinge ( Fig. 17). The linkage is housed in the
intercondylar region, preserving the condyles on each
These designs can either be unlinked, as in the superstab- side. Variations of this type include the Attenborough,
ilizer-constrained condylar (CCK ) designs, or linked as the St Georg Endo model [49], the rotating hinge, the
in fixed or rotating hinges. The former can be regarded Sheehan, the Rotaflex, the Gshwend and the PFC
as an extension of designs with intercondylar guide sur- S-ROM. The bearing surfaces can be extended to the
faces described above, but where the plastic post is elev- full width of the knee, an advantage for reducing
ated [39]. This arrangement provides for posterior the varus–valgus bending moments on the linkage. The
displacement of the contact point in high flexion, some patello-femoral resurfacing has been absent in some
anterior displacement towards extension, and partial designs but ideally should be included. Advantages of
control of A–P displacement in the mid-range. In Fig. 16, an intercondylar design are preservation of bone and the
this type of design is generated using the software shown fact that the axis of rotation can be placed in an anatom-
in Fig. 9. The main value of this design is in providing ical location. The disadvantages of an intercondylar
varus–valgus stability. An important design goal is to design are the relatively restricted size of the bearing
maximize the area of the plastic post for bending stiffness components, with an increased potential for wear and
and strength. However, under extreme varus loading deformation, the possibility of fracture of the femoral
activities, a plastic post bends and deforms over time, condyles on either side of the intracondylar housing, the
making it unreliable in sustaining such loading in the difficulty of linking the components at surgery and the
long term [48]. In most designs a metal reinforcing pin possibility of dislocation in extreme loading conditions
within the plastic post is provided, which improves the (in some designs).
stiffness and strength. However, to provide more rigid The least conservative type of linked TKR is the fixed
and long-lasting support to varus–valgus moments, hinge or rotating hinge. Placement is achieved by resec-
some type of linked device is needed. tion of about 25 mm from the distal femoral condyles,
Over the years, numerous different linked designs have and 10 mm from the upper tibia. An axle is then used
been introduced. The largest numbers continue to be to connect the femoral and tibial components, usually
used in Europe whereas, in the USA, the tendency is to with plastic bushings to act as the bearing. The total
use the most conservative design whenever possible, even thickness of the implant is dictated by the required
if substantial soft-tissue balancing is required. The dimensions of the axle and bushings, and the ideal
characteristics of a linked design are as follows: placement of the axle, which is close to the level of the

Fig. 16 The method shown in Fig. 9 used to generate a superstabilizer–constrained condylar design. Motion
control is provided in early and late flexion, while in the mid-range there is contact near the top of
the plastic post. The parameters R1, R2 and R3 were chosen to maximize the area of the plastic
post, thus maximizing bending stiffness and varus–valgus stability
H02599 © IMechE 2000 Proc Instn Mech Engrs Vol 214 Part H

Downloaded from pih.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on March 12, 2015

116 P S WALKER AND S SATHASIVAM

Fig. 17 Designs of rotating hinges with different bone resections and rotational schemes: (a) Link-Endo
model ( Waldemar-Link), a condylar replacement with an intercondylar rotation mechanism;
(b) kinematic ( Kinemax) rotating hinge (Stryker-Howmedica), posterior-distal straight-across axle,
partially conforming rounded cylindrical surfaces providing progressive rotational restraint; (c) Finn
knee system (Biomet), with similar axis system to (b), flat plastic on metal plate, rotation limited
by stud in groove; (d ) SMILES rotating hinge (Stanmore Implants Worldwide), tibial fork for axle
allowing continuous patellar flange, with similar rotation mechanism to (b)

epicondyles. A lower or more posterior axle location will dished plastic surface. The latter is preferable because it
reduce bone resection but will result in abnormal track- provides a ‘soft’ limit to rotation, reducing the possibility
ing of the patella. It should be noted that an important of instability or patello-femoral subluxation. Clinical
requirement in the placement of hinged replacements is follow-up studies have demonstrated improved perform-
that the patella locates at the correct level on the femoral ance of rotating hinges compared with fixed hinges [51].
flange with the knee in extension. Cases with bone loss, including revision, need special
The fixed hinge is the simpler design, applicable to consideration. Augments, such as spacers and wedges,
patients of low demand who require only a stable knee. are useful for filling bone defects and for accurately
Examples are the Guépar, the St Georg, the Blauth and reproducing the joint line. The augments can be screwed
the Stanmore. Some of the long-term clinical follow-ups or cemented to the main components. For larger defects,
have shown survivorship similar to that of condylar space fillers made from metal or plastic are an alternative
knees [50]. The rotating hinge, however, results in a to bone grafting. Stems are useful for bypassing such
more ‘natural feel’ to the patient and is more durable in larger defects or for protecting against fractures. When
the long term. Examples are the kinematic rotating the TKR design carries varus–valgus moments, intra-
hinge, the Finn, the Lacey, the Noiles (now the PFC medullary stems are needed. In older patients, cemented
S-ROM ) and the SMILES. More bone resection, how- stems are preferable. Uncemented fluted stems can be
ever, is required to accommodate the extra bearing sur- used if there is sufficient cortical thickness in order to
face compared with the fixed hinge. The rotation can be reduce the stress shielding of the cancellous bone near
achieved by a flat polished metal surface pivoted on a the joint. Empirically, suitable stem lengths for super-
flat plastic surface, or by a convex metal surface in a stabilizer types are 100–120 mm, while for rotating and
Proc Instn Mech Engrs Vol 214 Part H H02599 © IMechE 2000

Downloaded from pih.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on March 12, 2015

 DESIGN FORMS OF TOTAL KNEE REPLACEMENT 117

fixed hinges 120–150 mm is needed. Revision of a failed compared with the natural knee. To what extent this
cemented stem requires a new stem at least 50 mm longer adversely affects function is debatable, but it is not con-
[52]. In all cases, stem centralizers are needed to prevent sistent with the design criterion stated at the start of this
the stem tip from impacting the cortical wall, which fre- article. A possible solution is to use a posterior stabilized
quently produces osteolysis, penetration and even bone or saddle type but with the plastic component as a
fracture. For cases with abnormal geometry, especially rotation platform. Several designs of this type are
of very small size where bone preservation is a pre- already in clinical evaluation. However, increased wear
requisite, custom superstabilizers have been designed is a possibility because of the introduction of the
and made using computer aided design–computer aided additional bearing surface.
manufacturing (CAD/CAM ) [53]. Concerning the patellar flange, because not resurf-
acing is preferred in many cases, a flange with an ana-
tomical skyline view is suitable. To maximize the contact
5 DESIGN SELECTION area and to allow for interchangeability, a conical
patellar resurfacing with a medially biased peak has
5.1 The ideal design of condylar replacement advantages.

The main advantages of a fixed-bearing condylar

replacement are its simplicity and, as the clinical experi- 5.2 The ideal ‘alternative’ design
ence has shown, its reliability with good to excellent
function in the large majority of cases and a survivorship It is much more difficult to compare designs in the
of around 95 per cent at 10 years. However, it would be ‘alternative design’ category than in the fixed-bearing
erroneous to assume that all condylars can be regarded category because of the wide variety, the limited
as equivalent. As the above discussion has shown, there follow-up with several of the configurations and the
are major differences in function depending upon the likely introduction of yet more designs. The advantages
surface geometry and whether or not the cruciates are of the mobile bearing designs are low wear volume and
preserved. Concerning the surface geometry, shallow low deformation, freedom of rotational and/or A–P pos-
surfaces with preservation of both cruciates must remain ition, and the ability to determine a ‘home position’ for
as the ideal. The realization of such a concept can range each knee in particular activities.
from a unicompartmental design used on both sides of However, the disadvantages are less inherent con-
the joint, to a total condylar type with cut-outs for the straint than in the natural knee due to flat plastic–tibial
cruciates. However, in order to obtain the asymmetry of plate bearing and lack of smooth motion (A–P and
the kinematics, namely internal rotation as well as pos- rotation) in function, due to periods of stick and to the
terior translation with flexion, the tibial surface should inability of the femoral–tibial surface to roll as in a fully
be dished on the medial side to limit the A–P translation conforming design. If the introduction of cross-linked
to a few millimetres, while the lateral surface should polyethylenes substantially reduces the wear and delami-
be shallow. nation in partially conforming designs, one of the advan-
Designing for preservation of the PCL only is prob- tages of mobile bearing would be removed.
lematic because, while, with accurate surgery, femoral Guided motion knees are an attractive option because
roll-back with flexion can be achieved, there is no driver the femoral–tibial position can be defined within limits
for roll-forward on extending the knee. Hence tibial sur- in both A–P and rotation. For example, a saddle design
faces that are flat or shallow in the sagittal plane with ( Fig. 10) can control A–P motion over a substantial part
the intention of allowing normal A–P translations result of the flexion range. Such an arrangement seems prefer-
in unpredictable contact locations and motions. able to a traditional PS design, which has a plastic post
However, a compromise can be achieved by moderate of restricted size, and which only provides for a single
dishing of the tibial surface so that, on extending the action, namely roll-back in high flexion. If it was
knee from a flexed position, the femoral surface will slide required to provide internal rotation with flexion (as in
and roll forwards to locate at the bottom of the dish. the natural knee), intercondylar guide surfaces could be
With this scheme, reasonably normal laxity character- combined with bearing surfaces such as in the medial
istics can be achieved. pivot knee (Fig. 11).
For cruciate resection, generation of the surfaces to In the formulation of guided motion knees, it is evi-
produce normal laxity and stability under functional dent that intercondylar guide surfaces need to be such
loads results in moderate constraint. However, when the that dislocation could not occur except under extreme
compressive forces are reduced, the stability is situations, that cam contact points are not so high that
insufficient. Hence increased constraint is required, but component tilting and loosening could occur and that
this has the effect of reducing the laxity under weight- insertion is feasible. Such criteria can be avoided by
bearing conditions. For that reason cruciate-sacrificing adopting the approach of the Kyocera Bi-Surface design
and cruciate-substituting designs are overconstrained ( Fig. 13). Here, the bearing surfaces are shallow and rely
H02599 © IMechE 2000 Proc Instn Mech Engrs Vol 214 Part H

Downloaded from pih.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on March 12, 2015

118 P S WALKER AND S SATHASIVAM

for stability on the principles of the original total condy- P. S. Relation between knee motion and ligament length
lar design, namely by providing sufficient constraint patterns. Clin. Biomechanics, 1991, 6(4), 213–220.
from dished bearing surfaces. 3 Zavatsky, A. B. A kinematic-freedom analysis of a flexed-
One of the key issues in all the above designs is knee-stance testing rig. J. Biomechanics, 1997, 30, 277–280.
whether the laxity should be ‘allowed’, ‘constrained’ or 4 Hollister, A. M., Jatana, S., Singh, A., Sullivan, W. W. and
Lupichuk, A. G. The axes of rotation of the knee. Clin.
‘guided’. Concerning the femoral–tibial A–P position,
Orthop., 1993, 290, 259–268.
there is a strong rationale for the TKR, with or without
5 Churchill, D., Incavo, S. J., Johnson, C. C. and Beynnon,
the assistance of ligaments, providing roll-back with B. D. The transepicondylar axis approximates the optimal
flexion and roll-forward on extension. Nevertheless, flexion axis of the knee. Clin. Orthop., 1998, 356, 111–118.
mechanical analysis and experimentation on each design 6 Feikes, J. D., Wilson, D. R. and O’Connor, J. J. The unique
are needed to show that this indeed provides the desired track of intact passive knee motion as a kinematic baseline.
advantages in muscle forces. The situation regarding Trans. Orthop. Res. Soc., 1998, 23, 170.
internal–external rotation is not so clear, namely as to 7 Williams, A., Vedi, V., Gedroyc, W., Hunt, D. and Freeman,
what advantages are provided by either allowing free- M. A. R. ‘Dynamic’ M.R.I. scanning of the weight-bearing
dom of rotation or guiding rotation with flexion. Further asymptomatic knee–tibio–femoral motion during flexion.
studies are needed on this issue. In Proceedings of the European Federation of National
Associations of Orthopaedics and Traumatology
(EFORT ) Congress, Brussels, 1999, p. 44.
5.3 The ideal varus–valgus design 8 Blankevoort, L., Huiskes, R. and De Lange, A. The envelope
of passive knee joint motion. J. Biomechanics, 1988,
When varus–valgus constraint is required, the choice is 21(9), 705–720.
between the superstabilizer CCK type and the rotating 9 Piziali, R. L. and Rastegar, J. C. Measurements of the
hinge. The clinical results of the former, and the small nonlinear, coupled stiffness characteristics of the human
amount of bone resection required for insertion support knee. J. Biomechanics, 1977, 10, 45–51.
its viability, although there can be an incidence of pro- 10 Reinschmidt, C., van den Bogert, A. J., Nigg, B. M.,
gressive varus deformity due to deformation of the plas- Lundberg, A. and Murphy, N. Effect of skin movement on
tic post. In addition, stability in high flexion may be a the analysis of skeletal knee joint motion during running.
J. Biomechanics, 1997, 30, 729–732.
problem. Hence, specific designs where the strength of
11 Lafortune, M. A., Cavanagh, P. R., Sommer, H. J. I. and
the plastic tibial post is maximized and the femoral–
Kalenak, A. Three-dimensional kinematics of the human
tibial bearing surface constraint is maximized are an knee during walking. J. Biomechanics, 1992, 25(4),
advantage. However, there are indications where a 347–357.
linked design such as a rotating hinge is preferable. These 12 Dennis, D. A., Komistek, R. D., Hoff, W. A. and Gabriel,
indications are where there has already been consider- S. M. In vivo knee kinematics derived using an inverse per-
able bone loss, where there is a serious soft tissue insta- spective technique. Clin. Orthop., 1996, 331, 107–117.
bility and where guaranteed stability is needed. Long- 13 Morrison, J. B. ‘Function of the knee joint in various activi-
term fixation appears to be better when metallic rather ties. Bio-Med. Engng, 1969, 4(12), 573–580.
than plastic intramedullary stems are used. Given that 14 Morrison, J. B. The mechanics of the knee joint in relation
the durability of knee hinges can be comparable with to normal walking. J. Biomechanics, 1970, 3(1), 51–61.
that of standard condylar replacements. 15 Kuster, M. S., Wood, G. A., Stachowiak, G. W. and
Gachter, A. Joint load considerations in total knee replace-
ment. J. Bone Jt Surgery, 1997, 79B(1), 109–113.
ACKNOWLEDGEMENTS 16 Taylor, S., Walker, P. S., Perry, J., Cannon, S. R. and
Woledge, R. The forces in the distal femur and the knee
during walking and other activities measured by telemetry.
The input from the staff of the Centre for Biomedical
J. Arthroplasty, 1998, 13(4), 428–437.
Engineering, University College London, Royal 17 Sathasivam, S. and Walker, P. S. Optimisation of the bear-
National Orthopaedic Hospital Trust, is gratefully ing surface geometry of total knees. J. Biomechanics, 1994,
acknowledged. Thanks are due to Marilyn Englehardt 27(3), 255–264.
and Jarmila Dodd for the preparation of this 18 Garg, A. and Walker, P. S. Prediction of total knee motion
manuscript. using a 3-dimensional computer-graphics model. J. Bio-
mechanics, 1990, 23, 45–58.
19 Andriacchi, T. P., Yoder, D., Conley, A., Rosenberg, A.,
REFERENCES Sum, J. and Galante, J. O. Patellofemoral design influences
function following total knee arthroplasty. J. Arthroplasty,
1 Zavatsky, A. B. and O’Connor, J. J. A model of human 1997, 12(3), 243–249.
knee ligaments in the sagittal plane. Part 1: response to 20 Stiehl, J. B., Komistek, R. D., Dennis, D. A., Paxson, R. D.
passive flexion. Proc. Instn. Mech. Engrs, Part H, Journal and Hoff, W. A. Fluoroscopic analysis of kinematics after
of Engineering in Medicine, 1992, 206(H3), 125–134. posterior-cruciate-retaining knee arthroplasty. J. Bone Jt
2 Rovick, J. S., Reuben, J. D., Schrager, R. J. and Walker, Surgery, 1995, 77B(6), 884–889.
Proc Instn Mech Engrs Vol 214 Part H H02599 © IMechE 2000

Downloaded from pih.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on March 12, 2015

 DESIGN FORMS OF TOTAL KNEE REPLACEMENT 119

21 Walker, P. S., Wang, C.-J. and Masse, Y. Joint laxity as a 37 Walker, P. S. Design of a knee prosthesis system. Acta
criterion for the design of condylar knee prostheses. In Orthop. Belgica, 1980, 46(6), 766–775.
Proceedings of the Total Knee Replacement Conference, 38 Wilson, S. A., McCann, P. D., Gotlin, R. S., Ramakrishnan,
London, 1974, 1975, pp. 22–29 (Mechanical Engineering H. K., Wootten, M. E. and Insall, J. N. Comprehensive
Publications Limited, London). gait analysis in posterior-stabilized knee arthroplasty.
22 Walker, P. S. Bearing surface design in total knee replace- J. Arthroplasty, 1996, 11(4), 359–367.
ment. Engng Medicine, 1988, 17(4), 149–156. 39 Walker, P. S. and Sathasivam, S. Controlling the motion
23 Seedhom, B. B., Longton, E. B., Dowson, D. and Wright, V. of total knees using intercondylar guide surfaces. J. Orthop.
The Leeds knee. In Proceedings of the Total Knee Res., 2000 (in press).
Replacement Conference, London, 1974, 1975, pp. 108–114 40 Deane, G. A new concept in knee joint design. In
(Mechanical Engineering Publications Limited, London). Proceedings of the Total Knee Replacement Conference,
24 Walker, P. S. Design of Kinemax total knee replacement London, 1974, 1975, pp. 140–152 (Mechanical Engineering
bearing surfaces. Acta Orthop. Belgica, 1991, Publications Limited, London).
57(Suppl. II ), 108–113. 41 Elad, D., Seliktar, R. and Mendes, D. Synthesis of a knee
25 Walker, P. S., Ambarek, M. S., Morris, J. R., Olanlokun, joint endoprosthesis is based on pure rolling. Engng
K. and Cobb, A. Anterior–posterior stability in partially Medicine, 1981, 10(2), 97–105.
conforming condylar knee replacement. Clin. Orthop., 42 Draganich, L. and Pottenger, L. TRAC PS knee design and
1995, 310, 87–97. clinical outcome. In Abstracts of the Proceedings of the
26 Unsworth, A. Tribology of human and artificial joints. Proc. Annual Symposium of the International Society for
Instn Mech. Engrs, Part H, Journal of Engineering in Technology in Arthroplasty, Marseille, 1998, pp. 96–99.
Medicine, 1991, 205(H3), 163–172. 43 Bradley, J., Goodfellow, J. W. and O’Connor, J. J. A radio-
27 Sathasivam, S. and Walker, P. S. A computer model with graphic study of bearing movement in unicompartmental
surface friction for the prediction of total knee kinematics. Oxford knee replacements. J. Bone Jt Surgery, 1987,
69B(4), 598–601.
J. Biomechanics, 1997, 30(2), 177–184.
44 Psychoyios, V., Crawford, R. W., O’Connor, J. J. and
28 Berchuk, M., Andriacchi, T. P., Bach, B. R. and Reider,
Murray, D. W. Wear of congruent meniscal bearings in
B. R. Gait adaptations by patients who have a deficient
unicompartmental knee arthroplasty. J. Bone Jt Surgery,
ACL. J. Bone Jt Surgery, 1990, 72A, 871–877.
1998, 80B, 976–982.
29 Weinstein, J. N., Andriacchi, T. P. and Galante, J. Factors
45 Weaver, J. K., Derkash, R. S. and Greenwald, A. S.
influencing walking and stair-climbing following unicom-
Difficulties with bearing dislocation and breakage using a
partmental knee arthroplasty. J. Arthroplasty, 1986, 1,
movable bearing total knee replacement system. Clin.
109–115.
Orthop., 1993, 290, 244–252.
30 Blunn, G. W., Joshi, A. B., Lilley, P. A., Englebrecht, E.,
46 Polyzoides, A. J., Dendrinos, G. K. and Tsakonas, H. The
Ryd, L., Lidgren, L., Hardinge, K., Nieder, E. and Walker,
Rotaglide total knee arthroplasty. J. Arthroplasty, 1996,
P. S. Polyethylene wear in unicondylar knee prostheses: 11(4), 453–459.
106 retrieved Marmor, PCA and St. Georg tibial compo- 47 Des Jardins, J., Walker, P. S., Haider, H. and Perry, J.
nents compared. Acta Orthop. Scand., 1992, 63(3), The use of a force-controlled dynamic knee simulator to
247–255. quantify the mechanical performance of total knee replace-
31 Murray, D., Webb, J., Price, A., Topf, H., Dodd, C. and ment designs. J. Biomechanics, 2000 (in press).
Goodfellow, J. Minimally invasive Oxford unicompartmen- 48 Robie, B. H., Daellenbach, K. K., Bolanos, A. and Wright,
tal knee replacement. In Proceedings of the International T. M. Retrieval analysis of constrained condylar tibial com-
Conference on Knee Replacement 1974–2024, London, 1999 ponents. Trans. Orthop. Res. Soc., 1997, 22, 644.
(Professional Engineering Publishing, London). 49 Engelbrecht, E., Nieder, E. and Klüber, D. Ten to twenty
32 Toksvig-Larsen, S. Minimally-invasive surgery for the next years of knee arthroplasty at the Endo-Klinik: a report on
decade? In Proceedings of the International Conference on the long-term follow-up of the St Georg Hinge and the
Knee Replacement 1974–2024, London, 1999, pp. 261–263 medium-term follow-up of the rotating knee Endo Model.
(Professional Engineering Publishing, London). Report, Endo-Klinik, Hamburg, Germany, 1996.
33 Soudry, M., Walker, P. S., Reilly, D. T., Kurosawa, H. and 50 Böhm, P. and Holy, T. Is there a future for hinged pros-
Sledge, C. B. Effects of total knee replacement design on theses in primary total knee arthroplasty? J. Bone Jt
femoral–tibial contact conditions. J. Arthroplasty, 1986, Surgery, 1998, 80B(2), 302–309.
1(1), 35–45. 51 Unwin, P. S., Walker, P. S. and Blunn, G. W. Rotating
34 Walker, P. S., Blunn, G. W., Broome, D. R., Perry, J., hinge, hydroxyapatite coated versus fixed hinged, uncoated
Watkins, A., Sathasivam, S., Dewar M. E. and Paul, J. P. distal femoral replacement. A study of 402 cases. In
A knee simulating machine for performance evaluation of Proceedings of the Tenth International Symposium of the
total knee replacement. J. Biomechanics, 1997, 30(1), International Society of Limb Salvage, Cairns, Australia,
83–89. 1999, p. 15.
35 Luger, E., Sathasivam, S. and Walker, P. S. Inherent 52 Inglis, A. E. and Walker, P. S. Revision of failed knee
differences in the laxity and stability between the intact knee replacements using fixed-axis hinges. J. Bone Jt Surgery,
and total knee replacements. Knee, 4, 1997, 7–14. 1991, 73B(5), 757–761.
36 Banks, S. A., Markovich, G. D. and Hodge, W. A. In Vivo 53 Sathasivam, S., Walker, P. S., Pinder, I., Cannon, S. and
kinematics of cruciate-retaining and -substituting knee Briggs, T. Custom constrained condylar total knees using
arthroplasties. J. Arthroplasty, 1997, 12(3), 297–304. CAD-CAM. Knee, 1999, 6, 49–53.

H02599 © IMechE 2000 Proc Instn Mech Engrs Vol 214 Part H

Downloaded from pih.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on March 12, 2015